Soyuz Clock Part 7: Taming the Glitches in our Space Clock Crystal Oscillator
Ғылым және технология
In this rather longish video we get to the bottom of the glitching issues we had seen in the crystal oscillator of our Soyuz space clock. Which gets us on a detour to reverse engineering a Soviet TTL chip, look at a quartz crystal on the VNA, and fight general RF electronic mayhem.
00:00 Glitch investigation
04:39 Inside a vintage Russian TTL chip
08:16 Quartz crystals and oscillators
21:15 Debugging the glitchy oscillator
32:44 Success and conclusion
Ken Shirriff article on reverse engineering the Soviet TTL chip:
www.righto.com/2020/03/looking...
Fantastic reference book about crystal oscillators:
Crystal Oscillator Circuits, Robert J. Matthys, Kreiger Publishing, ISBN O-89464-552-8
Our sponsor for PCBs: www.pcbway.com
Support the team on Patreon: / curiousmarc
Buy shirts on Teespring: teespring.com/stores/curiousm...
Learn more on the companion site: www.curiousmarc.com
Contact info: kzread.infoa...
Пікірлер: 310
So where’s the noise coming from? Could it be that the noise has developed over time due to components ageing and perishing? Or was this as built?
@CuriousMarc
3 жыл бұрын
Could be either. We actually don’t know, except that it comes from the power supply. Or a third possibility, some component was fried in a previous power up attempt: we had found added wires across the 5V and the switching transistor that should not have been there, leading to the outside. Thinking of which, these could also have been also a diagnostic attempt to look at a power supply noise problem while the clock was closed. So a mystery still remains.
@letsgocamping88
3 жыл бұрын
CuriousMarc either way it was a great explanation of crystal oscillators.
@CameronTacklind
3 жыл бұрын
Marc mentioned the noise looked like spark gap noise... confirmed my thinking, what if there is a tiny gap forming somewhere?
@dlugiprogres
3 жыл бұрын
@@CuriousMarc Badałem amerykańskie, japońskie i sowieckie zegary z lat 70 i 80, oscylatory obecnie podlegają jakiejś anomalii.
@robbyddurham1624
3 жыл бұрын
@@CuriousMarc you guys are very smart. I was wondering if you'd tried a dc battery in place of the power supply? I know you tried filtering it out. I wish I'd had youtube when I started working in electronics 35 years ago.
In USSR we make reverse engineering of import chips often , now you make reverse of reverse engineering, it's really silicon art))
@Jmzz542
3 жыл бұрын
In USSR we reach space first you second.
@garyzod8818
3 жыл бұрын
Warp factor 4 Mr sulu.
@BalticLab
3 жыл бұрын
Would reverse engineering of reverse engineering be considered forward engineering?
@oleglithin5101
Жыл бұрын
@@Jmzz542 In USSR without Korolev(Ukraine) and Ghlushko(Ukraine) you only reached your private ass ..
@MichaelLenz1
Жыл бұрын
@@BalticLab backward reverse
"2 free resistors for the people" :)
@grebz
3 жыл бұрын
I had a good giggle at that. And the peoples pull-up resistor. :D
@physnoct
3 жыл бұрын
@Peter Mortensen 7:52
@khatharrmalkavian3306
3 жыл бұрын
"Free resistors, comrade!" 6:20
“The people’s pull up resistor” 🤣
@brocktechnology
3 жыл бұрын
It's a workable T-shirt.
@yorgle
3 жыл бұрын
Really clever idea too. i'd have never thought of doing that technique. ;D (Inverter from ground vrs pullup resistor... erm... people's pullup vs capitalist pullup.... i suppose i never thought of it because Big Resistor wants you to buy more resistors.)
@staglomagnifico5711
3 жыл бұрын
My favorite wrestling move!
Somewhere in Russia a retired Soyuz engineer just said “D’Oh!” 🤦♂️😂
@vidasvv
3 жыл бұрын
I think he's still in Siberia
@dimakrayushkin
3 жыл бұрын
Спасибо за совет по доработке, я уже написал рацуху.
@user-ec2jr6pv1s
3 жыл бұрын
He probably died in 90's from vodka...
@MrPegge
3 жыл бұрын
@@user-ec2jr6pv1sIf not migrated to USA
@gpa30051984
3 жыл бұрын
"... а-а и так пойдет, сдавай работу"
Using the fourth gate is deliberate. It is to prevent an unused gate self oscillating at a high frequency, which can create all sorts of crazy parasitic effects, Noise and chip instability.
Excellent video..!! I'm a physics professor at a university in Buenos Aires, and I'm going to ask my students to watch this video, simulate your circuits and analyze your fixes. Very didactical..!! Thank you Marc..!
Long experience with crystals. In my professional designs, never use the logic gates, because they can stop working when the quartz ages and become "deaf", requiring higher excitation. I use a single jfet - which allows for proper loading of the crystal. Since you have the network analiser, you can plot the caractheristic of the quartz. Another possibility, have a dozen of crystals and test the circuit until it works. This happen because of the real Q depends on the cut of the quartz wafer and is random. Also, don't forget to ground the quartz case, otherwise the correcting capacitor value has no sense. Last, you may want to replace the TTL with a CMOS, so all the power supply noise problem goes away. The linearizing resistor value for CMOS is 2.2 MOhm, the quartz load resistor is from 4.7 to 10 KOhm. Don't use the 1 KOhm value suggested by data sheets. Also, check the ferrite cores in the switcher circuit are not cracked, and the isolating varnish on the copper winding is still valid. Also, measure the resistance of the PCB tracks, sometime they erode under the photoresist paint. Something in your PS is no longer within specs - and may not necessarily be a component.
@peterfitzpatrick7032
3 жыл бұрын
Antonio, you are talking about redesigning the osc. from scratch but thats not the goal here...personally, I would like to see if the power-rail spike could be traced, he did mention a switching transistor as a possible culprit. Interesting comment from you , thanks... 😎👍
@antoniomaglione4101
3 жыл бұрын
@@peterfitzpatrick7032 Switching transistors, especially that old, have shown a tendency of an increase of the recombination time within the Base. From a circuit perspective, this is equivalent to an increase of switching Time and a reduction of bandwidth. The driver transistor can present the same kind of derating. The net effect is that the driver and final transistors may not be able to switch all the energy contained in the coil, which is then radiated as an interference. So, the output transistors (and any eventual recovery diode) should be fully parametrically tested, in view of restoring the original functionality - and not hardening the crystal oscillator against the fault.
@peterfitzpatrick7032
3 жыл бұрын
@@antoniomaglione4101 Antonio ... when you say bandwidth, is that the same as the unity gain transition frequency fT ? And would a reduction in fT not reduce hf harmonics ? Or am I confusing noise with RF ? I believe he said the SMPS is runninig at 40Khz but the noise is up in the Mhz.. I know sq. waves can produce many harmonics 🤔 Thanks for the last reply btw.. 😎👍☘🍺
@antoniomaglione4101
3 жыл бұрын
@@peterfitzpatrick7032 Yes, increase in switching time reduce the ft and conseguently the bandwidth of entire stage. Reduced switching time can lead to many conseguences, one of which is the creation of stationary wave patterns, with unforeseeable results in the load.
@taldmd
3 жыл бұрын
>when the quartz ages How's that the quartz ages? Formation of quarz on Earth took literally millions of years. Explain "quartz aging" please.
I am an oscillator nerd and I learned a few things, I had never seen such a clear explanation of the series vs resonant osc difference :-) (to me they were about the same anyway) Seeing the narrowing of the band with the capacitor is also eye-opening :-)
Just a little bit of additional information on Soviet electronic components. If you find letters OC on it, then it means that this component belongs to a special series (ОС means Особая Серия, i.e. Special Series). Special series components may have slightly different specifications than those given in handbooks. Your IC 134ЛА8А has precisely those ОС letters
У Вас очень хороший парк приборов. Успехов в работе!
Great description of quartz crystal theory, Mark! When I learned all that stuff as a young radio technician 50+ years ago we didn't have VNAs to help us to visualise the serial/parallel resonances and the phase shifts - it was an entirely theoretical exercise. Wonderful to see that theory demonstrated visually on modern kit. I think you're right - the 'scopes of that era would have struggled to "see" those glitches, so they were probably blissfully unaware that there was a problem. I love the way you approach these challenges and how you explain what you are doing.
You guys never fail to make me smile. "It's a very pretty chip" shows you guys are truly passionate and I have so much respect for for that. You make it easy to understand complex designs. Keep up the good work.
I'd love to have you track down and diagnose the origin of the noise in a future video!
Master Ken strikes again...This time reverse engineering an IC... GENIUS👍
@Digital-Dan
3 жыл бұрын
And not for the first time.
@letsgocamping88
3 жыл бұрын
Hope he pinged an image to his French friend. With the IC de-capping habit.
134 серию помню, малопотребляюший аналог 133 серии. Микросхемы на керамике, 1975 год. Девочки в белых халатиках, под микроскопами приваривали проволочки от кристалла к выводам....
10:54 This is the best and shortest description of Resistance, Inductance and Capacitance I have ever come across! I will use this next time I explain electronics to a novice!
@alicangul2603
3 жыл бұрын
Yes he is the first person to come up with mechanical - electrical analogy lol
I like to see actual experts are using those little Ada fruit proto boards. I love those.
These days, Whenever I have to fix a Soyuz clock oscillator I always reverse engineer the IC's just like you too 😜
Hello CM, I love your instructional videos. Every video is a deep dive into something fascinating. Thank you for all of your efforts! You do a superb job! Thank you!
Awesome. Thanks for taking us through the debugging process. That was really enlightening!
Amazing troubleshooting, and thanks for the instructive discussion about crystal oscillators. Very interesting how a piece of quartz has all those electronic properties interacting.
I love everything about this video! Perfect explanations, beautifully executed engineering, and a great decision to not modify a historic piece of equipment, since it works already as intended.
Great video! I appreciate the analysis and explanations. You gave good solid theoretical explanations and followed it with practical hands-on fixes.
Marc, yet another amazing repair and explainer video! Keep up the great work! *(with all that awesome HP test gear!)
I am just a novice hobbyist, but you make things understandable and enjoyable! Thank you for your efforts to teach us. I know that it takes a lot of your time, energy and expertise. We do not count that lightly! It is much appreciated.
Excellent! Really enjoyed that one. I had to learn about this type of oscillator while working on a phone design that needed an ultra stable clock without costing too much - this brought back a lot of memories :)
Top quality content again. Thanks Mark.
Soyuz figured out how to tame the glitches? Excellent! Glad you weren't Russian the job and did it properly.
Once again a 30 minute video from you is magnitudes better than the three years of electrical engineering and solid-state physics that I did. You have a gift for teaching. Or, maybe I’m getting old and I’m better at listening.
Wow another great video. I actually learnt more than i thought i would. Also understand things clearer. Thank you.
Very interesting lesson in electronics. Thank you Marc.
Just yesterday I picked up my very own 8753D... It is fast becoming my pride and joy.
Love your arsenal of test gear! 👍
Fantastic video! Learnt a few things about dealing with jitter there!
My mind is blown! This episode was much more spicier than i expected.
You are VERY intelligent, we see that you a solid and deep knowledge of electronics. Congrats!👍👍👍
Outstanding side by side equivalency circuit!
Fast forward a year and they are ready to deliver a completely reengineered Soyuz to Baikonur. Top notch quality content!
This was fascinating, thank you.
Marc, Great series of videos on Soviet electronics. Your demonstration of increasing the radiated EMI immunity of the gate (EMI receptor) is very clear. Please also similarly demonstrate reducing the radiated EMI at the source. The 10MHZ ringing EMI source could be either the power transistor switching (try inserting a resistor or lossy bead in its base ) or the recovery diode not being soft enough or by using a transistor with lower Ft or...
Damnit, I love your vids, your lab and all your test equipment!
Were you tempted to go back to find the source of the noise? It may be that there is still an actual fault. Where they used a gate as a resistor, I wonder if that was a disable signal that they used in debugging and then just grounded the input to the gate for normal operation.
@SidneyCritic
3 жыл бұрын
That's what I was thinking, ie, it's patching a problem that is still there.
@antoniomaglione4101
3 жыл бұрын
Sometime is because the old ferrite core in the switcher circuit develops an hidden crack.
@dosgos
3 жыл бұрын
I thought deteriorating bits perhaps.
@curiouscrandall1
3 жыл бұрын
Me too on the "what about the source" thing. If the drain of the switching FET is ringing at 10MHz, I'd be inclined to investigate adding a little snubbing. Plus as Antonio says, have a look at the ferrite core.
@CuriousMarc
3 жыл бұрын
@video99: Absolutely, you could work the problem from the other end. I mostly wanted to know if there was a fault in the oscillator or not, and if not, how hard it would be to fix the design. Turned out not very hard. On the other hand, the power supply definitely could have an actual fault. But it’s harder to debug, quite risky if you mess up, and would require larger components to fix.
You guys are absolutely awesome!
Watching this 30 plus years dissapeared and I was back at Uni and National Semiconductor in Scotland looking down a microscope at dies and manipulating probe stations - thank you!
Good job! Very thorough reverse engineering!
I'm so excited for this video!
Can I have the name of the book at 8:53? Tried looking for "Oscillator Smörgåsbord" but found nothing hahaa edit: It's Crystal Oscillator Circuits by Robert J. Matthys
@CuriousMarc
3 жыл бұрын
Yes, it's an excellent book. Added the reference in the video description. Not sure if it's legit, but you can Google it up and download it in pdf form.
@iamjadedhobo
3 жыл бұрын
It is for sale on Amazon www.amazon.com/Crystal-Oscillator-Circuits-Robert-Matthys/dp/0471874019 for a measly $795.64
@stubell2363
3 жыл бұрын
@CuriousMarc: Good thing the book can be downloaded in PDF form - Amazon wants $795 for a hardcopy version. Wow!
Very informative! Thanks.
I want one! A replica clock: faithful panel, display, functions, would be quite nice. I suspect that clock works quite nearly the same now as when originally installed. ‘Good enough’ was good enough. Reliable, even with those performance issues, was the deciding factor. Glad to know that the clock will remain in its original configuration. Hopefully the hay-wire hacks that were applied by others isn’t the cause of the noisy behavior. Impressive diagnosis and engineered remedy. First class investigation.
That is a great analysis of the circuit. I'm not an engineer, but what I take away from this is the feedback loop in the stage after the crystal is, or should be, a DC control of the gain of the circuit with the AC component of the signal removed. I enjoyed this video as always!
@jerryeiting5367
3 жыл бұрын
@Peter Mortensen Good point. Maybe I should say the feedback loop should be low pass filtered at a frequency much lower than that of the oscillator.
im learning so much , nice video and exploration !!
This deserves two thumbs up!
I'm under the impression that rf stuff is basically black magic, and this video, while interesting and reasonable understandable, does not change my opinion 😁
One of your best videos yet! Love these! But was a bit of a kicker when you said you won't improve it after all this effort. Also curious where the noise originated from and if that was due to age or also part of the design.
Thank you Marc !
This is the first explanation of a crystal oscillator I've actually understood. Mass, springiness, and mechanical losses.
Excellent advises.
Very nice video. "It is within spec" => no need to make it better or fix the issues it has. I wonder if the people that designed this unit will see this video. Also, it would be interesting to use a really old oscilloscope on that clock signal to see if the issue appears.
@AngDavies
3 жыл бұрын
Jittter would just look a bit... Smudged, no? But a cro has bloom anyway at higher brightness, and the smudging might not be visible at higher (it's why the CRT traces look so smooth)
Ha ha ha... I thought it was your phone ringing at the end. Nice sleuthing on it... glad those chips worked out for you.
I wonder if grounding the crystal can might be an even simpler fix?
I studied electrical engineering about 32 years ago, so I really enjoyed that
Thenks Marc, very interesting. Have you considered grounding the quartz case?
@repairitdontreplaceit
3 жыл бұрын
i was going to suggest the same thing
@CuriousMarc
3 жыл бұрын
Just grounding the oscillator to the clock ground made it worse, so I'd expect it would make it even worse if you tied the crystal case to that bad ground. But I am just speculating, I did not try it. It would have been an interesting experiment.
Well done getting it working. Was the xtal can grounded? Wondering if the can was picking up a lot of that interference.
Hello. Would you like to take a look at MK-52 programmable calculator? "In November 1988, the MK-52 went into space on the Soyuz TM-7 spacecraft, where it could have been used to calculate the trajectory of landing in the event of an onboard computer failure." (citation from wiki). It have EEPROM and external connectors - one for ROM cartridges, and one unexplained in the manual. Also there's cheaper and slightly faster version with the same instruction set and registers but without EEPROM and slots - MK-61, it was bestseller in USSR, programs and games for it were printed in magazines and books. A lot of soviet kids started programming on MK-61.
“The People’s Pull Up Resistor” Hahahahaha. Got a chuckle outta that.
В чем проблема, мужики? В свое время мы такие штуки чинили с помощью лампового осциллографа, и какой то матери....все получалось.
@FindLiberty
3 жыл бұрын
Agree. A tube scope and some skills would also be enough to work it out.
Dammit, how is it that you don't have 100k subscribers?!
@crazystuffproduction
3 жыл бұрын
Hes pretty close!
Marc, are you going to look into what part of the power supply is causing the noise shown on the scope. I don't mean fix it, just show what's causing the noise.
@CuriousMarc
3 жыл бұрын
I should, but I am a bit tired of the clock right now. Plus working on the power supply in the live clock is very risky. I just wanted to make sure the oscillator worked as designed, then button the clock back up so it's off the bench and back to its beautiful self...
Fantastic as usual, thanks Marc :)
SUPER !!! Автор молодец! Хорошо разобрал причины сбоя в работе оригинального кварцевого генератора часов Союза. Если бы такие приборы были в 70-е когда разрабатывали конструкцию часов (цифровые осциллографы) - тогда бы все были гениями. Но делали с той элементной базой и с помощью тех приборов, которые были. Вопрос автору: Откуда столько свободного времени на изучение советской техники?
Great video and investigation 😀 The T feedback network is used to isolate the input impedance from the DC bias feedback. Without it the input impedance will be dependent on the closed loop gain of the stage, because a direct bias resistor will have its resistance reduced by the stage gain. I have a in-depth video about CMOS gates as linear amplifier, take a look! ☺️☺️
Hey Marc...The AGC series was terrific...well done and interesting to compare technologies. I caught a gimps of your Vector NW Analyzer as you worked in the Crystal Oscillator. I would really like to get hold of one....are they very pricey...How much should I expect to pay in a repairable/serviceable condition...?
@CuriousMarc
3 жыл бұрын
There is something stuck in my mind from when I saw my first HP Network Analyzer at the University. It was a 40 GHz HP 8510B VNA. On it was a sign in big letters for warning grad students. It said: "THIS INSTRUMENT IS WORTH MORE THAN YOUR LIFE". No kidding. They were in the 6 figures new, and you would consider yourself happy to get a much slower version in the low 4 figures figures now. Fortunately for amateurs, there is now a modern, budget-priced alternative: the NanoVNA. It's no HP, but for under $100 new, this thing goes up to 1.5 GHz, and can do all the basic measurements. Which is truly mind-boggling.
Marc thx for the insight to designing a dual inverter quartz oscillator and the tradeoffs to be decided. Assuming that the unit was operating properly when it was built I believe that a component around that switching PS has deteriorated. It's been a very long time since this unit was built and components don't always age gracefully. I'm with Ben Harvey on this one. Your thoughts. Again thx for sharing.
@CuriousMarc
Жыл бұрын
Yes, most likely something must have gone wrong in the power supply.
Excellent video ! What was the output level of VNA when you were checking the crystal ?
Кварцевый резонатор даёт погрешность порядка 10е-6...10е-7, разработчики, видимо, сильно не напрягались, ни термостатированного кварца, ни блокировочных конденсаторов на ТТЛ логике... Просто можно раз в сутки синхронизировать часы по внешнему источнику из ЦУПа.
Thank you!
I'd like to see what it would take to keep that noise from getting out of the power supply. Would a sheet of grounded & insulated alum foil between boards be enough? I appreciate not changing this original design, but it would still be interesting to see how difficult it would be to improve it.
It would be interesting to see the signal on an old scope like they would have used back then.
@paulstubbs7678
3 жыл бұрын
Even if they could see a little 'fur' on the signal, they would probably never have been able to capture it and display it as in this video. Boy have scope's come a long way.
@user-ec2jr6pv1s
3 жыл бұрын
Jitter is visible as good as in dso. Even better, cause of true luminophore
Who remembers the mnemonic for the phase difference between voltage and current in components with inductance or capacitance? "ELI" the "ICE" man, with L for inductance and C for capacitance.
@FindLiberty
3 жыл бұрын
Yep
The now very old clock just has bad caps, it's always bad caps 😁 I find it highly unlikely it acted like that brand new. Good video.
I ever read an application note from NS. inside that doc said about this kind of 2 inverters oscillator is "unstable" . what will happened is sometimes it may failed to start the oscillation.
What is that end tune? Great music!
So you've found a way to remedy the problem, and you've found out broadly where it emanates from - but I'd still love to know the exact cause of the problem and if it _could be_ fixed at its source (even if you won't actually do that to this historical artefact).
Where can I find the music that is played at the beginning?
'Unfortunately this isn't going to be an easy one' ... YES!!
BTW: the 7400/7401 are Quad 2-input NAND with totem-pole/O.C. outputs, respectively, while the 7402/7403 were quad 2-input NOR gates with totem-pole/O.C. outputs. Finally the 7404/7405 are hex inverters with totem-pole/O.C. outputs. So the even numbered ones had totem-pole outputs, while the odd numbered ones had open-collector outputs.
Awesome debugging! I probably, first, would have tried to eliminate the Power Supply switching flyback and/or shield that EMI at the source (agressor) switching power transistor(s)/transformers. If not successful in eliminating the EMI at the source, then I would have gone on to making the victim circuit less suseptable to it. The Crystal is in a steel package with tiny insulators around the leads, so I doubt 10MHz is getting into the crystal directly through those non-ferrous gaps, so, the coupling is from the wiring and then feeding back into the crystal output and being amplified by the first stage NAND amp. As a non-intrusive change to the Soviet clock, could thin steel plates (insulated with kapton tape) sandwiched on each side of the whole oscillator circuit work?
dream life, sold his company pursues his hobby vintage pc stuff
Can you fit a 470UF Electrolytic Capacitor in conjunction with a Ceramic Capacitor.
Is there a simple patch that could be implemented in the clock to improve its performance by even a little bit, that could still be considered in keeping with the original design esthetic?
Did you consider fixing the power supply? Since it stimulates the problem and potentially could cause additional problems with other electronics in this clock.
Hello! Interesting video! I don`t understand - WHY so many IC on the bord 1:04? There needs only something like CD4060 and a thermostat circuit (one op amp plus power BJT) for the crystal. It`s less relabilty and more weight..
The fact that the eastern blick scientists were able to reverse engineer so many wester chips is impressive to me the even more so when it comes to the cpu's easter german Z80 clone. Czechoslovak 8080 clone and so on.
It's very interesting where the source of this interference is. I think these are bursts in the power chain. Something like ringing in wires after switching in a pulse converter. Works like radiation of radio waves.
Would a high frequency capacitor tied from the crystal oscillator case to ground have alleviated the majority of the problem?
Hmm, how would the waveform look with an older Analog scope, we would know if they saw the glitch or did not bother.
I wonder how much improvement would be seen by only installing the bypass capacitor across the first NAND gate.
Very interesting! Every one of your videos shows me how much I still have to learn as an Electronics engineer :) Also: Wouldn't that also be problematic with other devices in the vicinity? Than amount of EMF would probably influence a lot of other devices around. Might be why they replaced it.
@CuriousMarc
3 жыл бұрын
When the clock is in its metal enclosure it would not radiate outside so it would be fine. Unfortunately the poor oscillator is inside...
Made me laugh, "here are your resistors comrade".