EEVblog
ฝัง
- เผยแพร่เมื่อ 5 ต.ค. 2024
- Part 1 here: • EEVblog #564 - Tektron...
Dave follows a more methodical troubleshooting procedure in order to track down the fault in the Tektronix TDS3054 oscilloscope.
Measuring the low and high frequency differential outputs of the hybrid front end, the voltage rails and other parts. And has the magic smoke escaped?
And he couldn't resist playing with his new Flir E8 thermal imaging camera .
Forum: www.eevblog.com...
Memory Datasheet: www.datasheetar...
EEVblog Main Web Site:
www.eevblog.com
EEVblog Amazon Store:
astore.amazon.c...
Donations:
www.eevblog.com...
Projects:
www.eevblog.com...
Electronics Info Wiki:
www.eevblog.com...
Dave, you've saved my day! Your videos are definitely more educational and entertaining than anything on TV and they are actually out there for free! You're brilliant, man!
He's a Bobby Dazzler!
Absolutely!
*****
Urban Dictionary comes to the rescue:
www.urbandictionary.com/define.php?term=bobby%20dazzler
SysGhost oh god... number 4 on that list...
Techy FourOneNineEight Well... that's the flip side of Urban Dictionary. People just throw in their own versions, no matter how right or wrong it is.
Rule of thumb: If a suggestion has more "downs" than it has "ups", it's most likely an incorrect description.
If it was the 9th bit of the digital signal, wouldn't it be interesting to test what happens if you have an input amplitude of more than half of the maximum input of the scope, so that the 9th bit is actually supposed to be switching between 0 and 1, instead of being 0 all the time as with the signal you currently have.
That way you could identify whether it's a faulty connection somewhere in the digital processing.
As I can see by the other comments, I'm not the only one enjoying this like a tv-show. Although I have no EE skills it is fascinating to watch you narrow down the problem, thereby commenting every detail in an "edutainment" fashion. Thanks mate!
That thermal camera is freaking awesome !! USE IT !
Thermal image camera looks really fancy but in practical it's not that helpful. Maybe I am too stupid to use it properly. Anyway, for personal use I will invest my money for something else.
Well, it costs a lot and doesn't do that much. But if you get it for free like dave, i would use it a lot.
I know it's a bit risky and some work, but you could just swap two ADCs and see if the problem swaps to the other channel. If it is the ADC you'd have shot the trouble pretty fast and otherwise you could exclude the ADCs from the possible faults. So it's like a win-win situation.
The freezing sequence totally worth your efforts, so beautifull, amazing.
1. Schematic time. Lots of assumptions made which may send one down a sorry path.
2. As other pointed out, diode failed for good reason. Schematic will help to better follow.
3. I am not clear as to fault conditions. You went (for me) too fast. Unclear what input signal state and settings lends to the problem.
4. Try the old "Apollo" method. Works for me. Tap on pcb to gently flex at various points along chain and amongst supporting rails. When small solder blob floating in ABORT switch nearly scrubbed landing on Moon, a tap helped diagnose.
5. You were also a bit fast with the freeze spray. I noticed improvement at one of the first tries.
6. Might definitively rule out input module by (eek!) trace cutting and using nice coax to drive suspected broken channel with suspected working channel.
7. With all the high quality parts and (presumably) design, perhaps component failure more likely to happen when signal applied to channel way out of spec. Was scope supplied with all probes? Might check to see if one is bad. Leads back to input module.
Don't give up. I think we are kindred souls in this: there's great satisfaction in bringing something back from the dead.
The high freq noise on channel 3 could be caused by a dead cap along the 3rd ADC. That could explain why there is a offset when you spray on it.
I am so darn jealous!! I really love all your gear, wish I could afford even a fraction of it!! I can't believe FLIR just hands over a many thousand dollar thermal camera. I hope my channel does well enough to attract donations!
Your channel will fill too Frozen Electronics :-) But you really need that website!
It actually doesn't cost them much at factory cost. Stuff like this can be cheaper than normal advertising, and get better results.
EEVblog Why don't you swap the LCD out with the other good scope that you have. That will eliminate the display as the cause immediately.
I would put a heatink on that chip.
100C is not good for the transistors.
it normally has a fan blowing over it, hes got it open and the fans not on it during that flir thermal pic, he said that too
Hi Dave. the hybrid circuit does signal processing and a fault here would be random noise, so not hybrid module fault.
the errors appear above and below wave form full display not just in wave form.
This would suggest stuck data bit in video display module part of the ADC i.e. you get full scale "noise" regardless of wave form height.
I think the diode are just overload protection.
Best wishes for new year Barry
Hmm. could it be noise on the ADCs ref voltage?
Depends.
Whew! Must be nice to have someone GIVE you a US$5,995.00 FLIR E8!
Also, lifting the common lead on a diode pair that isn't working in the first place may not be the best way to determine if that defect was causing the problem. Might want to replace it anyway.
Dave, why don't you mount a new diode ? It's suppose to be there.
I can see from the PCB layout that the two diodes are in parrallel, so it's likely a double diode which runs more currents as a single one could take. Just mount a 1N4148 or similar and see if it makes any difference.
Big thumb up for you Dave and your work. Thanks for video.
i dont do electronics i have always wanted to but i am out of my depth. If i can see it blown burnt or popped i can repair it but trying to understand what you are doing is beyond me. BUT i find your repair vids completely interesting and i very much do enjoy the knowledge you have and how you put it across to people like me.
Thank you so much for sharing.
Remove the broken diode first, and check if the problem persists, the diode probably has build up a resistance now and couples two traces that shouldn't connect. Then check the trace on that diode that carries 3.3V on the working channels, find out why it has only 2.5V, or if that is ok again when diode is removed.
Hello Dave, in response to this your repair attempt I would suggest checking video strip contacts, that white strip runs across the LCD back and has two black peg press connectors. Now these can be tricky and have to be pretty hard pressed until they click. The one at the main PCB usually is OK ie. hard pressed in but the one at the other side maybe isn't at home position as yet. Trying to hard press it bends the display PCB so perhaps the tech didn't push it hard enough when assembling and so it's barely holding. Support the PCB and press the peg either by finger or a eraser pencil or similar -- a screwdriver would easily slip and damage nearby SMD elements. If loose, the connector should click home. I've got no display when I first got a TDS 3032 -- luckily this has been the case, obviously because things fly during transport and not only overseas. Now everything is fine. So perhaps this fault could be also due to poor contact there, just try it out. Best of luck, Edi
Im from a latin tong, if you speak a llittle slowly i can hear you cleary, great show is in a easy way to learn alot.
Great cat and mouse for a very difficult problem!
Absolutely fascinating as ever. Personally I prefer these fixes to mailbag.
I'd go with what a few of the chaps have already said. I've you've got a broken diode then disconnecting it isn't going to make its role on the circuit any more broken. Swap it with a known good one and see what happens.
I'm also starting to see similarities here with that other board you first IR'd and then chomp() the whole ASIC off a few months back, legs and all!
Hi Dave, just a note around measuring DC voltage rails, I personally also measure AC ripple to see if there is some noise at that level, that sometimes identifies issues with filtering... Just my 2 cents :-)
I have. Nothing wrong with the rails.
Dave, another suggestion - can you get at the raw / scaled sample points out of the scope to see if that reveals an interesting / telling pattern. When you zoom in in the first video, it looks like the superimposed noise is 'beating' against either the sample frequency, or the way it renders the points to the LCD.
You could swap diodes from another channel. Thanks Thumbs up.
I would try replacing the Diode, it is not uncommon for such a simple part to have a major role.
How would one calibrate something like that? I see no trimmers etc... I'm guessing laser trimming..?
Unique Dave Unique repairs and educational tutorials! Thank you!
You should replace that diode just for shiggles :P
As i said on the last vid, the artefact extends one way from the waveform and which way seems to depend on the waveform's vertical position. That's definitely not a high frequency noise issue and I'd say not an analog issue unless they're doing something really unusual. Coupled with the fact it must be before the point where the channels come together for processing it really narrows the possibilities down...
Hey Dave, How are you finding those Pamona test leads? I bought some after seeing them in your mailbag video (from Farnell/Element 14) and I have found them to have an issue where the resistance creeps up over time (to as much as 4-5 ohms when shorted together) and you have to flex and bend the bodies of the probes gently to get the resistance to go back down to zero. Pretty disappointing :(
No problem with mine, but I haven't used them much.
We bought few from digikey and they have similar issues. It seems all happen to the red probe, so I have to get my colleague's black probe to make it a workable pair.
Wei Li It's a shame as they are very nice precision probes (if not for the creeping resistance issue) It's interesting to hear I am not the only person having issues with them. I might see if i can get them swapped and see if a second pair is any different
After doing some google searching, it appears that these are also exactly the same as the Agilent 34133A and also the Fluke TL910
Just be happy with your new THREE channel scope.
Does the fault signal change with input amplitude?
mV, V, tens of Volts...
I suspect the diode fault is another symptom... I wonder if something took it out and took out something else (the actual fault) as well?
That is what I am thinking, someone blew the protection diode and damaged the ADC chip.
Hi Dave! For the thermal camera review, can you show how a cooling fan affects the temperature of components? It would be nice to see the color change happening.
No msx, "pretty boring" No way dude, for being that close thermal mode is exactly what you want, its so much better without the confusing overlay not lining up with the actual thermal data, what you care about. At farther distances, totally useful, not gonna argue with you there.
The thermal can't focus that close anyway either. This is not the right tool for close up macro work. You can play with the lens focus adjustment though if you dare.
I knew you would not give up.
Hi Dave, try to solder the '27' diode from the working channel to faulty one. Good luck!
That diode blew out for a reason. try following the traces and see what that gets ya. Try replacing the diode if it doesn't blow again it probably was overloaded by spike of some kind and other things could be broken
Can you short the high frequency output of the CH3 front-end? Also, the blown diode is definitely a hint...
Dave at 22:00 is just like Dr Bunsen Honeydew in Muppet labs, when he cuddles with a tenderized plate 😂
Did you ever repaired anything?
Pretty awesome reverse engineering there! Keep it up mate.
It's great to recap, I haven't seen this video for ages but I had to laugh when this time while watching you use another scope to fault find on the bad scope..... remember the film 'Silent Running' when the astronaut used one of the three robots to try to repair the injured robot which was knocked down when the other crew members were larking about in the 6 wheeled buggies? ( he could have done it in half the time if he had your Rigol)... lol
Long shot: 'noisy' attenuation relay contacts? Follow input signal from input jack with your scope up to the ASIC on the hybrid module Maybe even one of the rail to rail clamps in there might be introducing noise.
Scope the "diode" (or transistor, as it may be.) Since the shield is right over the part, clearly it is passing a signal through the shield. See what the signal looks like on it. I suspect that this may really be a transistor. Usually 2 diode arrays have 3 pins, as do transistors. Problem is getting a part for it. Number on the device is proprietary.
This video is the same number of Bach's famous pipe organ Toccata & Fugue in d minor BWV 565
Hey Dave, what's about changing the ADC chips of the channels? You can preheat the board in your (reflow) oven and take the ADC chip off board with your atten SMD Rework station. And then solder the chip of a working chanel to the faulty.
I like the look of that Flir E8, I imagine it costs a pretty penny, would be useful for laptop repair for tracing broken solders too
Heh, £5,500, yeah it's a fair amount
*****
The only difference between models in the FLIR E series is the software. The E4 (as an example, as it's the cheapest one) can be hacked to an E8.
Thank you , how easy is the hack though. Looking at prices the E4 is still a £1,000 device :)
*****
There is a description on the EEVBlog forum.
Thank you I'll have a look :)
I suspect those ADCs employ internal time interleaving between parallel paths. A failure in one interleaved path would show a failure characteristic similar to what is observed. Just a guess, but likely BER in any event.
Nice video Dave, I hope there will be part 3 soon :-).
Have you measured (and compared to other channels) the reference voltage pins on the front BNC connector? Maybe one has shorted the test points and blown the diode away?
I would also try to solder a new diode in.
Nice to see whats the steps are to fix this thing :) Thx
I know this is an old video: But I think it is faulty memory.. purely on the digital side.. does not matter how many chips there are.. I bet the memory is segmented per channel by the processor. Seeing different behavior at different sample rates is a giveaway. Somehow it is not triggering the memory check diagnosis. Replacing the memory chips may be worthwhile.
32:40 this thermal stuff was really cool though!
Always ending with a cliffhanger...
Dave, what if the sucker is U330 at 24:44 ?
It's an opto coupler as i see.
Have you checked it ?
Maybe e failed resistor somewhere ?
This is a real stab in the dark, but have you tried resetting it? It could be something as simple as a software issue/bug/feature which has gone wrong...
Its bad, that there is nothing about this ADC chip in internet, because I think that this diodes protects some signal line not only from external peaks, but from internal ones. I suggest that first of all is necessary to swap or get new diodes and check the signal. And I didn't memories - what was voltage differences between anode and cathode with diodes and without. And check the signal on diodes lines during work - that could be informative.
Yup seems like the ADC is the culprit and the sampling is not working so well. Maybe it got damaged by the faulty diode. I guess next try to connect the chip to another channel to be sure.
Find out what made the diode short. That will be a move in the right direction.
Fascinating stuff !
Dave, be very careful when you store those Pomona probes. They seem to be very poor quality. I've had three or four of those probes fail where the wire exits the probe body.
Hi Dave. So long time since solve problem. Did problem solved? I think may be some thing wrong by supply filter capacitor for ADC. Why not replace them. Thank you. You are great man. I love your LAB.
What are the four chips above the large white ribbon cable connector? Is there one per channel? Could one of those be involved here? Great video Dave! Thank you.
smd codes: www.marsport.org.uk/smd/mainframe.htm
I would try to swap a diode. removing a non functioning diode from the circuit making no change, doesn't mean it wouldn't fix it by replacing it with a working one. but as low the chances are it's done within one minute
Cool! There is something perverse about probing a scope with a scope...
If your A/D reference goes low, you'd expect the result to go high... toward either rail, depending.... but both rails? Look for problems in reference voltages!
1. What is the noise frequency? 2. Doesn't noise on the signal only follow the signal swing direction? 3. Could signal be coupled to, shorted to, and driving a faulty oscillating reference voltage? 4. Make a sniffer coil from a few turns of wire, and track the source/general location of the noise frequency. 5. Freeze/heat nearby components.
Did you ever find what was wrong with this oscilloscope?
17:56 and if you measure the resistance?
Is it possible to flex the pcb and watch the display? I find that a good way to locate dodgy solder joints.
Whats the diode feed into/out of ? Regardless whether the diode is cause or effect, there is a good chance it's neighbors are also feeling unwell. Does it go directly to a ADC pin?
21:44 well the magic smoke might have escaped all right!
Did you try to do a math function with channel 3 combined with another channel? At least that will rule out the display software for that channel
Also, on the trigger menu it seems you can turn HF Reject under Coupling. Not a fix but at least you could see if it's the HF (don't own one, but checked the manual)
I guess it's the ADC. Could be floating MSB data line, caused by fractured solder or semifried ADC. Maybe zoom into the screen and see if there's a pattern, like every, every second or every forth sample => one bad MSB line with 9, 18 or 36 data lines. Find out how many data lines it has(9,18 or 36, maybe +1 for parity?) and where they are. I guess 18 or 36. Maybe you can decrease sampling speed, so you can capture traffic with common logic analyzer. Or get high speed LA, like HP 16517A+16518A expansion card and caputre traffic. Other simple test would be to swap ADC of channel 3/4, since it's only QFP and easy to solder. But it's kinda pointless since you can't buy new ADC anywhere, so maybe better to call it a day.
Hi Dave. Is it possible that it actually is a problem in the front end and your Rigol is not quick enough to see the noise?
Is it normal operation in this case that the chips are running so hot?
good video shame you could not find it
I like those switching transformers.
Could the diodes be swapped from a working channel to the faulty channel?
Did find anything yet? I was wondering if in such a case you could get another (probably better specced scope) and compare ch3 to another channel with the same input on the whole path of all traces from frontend through adc and maybe spot some difference? would that be a viable thing to do?
could the through hole solder bridge thing that appears to connect the two shorted pins of that diode to the rest of the circuit be damaged simply re flowing it could eliminate the possibility
Hi Dave RayB. here, can you use your FLIR and check another Tek scope to see if the same chip is getting hot in the same place as the bad unit? Probably would be good to get a similar scope TDS3054. Also, try to replace the bad diode, or swap it with channel 4. Trying to get speciality parts from Tektronix is very hard, and costly. I know I use to work for them as a Technician on the Analog Scopes.Maybe you could get a used good main board from someone. Unless it is a easy fix. Later RayB.
Obviously, the diode is blown! You should replace it and see if that leads to an improvement. You have a couple of choices--you could remove it and try the circuit but that is a little chancy if it is a biasing diode. I'd remove and replace it. You could replace it with another diode from another channel. If it is an MMBD1204, then it is not likely a protection diode--that pin out configuration is just not very useful in a protection circuit (woudl rather see a MMBD1203). And, from your vid, it looks like the two anodes are in parallel. I certainly would measure a good one to verify that it realy is an MMBD1203 (you should be able to tell from the voltages).
It is pretty strange that it would be blown and I doubt that it has to do with signal overload. The diode is rated at 100V, 1A, 2A-pulse and it should be able to do that at 150C junction temp. It could be a biasing circuit, or a control line. Maybe it has somethign to do with switching on the preamp. My guess is that it is associated with the relay, switching, or a bias option on the IC. I don't think it is in the signal path.
I am really having trouble thinking what might be wrong with the circuit. Sure you have lots of noise, but the noise is not there on the signal going into the A/D chip, right? If so, then it is not likely the A/D. I can't imaging a diode like that in the signal path--that makes no sense at all! You should use your ohmeter to try to figure out where the diode is in the circuit. I know it is difficult (to impossible) with the multilayer board and buried beneath IC feedthroughs, but it just takes time. See if you can find the problem.
The hybrid circuit is the preamp circuit along with the attenuators. Look closely for damage. THe IC is the amplifier and all the long black gadgets are carbon resistors, laser trimmed in production. There are caps, and even laser trimmed (zapped) reactances. Very cool! A really nice design. But, if it is all working, and the signal looks good, then that is not the problem.
I really hope it is the diode, because otherwise your Australian Defense Force scope is going to be really expensive. You can't buy the A/D chip--they are custom and Tek will only sell the board. There are some places in the US that will replace them (Norway Labs) as they have schematics and they use parts from salvage units. However, it might not be worth it for the 3054.
i sware u did demo the flir in an oldr video
what is this magic smoke he is talking about?
Hi, something wrong with my tds3054b, after plug in power cable , it automaticly reboot again and again.i checked the power board,the output voltage seems not stability issue. do you have any experience on this problem?
Very nice thermal camera. However, considering the price of that "toy", I sincerely doubt that a review of it would be of much help. Let's face the fact, this is far from being in the common hobbyists budget. Personnaly I prefer reviews on things that I can see myself buying in the future but that Flir thingy is way past my buying plans. Personnal opinion, no more, no less. Keep up the good work buddy !
Why do they split the frequencies? Is it for different timebases, or do they recombine the bands (So to speak)?
nope, they recombine them
How long would it take you to remove two of those DACs, and solder them back? (put no.3 in no.2's place and viceversa)
***** You don't fix computers you just move around peripherals.
With the "protection" diode burned, it is likely that the ADC input was overloaded. Dave, just swap 2 ADCs. You have the tools and the skill.
Holy crap, did he say that FLIR gave that thermal camera to him for free? That's freaking 6 grand...
32:00 "Getting your oscilloscope high on CO2"
If it's the same stuff here in 'Murica, it's Tetrafluoroethane or R134a, the same stuff in the air conditioner in most cars.
What model flir camera is that?
Are differential pairs like balanced audio signals?
Yes. It's exactly the same thing.
Except it allows cranking the frequency way up.
Today on EEVblog: How to make a oscilloscope drunk
Holy shit, FLIR sent you a $6000 (USD) camera for FREE!?!?!?!??!
Dave says he's never that lucky?? He got a $6,000 IR camera for free, he's pretty damn lucky :/
Yup that's burnt.
Maybe fix the next one ;)
try the screen...
Mint!