EEVblog #565 - Tektronix TDS3054 Oscilloscope Repair - Part 2

[deephaven]

Simple suggestion: Try pushing down on all the chips and run your finger nail around the pins of the chips. It could just be a dry joint somewhere.
 

[Bored@Work]

Dave, wouldn't this be something for a live show?

You take diagnosis suggestions live from the audience (or from this thread), the audience maybe quickly votes which suggestions to take first, and then you perform the suggested diagnosis. Of course, at the end of the live session you can junk the whole instrument. But hey, you can't have everything

[EEVblog]

Dave, wouldn't this be something for a live show?

You mean live debugging with hundred of nerds watching over my shoulder all screaming something different? Yeah, that sounds like fun 

[electronics man]

just a sugestion but the diode doesnt seem to be the problem but may it be the symptom of one of a problem whith one of the ics the diode could have just prevented the magic smoke from escaping from another device whitch may have got damaged anyway.

[Bored@Work]

Dave, wouldn't this be something for a live show?

You mean live debugging with hundred of nerds watching over my shoulder all screaming something different? Yeah, that sounds like fun 

You get the idea. How to Entertain a Nerd

[joseph.anand]

Back seat driving with electronics debugging. Fun indeed. BTW, do the ADC's have an internal reference/common external reference  or do they use individual voltage references (I'm too lazy to check through your videos).

Also why do we get service manuals with not schematics these days (http://exodus.poly.edu/~kurt/manuals/manuals/Tektronix/TEK%20TDS%203000%20series%20Service.pdf)

You mean live debugging with hundred of nerds watching over my shoulder all screaming something different? Yeah, that sounds like fun 

[xrunner]

But I expect not down to component level, I did not asked, but pretty sure that for electronics parts they can just sell a complete board assembly only, for some big $$$, as that's where the all magic smoke located.

Yep, and unless it's a broken BNC connector or wire or some standard blown cap or the like, we're back to the landfill or recycling center. It's just not economically worth it to fix complex electronics these days.

[snowl]

How does the Error on the sine wave look like? Has it the same Amplitude as with the square signal or do the peaks follow the sine wave curve? What is the Frequence of the Error spikes?

[vlad777]

I think it is a SDRAM problem, like when your (old PC) graphics card starts displaying gibberish.
 
There are four SDRAM chips (NEC  D4516821) and they may be in address space of Motorola XPC
but channel three may be statically addressed always in the same chip.

Or it is a chip per channel than also same error.

[electronics man]

How old is the scope?

[Simeon]

Hi everybody !
Dave, check for open resistors.
Those SMD fuckers like to go open circuit very easily ! 
My cheapass multimeter died because of such a sucker.
I have looked all over it, to find it is a reference resistor for the resistance measurement range.
I released the magic smoke when i accidently probed the 320V charged caps on a SMPS... 
So, check for open circuit resistors. 
I have posted in the youtube comments- what if is the optocoupler U330 at around 24:44 ?
Check all the IC's and opamps if there are any.

 

[Rufus]

I say again a more detailed study and intelligent analysis of what is seen on the screen would get you closer to locating the fault than we are now without even opening the case.

I initially suspected a digital intermittent hot bit fault but from the few more seconds of screen we see in the video it now looks more like noisy/intermittent gain. Just sweeping a dc level across the input would be telling and how hard is that?

Presumably all attenuation/gain is in the hybrid and the output from that looked OK so it is probably the ADC reference voltage which is noisy or intermittently collapsing or possibly one of the ADC supplies. If you are lucky it might have an external cause but the ADC is probably knackered. I would put the same signal on two channels and compare ADC signals between channels pin by pin. They are TQFPs so swapping ADC chips between channels isn't a terrible job either.

I guess Dave needs to decide if he wants a working 3 channel scope or is willing risk trashing it.

[Co6aka]

After thinking-while-sleeping on this funky problem... Here's something that has yet to be considered:

Maybe it's a symptom of NSA spyware!   (It was an ADF scope, so...)

[bitwelder]

My guess is that you have a most or least significant bit of the ADC floating. probably unrepairable if you can't find that part. At least it is a 3 CH usable scope.

Hmm... in that case would it be possible to simulate the oscilloscope acquisition chain e.g. in Matlab, then try to reproduce a flipping/noisy LSB on the ADC, and see what is the result on the 'simulated oscilloscope' ?

[RandomLogic]

I wonder, is the CH3 data just displayed incorrectly or is the data itself wonky. Is the purple noise also visible on possibly a screen capture or an USB data capture? Is the noise really random? If the noise is not random I would suspect the DRAM. Can one possibly activate the FFT on CH3 to check for possible anomalies in the frequency domain?

[vlad777]

Lets say every second memory location is corrupted (and it is a memory problem).
Then your "noise" signal should have upper and lower bound since 2^9=512.
So set vertical scale to most volts per division and see if the noise is bounded.

[owiecc]

Do a single shot, dump the data and see if there is a pattern.

[Stonent]

How old is the scope?

The chip dates that I saw were late 90s.

[electronics man]

Mabe dave should save a waveform to a floppy and open it to see if the problem is still there if not then it would be a problem whith the display controller might that be the problem?

[calexanian]

Mabe dave should save a waveform to a floppy and open it to see if the problem is still there if not then it would be a problem whith the display controller might that be the problem?

I thought about that as well, but it is unlikely the controller would only affect one channel. It would most likely effect all In the same manner. That virtual phosphorus algorithm is nagging at me though. 

[electronics man]

i still think he should try that just to be 100% sure

[vaualbus]

Despite all that it has been said on the blog I still think that the problem is either logic or ralated to nvram corruption (try to run a couple of time SPC and restart the scope. I manged to restore option 1M on my TDS540 using this technique), I will exclude the voltage reference noise, repetitive noise, burn op amp, becuase in this cases the self tests shouldn't pass but it passes.

Also I take this opportunity to share a photo of my attunuation hybrid on my TDS540, they are completly made by ceramic, on the other side there is a similar hybrid ic and a couple of relays and some capicito and resistorsr.
The interst thing to notice is the connection of the hybrid. There is a little pieces of gold in the plastic shell that than make contact with the attenuator pcb.
This hybrids are very fragile, I've broken one when I remount the attenuation pcb board in place, than another one in the same way.
Fortuanlly I was able to repair the second hybrid!
Only god know hoae much those hybrids cost when they were in prduction. Any ideas on the original price?
And why they use this type of connections and not use direct pins?
The TDS540 was just only a 500Mhz scope!

[burra7]

Home with with flue so I tried to make a block schema. I think we have some voltage problem. U300 (ADC3) does not get correct voltages. Or some bypass has gone bad. The C929 tantalium might gone bad when it is so close to the hot chips.

But have a look yourselfs and improve on schema
https://www.dropbox.com/s/i4kxi8sqwzslniu/tek%203054%20blockschema%20r1.odg
https://www.dropbox.com/s/vylmxa4nonqpr1h/tek%203054%20blockschema%20r1.pdf

some images
https://www.dropbox.com/s/wpwertu1x05t71d/Capture3.PNG
https://www.dropbox.com/s/027hj2wut8xjyoa/Capture4.PNG
https://www.dropbox.com/s/5lzzpv4hle8aqzn/Capture5.PNG

[ivanjh]

I'm sure Dave's driving to Chester Hill on Monday to pickup a MMBD1204. 

[burra7]

Looking really hard at time 23.10 in the video one can see more components.

U320 and U330
I can't see what the values are. These might be in the signal path.

https://www.dropbox.com/s/netuerv42t9vojs/tek%203054%20blockschema%20r2.odg
https://www.dropbox.com/s/9bwbkhca8j2p5z3/tek%203054%20blockschema%20r2.pdf

Edit:
looking some more video. One can see that U320 and U330 are not in the signal path. But one can make out the markings on them

U320 AAAG 5pin
U330 A00A 5pin

Then I also noticed that there is actually a some differance between channel3 and the others. One can clearly make out a Ferrite beed time 28.11 that is only on channel 3 and none of the other channels. Might be a problem somewhere else that only shows on channel3.