EEVblog #565 - Tektronix TDS3054 Oscilloscope Repair - Part 2

[TiN]

27 code SOT-23-3 is likely ON Semi MMBD2104, common cathode dual diode.

I'd probe with 87V to see where those diode pads go, and check those curcuitry if it's differ from other channels

P.S. Good thermal cam, parallax still not as bad as on Fluke Ti32, they put webcam much more far on it 

[JoeO]

From the comments made here, some people are not watching the video all the way through?

[tru]

The main ASIC is too hot. 100°C is a high temperature. It might be dangerous to operate it without a cooling fan near. And I would expect a heatsink, too. Heatsinks are very common at most scopes today. Agilent DSOX2000, Rigol DS2000, GW Instek GDS-2000A...

It's probably too hot because it's outside its enclosure and not getting air blown over it by the fan.

I'm pretty sure he meant that it would be better if Tek had designed in heatsinks for those hot ICs.

I agree with JoeO, you should see video all the way through, and cross check your questions to see if already can be answered.  Near the end of the video you can see clearly that there is a green trace going from Pins 1 and 2 of the sot23, so is a designed short, Dave even mentions the trace later on.

@SeanB & TiN:
I don't think the problem is with the analog frontend because Dave probed the outputs of the differential pairs and you can see in the other scope the signal is clear (no funky HF crap).

[Chupacabras]

Man, that is thrilling...
Can't wait to see next parts

[SeanB]

I was talking about the ADC units, as it is likely that each ADC channel in the chips has a power rail set, and it is possible that the diode dying drew enough current to pop an internal board via or blow open a trace that powers part of the chip, crippling the one part. I saw the hybrids are fine. Thus taking the 2V5 rail, soldering a flylead onto the capacitor and doing a simple resistance check on the pins WRT this supply and on another working channel will likely show up a broken trace. Probably you will find each channel supply has a decoupling resistor on it. Simple enough to do, and not dangerous or tedious. This was brought by the chip diode having the negative supply on both sides, while the others have a positive supply on the one end. Losing a power rail probably did not affect the hot running digital logic of the chip, it could have been a supply to an ADC front end amplifier that went out, and this barely drew any power in the first place so made little difference thermal wise.

If this unit is fixed a set of glue on heatsinks on the bigger chips to cool them down would not go amiss though.

[xrunner]

A valiant effort in two videos, but it shows why our landfills or recycling centers are filling up with electronic waste - too much non (or not easily) repairable equipment.

[Co6aka]

...it shows why our landfills or recycling centers are filling up with electronic waste...

That's because in vampire capitalism everything is about the Benjamins; make it however for less, and so they'll break it, then sell 'em a new one; rinse and repeat. 

Anyway... I still think the "noise" changing polarity and being on only one side of the signal is the clue; looks like the ADC has "issues."

[Ronald1962]

Hi,

It is unbelivable what kind of equipmnet Dave is using!

The newest "toy" is the FLUKE E8 (not the E4 which is linked
in the text of the 2nd repair video).

This funny thing cost almost 6000 EURO...

Even I will never ever use these things, it is incredible interesting
to see them working...

Go ahead Dave!

Regards

Ronald

[AndyC_772]

The newest "toy" is the FLUKE E8 (not the E4 which is linked
in the text of the 2nd repair video).

I think it's quite noteworthy to see that having expensive diagnostic tools is no guarantee of a quick fix.

The thermal image didn't show up anything unexpected - but even if, for example, the channel 3 ADC were a different temperature to the others, it still doesn't give any clue as to why. Could be power, or clocking, or silicon damage, or some statistical effect to do with the bit pattern it's generating (for whatever reason).

[Frost]

This funny thing cost almost 6000 EURO...

Not almost, the E8 is more around 7000Euro 
http://www.reichelt.de/index.html?ACTION=3;ARTICLE=138176;SEARCH=FLIR%20E8

[lewis]

I'm sure there's a thread on here somewhere about someone turning an E8 into an E4 or something like that...

[xrunner]

I think it's quite noteworthy to see that having expensive diagnostic tools is no guarantee of a quick fix.

There's prolly some technician at the main repair facility that knows exactly what's wrong, he/she probably seen it a hundred times. "Oh yea that fuzzy display on a channel - yea that makes us a lot of money in repairs LOL".

If you could only find that person - but of course you can't get to them. I don't know about Tektronix, but for example Yaesu - an amateur radio manufacturer, sells all the parts for their radios down to the individual resistor and screw, I've ordered several parts for my FT-450 to bring it back up to snuff. The parts are even quite affordable. Does Tektronix even sell repair parts to individuals?

[TonP]

Put the 100% working one (you have a couple of these) next tho the broken one and go measure to find different voltage or signals.

[delmadord]

I'm sure there's a thread on here somewhere about someone turning an E8 into an E4 or something like that...

You think the one from Mike's electric stuff? Is that the same thermal camera?

[envisionelec]

I didn't see a cracked diode. I saw a blob of dried glue/flux on it.

But what do I know?

[snoopy]

At the very least a schematic or SM would be helpful.

Are these available ?

[victor]

I don't know if anyone suggested that, but I would try to guess some of the input and output pins on that supposed ADC chip, and then jump wire it to the other ADC's channel, and see if you can mirror that issue on the other channel.

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.

[xrunner]

OK shot-in-the-dark time.

I noticed in the 1st video that there was some, what looked like fibers, right there on the BNC input. If that isn't a fiber, but a conductive piece of whatever, it might cause a frizzy signal if it was causing a resistive weird path to ground.

Like I said - shot-in-the-dark. 

[Co6aka]

...it might cause a frizzy signal...

But not one that's "fuzzy" only either above or below the input waveform.

I think what we're looking at is a problem during or after the signal has been digitized. 

[Rasz]

its just a fur from Daves  my little pony costume not to mention signal was clear behind frontend.

I spend good 10 minutes of that video screaming "just lift it off already" at the monitor , that was after feeling all smug because I spotted pcb trace connecting two pins of a diode
Why not solder random diode in that spot just to check? or lift one from other channel? arghhh the anticipation is killing me

[TiN]

Does Tektronix even sell repair parts to individuals?

I bought some mechanical parts (casing covers, screws, knobs, handle) for my CSA7404 scope repair project, so Tek do sell parts.
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.

[Stonent]

In the first video, I was thinking Dave should attempt to swap hybrid modules to see if the problem followed it.

But after this one I'm wondering if the problem isn't even in the analog at all, maybe there's something between the ADC and the display?

[peter.mitchell]

it almost certainly isnt a analog issue and it seems like the adc is functioning correctly, i would expect something far worse if it were to be the case.

I'm leaning haaard on it being memory. In 1 off qty, avail stock, compatible chips, digikey, you're looking $10-11 per chip. If you've got a order coming from digikey, may as well throw some in anyway.

[bktemp]

If it is a purely digital bit issue I would expect a distorted sine wave at the MSB boundary and not only noise outside the waveform.
The SRAM ist much too slow for 5GS/s (3x 18bit 100MHz -> 600MByte/s), so there must be some other sample memory. Does somebody have information about how the scope works?
The scope is quite old and 5GS/s is much for such an old scope and the ADCs do not look as special as I would have expected from a 5GS/s 9bit adc for that time. So my guess is, the adc is actually running much slower and there is some sort of analog sample memory in front of it, slowing the signal down befor it goes into the adc.
Another guess: Could this scope use 2 interleaved adcs and one of it beeing completely faulty? That would explain the good shape if the waveform (coming from the working adc) plus the additional noise (every odd sample) and also the disappearance of the waveform at slow samplerates when it uses only the one faulty adc.

[peter.mitchell]

watch the section of the video with the scope in dots mode, it shows a bit more clearly what is happening, would be nice if dave used a ramp or a sawtooth wave.