Topics

Tek 7854 intermittent issues


Nick Corvid
 

I recently got a Tektronix 7854 in seemingly great condition from a surplus lot. On it's first power on it functioned absolutely perfect, but now in following uses it has started to be plagued with strange intermittent issues (No trace, no saw-tooth, no intensity control, Cal signal wildly out of spec, ect.) but all of these issues generally subside after a short while making it difficult to nail down the cause. Currently the issue that is causing the most issues is the fact that the digital acquisitions are no longer working properly when only a few days ago they were. Now if I do a single shot acquisition it produces an error and no saved waveform, meanwhile if I get an averaged acquisition it will still error out but save a mangled waveform. All this is while the analog side seems to be just fine (at the moment)

Here is a photo of the same waveform (produced by the cal. out) , both live and average 10 digital.

Analog : https://groups.io/g/TekScopes/photo/252782/0?p=Created,,,20,2,0,0
Digital : https://groups.io/g/TekScopes/photo/252782/1?p=Created,,,20,2,0,0

Does anyone have any suggestions on what is wrong? I am happy to solder in new parts if they can be had. (please don't say the tek IC's, please I beg of you)


Dave Casey
 

Check the power supplies for level and ripple. It looks like you have a
mostly digital problem, and the digital side is more susceptible to noisy
supplies than the analog.

Dave Casey

On Fri, Aug 28, 2020, 5:06 PM Nick Corvid <awsomedrack@...> wrote:

I recently got a Tektronix 7854 in seemingly great condition from a
surplus lot. On it's first power on it functioned absolutely perfect, but
now in following uses it has started to be plagued with strange
intermittent issues (No trace, no saw-tooth, no intensity control, Cal
signal wildly out of spec, ect.) but all of these issues generally subside
after a short while making it difficult to nail down the cause. Currently
the issue that is causing the most issues is the fact that the digital
acquisitions are no longer working properly when only a few days ago they
were. Now if I do a single shot acquisition it produces an error and no
saved waveform, meanwhile if I get an averaged acquisition it will still
error out but save a mangled waveform. All this is while the analog side
seems to be just fine (at the moment)

Here is a photo of the same waveform (produced by the cal. out) , both
live and average 10 digital.

Analog :
https://groups.io/g/TekScopes/photo/252782/0?p=Created,,,20,2,0,0
Digital :
https://groups.io/g/TekScopes/photo/252782/1?p=Created,,,20,2,0,0

Does anyone have any suggestions on what is wrong? I am happy to solder in
new parts if they can be had. (please don't say the tek IC's, please I beg
of you)




Albert Otten
 

Hi Nick,

The error is probably indicating that insufficient horizontal waveform positions were filled during acquisition. Do you also have the calculator keyboard? If so, you can set the display in "dots" mode in stead of the default "vector" mode. I guess you will see some very large horizontal gaps in the dot pattern; the slowly up going line in your second photo is merely an interpolation line (vector) between 2 dots at the end points of that line. Similar for the large horizontal line. It could, as suggested by Dave, be a power supply problem.
Are the other issues only present in digital mode or also in analog mode?

Albert


Nick Corvid
 

Thank you for the suggestion Dave, I'll have a look at where the power supply points are located in the manual and see if anything seems out of spec when I'm next able. And in response to Albert, you are correct that it's missing chunks of the horizontal in dots mode, or were until I had set up my camera to take pictures and the behavior changed again.(this thing is making me crazy!) Leading to these pictures:
Vector mode: https://groups.io/g/TekScopes/photo/252782/1?p=Created,,,20,2,0,0
Dots mode: https://groups.io/g/TekScopes/photo/252782/0?p=Created,,,20,2,0,0

Some of the issues can be found in analog modes, but usually only very intermittently, for example the trace jittering horizontally even with a solid trigger (and the intensity for the A time-base but I'm pretty sure that is just a dirty pot). Also this may be unrelated as I don't know exactly what operating as intended is but one of the transistors on what I am fairly certain is the horizontal amp board can get very hot to the touch while the others stay cool when the machine has been running for over 10 minutes.
This is the transistor in question: https://groups.io/g/TekScopes/photo/252782/2?p=Created,,,20,2,0,0 . Haven't had the time to find where it lives on the circuit diagram


Dave Casey
 

Q141, sheet <11> coordinates E2. The other transistor in the clamp is Q41,
coordinates E5.

Dave Casey

On Sun, Aug 30, 2020 at 6:12 PM Nick Corvid <awsomedrack@...> wrote:

Thank you for the suggestion Dave, I'll have a look at where the power
supply points are located in the manual and see if anything seems out of
spec when I'm next able. And in response to Albert, you are correct that
it's missing chunks of the horizontal in dots mode, or were until I had set
up my camera to take pictures and the behavior changed again.(this thing is
making me crazy!) Leading to these pictures:
Vector mode:
https://groups.io/g/TekScopes/photo/252782/1?p=Created,,,20,2,0,0
Dots mode:
https://groups.io/g/TekScopes/photo/252782/0?p=Created,,,20,2,0,0

Some of the issues can be found in analog modes, but usually only very
intermittently, for example the trace jittering horizontally even with a
solid trigger (and the intensity for the A time-base but I'm pretty sure
that is just a dirty pot). Also this may be unrelated as I don't know
exactly what operating as intended is but one of the transistors on what I
am fairly certain is the horizontal amp board can get very hot to the touch
while the others stay cool when the machine has been running for over 10
minutes.
This is the transistor in question:
https://groups.io/g/TekScopes/photo/252782/2?p=Created,,,20,2,0,0 .
Haven't had the time to find where it lives on the circuit diagram




Albert Otten
 

On Mon, Aug 31, 2020 at 01:12 AM, Nick Corvid wrote:

And in response to Albert, you are correct that it's missing
chunks of the horizontal in dots mode, or were until I had set up my camera to
take pictures and the behavior changed again.(this thing is making me crazy!)
Leading to these pictures:
Vector mode: https://groups.io/g/TekScopes/photo/252782/1?p=Created,,,20,2,0,0
Dots mode: https://groups.io/g/TekScopes/photo/252782/0?p=Created,,,20,2,0,0
Hi Nick, the dot pattern is not what I expected. It seems that still linear interpolation is done to fill missing horizontal positions. I expected to see no dots at all during some wide horizontal gaps. Anyway, in my opinion it's clear that data are missing. In your new photos there is also a negative outlier and, very strange, a negative trace near the end of the sweep which looks like a inverted portion of the correct waveform. In the correct waveform the rising edges end more or less rounded while the falling edges are sharp. Wrong sign bit?

I thought I would see horizontal gaps when I do a too fast single sweep acquisition. But my 7854 simply refuses AQS with a nasty beep.
Albert


Nick Corvid
 

No pictures this time but, (YET AGAIN) it's changed behavior, now its getting AVG acquisitions (mostly) fine (As in not erroring out but still occasionally having some weird defects to the wave but now everything is the right size and, mostly intact). Still refuses to do any AQS's at all. I did notice though that during an AVG acquisition (when the issue was active) if I already have it in digital mode. I could actually see it missing chunks as it gets data and then just filling it in with weird stuff at the end, making those slopes and spikes (badly trying to interpolate the missing sections of the wave?) *see first video for alright example.


Never mind, in the middle of writing this I turned on the scope to check ( sat for a few hours powered off ) and it was doing it again, but I've discovered something very strange that may have solved (at least a small part of and temporarily) the mystery of this thing when I tapped on the transistors in the horizontal section, suddenly it worked just fine! This thing is weird, doesn't make much sense to me as they aren't even the metal potted ones but the ones in the heat sink clamps. Luckily I took some video this time, first before the miraculous (and likely temporary) fix and again when it started making proper acquisitions. Still won't do AQS under any settings though.

Broken video: https://www.youtube.com/watch?v=QYEH3q-4Dmc
Working video: https://www.youtube.com/watch?v=vm189Z5WirA (flashing on analog side is from a dirty pot I'm fairly sure)

Also Dave, you are definitely right about which transistor that is, thank you for finding it for me. According to my service manual A17Q141 to be exact . Although Q141 (and it's paired Q41) are PNPs ( tek part no.151-0220-00 ) but I can't really find any other specs than some incomplete ones I scraped together, having some difficulty locating an equivalent replacement if case I need it, but I may just be bad at looking (don't know if you can fudge the numbers a bit like some caps)

Tomorrow, if I have time after work, I'll finally check the power supply and see if it's correct or not. Sorry if this was a little rambly, I should have called it a night hours ago.


Albert Otten
 

Still refuses to do any AQS's at all.
You probably tried AQS because I pointed to that. Something very strange is going on with AQS in my 7854 (or with my brains?) AQS worked at 0.5 ms/div but not at 0.2 ms/div and faster. OK. Later on I thought to remember a trick to mislead the scope: set trigger level for not triggering, then do AQS (the scope waits...), then switch to faster sweep rate, then turn trigger level produce a sweep. Somehow it didn't work, error all the time. I switched off the scope and read the operators guide. AQS: works ONLY with 7B87! But I used a 7B80 (no 7B87 at all in the scope). And AQS just worked fine at 0.5 ms/div.
Back to the 7854 to do it once more. And this time AQS gave an error beep all the time!??
I have a 7B87 but didn't succeed to AQS a waveform with gaps. The manual says that "empty" horizontal locations are filled with the last valid data at the left side, so not by interpolation between two valid data points.
I did notice though
that during an AVG acquisition (when the issue was active) if I already have
it in digital mode. I could actually see it missing chunks as it gets data and
then just filling it in with weird stuff at the end, making those slopes and
spikes (badly trying to interpolate the missing sections of the wave?) *see
first video for alright example.
[this one: Broken video: https://www.youtube.com/watch?v=QYEH3q-4Dmc ]
Very nice video. Indeed seems to show interpolation afterwards.
.... when I tapped on the transistors in the
horizontal section, suddenly it worked just fine!
You might "jiggle" circuit boards and connectors as well, who knows...


Albert Otten
 

Hi Nick,
It doesn't help you any further, but in a rather tricky way I managed to create a horizontal gap in AVG waveform acquisition. And indeed a (dotted) linear interpolation line between the dots at the ends of the gap was shown. Since the vertical digitizing in your video seemed to miss or add a sign bit during part of the trace it seems that both vertical and horizontal digitizing suffer from DAC errors.
Albert


Paul
 

Hi Albert -
On Tue, Sep 1, 2020 at 03:02 AM, Albert Otten wrote:
You probably tried AQS because I pointed to that. Something very strange is
going on with AQS in my 7854 (or with my brains?) AQS worked at 0.5 ms/div but
not at 0.2 ms/div and faster. OK. Later on I thought to remember a trick to
mislead the scope: set trigger level for not triggering, then do AQS (the
scope waits...), then switch to faster sweep rate, then turn trigger level
produce a sweep. Somehow it didn't work, error all the time. I switched off
the scope and read the operators guide. AQS: works ONLY with 7B87! But I used
a 7B80 (no 7B87 at all in the scope). And AQS just worked fine at 0.5 ms/div.
Back to the 7854 to do it once more. And this time AQS gave an error beep all
the time!??
Are you doing single-sweep Acquisitions with the 7B87? Single-sweep uses the 7B87's internal sample clock, and I just checked on mine and confirmed that at .5mS, the acquisition works fine, and beyond that, it's clocking too fast to fill up the memory. If I do a single-sweep acquisition at .2mS, I'll see the wave, but I have to hit stop for it to complete, and it beeps with an error.
I've got a seemingly working (knock on wood!) 7854 w/ a 7B85 & 7B87, so y'all let me know if you need assistance checking measurements or behavior.

cheers,
Paul


Albert Otten
 

Hi Paul,

I think the "official" method is: set the 7B87 for single sweep [and wait until the Ready light is off], then press AQS at the 7854, then activate the 7B87 again by pressing the single sweep Reset button.
When I do so at 1024 P/W, then 0.5 ms/div works fine but 0.2 ms/div immediately gives a beep and error, even before I press the Reset button. No busy light and of course no need to stop acquisition. The previously acquired waveform is still displayed.
When follow the procedure at 0.5 ms/div, but just before pressing Reset change to 0.2 ms/div, then a waveform is acquired at 0.2 ms/div without error. But the waveform is far from perfect, with dots scattered around it and even with obvious outliers (as in one of Nick's photos).

According to my observations these results can also be obtained with the 7B87 in Norm mode, AQS at scope, and playing with trigger level to generate a single sweep. But this is a far less comfortable way since the trigger event is not well defined and switching from 0.5 ms/div to 0.2 ms/div [and faster] often produces a "false" trigger event.

Perhaps you can help Nick with his perhaps too hot transistor in the vertical amplifier. My 7854 has an EMI shielding option; the panels are difficult to remove and to install properly, so I had rather not do that.

Albert

On Wed, Sep 2, 2020 at 04:12 PM, Paul wrote:


Hi Albert -
On Tue, Sep 1, 2020 at 03:02 AM, Albert Otten wrote:
You probably tried AQS because I pointed to that. Something very strange is
going on with AQS in my 7854 (or with my brains?) AQS worked at 0.5 ms/div
but
not at 0.2 ms/div and faster. OK. Later on I thought to remember a trick to
mislead the scope: set trigger level for not triggering, then do AQS (the
scope waits...), then switch to faster sweep rate, then turn trigger level
produce a sweep. Somehow it didn't work, error all the time. I switched off
the scope and read the operators guide. AQS: works ONLY with 7B87! But I
used
a 7B80 (no 7B87 at all in the scope). And AQS just worked fine at 0.5
ms/div.
Back to the 7854 to do it once more. And this time AQS gave an error beep
all
the time!??
Are you doing single-sweep Acquisitions with the 7B87? Single-sweep uses the
7B87's internal sample clock, and I just checked on mine and confirmed that at
.5mS, the acquisition works fine, and beyond that, it's clocking too fast to
fill up the memory. If I do a single-sweep acquisition at .2mS, I'll see the
wave, but I have to hit stop for it to complete, and it beeps with an error.
I've got a seemingly working (knock on wood!) 7854 w/ a 7B85 & 7B87, so y'all
let me know if you need assistance checking measurements or behavior.

cheers,
Paul


Paul
 

Albert -
Yep, those are the steps I'm following. Oddly though, my 0.5ms/div works as reported at 512 P/W, and fails gracefully as described at 0.2ms/div, but at 1024 P/W, it works at 2ms/div & fails at 1ms/div. I never get the immediate error. I've got a 'newer' model w/ combined RAM/ROM board, you?

Nick -
Happy to poke around & take readings for ya - let me know if I can be of help.


Nick Corvid
 

It seems like my issue might be thermal related. When it cold starts (everything is at room temperature) it almost always is missing parts of the digital waveform, but if I set it to AQR (and since parts are missing it won't complete) and watch it in stored mode, after a few minutes the missing parts start to fill with noise then the correct data starts to fill in a short while after that, once that has occurred, it seems (most of the time) that the digital side will capture properly until I turn it off and it cools down for a while again. When I'm able I'd like to capture a video of this in action to better describe what's happening.

Also I genuinely didn't know you were supposed to use a specific plugin for AQS until just now, when I very first set up this machine to test it, it would AQS for anything, in auto trigger mode, using just a 7B85 in the left horizontal slot, at any time per div I selected. When this stopped I thought there was an issue.

Finally have the power supply slid out as well so I can finally make some measurements to see if I'm getting some noise or something that's messing with everything until it's "warmed up". Though the fact that the issue disappears randomly makes me not very confident in any correct looking readings.

Still have some occasional other issues as well, like the cal out reading (both on the scope and on another scope) around 2.6 times what it should, but only for a few minutes and SUPER intermittently (like I might not see it for 20 or 30 power cycles or more)

This thing causes me great confusion.


Nick Corvid
 

Alright, sorry for the quality I had it on the bench rather than in a rack so my angle is quite bad (looking down from the top) but, I have a video of the behavior!

Video: https://www.youtube.com/watch?v=o7OC9Qm7inE

If you have time watch the whole thing, it shows some strange artifacts as it goes, if not have some time stamps of the more interesting things.

Immediately after power up: 00:00

Switch to stored: 00:05

Start of first AQR: 00:08

Things start happening: 02:11

End of first AQR: 02:25

Final result after 3 more AQR's: 02:32


Albert Otten
 

Hi Nick,
My 7854 is an earlier version. Strange, if that makes any difference I would expect the newer version to have a faster sampling rate.
Albert

On Thu, Sep 3, 2020 at 04:48 AM, Paul wrote:

Albert -
Yep, those are the steps I'm following. Oddly though, my 0.5ms/div works as
reported at 512 P/W, and fails gracefully as described at 0.2ms/div, but at
1024 P/W, it works at 2ms/div & fails at 1ms/div. I never get the immediate
error. I've got a 'newer' model w/ combined RAM/ROM board, you?

Nick -
Happy to poke around & take readings for ya - let me know if I can be of help.


Albert Otten
 

Ah, I meant Hi Paul (not Nick) in my previous message. Albert.


Albert Otten
 

Hi Nick,

I watched your video https://www.youtube.com/watch?v=o7OC9Qm7inE and again some earlier videos and photos.
As far as I can see your calibrator is OK in analog mode? Photo _MG_2320 in https://groups.io/g/TekScopes/album?id=252782 shows 40 mV amplitude in analog mode.
But in AVG photo _MG_2318 the amplitude is 20 mV; also mainly 20 mV in _MG_2323 except for an "outlier" and the right most part.
You didn't show where the zero level is. My guess is that the zero input values are digitized to 20 mV in stead of 0 mV. Except for the "outlier" and right most part where correctly 0 mV is displayed.
If you repeat things with the calibrator at 0.4 V or 4 V and the scope at 0.1 mV/div or 1 V/div you might see other anomalies.
Could be a digitizer problem but also RAM problem I think.

Albert

On Thu, Sep 3, 2020 at 09:55 AM, Nick Corvid wrote:


Alright, sorry for the quality I had it on the bench rather than in a rack so
my angle is quite bad (looking down from the top) but, I have a video of the
behavior!

Video: https://www.youtube.com/watch?v=o7OC9Qm7inE

If you have time watch the whole thing, it shows some strange artifacts as it
goes, if not have some time stamps of the more interesting things.

Immediately after power up: 00:00

Switch to stored: 00:05

Start of first AQR: 00:08

Things start happening: 02:11

End of first AQR: 02:25

Final result after 3 more AQR's: 02:32


 

On Thu, Sep 3, 2020 at 04:43 PM, Albert Otten wrote:


Could be a digitizer problem but also RAM problem I think.
I've only superficially followed parts of this thread and now watched the video.
It's normal that during acquisition, the readout characters are being garbled in a way that looks random, we know that. The video however seems to show a repetitive pattern with a repetition frequency of about 3 Hz for a rather long time, the waveform display is updated, then proceeds to "stop" (sort of), next "busy" is displayed and the procedure seems to continue a few times until it finally comes to a stop, showing the correct waveform. This all seems to happen without human intervention.
To me, that doesn't look like a data (RAM, digitizer) problem but rather a "randomizer/stop/ready" criteria problem. I have no idea about the internals of that in the 7854. "Randomizer" because of the random sample acquisition in the 7854.

Raymond


 

On Thu, Sep 3, 2020 at 05:32 PM, Raymond Domp Frank wrote:


This all seems to happen without human intervention.
Ouch, it happened again! Nick tells us about his intervention so that part of my answer falls through (RTFM).
The repetitiveness in the random garbling remains though (I guess...)

Raymond


 

On Thu, Sep 3, 2020 at 08:33 AM, Nick Corvid wrote:


and since parts are missing it won't complete
How does the 'scope decide AQR is complete, if samples are taken randomly, out of order? Or are they not taken randomly, just at a slow pace, just at subsequent scans? I guess it's RTFM-time again for me.

Raymond