IMP D&E /AIMP/ PFM encoding system interface document NASA Technical Memorandum X-'/11-66-384 55652 I N67 166 74 IACCKS_ION NUMBER_ (THRU) == ' 7,/ (PAGES) / (CODE) t= _/_-#-j4s. (I_IASA Cr. O_ TMX OR AD _" ._ NUMBER) O' r {'CATEGORY) IMP D&E (AIMP) PFM ENCODING SYSTEM INTERFACE DOCUMENT REVISION A GPO PRICE $ CFST| PRICE(S) $ Hard copy (HC) , _-#'-';_J) Microfiche (MF) /'- _+3V <+7V and most sig. bit is IV Expected tolerance + 0.2%, - 0.1% Extreme freq. limits • 1.0% Expected stability better than. 01% Figure II-2 6 | I ? .f I "l $ IMP's D&E, as in previous IMPls, contains a sequence counter ("satellite clock") that will have a capacity of 60 days. This clock, of course, may be used by the computer, in conjunction with optical aspect data, to resolve timing problems. The clock may be put to many other uses by programming techniques. C. READOUT OF EXPERIMENT DIGITAL WORDS Some digital experiments are not amenable to accumulation and must be readout into the telemetry. Various methods have been used in the past to ac- comflisll the readout. Since the number of bits per channel has im.reased from 3 to 8, the number of wires required for some readout techniques _ill become plohibitive since the encoding system packing density is becoming c.mnector- bound. All methods have disadvantages and advantages and each experiment will be treated on an experiment-by-experiment basis. Four methods will be discussed assuming 32 bits are required for readout. : 1. Method I VrDirect Scan" _- In this method, very commonly used in the past, a wire is received from _ each bit and directly scanned in the encoder. Thus, 32 wires are required for 32 bits, which is the main disadvantage of the direct scan method. The advan- tages of the method are that the experimenter needs no complicated matrix or shift register in his package although he must work with a negative four-or- _- _: five-volt supply to obtain the best system performance. Another advantage is _ that the readout is directly slaved to the encoder clock and countdown, and thus _ there is no chance of "slipping bits". A further advantage is that the bits are '_" scanned in order. '7 2. Method II 'q:ndirect Scan" (See Figure II-4. ) i In this method the wires are reduced from 32 to 12 for 32 bits. This ._ method has also been used in the past with excellent success (Optical Aspect in IMPVs A, B & C. ) The main advantage of this system over the direct scan method is the saving of wires. The readout is also slaved to the encoder. The disadvantage over Method I, from the experimenter point of view, is that he _ must do the "ending" of his bits with the encoder supplied sync pulses four bits : at a time into the encoder. The encoder sync pulses are negative when the four _ • bits associated with it are to be readout and positive at all other times. In * addiUou each sync pulse sometimes must be conditioned by the experimenter. 9 J I 2 > q SYNC 1 3 4 -------_ 6 _ SYNC 2 7 8 ._ 9 SYNC 3 10 _ q 11.-_.-_ 12 > 13 > ENCODER 14 _ SYNC 4 SYNC 15 -_ MATRIX 17 18 > SYNC 5 19 ._ 20 > 21 :- SYNC 6 22 > e ,, 23"_"_ 24"------_ 25 = SYNC 7 26 _ _ . 27_ 28 29 > 30 : . SYNC 8 DIGITAL 31 _ - ' COMMUTATOR 32 I ; I I 11 i I I - DATA [] _ INPUTS IV I III Figure ]I-4. Indirect Scan Method (Readout Method H) 10 { I "1 Another advantage of this method is that the bits are scanned in order and that the experimenter need not have a shift register. The sync pulse generation for the indirect scan is more complicated than those required for the direct scan; therefore, the total electronics required for the indirect scan is generally more complicated than for the direct scan. 3. Method III "Countdown Reconstruction" (See Figure 11-5.) In this method the encoder supplies the experiment the channel rate, called S, at the rate of 50/16CPS and the sequence rate, called _,, at a rate of 1 cycle every 81.92 sp,'_nds. The experimenter then drives an 8 binary stage in his package with S and resets them once/sequence with _, to get his binaries (called H' through A') in phase with the encoder countdown. Once the countdown is reconstructed, the experiment can generate any required pulse to scan his word four bits at a time into the encoder on just four wires. This method saves wires in that only 6 interface wires are required for any number of bits. The main disadvantage is that the possibility exists that a failure could cause the experiment countdown to get out of phase with the encoder countdown, resulting in garbled data. This method becomes very attractive if the experiment requires much commutation of intern._l electronics in addition to the readout problem. In this case, the encoder would have to supply many additional sync pulses in addition to those described in previous methods. With this method, the experimenter then generates all sync pulses in his package and reduces the interface problem and his checkout. In other words, his package is almost self-contained. -4 Method 3 is not recommeT:ded to solve the problem of just reading out ex- 1_ periment digital words but should be considered if many additional switching functions are required by the experiment. '_ 4. Method IV "Shift Signal Method" (See Figure 1I-6. ) This fourth method causes the experimenter to have a shtR register in his package and may be used for any size word. Figure H-6 shows a 32 bit series shift register that is shifted 7 times, one bit at a time, in order to read " out the 32 bit word. The first readout, bits 8, 16, 24 and 32, are readout by the encoder before a shift signal is applied. The 8th positive going waveform may cause his register to shift or not, at his discretion, since the eneoder will • not readout this experiment at this time. 11 I .! EXPERIMENT ENCODER CHANNEL RATE 50/16 CLOCK J EXPERIMENT ENCODER LOGIC LOGIC FUNCTION FUNCTION MATRIX MATRIX SEQ RATE J TO CLOCK I I _ DIGITAL DATA J 4 Figure II-5. Countdown Reconstruction Method (Readout Method m) 12 13 I It is essential that the experimenter understand two timing considerations: a. His shifting must be complete in less than 100 x 10 -6 seconds from the time the positive going shift signal reaches +3 V DC. b. The shift register must not change state again until the next positive going swing of the shift signal. These timing limitations _re required so that the sexadecimal burst will never change during the readout. It is realized that modifications of this scheme are possible, i.e., four parallel 8 bit shift registers could be ased. A possible disadvantage of this scheme is that the bits are not scanned in order. It may be possible, however, for the experimenter to arrange his bits so that the sexadecimal readout is a meaningful entity. As previously men- tioned, disordered bit structure may cause a checkout and data processing problem. The shift signal characteristics will be as follows: a. Output Z approximately 20 K ohms, this is an exception to the rule for sync pulses and is required in this case to keep rise times down to a reasonable value for the shifting. b. When not being readout, the shift signal is approximately +5.5 V DC. c. During readout the shift signal will be an approximate square wave as shown on Figure II-6 that goes from approximately -2.8 V to approximately +5.5 V. The experimenter is to initiate his shift pulse _ on the positive going part of the square wave. d. The shift signal may have noise spikes of duration less than 10 x 10 -6 _: seconds. :_" This method is attractive for experiments that already have shift registers and requires the least amount of encodsr sync pulse generation and wires in • the cable. For large words in the experimenter package (greater than 16 bits), this method is suggested. t The disadvantages of the method are numerous, some of which are listed: a. Experiment may have to use a shift register that would otherwise be unnecessary. 1 : 14 . I 'I b. Possibility of experiment word getting out of syne with the eneoder exists. • c. Possible problem with bit ordering exists. Timing should not be a problem since 100 microseconds should enable low power shift register to be used. +6.5V 100K TO DIGITAL BIT I, 1I, Ill OR IV COMMUTATOR FROM EXPERIMENT I_ "0" STATE= 0 V 20 K '°1" STATE= > +3 V WHERE i NT. S291 Figure II-7. The Digital Scan Interface Circuit D. ANALOG DATA Two experiments will contain analog data. The Ames magnetometer and part os the I_T data will be converted from a 0-to+5 volt signal to a 15 KC-to- 5 KC frequency, and then the video will be divided by 16. The analog performance parameters will be diglt/zed and transmitted as sexadec/mal frequencies. There are several advantages .to diglt/zat/ctz of these parameters. • a. Many of these parameters are needed to make decisicm as t( when the fourth stage should be fired. The ground support equipment can decode these sex_cinud munbers even in noisy signals, whereas analog data . would not be as easily decoded. b. The nml_Ix _ whether analog or digital data should be decoded is simpler, hence more relinble. 16 c. The oscillator calibration will be independent of comb filter anomalies, instead of varying due to the filters. d. The only disadvantage is that the parameters are sampled less fre- quently, but those infrequently sampled do not change quickly, so no information is lost. e. Pre-flight checkout can be accomplished nmch more easily tka9 if the data were telemetered as raw analog. f. The frames are either all analog or all digital (except for channel 0 sync pulses), thus making the system more standard. 1. Analog Experiment Output Characteristics (except for P. P. ) The analog experiment should vary in the range of 0 V to +5 V full scale and should never exceed +5.5 V when saturated. The experiment ,,oltage output can vary slowly (. 05 volts/burst) within the above voltage limits, this dynamic voltage restriction is necessitated by comb filter response. If the experiment output impedance is less than 1K ohm, the encoder im- pedance will have a negligible effect on the number being telem, tered, thus the experimenter should be able to match his output voltage versus parameter readings with the encoder input voltage versus telemetered number without taking into account the encoder input impedance. 2. Encoder Input Versus Telemetered Number Characteristics for Performance Parameters Figure H-8 is the ideal curve of analog input voltage versus decimal conversion of the double sexadecimal channel bursts representing the analog input voltage. Thus, an analog input voltage of 0V would be changed to 15KC and converted to tt,e number 219 while an analog input volt_ge of +5 V would be changed to 5KC and converted to the number 19. In the r.eferenced curve, the control volt versus number cut is linear, so that +2.5 V input voltage will re- sult in a telemetered number 119. The dynamic range of the system is 200 counts for a voltage swing of 5 volts. The counts will be limited by th_ system characteristics for voltages greater than +5 V DC. Note Figure II-8 is the theoretically perfect curve and the actual curves will not be this good. Cal/bration curves will be made _ Inlmt voltage (at 1 K ohms out) versus telemetered deoimal numbers at various temperatures and In-flight cal/brat/on will be made. It/s hoped that system accuracy _ _I_ over the temperature range d 0_0 to +40' 0 _,tll be ma/ntained/n this manner. 16 1 I 1 229 219 15KC 209 CALIBRATION 199 189 179 ACTUAL CALIBRATION CURVES VERSUSTEMPERATUREWILL BE 169 AVAILABLE FOR EACH PAYLOAD 159 149 139 O 129 IBRATION "_ I0 KC .< 119 L. ..... 109 e_ ; Z :::) 99 I 89 0 : 79 u I 69 O I _: )N s9 L 49 0 Z I i 39 I 29 I i I _ "ION I I I I I ' O0 PERFORMANCEPARAMETER I 2 3 VOLTAGE 4 INPUT 5 6 ! Figure H-8. Performance Parameter Voltage Versus Digital Readout 17 t ! | 3. Description of Operation (see Figure II-9) The two analog experiments and the performance parameters will be commutated into the single analog oscillator. During channel 0,000, a 20 milli- second gating Jziterval allows the pulses generated by the analog oscillator to be accumulated by the 8-bit accumulator. The 8 bits are readout during channel 8 of each odd frame, as one performance parameter. The average of the performance parameter is taken during the 20 MS sample. Since 15KC for 20MS is 300 counts ana since 5KC for 20MS is 100 counts, the system has a dynamic range {resolution} of 200 counts. It is clear that the 8-bit accumulator will overflow if greater than 256 counts enter it. Thus, it is preset to 175 before the accumulation occurs. This is a strange number, but is chosen because it results in the preset word having the maximum number of "l's" and still being useful. The number 175 was chosen then, to reduce the "race problem" and not for ease of computer usage. If all works according to plan, with the accumulator preset to 175, exactly 81 additional pulses will cause it to overflow to all "O's". Thus an input fre- quency of 4.05 KC will result in the number 0 being readout and an input fre- quency of 12.75 + 4.05 = 16.8KC will cause the readout number to be 255. One immediately asks, why not make use of the entire accumulator capacity and have a resolution of 1 part in 256 instead of 1 part in 200 ? The answer basically is one of hardware, the analog oscillator is tried and proven and is in the 5 to 15 KC range (see reference B} while 20 MS is a 2" division of the 409.6KC Xtal, thus 10KC for 20MS results in the 200 count dynamic range. Another reason is that it is nice to have some guard bands to insure against overflow in the event of a negative input voltage. The oscillator limits at a i frequency slightly below 5 KC (control volts greater than +5 V), _ 4. Reliability Considerations One familiar with previous PFM encoding systems will note that this is a very radical departure in that only one analog oscillator is used instead of several (one for each experiment). Indeed, only one digital oscillator is used. A few of the concepts used in reliability considerations will be listed and a • mathematical comparison will be left to the reader if he wishes one. a. The cross talk problem among analog gates has been resolved by the addition of the SEQ gates. {page 18) This was previously taken care of using multiple "fail safe" OFF analog oscillators. b. One oscillator takes less parts than more than one oscillator. 18 ! ! , .! I ! AMESIN [_ G AME_-_/ MIT IN [_ G MIT PPI PP2 G____ PP3 GATES GATE PP4 SEQ .._ PP5 FRAME I GATES PP6 PP7 GATES GATE PP8 SEQ .__ PPI7 SEQ _'_ PP18 GATES GATEJ.} PP19 LJ PPI5 _ GATES GATES PP26 t GATES GATES OSCILLATOR GATES _ CALIBRATION "._ J SEQ .._ , ANALOG "_ OSCILLATOR .-i TO MIXER ACCUMMULATOR DIGITALCOMMUTATOR • 1 il !11 IV CONVERTER r GATEFOR TO DIGITALOSCILLATOR PP Fibre TT-9. Analog Data Handling lg k ! ! • I t I c. Although the oscillator has been designed wi_h "fail safe" OFF cir- cuitry, multiple oscillators require multiple gating matrices for which a "shu_" may cause two oscillators to be gated "on" at the same time. This type of failure may be catastrophic with multiple oscillators. In this case, the single oscillator is on for analog data and for 20 MS every odd frame and required gating is done at the accumulator output with a simple NOT digital function. d. From a systems point of view this single oscillator approach has merit in that in-flight calibration is practical. 5. Analo_ Oscillator Calibration Procedure A check of the analog oscillator will be made in-flight, calibration points will be as shown in the following table: Readout in Voltage Source Channel 8, Frame 15 Purpose Sequence - t Ground, 0 Volts 2 Check Slope & Bias Thermistor 4 Temperature Slope and Bias from Voltage Standard, 4 Volts 6 Absolute Reference Detect 7.5 volt change, 4 Volt Divider 8 slope & bias change & compare with standard Ground, 0 Volts 10 Check Slope & Bias Thermistor 12 Temperature 25 Volt Divider 14 Check Llnearity b 5 Volt Divider 16 Check Knee & Linearity The data processing people will bergiven the calibration of the oscillator as determined during encoder burn-in. Comparison of these numbers, with those experienced during spacecraft calibration and flight should be made. Deviations of more than one count indicate a shift in some parameters. The amount of 2O ! I I l shift experienced by each calibration point may define the type of parameter shift, and allow data to be corrected to within i1% of proper value. The addi- tion of an analog calibration flag, based on the correct comparison of all 4 calibration checks {with possible correction due to temperature) should be con- sidered by the data processing people. With consultation of the project staff, correction of analog data in excess of 1% should be found possible. Table of Oscillator Calibration Capability Showing Minimum Detectable Variations at cach Calibration Point, Independent of all others: I Apparent Voltage Shift {Expressed in Percent of Five Volts Full Scale), Sufficient to Cause a One Count Type of Change in Oscillator Change at the Accumulator Output. Circuits Ground 2.5 Volt 4 Volt 4 Volt 5 Volt Divider Divider Standard Divider 7.5 V Supply Apparent Flux +0.5% -_0,8% :el. 1% -_0.5% +1.5% Change (func- +0.5% +0. 5% _0.5% +0. 5% +0.5%* tion of Temp} Limiter Voltage - ?%** Nonlinear Curve 0 ±0. 5% 0 0 +0.5% Slope Changes m A. About Gnd :el. 0% 4-0.6% :e0.6% +0.5% : B. About +5V +0.5% +1.0% +2.5% :e2.5% C. About +2.5V +1.0% 4-1.7% +1.7% +1.0% Shift in 2.5V Divider +0. 5% • Shift in ,Standard +0.5% • *The +5 V divider reading will be -. 5% unless the apparent shift appears to reach the limiter voltsge, in which esse the % change will be smaller. **Depends on sharpness of knee of ltmiter, and ff it approaches the 5 volt input. I I { ! I i E. SYNC PULSES SUPPLIED TO EXPERIMENTER No standard sync (timing) pulses are available in a given encoder because the various possible combinations of sync pulses are tremendous. For any given satellite encoder, the experimenter's requirements are analyzed during the initial planning stage and the experimenter requests those pulses he desires. In the case o', D&E, the encoding system may become connector-bound, so it is valuable to save wires. (See Section II-C.) Another limitation in IMP's D&E is that of the experimenter no longer having a "blank" time in which to perform his switching since the blank is filled with data. In general, the experimenter has 100 x 10 -6 seconds to perform his switching from the time the sync pulse goes to +3 V (if + going) or to -1.8V (if - going) before any trouble may occur. Experimenters using the DDP may switch during the "freeze" time without causing any problems to the telemetry. The analog experimenters have at least 1 MS in which to switch from the beginning of any channel. Since it is essential that the switching does not cause the PFM burst to change during a channel period, a delay has been introduced into the video divide unit to allow switching times specified above. For the most part, any sync pulse can be generated in the eneoder so long as it starts at the beginning of any channel (e. g., at the beginning of the old "blank") and ends at the end of any channel. It is obvious that some sync pulses are more easily generated than others so that various trade-offs are sometimes desirable. All sync pulses, of course, are "timed" or synchronized with the encoder. The main requirement is that the experimenter makes known his requirement before the design freeze date because it is very difficult to make design changes after the encoder is built. Please note that the experimenter should feel free to change his mind about sync pulses up to the design freeze if it simplifies his experiment. Most pulses have "standard" output inpedances as follows: OFF condition _ +8.5 V at 200 K ohm. +6.5 V ON condition _ -3 V at 100 K ohm. The output circuit will be as follows: _|00K SYNC i PULSETO f(x)_ 100K EXPERIMENT " I _ _ _ "_ -3V -4V 22 ! • I { I I The rise times of the pulses are very poor. NOISE _ -1.8V- %_--k _-3V ZF o-oc J As shown in the above figure, the rise and fall times are less than 30 microseconds when open circuited at the experiment end. These pulses have a tendency to be noisy (sharp spikes of a short duration less than 10 microseconds) so that the experimenter receiving these pulses should take advantage of the large voltage swing between the two states to per- form his switching. If differentiation of leading or trailing edges of the sync pulse is required, it must be done in the experimenter's package by the experimenter taking into account the noise problem mentioned above. The Shift Signal has different characteristics. (See Section IIC-4, page 10). The following table gives basic timing nomenclature. ! IMP D ENCODER COUNTDOWN NA ME FRE QUE NC Y PE PlOD FORMAT _, mmJ 4C 1 409.6 kc 2.44_s Clock ! 2C 1 204.8 kc 4.88 _s C1 102.4 ke 9.77 ps C2 51.2 kc 19.5 ps • C3 25.6 kc 39.1 _us C4 12.8 kc 78.1 ps C5 6.4 kc 156.0 ps • C6 3.2 kc 313.0 ps C7 1.6 kc 655.0 ps C8 800; 0 cps 1.25 ms continued 28 IMP D ENCODER COUNTDOWN (continued) NAME FREQUENCY PERIOD FORMAT C9 400.0 cps 2.5 ms C10 200.0 cps 5.0 ms Cll 100.0 cps 10.0 ms C12 50.0 cps 20.0 ms C13 (8S) 25.0 cps 40.0 ms 1//8 Channel C14 (4S) 12.5 cps 80.0 ms 1/4 Channel C15 (2S) 6.25 cps 160.0 ms 1//2 Channel S 3.13 cps 320.0 ms 1 Channel H 1.56 cps 640.0 ms 2 Chaamels G 781.0 mcps 1.28 seconds 4 Channels F 391.0 mcps 2.56 seconds 8 Channels E 195.0 mcps 5.12 seconds 1 Frame (16 Channels) D 97.7 mcps 10.2 seconds 2 Frames C 48.8 mcps 20.5 seconds 4 Frames B 24.4 mcps 41.0 seconds 8 Frames A 12.2 meps 81.92 seconds 1 Sequence (16 Frames) a1 6.10 mcps 2.73 minutes 2 Sequences a2 3.05 mcps 5.46 minutes 4 Sequences a3 1.53 mcps 10.9 minutes 8 Sequences a4 763.0 jucps 21.8 minutes 16 Sequences a5 381.0 _cps 43.7 minutes 32 Sequences a6 191.0 _acps 87.4 minutes 64 Sequences (1.47 hours) a7 95.4 _cps 2.91 hours 128 Sequences a8 47.8 tucps 5.83 hours 256 Sequences a9 23.8 _ucps 11.7 hours 512 Sequences alO 11.9 bucps 23.3 hours 1024 Sequences all 5.96 _ucps 46.6 hours 2048 Sequences (1.94 days) a12 2.98 _cps 93.2 hours 4096 Sequences (3.88 days) a13 1.49 _ucps 7.77 days 8192 Sequences (I. Ii weeks) a14 745.0 ncps 15.5 days 16384 Sequences (2.22 weeks) a15 373.0 ncps 31.1 days 32768 Sequences . 14.44weeks) (1.02months) a16 136.0 ncps 2.04 months 65536 Sequences 24 F. CIRCUIT DESIGN AND FABRICATION TECIINIQUES I. Approaches Taken The IMP D&E encoding system is functionally much more complex than that of IMP's A, B and C. If a measure of the functional complexity is the num- ber of binary stages required, then IMP D is approximately 2 times as complex as IMP A. IMP A has about 140 binary stages and IMP D has about 250, II\IP A had approximately 5,000 electrical parts, of which 3,000 were not resistors. If IMP D used the same circuit design and fabrication technique it would have approximately 9,400 electrical parts of which 5,600 would not be resistors. None of the above figures include welds. Since the IMP A encoding system pushed the state of the art, using conventional components, it would be very desirable to go some other route with IMP D. Several approaches have been taken to reduce the IMP's D & E parts count and the main ones are listed: a. Redesign of electrical circuits (binaries and logic circuits) and im- proved system design (sexadecimal bursts instead of octal bursts, to make the logic based on a system divisible by 2 instead of 3) in order to reduce the number of parts. This has been done and it is estimated that approximately 2,700 parts can be saved resulting in a :'conven- tional" design with 6,600 parts of which 4,300 are non-resistors. This is still a large number of parts. b. A monolithic solid circuit approach where two types of low power elements would be used. The elements were a BINARY BLOCK (at 1/2 row/bit) and a low power LOGIC BLOCK (at approximately 1/10 mw/logic function). The two blocks would be used in approxi- mately 65% of the system and "conventional" component., _n the rest. Both blocks were based on the improved circuit de_ , eve. At this writing, this approach is still in the developme: ,age and does not appear to be feasible for the IMP D & E schedule. c. The MOSFET (Metal Oxide Silicon Field Effect Transistor) approach. " Here again two basic building blocks will be used, a MOSFET BINARY (with resistors added externally). Two promising things happen with i this approach, one is that about 93_ of the system will be MOSFET 1 " blocks or resistors and the other is that the total parts count goes | down to less than 4,000 parts where less than 1,000 are non-resistors. Thus, this approach results in about the same number of total parts used in the IMP A encoding system but has even less non-resistor 25 parts than IMP A. Please note again that IMP D has about two times the functional complexity of IMP A. In all the above, .the assumption is made that resistors are in order of mag- nitude more reliable than non-resistor (capacitors, diodes, transistors or MOSFETWs) parts and may be weighed accordingly in a mathematical reliability analysis. 2. Status of Approaches a. Conventional Design (using IMP A circuits and logic) is out of the question (about 9,400 parts required). b. Conventional Design (using simplified binary circuits) is better than the above and will be possible but not desirable (6,600 parts required). c. Monolithic solid circuit approach for 65% of the system and conven- tion design for the rest is desirable but cannot be used due to schedule problems. (Not feasible due to schedule.) d. MOSFET approach. Appears to be by far the best from its electrical properties, reduction of part_, and schedule. Its disadvantage is that "SUPPOSEDLY" less is known about its long term reliability than is known about the conventional approach. (Approximately 3,100 parts required where less than 1,000 are non-resistors. ) 3. MOSFET Approach Expanded Since all other approaches except the MOSFET approach are familiar to most readers of this document and since the MOSFET is a new animal, this paragraph is intended to give some of the electrical properties of the device that are useful in IMP encoding system design. A primary design constraint is power, it must be relatively low. Since we do a great deal of accumulation and storage in the Encoding System Digital Data Processor, binary counters are used. One primary problem then, iR how do you make a low power binary, at reasonable speeds (up to 500KC), that is elec- trically quiet and that will perform for a year in orbit over the temperature range of -30°C to +60Oc. In the past (IMP A) we have used complimentary flip-flops with considerable success. _Ve have developed a 2 transistor "low power" f'dp-flop to el/m/hate half the transistors. Either of these devices op- erate on the princ/ple of using low voltages (supply) and reasonably low resistor values up to about 200 K ohms. The fact that the suPPlY voltages are low implies 26 ! I that the "trigger" voltages must be low and this means that the noise rejection threshold is low. The MOSFET binary works on a different principle to achieve low power, it uses large voltages and large resistors. The supply voltages have a difference of potential of 10.5 volts and it takes a "trigger" of about 5 volts {large) to flip them. This gi,es excellent noise rejection. It should be pointed out here that all capacitors and diodes have been eliminated in the bi- nary design and the device uses a double '-'nl)utthreshold level "flipping" tech- nique. Indeed, a binary stage consists of one TO-5 can and four large value resistors where a complimentary flip-flop requires 4 transistors, 6 resistors, 5 capacitors and either 4 or 8 diodes depending on speed requirements. Thus, the low power MOSFET has 5 each discrete parts and the low power complimen- tary flip-flop has up to 23 parts; this is a parts reduction of better than 4 to 1, and is a "non-resistor" parts reduction of up to 17 to 1! ! The fact that the re- sistors are external enables the designer to obtain an efficient speed versus power profile. Thus, a 0.1 mw low speed MOSFET flip-flop is obtained by using 2 megohm resistors. The external high value resistors used with the MOSFET binary protect against a "short" in a MOSFET binary causing a cata- strophic failure by "deadheading" the line. This is particularly important since 80%of the binaries are such that a failure will cause loss of data from only part of a single experiment. The rest of the experiments would still provide useful information. An especially exciting feature of the MOSFET approach is the elimination of many stages in the electrical testing of modules and of the individual dJacrete components that go into the modules. Another desirable quality of the MOSFET is that it is a voltage (not current) operated device. The input impedance at its gate is many megohms, thus no _ steady state power is required to operate them as a logic function (e. g., it is _ not necessary to supply "base current"). The MOSFET essentially has no "offset" voltage; it acts like an open cir- _ cult when off and a resistor (less than 3 K) when on. When off the leakage is very low. This feature is very useful in commutating devices. Another unique propriety of the MOSFET is that it is symmetric (one cantt tell the source from dra/n). This, of course, means that a simple monolithic logic block can be made from MOSFETts and can be arranged in many different logic configurations. The blocks used are as follows: ! I • ) I l It is left to the designer to utilize the many ways that the blocks can be connected and he will be pleasantly surprised at the things he can do with MOSFET logic blocks with very few parts and with considerable power savings. On some occasions transistors are required and can be mated very nicely with MOSFET logic when required. The writer is not qualified to comment on semiconductor manufacturing processes, but has been assured by competent personnel ;hat the manufacturing process steps required to make an all MOSFET monolithic chip are considerably fewer than those for conventional integrated circuits. The two basic MOSFET blocks are made by the General Micro Electronics Company (GME) in California. The cooperation received from that plant has been outstanding in thai "prototype" blocks of both types were delivered within a month after they received our drawings (FIBS-2 Figure 7). The blocks worked very well and the above exam- ple is given to illustrate that the manufacturing processes must indeed be simpler than those of integrated circuits. 4. Evaluation of MOSFET Approach The various stages in evaluating the MOSFET approach are and will be as foItows: a. Circuit Deslb_n Using Individual MOSFETts: Several different binaries were designed and breadboarded at Goddard. These binaries were operated in an extremely simple sys- tem to check their noise rejection, and speed versus power profile. The one chosen is that shown on Drawing FIBS-2, Figure 7. The disadvantage of this binary is that it takes 12 MOSFETts a_ld 4 resis- tors and that its standby power is twice that of other binaries bread- bearded. The advantage gained is that it is a threshold double input direct coupled device that can be manufactured with no capacitors or diodes. Good speed versus power was not obtained with the bread- beard due to excessive capacities but this problem was largely solved by GME in their monolithic package. b. Specification for the Two Basic Building Blocks It was decided to have the 12 MOSFET's of the binary in an integrated package and to mount the chip in a TO-5 header. The four resistors were to be placed external to the package for two basic reasons as follows: 2S (1) The speed versus power can be tailored and (2) The MOS technology has not advanced to the point where it is practical to put large value MOS resistors on the chip. In addi- tion, the binary cannot "deadhead" the line if the large value resistors are external. The TO-5 package was chosen because it is easier to weld to than the flat pack. c. System Check Using Many MOSFET's (Breadboard) A representative encoding system has been breadboarded. This sys- tem is about two-thirds the functional complexity of IMP A and con- tains the circuits that produce the new functions to be used in IMP's F&G (e.g., sixteen level oscillators, A/D converter, "S-T" accumu- lators, etc. ) It contains 127 binary blocks and 208 logic blocks with only 17 transistors. The breadboard was made with 8 pin "tube -ockets" such that the production MOSFET's were "plugged" in as soon as they arrived. Breadboarding with MOSFET blocks turns out to be much nicer than with conventional design. The electrical design and MOSFET performance was evaluated with this system. The system worked very well indeed. d. Welded MOSFET Encoding System Package The breadboard MOSFET's were removed and were welded into a package (Delta Pack) and complete system checks have been per- formed using simulated experiment inputs. The package has been potted. The welded modules were layed out and fabricated by .i Goddard's Modular Techniques Section. The MOSFET card has been evaluated at T&E. The following tests were run, and the unit passed all successfully: "_ % a) Initial magnetic tests b) Acceleration tests c) Vibration " d) Humidity e) Thermal vacuum 0 Final magnetic tests ? Modules for IMP D&E are now bein$ fabricated. 29 .l I ........ '} PRECEDINGPA-GE BL -ANK'NOTF!LNED. PART III. IMP D&E ENCODER INTERFACES AMES MAGNETOMETER INTERFACE The Ames Magnetometer experiment will read out 30 analog samples during channels 1 through 15 of frames 0, 4, 8 and 12 and digital data flags wPl be read in channel 12 of frame 7. 1. The analog voltage data line, called "Ames In" will vary between ground and +5 volts with an output impedance less than 1 K ohm. The experimenter shall insure that the voltage never exceeds +5.5 volts and does not go more negative than -0.5. Variation during readout burst less than . 5% is required for accurate data processing. 2, 3, 4, and 5. The digital data flag lines will be defined in the following manner: The "true state" will be defined when level is 0 V +. 25, -3 volts, this will be encoded as a "0" The "false state" will be defined when 5 V +1, -2 volts, this will be encoded as a "1" Bit I is least significant and Bit IV is most significant. The data flags will mean: .,. a. Calibrate flag axed perm. bit 2 :" L- b. Flipper flag and perm. bit 3 :_ y:, c. Channel switch flag and perm. bit 4 _: :_ d. Perm. bit 1 and perm. bit 5. :_ One set of flags will be gated by the experiment onto the four lines during " the first half of every channel, and the other set of flags will be gated, by the experiment, onto the four lines during the second half of every channel. ' The synchronization pulses will be, ._ore negative than -2 volts through 100 K ohms during the specified time, and more positive than +5.5 volts through 200K ohms at all other times. (See II-E) 31 t l • l Y "I The synchronization lines required are: 200 6. "4S" (16 cps) square wave, positive at beginning of charnel 100 7. "2S" (16 cps) square wave, positive at beginning of channel 5O 8. "S" (16 cps) square wave, positive at beginning of channel 9. "H': square wave, positive at beginning of channel 10. "G" square wave, positive at beginning of frame 11. "F" square wave, positive at beginning of frame 12. "E" square wave, positive at beginning of frame 13. "D" square wave, positive at begmmng of sequence 14. "C" square wave, positive at beginning of sequence 15. "ato" square wave period approximately 23 hours 16. "B" equare wave, positive at beginning of sequence 17. "A" square wave, positive at beginning of sequence . 18. "a t ,t square wave period of two sequences. . • 82 33 EXPF:R IMENT ENCODER AMESIN ANALOG OSCILLATOR COMMUTATOR 4S 2S S H G F E SYNC _, PULSE D MATRIX C B A ,q al alO BITI - LEASTSIGNIFICANT BITTI b DIGITAL OSCILLATOR BIT1"n' Ii COMMUTATOR BIT I_- MOST SIGNIFICANT , ) nn ! Ames Interface Diagram 84 { • I UNIVERSITY OF IOWA INTERFACE The University of Iowa Experiment is allocated eight accumulators in the DDP as defined below. Number Readout In Maximum Accumulator of Bits - Count Rate Channels Frames (KC) ,m 1. la 20 4, 5, 1/2(6) 1, 9 250 2. lb 12" 1/2(6), 7 1, 9 250 3. 2a 20 12, 13, 1/2(14) 1, 9 250 4. 2b 12" 1/2(14), 15 1, 9 250 5. 3a 20 4, 5, 1/2(6) 3, 11 250 6. 3b 12"* 1/2(6), 7 3, 11 250 7. 4a 20 12, 13, 1/2(14) 3, 11 250 8. 4b 12"* 1/2(14), 15 3, 11 250 Each accumulator will be reset after every readout, in groups of 32 bits. Pulse duration will be between 2 and 3 p_seconds. The synchronization pulses will be more negative than -2 volts through 100 K ohms during the specified time, and more positive than +5.5 volts through 200 K ohms at all other times. (See II-E.) \ The synchronization pulses required are: !. ! 10. Channel (0------_15) - Frame 1, 2, 3 _ ii. Channel (0----o-15) - Frame 9, I0, II The accumUlator input pulses are counted when they change from ground to a voltage greater than +3 volts. The voltages should not exceed +7 e • The readout of Ib represents the least sipifieant 12 bits and readout of 2b represents the most sipificant 12 bits of a '_4bit accumulator. "The readout of 3b represents the least sipificsnt 12 bits and readout of 4b represents the most sipifi,'_mt 12 of s 24 bit sccmnulator. 35 .l EXPERIMENT DDP AND ENCODER ACC lalNPUT - ACC la J RESET ACC lb, 2b INPUT ... ACC lb f 11 ACC 2b j RESET . _ ACC 2a ACC 2a INPUT j ACC 3a INPUT ._ ACC "_ "_ J,RESET " ACC 3b, 4b INPUT _ ACC 3b J_ ---_- Jl_R Acc I RESET --------=1 ' " ACC4a J IJllJ (0"15) 1, 2, 3 II (0_15) 9, 10, 11 SYNC - PULSE MATRIX University of Iowa Interface volts or-3 volts at the accumulator input. The accumulator input impedance will be at least 10K ohms. The experiment must not allow pulses to activate line "lb, 2b in" during channels 4 ------15 of frames 1 and 9 or line "3b, 4b in" during channels 4 -----_15 of frames 3 and 11 as erroneous readout of the accumulators will occur. See Section IV-D for encoder freeze information of all accumulators. r-I" EACH SEQUENCE FRAME NUMBERS +V --1 I (o-.15 1,2,3 -v +V (o--.15) 9,lo,11I I -v • • aimmo •in0 • alnoo o o m• mo • • • • • • • no • • • • • • moo • • • • • • • ma no m o •• o aaln• • • o o • m o • •• • Io l i• • ou • • • • B •• • •• • • • om• allm_• • • 10u• • • • ONE SEQUENCE _I-_ ONE SEQUENCE *'1 University of Iowa Sync Pulses MASSACHUSETTS INSTITUTE OF TECHNOLOGY INTERFACE The MIT experiment will read out 30 analog samples during channels 1 through 15 of frame 6 and 30 4-bit digital words during channels 1 through 15 of frame 14. _ 1. The analog voltage data line, called "MIT IN" will vary between ground and +5 volts with an output impedance less than 1 K ohm. This voltage shall not be able to exceed +5.5 volts or -3 volts (limiting to be done in experiment). Variation during readout should be less than. 5%. i 2, 3, 4 and 5. The digital data lines will be defined as follows: One state will be defined when near ground. The other state will be de- . fined when greater than +3 volts. (See p. 14) Bits to be held constant during each burstt within lOft _ seconds after burst begins. Bit I is least stgrdficant and Bit IV is most significant. EXPERIMENT ENCO DER ANALOG INPUT ANALOG OSCILLATOR COMMUTATOR BIT I - LEASTSIGNIFICANT lm BIT IT DIGITAL BIT IU OSCILLATOR II BIT IV'- MOST SIGNIFICANT COMMUTATOR 2S SYNC PULSE {0_15) -5 MATRIX (0--15)-6 (o-.15)-14 II r ii Massachusetts Inst/tute of Technolo_ Interface 38 The synchronization pulses will be more negative than -2 volts through 100 K ohms during the specified time, and more positive than +5.5 volts through 200 K ohms at all other times. (See II-E) The synchronization lines required are: 6. Ci-- _ - 50cps square wave, negative at the beginning of each data burst. 7. 2-'S - 50/8 cps square wave, negative at the beginning of each data burst. 8. S - 50/16cps square wave, negative at the beginning of each channel. (Unused in AIMP D; used by thermal ion and electron experiment in AIMP E) 9. (0------_15) - 6 - negative during all of frame 6. 10. (0------+15) - 14 - negative during all of frame 14. 11. (0-------15) - 5 - negative during all of frame 5. L. i_ EACH CHANNEL _I w I +V C°2 -V I L,, I ' ollllp oiD o IDio 00o o e ollllllo igloo j ilO000DOggllOlOiOgaBSIDlOOOIBOOOlO OI OlO a • i a mgo • • m • DDO mgOIDOOalmmBg ,[ • I.. EACHSEQUENCE _o Io I' I 21314Is 161;' 18I_'I'°1" 1':_1'31_41'sl FRAMES i -5 (0--15) ..... j j, -V +V , +V .:.z (o..,._5)-6 I I -v -_ C_ --15) -14 • , L___'_ +V -V _ Maslmehueette Im_Utute of Technology 8ync 1Atleee _ L 3g _- , ! THERMAL ION AND ELECTRON INTERFACE FOR AIMP D TEMPLE UNIVERSITY INTERFACE FOR AIMP E The Temple University Experiment digital data will be scanned during channels 5, 6, and 7 of frames 7 and 15. Four bits will be scanned during each half of the channel. These bits shall be shifted and held steady within 100 microseconds by the experiment after the shift command is sent. (See p. 10) 1, 2, 3 and 4. The digital data lines are defined in the following way: One state will be defined when near ground. The other state will be defined when greater than +3 volts. (See p. 14) Bit I is least significant and Bit IV is most signii._,_, Synchronization pulses will be supplied more negative than -2 volts through 100K ohms during the specified time, and more positive than +5.5 volts through 200K ohms at all other times. The pulses are: 5. Channel 4 _ 7 of frames 7 and 15 (B. C. D. F. F. ) will be negative only during the second quarter of frames 7 and 15 (to be used to inhibit experiment during readout). 6. _, negative during first half of sequence. 7. F 5 32 seq. period 8. Shift pulse. A shift pulse will be supplied to the experiment, changhlg from a voltage more negative than -2 volts through 1.0 K ohms and rising to a voltage more positive than +5 volts through 20 K ohms. The output is an emitter follower with 20 K ohms output series resistor. The shift operation is to be perforr._d when the level shifts from negative to positive. The shift pulse should not be directly differ- entiated because of possible noise on _e shift line causing interference. 40 FRAMES7 AND 15 _-I CHANNELS IIIIIII +V SHIFT PULSE I_ -V TIME OF SHIFT A,,,,, " II BITS1, 2, 3, 4 I BITS5, 6, 7, 8 TIME OF • BITS9, 10, 11, 12 READOUT OF 24 BITS • BITS 13, 14, 15, 16 • BITS17, 18, 19, 20 : • BITS21, 22, 23, 24 I J EXPERIMENT ENCODER SHIRTPULSE .1[. " _ 21 _ A SYNC 17 a"_ PULSE : MATRIX 13 9 _ (4--*.7)7 + 15 5 ! T BITI - LEASTSIGNIFICANT i,,, _L 22 18 14 !0 6 _" 2 __ w "1- BIT 11 DIGITAL iv ,I';" .._ DATA COMMUTATOR :z: 23 19 J 15 IIw 7 BIT!11 H BIT1_'- MOST SIGNIF:CANT _i " _:_ :_ _ Thermal Ion (AIMP D) & Temple Interface (AIMP E) ,,+ ' 42 UNIVERSITY OF CALIFORNIA INTERFACE The University of California experiment will feed four accumulators as defined below: I Readout In Accumulator Accumulator Type t Speed (KC) Channels Frames 1. 5a - 16 bit S-T 100 4 & 5 5, 13 2. 5b - 16 bit Count 10 6 & 7 5, 13 3. 6a - 16 bit S-T 100 12 & 13 5, 13 4. 6b- 12 bit* ] Count 10 14 1/2 & 15 5, 13 _,ceumulator 5 will De reset after channel 7 on frames 5 and 13. Accumulator 6 will be reset after channel 15 on frames 5 and 13. The synchronization pulses will be more negative than -2 volts through 100K ohms during the specified time, and more positive than +5.5 volts through 200K ohms at all other times. The sync pulses required are: (See II-E) 1. C7 - 1.6 kilocycle square wave 2. (Channel 4-------7} Frames 5 and 13 3. (Channel 12-----*15} Frames 5 and 13 The accumulator input pulses are counted when they change from ground to { a voltage greater than +3 volts. The voltages should not exceed +7 volts or -3 volts at the accumulator input. The accumulator input impedance will be at | • least 10 K ohms. The accumulator input pulses will be 2 _second or greater. _' (See Iv-c) *The position in the telemetry format for the first four bits of ace. #6b will be used to repeat the four most significant bits of accumulator 6a. 4B • I I 1.6 KILOCYCLES i_,_ EACH SEQUENCE =J I0 I, 1213I_ 15I_ I_ 1819I,°I I_1,21,_I,41151 I I FRAMES I I . +V CHANNELS 4-_7 U FRAMES5, 13 U -V I I I I +V CHANNELS 12-*'15 U FRAMES5, 13 U -V ||i|iiiimoIl|iilllO|l|ill ii IIIlllllgiO|llliOli||||lil|l|lli|i|lll||nl||||i|||||i||i||||lii|||ll|i||il||i BIT ORDEROF ACCUMMULATOR//6 "S-T" "S" FRAME5 + 13,____.__._ 12 13 14 15 IN CHANNELS 14 14 ACCBIT 4 I 8 12 I 16 16 I 4 8 I 12 LOCATIONS _ I" PARTOF ACC 6a University of California Syne Pulses 44 -i EXPERIMENT DDP AND ENCODER I ACC 5a INPUT [ S-T j "- ACC 5a J RESET ACC.5b INPUT r,,,. ACC 5b I ,_CLOCK INPUT ACC 6a INPUT S-T l ACC 6u ACC6bINPUT _" _] 6b I RESET z_ C7 -1.6 KC •11 y ACC 5 FREEZE SYNC { PULSE i MATRIX ACC 6 FREEZE -_: •. iii i University of California Interface 45 I ,! ) i GODDARD MAGNETOMETER INTERFACE The Goddard Magnetometer experiment willbe scanned on four digitaldata lines. The four digitaldata lineswillbe definedin the followingway: "0" statewillbe definedwhen near ground. The "I" statewillbe definedwhen greater than +3 volts.(Seep. 14) BitI is leastsignificant and Bit IV is most significant. Bits shouldnot change during readout. The encoder inputimpedance willbe greater than 20 K ohms. The synchronizationpulses willbe more negativethan -2 voltsthrough 100 K ohms during the specified time and more positive than +5.5 volts through 200K ohms at all other times. (See II-E) The sync pulses supplied by the encoder will be: i. Channel 1 and 9, every odd frame. 2. Channel 2 and 10, every odd frame. 3. Channel 3 and 11, every odd frame. 4. Channel 4 every frame. 5. Channel 12 frame 15. 6. S - negative during first half of every channel and positive during the second half. 7. A - negative for first half of every sequence, positive for second half of every sequence. (unused AIMP D; used by thermal ion _r exp. on AIMP E) 8. _ - negative for 256 sequences, positive for next 256 sequences. 46 • ! 9 EXPERIMENT ENCODER BIT I X FIELD r LEAST SIGNIFICANT ID 8 BITS Y FIELD BIT 11 " DIGITAL _ DIGITAL 8 BITS DATA DATA Z FIELD COMMU- BIT [] COMMU- TATOR ' " TATOR 8 BITS BIT I_ MOST SIGNIFICANT 8 FLAG BITS MAX CHAN 12 FRAME 15 CHANNELS I AND 9, ODD FRAMES CHANNELS 2 AND 10, ODD FRAMES CHANNELS 3 AND 11, ODD FRAMES d CHANNEL 4 i| CHANNEL 12, FRAME 1.5 SYNC PULSE MATRIX J • I ao I • Goddard Magnetometer Interface 48 OPTICAL ASPECT INTERFACE The optical aspect detector will shift 120 digital bits, four bits per half channel, onto four digital data lines. These bits shall be shifted within 100 microseconds after the shift command is sent and held steady until the next shift commmld is sensed. The digital data lines will be defined as on page 14, . except the maximum voltage will be 3.5 volts. Minimum for "1" state will be +1.0 volts into the 10K base resistor, "0" state at ground to +. 5 volts. Bit I is least significant and Bit IV is most significant. The synchronization pulses will be supplied more negative than -2 volts through 50 K ohms during the specified time, and more positive than +5.5 volts through 100K ohms at all other times. (,See II-E) The pulses to be supplied are: 1. Clock pulse 800 cps - C s. 2. Clock pulse 6.4 KC - C 5. 3. (Ps)l (Channel 0------15) (0-----2, 8-------_10) sequence 1, 2, 3 FR (0------*2, 10) sequence 4. 4. (Ps)2 (All of FR0, 1, 2, 10) Seq l& 3, (All ofFR2, 10) Seq2&4. 5. (Pa)l (Channel 0-------15) (8 & 9) sequence 4. 6. (Pa)2 (Allof FR8, 9)Seql & 3, (Allof FR0, I, 8, 9) Seq2 & 4. q. Channel (0------15) FR (0, 1, 8, 9). A shift pulse will be supplied to the experiment, changing from a voltage _:• more negative than -2 volts through 20 K ohms and rising to a voltage more }_ positive than +5 volts through 20 K ohms. The output is an emitter follower with 20 K ohms output resistor. The shift operation is to be performed when the level shifts from negative to positive. The shift pulse should not be directly differentiated because of possible noise on the shift line causing interference. (See II-C-4) t 49 6.4KC +V Cs -V 800CPS c. ] rv h----SEQ 1 -_I_ SEQ 2 :I_ SEQ 3 ;I_ SEQ 4---_I 0 4 12 0 4 0 IZ 0 4 e 12 0 4 0 12 FRAMES ii il i ii il II II II li II II II II II II II FR. (0,1,8, 9)"]_J--'_ I ,t +V I I Ps, -L_f--il---! , r--I ,11-1 r-]I I-1 r----lr-l -V I I Ps='_ I I I 'I I I -V +V ! I t I I I I PAe ='1' I ,, I ,,' I ,,' I I;' I"I I " L__ -+v -V I t I I I I o i I I I ; I PA= ir-n r--1r u _ L_J -V SHIFT PULSE '_ FRAMES2 AND 10 "J Io1' 1213141516171819 I'o1"1'21'sl,41'51 tttt TIME OF SHift J BITS 1, 2, 3, 4 • BITS 5, 6, 7, 8 TIME OF READOUT OF • BITS 9, I0, 11, 12 FIRST 16 BITS i BITS 13, 14,'15, 16 Optical Juspeot 8yno Mses .,. " 5O l I EXPERIMENT ENCODER SHIFT PULSE q q FRAMES (o, 1, 8, 9) P,, Psz SYNC 30 BIT Pol PULSE MATRIX SHIFT • REG = Po, C5 Ce 4 J BIT [ - LEAST SIGNIFICANT 1 30 BIT SHIFT J BIT 11 1 30 DIGIiAL BIT DATA SHIFT COMMUTAIOR J BIT ITI,. . : I 3O BIT SHIFT i , " I lilT I_r - MOST SIGNIFICANT Op_ca2 Aspeot Znterface 51 i ! ADDITIONAL OPTICAL ASPECT INFORMATION A parity check of the optical aspect data will be attempted within the en- coder. This check will not only check the quality of the data when all is pres- ent, but will also allow single missing data points within a group of data to be corrected with a high degree of certainty, as shown in the following table: [ Approx. _, of ] Parity Data to Data to be Approx. P I Approx. I- Symbol be Checked Thrown Out (double error) I (correction) PCi FR2, CIt(1-----*7) 1500 Pe .50 Pe 2 Pe PC 2 FR2, CIt(8----*15) 1700 Pe .54 Pe a Pe PC 5 FR10, CH (1-----* 7) 1500 Pe .50 Pe 2 Pe ! PC4 ] FR10, CH(8-----*15) 1700 Pe .54 Pe 2 Pe l Where Pe = probability ofwronglydigitized data P (doubleerror) = probability ofnotdetectinga doubleerrorwhen all datapointsare digitized P (correction) = probability of inserting '.,he proper information for one missingdatapoint. For IMP D, thevalueofPe ifcomb filters are used insignaldetection is verymuch lessthan.01 (ifcorrelators are used,thevalueshouldstillbe less than .011. IfthevalueofPe were .01, then15% ofthedatawouldbe thrown outtogivetheremainingdataabout.00006probability ofbeinginerror. The probabilityofincorrectly inserting theinformation ina missingdatapoint shouldbe about.01. INTERFACE WITH SOLAR CELL EXPERIMENT The solar cells will produce voltages from 0 volts to +5 volts with imped- ances less than I00 ohms. The four output Ifues will be fed into the analog commutator, and be digitized before bering transmitted as performance para- meters d, ring the see.end sequence. Thus, they will be transmitted once even- lout _quences, as four, S-bit words. 52 Read Out PPg Seq. 2 FR1 ,$ PPlo Seq. 2 FR3 PPft Seq. 2 FR5 PPl2 Seq. 2 FR7 SOLAR CELLS ENCODER i SOLAR EXPERIMENT //1 PP9 IB SOL/R EXPERIMENT //2 PP 10 ' "_ _ ANALOG SOLAR EXPERIMENT //3 PP 11 .. COMMUTATION SOLAR EXPERIMENT //4 PP 12 i -z i m Interface with Solar Cell Lxperiment i ! INTERFACE WITH PERFORMANCE PARAMETER CARD t| The lines £hown on _e u_xt page will vary between 0 and +5 volts with im- . pedances less thsu 15 K ohms. The analog commutator will select the proper parameter, which is fed to the analog_)acillator. An S-bit accumulator is gated on for 20 milliseconds in channel 0 and readout in channel 8 of each even frame. (See H-D) $8 1 { I Jl Channel 8 (8 bits each) - Digitized Analog -- Frame Frame - No. ] Sequence 1 No. Sequence 3 1 PP1, +12 volt Buss 1 PPl, +12 volt Buss -: 3 PP2, Battery Voltage 3 PP2, Battery Voltage 5 PP3, Battery Current 5 PP3, Battery Current 7 PP4, Solar Array Current 7 PP4, Solar Array Current 9 PP5, Spacecraft Current 9 PP5, Spacecraft Current 11 PP6, +28 volt Buss 11 PP17, Transmitter Tem- perature 13 PPT, +7 volts Temperature 13 PP18, Battery Temperature PP's 15 PP8, UI voltage 15 PPl9, Prime Converter Temperature _ Sequence 2 Sequence 4 1 PP9, Solar Experiment 1 J PP20, Solar Array Tem- perature 3 PP10, Solar Experiment 2 3 PP21, Ames Temperature 1 5 PPll, Solar Ex-periment 3 _. PP22, Ames Temperature 2 7 PPl2, Solar Experiment 4 7 PP23, GSFC Magnetometer Temperature 9 PP13, Solar Experiment 9 PP24, UCAL Temperature Tempera_tre 11 PPl4, Low Bottle 11 PP25, M1T Temperature Temperature/Temple Temperature 13 PPIS, H/gh Bottle 13 PP26, UI Temperature Temperature/Fourth Stage F/rlng Time 15 PPl6, Encoder Oscillator 15 PP16, Encoder Oscillator Calibration Calibration . IMP-D Perforsmmce Parameter Assignments AIMP D PP CARD ENCODER • +12 VOLT BUSS PP1 _ - BATTERYVOLTAGE ._ PP2 ,, • BATTERYCURRENT .: PP3 > SOLAR ARRAY CURRENT PP4 -._ SPACECRAFTCURRENT PP5 > +28 VOLT BUSS -._ PP6 +7 VOLT BUSS :, PP7 _ ,._ Ul VOLTAGE _ PP8 e ; SOLAR EXP TEMP _ PP13 , LOW BOTTLETEMP/I & .ETEMP • PP14 ,, HIGH BOTTLETEMP/4THFIRING DURATIOk ANALOG -_ PP15 - COMMUTATION TRANSMITTERTEMP PP17 - BATTERYiEMP PPi8 _ "_" PRIME CONVERTERTEMP :_ PP19 , SOLAR ARRAY TEMP PP20 , , AMES MAGNETOMETERTEMP PP21 AMES ELECTRONICSTEMP PP22 ' GSFC MAG TEMP PP23 UN CAL TEMP PP24 MIT TEMP • PP25 " Ul TEMP ' PP26 - "11 r •_L1KIPD withp=.t=.=,m Channel 8 (8 bits each) - Digitized Analog Frame Frame No. Sequence 1 No. Sequence 3 1 PP1, +12 volt Buss 1 PP1, +12 volt Buss 3 PP2, Battery Voltage 3 PP2, Battery Voltage 5 PP3, Battery Current 5 PP3, Battery Current 7 PP4, Solar Array Current 7 PP4, Solar Array Current 9 PP5, Spacecraft Current 9 PP5, Spacecraft Current 11 PP6, +28 volt Buss 11 PP17, Transmitter Tem- perature 13 PLY/, +7 volt Thermistor_ 13 PP18, Battery Temperature 15 PP8, +20 volt Buss 15 PP19, Prime Converter Temperature Sequence 2 Sequence 4 1 PP9, Solar Experiment 1 1 PP20, Solar Array Tem- perature 3 PP10, Solar Experiment 2 3 PP21, Ames Magnetometer 5 PPll, Solar Experiment 3 Temperature 7 PP12, Solar Experiment 4 5 PP22, AC Line Temperature 9 PPl3, 3rd Stage Separation/ 7 PP23, GSFC Magnetometer Solar Experiment Temperature Temperature 9 PP24, UCAL Temperature 11 PPl4, Low Bottle Temperature/Temple 11 PP25, Attitude Control Temperature Tank Temperature 13 PPlS, High Bottle 13 PP26, UI Temperature Temperature/Fourth Stage Firing Time 15 PP16, Encoder Oscillator 15 PPl6, Encoder Oscillator Calibration Calibration IMP-E Performance Parameter Assignments ,, i J .¼ I '! AIMP E PP CARD ENCODER t +12 VOLT BUSS PP1 ,. "" r -. BATTERYVOLTAGE .._L ,._ PP2 _ ;_ • BATTERYCURRENT ._ PP3 '. _ - SOLAR ARRAY CURRENT -- PP4 _ -" SPACECRAFTCURRENT PP5 _' +28 VOI.T BUSS PP6 +7 VOLT THERMISTORS PP7 _,- _ +20 VOLT BUSS PP8 ,_ _ .. SOLAREXPTEMP/3RDSTAGESEPARATION :_ PP13 _- LOW BOTTLETEMP/TEMPLETEMP PP14 , , _- ANALOG HIGH BOTTLETEMP/4THFIRING DURATION COMMUTATION PP15 "_ TRANSMITTERTEMP ' PP17 '" ' BATTERYTEMP PP18 - "_ PPI9 SOLAR ARRAY TEMP PP20 "_I AMES MAGNETOMETERTEMP PP21 ATTITUDECONTROL LINE TEMPERATURE PP22 i,,, - GSFC MAG TEMP PP23 UN CAL TEMP PP24 ATTITUDECONTROL TANK TEMPERATURE • PP25 '" UI TEMP PP26 _e i.i X_e wi_ Pezfozmuoe _ete_ Cazd 5_ INTERFACE WITH PROGRAMMER CARDS _ : The binary performance parameters will have the following charac_._ristics: A voltage more positive than 3 volts into a 20 K ohm encoder impedance will represent "One" state. / • j A voltage more negative than +. 5 volts of ground will represeat the "0" logic state. 1. 4th stage ignition timer #1 1st decade (first burst, bit I) 2. 4th stage ignition timer #1 2nd decade (first burst, bit II) i 3. Separation Armed (first burst, bit IV) 5 4. 4th stage ignition timer #2 1st decade (second burst, bit I) :_ ,( 5. 4th stage ignition timer #2, 2nd decade (second tmrst, bit II) 6. Separation Armed (second burst, bit IV) ,_ The following sync pulses supplied to the programmer will vary from a voltage more positive than 5 volts at 10K ohms output impedance to a voltage more negative than ground at 110 K ohms output impedance: 7. "Flip Goddard Magnetometer" 8. '*Flip #1 Ames Magnetometer" 9. "Flip _2 Ames Magnetometer" An attempt will be made to filter out short duration noise spikes from lines 7, 8 and 9. 10. 3rd and 4th stage separation (first burst, Bit m) will be detected from a dropping resistor connected to the switched +12 V of the programmer card, t4mer section. (For AIMP D only) _... . tt 58 F 59 ENCODER 1 4TH STAGE IGNITION TIMER '_'1 BIT ] , 1ST DECADE 4TH STAGE IGNITION TIMER #I BIT 11 2ND DECADE 4TH STAGE j 3RD STAGE BIT ITI | SEPARATION SEPARATION .I. '-r- SEPARATION ARMED BIT IV" "_ DIGITAL 4TH STAGE IGNITION TIMER //2 BIT [ IST D_CADE i COMMUTATOR O O _" 4TH STAGE IGNITION TIMER //2 BIT ' '_ +V 2ND DECADE 4TH STAGE SEPARATION , ARMED BIT IV _tp I FLIP GODDARD MAG E.F. I ! .... q s, c , 1It. D j 6O i AIMP E _NCODER " 4TH STAGE IGNITION TIMLR #1 BIT T ii 1ST DECADE i 4TH STAGE IGNITION TIMER //1 BIT 11 J , ___ 2ND DECADE ATTITUDECONTROl_ SECTORS 8IT T1] 1000pfd 4TH STAGE SEPARATIONARMED BIT ]]Z _I _ 1000nfd r UJ | n 4TH STAGE IGNITION i TIMER //2 BIT I DIGITAL ,'-." 1ST DECADE COMMUTATOR O O " 4TH STAGE IGNITION TIMER //2 BIT 11' _ +6.5V 2ND DECADE 10K ATTITUDECONTROL SECTORS BIT rn' :.. 1000pfd * 4TH STAGE SEPARATION ARMED BIT EZ I FLIP GODDARD MAG - E.F. I I FLIP //2 AMES MAG .... -tI SYNC SIGNAL " _ E.F. II FLIP e //2 AMES MAG ' ' ' .... E.F._"1! MATRIX qf I _t,e_e wtt.EEhroiprammer Fm_tou 61 ! I ADDI'IIONAL INTERFACE WITH PROGRAMMEB CARDS FOR AIMP E I. Attitude control sector indicator (first buIst, bit IIl)- A voltage more positive than 3 volts into a 20 K ohm, 500 pfd encoder impedance will represent the "I" state. A voltage more negative than +0.5 volts into 44 K ohm, 500 pfd to ground will represent a logic "O". 2. Attitude cot_trol sector indicator (second burst, bit ii_ - A voltage more positive than 3 volts into a 20K ohm, 500pfd encoder impedance to ground and 10 I': ohm impedance to +6.5 volt., within the encoder will represent the "I" state. A voltage more negative than +0.5 volts into a 44 K ohm 500 pfd to ground and 10 K ohm to +6.5 volts encoder impedance will represent a logic "O'. INTERFACE WITH TRANSMITTER The output to the transmitter is a complementary inverter with a 1 K ohm series impedance. The signal will be a series of unsymmetric square-wave bursts varying from ground to -3 volts (+570) in which the unsymmetry produces an average of -1.5 volts over a period of 40 milliseconds. That is, the duty cycle of the pulses varies from greater than 50% to less than 50% for equal periods of time and equal amounts within the 40 millisecond periods of digital data. The analog cLata will be symmetric at all times. A 24 kilohm resistor isolates the 'qow speed video test" output from the 4.4 modulator input. The minimum expected frequency is -_- KC _270 cycles and the maximum expected frequency • ls-__5 KC = 940 cycles. The square-wave will have rise and fall times less than 5 microseconds, and "noise" spikes less than 1% of video sib_mal. 4.4"15 KC TO MODULATOR WAVE 24K h -3:1:5,%V -4V _ 100K VIDEO TEST (FOR GSE) , Transmitter Interface Circuit 62 p t INTEIIFACIC _,_q'Ftl ENCODEI{ C()._ I'_I1"I'I':I{PACKAGE 1. The encoder converter pack_,re will contain the electronics for providing 5 voltages to the encoder electronics and will also house a voltage standard used for analog oscillator calibration. The estimated requirements for the eneoder voltages are shown in table form on the following page. CONVERTER ENCODER GROUND +4V STANDARD +7.5 VOLT,_ i r +6.5 VOLTS i | -3.0 VOLTS Iii -4.0 VOLTS ill -9.0 VOLTS Ii i i i 68 I i • -h I_. _ -H 0 I C_1 C_1 C_I ! i INTERFACE WITH ENCODER TEST FOR GSE FOR AIMP D The following syr, c pulses will be provided for GSE. e i. 2S - negative at beginning of each burst 2. A - negative at beginning of each sequence 3. _22 - negative at beginning of "sequence one" 4. C8 - o00 cycle square wave The following outputs will be provided. 5. Low Speed Video Out (268------1000 cycles), an emitter follower with 24 K ohm output impedance isolated from the signal to transmitter. 6. High Speed Video Out (4.4KC----*15KC). To be used for DDP machine only (not to be used in T&E). <_+6o 5V TO GSE : f(x) 330K f(x) L 1_ 100K I-4 V 1-3 V f(x)--2s, A, az, Ce oeo • illallallogoallmmlelmeomJe • elem • • mm • a_•• • •a • am • • malI• • •• • am • •J • • eo •0 nm• wm I• iiaaDo •0 eamammme•m• iio ••• eomo •• • • • • •qmuiaw 120K LOW SPEED TBD k iK VIDEO_ _ TO OUT 120K J J _ MODULATOR j ,,, TBD _',___..___.. TO LOW SPEED i l-3V / .... VIDEO TEST i , 100K_ 6-4V Encoder Test Interface Circuit 65 AIMP E ENCODER I & E .S (OLD MIT S) ° THERMAL I & E A (OLD GODDARD A) ION AND ELECTRON EXPERIMENT GSE C8 GSE C8 SYNC PULSE GSE 2"-S MATRIX ,ill GROUND SUPPORT GSE EQUIPMENT .. , GSE L.S. VIDEO i ATTITUDE AC a-i (OLD GSE _) CONTROL _ ELECTRONICS iml i "_ AIMP E I & E, Attitude Control & GSE Interface 66 INTERFACES WITH THERMAL ION AND ELECTRON EXPERIMENT, ATTITUDE CONTROL, AND GSE FOR AhMP E 1. I & E S (old MIT S) - 50/16cps square wave, negative at the beginning of each channel (Encoder P_ - pin 3). 2. I & E _, (old Goddard A) - negative for first half of every sequence, positive for second half of every sequence (Encoder P3 - Pin 1). 3. GSE Cs - 800 cycle square wave with unknown phase relationship to channel when loaded. Will be shared with ground support equipment and Thermal Ion Experiment. ,_. GSE 2S - negative at beginning of each burst. 5. GSE _, - negative first half of sequence, positive second half of sequence. 6. AC _ (old GSE _) negative for sequences 1 and 2 positive sequences 3 and4. Each of these lines, 1 through 6, are 200K ohm to +6.5 volts when positive and 100 K to -3 volts when negative. 7. GSE Low Speed Video same as on Ar_;P D. 8. GSE High Speed Video to be used fol DDP machine only. Same as _. on AIMP D. f & 67 f PRECEDING PAGE BLANK NOT FILMED. PART IV. DIGITAL DATA PIIOCESS()R (DDP) FOR IMP's D&E A. GENERAL The DDP has been incorporated into the encoding system to accept experi- ment pulses that must be either accumulated or stored until readout The capacity is 188 bits that ore readout every 40.96 seconds, S bits/ channel. The 188 bits may be arranged in various combinations of aecumu!ator lengths. In addition, two types of accumulators are available; one is called an "S" type and the other an "S-T" type accumulator. The "S" type accumulator is a straight binary counter similar to those used in IMP's A, B & C while the "S-T" type accumulator is used for experi- ments that may cause an overflow of the accumulator. "S-T" strands for SIGNAL or TIME. The "S-T" type will accumulate until it overflows and will then count clock pulses for the rest of the accumulation period. Thus for low counting rates, the accumulator will readout the number of pulses counted and for high counting rates will give a measure of counting rate (e. g., the time required to overflow the accumulator). The maximum rate any accumulator will aecept is 500 KC but an effort will be made to tailor the speed versus power characteristics of the DDP. Thus, each experimenter is requested to give us the maximum possible rate his ex- periment will produce. We will build accumulators with maximum rates of 500KC, 250KC and 125KC. B. ACCUMULATOR GROUPINGS (SIZE) \ As shown on the telemetry format (SectionII-7),thebasic accumulator word is 32 bitsthatare readout every 40.96 seconds (4 channels). Since an entire 32 bit accumulator is frozen for the 4 channel readout, the minimum "dead time" will be 1.28 seconds out of 40.96 seconds for a ratio of dead time to accumulation time of 1/32. Each 32 bit accumulator may be broken up into - as many as 4 accumulators or as few as one accumulator where the number of ! bits in each accumulator is divisible by 2. In addition, "S" and "S-T" type accumulators may be combined into a 32 bit word. Some possible examples follow and they will be given: 5a = 16 bits "S-T"'type 5b = 16 bits "S" type 69 I or la :-- 20 bits "S" type lb :- 12 bits "S" type Many more combinations are possible, an example of an impossible combination follows: 5a = 19 bits any type 5b = 13 bits any type Accumulators not broken up into groups divisible by 4 will result in a sin- gle sexadecimal burst containing information from two different accumulators. Provisions mr, st be made to take care of this during satellite checkout and bit reordering will be required during data processing. In addition, the least significant bit will be readout first as was done in IMP's A, B & C. C. ACCUMULATOR INPUT CHARACTERISTICS As previously mentioned, an attempt will be made to tailor the speed power characteristics of the accumulator. The maximum rate an accumulator will accept is a 500 KC sine wave or pulses with a duration of greater than one microsecond. Over-all system performance will be increased if the experi- menter will inform the project staff as to his maximum possible input rates and minimum pulse duration. The input impedance that will be seen by the experiment pulse will be a resistor of value greater than 10 K. ?+V 12K TO DIGITAL COMMUTATOR :_ r ---7 >1o K = R, J_ i ENCODER Some noise reJe_ton is provided by the resistors R, and R 2 m the above ,v figure, The input signal will be reJect@d when its amplitude is less than 2 V • 0, 5 V depending an speed and temperature, This amplitude measured at the "in" terminal with the experiment connected, When the pulse is n_.otpresent 7O the experiment output nmst be less than +0.5 V and when the pulse is present, the experiment output must be between +3.5 V and +7 V. Thus the accumulators will not accept negative going pulses and will get confused if a positive vo!.tage . is present during standby oFeration. In order to solve any interface problems that may occur, a simulator will be sent to the experimenter, at his request, after the design freeze date. This simulator will contain the accumulator signal conditioner and the l'irst /'our bits of his accumulator. D. FREEZE This is not really an electrical interface since the experimenter has no control over it. However, an understanding of freeze may be useful to the experimenter in interpreting his data. The word freeze means, "will not accept pulses". Accumulators are a]- ways frozen during readout so that they will not change during the accumulator readout. Every 32 bit word will be frozen for the four channels during which it is readout. It is realized that this results in a dead time (1/16) that is undesir- able, but the circuitry required to freeze is very simple compared to a rapid transfer and hold in buffer storage scheme. Again, note that freeze may be made longer if experimenters time-share adjacent accumulators. E. RESET Each 32 bit group of accumulators has two options for reset: 1 1. The entire 32 bits are reset after readout (at end of freeze). _, 2. None of the 32 bits are reset after readout (at end of freeze). "S" type accumulators are reset to all l's (highest number). The first pulse received from the experiment will then flip the entire accumulator to its "0" J state. From then on it will count normally. Thus, the information telemetered from the resetable "S" type accumulators will indicate one less count than was actually received (e.g., if the accumulator reads 29,986 counts, then 29,987 counts were actually received). ? "S-T" type accumulators are essentially reset to zero as far as the experimenter is concerned. Thus, if the accumulator reads 29,986 counts, 71 I then 29,986 counts were received. The state of tile most significant bit will tell whether these counts were from the experiment or from the eneoder clock. F. "S-T" TYPE ACCUMULATOR This type accumulator willcount signalpulses (S)up to a maximum number and willthen count clockpulses (T)for the restof the accumulationperiod. Thus, eithersignalcounts (S)or countingrate (T)willbe telemetered. The purpose of the device is to take care of overflow, thus increasingthe dynamic range. Itshould be noted thatthe accumulationperiod and the clockpulses are slaves to the same 409.6 KC satellite Xtal. Although "S-T" accumulators may be made of any size following the, rules in paragraph B, a description will be given for a 16 bit "S-T" accumulator. If a different size accumulator is required, the princ'pal of operation is the same but the clock pulse frequency will be changed so as not to overflow in the count (T) mode. Referring to figure IV-l, the 16 bit "S-T" accumulator works as follows: 1. After readout (freeze), the entire 32 bit word, of which the 16 bit ;'S-T"accumulator ispart, is reset to alll's and allowsthe 800 CPS clock through. 2. The firstclock pulse willcome throughwithin2.5 MS and flipall12 accumulators to zero and allow the signal"S" inputto come through. 3. The accumulator will allow signal pulses (S) through from the experi- i ment until the 16th bit again changes to "1". This will take exactly 32,768 pulses, at which time the 800 CPS clock (T) will count until the accumulator is again frozen for readout. 4. At a 800CPS clock rate, the accumulator will acquire 800 x 39.68 - _ 31,744 pulses in the maximum accumulation time of 36.224 seconds. Since it would take 1,024 clock pulses to overflow in the T mode, an _,| 11% safety factor exists. Readout If the 16th bit of the accumulator is "0", then the readout is the number of _ pulses aoqu/red in the 39.68 second accumulation period. Note that it is unnec- essary to add the "1" to the oount. 72 i J ! 32,768 If the 16th ])it of the accumulator is "1", the first 15 bits of tile accumula- tor give the number of clock pulses acquired after the accumulator overflowed to a resolution of about 1.25 MS. Assuming an Xtal accuracy of exactly 409.6 KC, thus a sequence time of exactly al. 92 seconds the counting rate will be: Counting Rate = 32,768 (39.68 - . 00125 x T count) ±. 00125 in pulses/second Since the Xtal ,lccuracy is only 0.2_, although its stability is better than .01% and since, if it shatters, the clock will no longer be Xtal controlled, the above equation might be written in terms ef the actual sequence rate. The ratio of the clock frequency to the accumulation period will remain constant in the event of an Xtal failure so it probably isnlt necessary to take sequence rate into account in the equation. In the 'IS'v mode, however, actual sequence rate is important. 74