Tektronix TDS3054 data acquisition failure

[(*steve*)]

Look what I've been asked to fix now:



There's a few teardowns on YouTube (Dave Jones tried to fix a fault on one of these).

Bugger all documentation, and only a few voltages C/- Dave, so I'm not hopeful.

Everything after the data acquisition seems ok (not sure if that includes the memory).

The acquisition chips seem not to get anywhere near as warm as Dave's did (even allowing for the fact that I have the fan on).

My initial thoughts were a broken solder joint, but after feeling the chips, a failure of one or more rails is also a possibility.

This unit has a main SMPS power supply with an auxillary switch mode regulator board (possibly for extra rails), at least three (and maybe 5) switch mode regulators on the main board, and a couple of linear regulators. All up, the power supply stuff is hugely (overly) complex.

And yes, there are a couple of tantalum caps on the board :-D

Here's a better look at what the screen shows.



The vertical red lines flicker on and off. The yellow "trace" grows with time, although it can be cleared using the various functions (like reducing persistence). If you turn on other channels you get similar rubbish from them.

Connecting a channel to a signal doesn't seem to result in any difference in the display. If you try to use the math functions to give you frequency, period, etc, it tries, but you get meaningless and rapidly changing values.

I might try prodding at it with a wooden stick.

 

[bushtech]

This might help:

 

[(*steve*)]

Yeah, I've seen that before. This unit is very new compared to that one. The front panel looks similar, but the box is probably only 1/5th the depth. The 3054 has the form factor of most modern digital scopes.

I've done some prodding with a wooden stick, and nothing much happened.

I've used a bit of freeze spray on the acquisition side of the board and the pattern of flickering of the vertical red lines changed a little (maybe), and a spray on the CPU side of the board didn't charge anything.

On that basis I'm going to discount for now that the problem is a fractured solder joint.

Incidentally, this unit came with a report from the service agent. The diagnosis reads:

"The signal acquisition hardware has failed on the main board assembly.

For various reasons including calibration issues the manufacturer's repair strategy is for complete main board assembly replacement. This instrument is no longer supported by Tektronix who have confirmed nil stock of the required assembly worldwide. Requests via other sources have so far also been unsuccessful."
​

Unfortunately, the report also states (in part):

• All DC supply voltages are within specification.
• All power on and extended internal diagnostics pass.
• The Signal Path Cal utility fails.

That's actually not too bad because they may have only checked the DC rails from the main power supply and not the DC rails generated on the main board. Since all channels are affected and there are four separate sets of acquisition hardware it stands to reason that a common failure exists. Furthermore this failure would have to be early in the signal path or the self tests would probably fail. I'm still hoping it's a power supply issue.

 

[bushtech]

Let's hope it's a component and not a chip

 

[(*steve*)]

I think the odds are pretty good.

As in the previous repair, the fix may turn out to be simple and cheap, but finding it takes quite a while.

The cost of time is generally what makes these things BER.

However, the lack of schematics can mean that it's doubly difficult to know where to look. It's bad enough with older equipment, but stuff like this with multi layer boards and tiny, unlabelled surface mount parts can make even the most trivially simple problem almost impossible to find.

 

[(*steve*)]

Another thing that strikes me is that it is entirely possible that even though the DC rails checked out as having the correct voltage, a failed capacitor could make one or more rails noisy enough to affect the operation of something critical in the signal acquisition path.

I'll check that too.

 

[kellys_eye]

The 'interference' has a periodic value - can you associate that with any clock or timing within the device? Does it change with range?

Strange(ish) how it's only on the red signal line too - post D-A conversion?

 

[(*steve*)]

Nothing I do (so far) in changing horizontal or vertical timebase settings, or zooming on a "captured" signal does anything that changes the red lines or even causes the patterns on the display to change how one might expect given the changes.

I have not seen the horizontal spacing or position of the red lines change.

Changing the vertical position does have an effect, and I've seen a partial waveform appear in a single shot capture.

Setting the inputs to "ground" does not affect the "signal" that is displayed, not in not confident that whatever I'm seeing actually comes from the inputs.

Oh, the red line is the "math" output. I might try changing the vertical scale of that later today.

 

[kellys_eye]

Nothing I do (so far) in changing horizontal or vertical timebase settings, or zooming on a "captured" signal does anything that changes the red lines or even causes the patterns on the display to change

...which, to me, points to the DISPLAY circuitry (it being a fixed period, raster scan type of thing - SVGA?)

Is there any way to get the video on an external port or access the video signal from some internal take-off point?

 

[(*steve*)]

If it was just the display, I would expect that the functions which show the frequency, period, rise time, etc would still give valid results.

But I get your point.

 

[kellys_eye]

I was intrigued by your first picture that shows a red 'spike' outside the trace area and indicative of a video issue.

 

[(*steve*)]

I think that red spike is a problem with the display or the connection to it. Although typically such defects go the whole way across the display, not just part way.

 

[kellys_eye]

If it was just the display, I would expect that the functions which show the frequency, period, rise time, etc would still give valid results.

I think you might be looking at more than one fault - or, more likely - a fault that links the same clock frequency of the display to the sampling circuitry.

 

[(*steve*)]

A bit of a go-slow on this one.

I measured all of the supply rails and none of them have any substantial ripple. I can't comment on them being the correct value, but they seem sensible (+5, +/-15, +3.3, Etc).

I haven't checked the outputs of the regulators on the main board.

Oh, and the power consumption is about half of what the unit proclaims to be the maximum.

 

[CsakNorris]

A bit of a go-slow on this one.

I measured all of the supply rails and none of them have any substantial ripple. I can't comment on them being the correct value, but they seem sensible (+5, +/-15, +3.3, Etc).

I haven't checked the outputs of the regulators on the main board.

Oh, and the power consumption is about half of what the unit proclaims to be the maximum.

Good moorning,
Do you were able to resolve that problem maybe?
I've got a TDS3012B and sometimes doesn't boot up correctly and then doesn't appear the traces on the screen. If then I run the checking function that shows acquition failure. And then turn off and on again the device then sturtup normally and working.
Any suggestion maybe?
Thank you for your answer.

 

[(*steve*)]

Do you were able to resolve that problem maybe?

Unfortunately not :-(