Help Repairing Tek T922 Analog Scope - Mysterious Triggering Problem

[Smauel]

I need a little bit of troubleshooting and repair help.

I recently acquired an old Tektronix T922 analog scope. It seemed to be in fine working condition (there was some external damage in shipping, which I was able to get a refund for, and then repair mostly).

However, when I made a second order active low pass filter to turn a square wave from a 556 timer circuit into a sine wave, the triggering started acting up, on that signal source especially, and even more strangely, the more the potentiometer for the filter is turned (adding more series resistance into the input of the filter, filtering more high frequency out), the more it acts strangely.

Firstly, the region in which I can turn the trigger level and still get a stable trigger seems narrower than it should be. Not only that, but when I turn the trigger level, it is clearly changing something about the triggering, but I don't think its the trigger level, as the voltage level at the very start of the display when shifted all the way to the right is not constant, but always changing rapidly. Additionally, pressing the slope switch does not seem to change anything.

Here are two short clips of the weirdness:
https://photos.app.goo.gl/u8sVKzDo2Eq8g7mn7
https://photos.app.goo.gl/F8mwiigAFQ8rjpoC8

Changing the trigger settings have the result that it either triggers "normally" (a stable trigger, but the level where it triggers at is very binary, at either the very top or the very bottom, no in-between when adjusting the trigger level), or it triggers twice roughly 180 degrees out of phase with itself making two signals that slide around, or it triggers automatically at preset time intervals, making a chaotic display.

I can't make much sense of what I'm seeing. One observation that is halfway a guess is that it seems to only trigger at or near the extrema of the signal, if at all, but I don't know what that means.

Please tell me if you have ever witnessed anything like this, so I can see about figuring out what is causing it to try and fix it.

Thanks in advance for any help you can offer!

[jdragoset]

Exercise (hundred times or so) Triggering source and mode switches.
Switch the source to "ext" and jump the vertical signal to the ext trigger input.
Test for any change in triggering response.
Trigger appears to be in "Lf reject" as is on many newer scopes, as in no DC continuity in the source switch contacts.
This same type of problem will show up on vertical input as a + or - spike instead of each input square wave transition.
Very old wiping switch contacts will oxidize but often (not always) clean up with simple exercise without adding lube such as Deoxit fader lube. 

[wn1fju]

For what it's worth, I had triggering problems on my similar T912 scope (auto/norm wasn't right and multiple traces).  I worked my way through the triggering schematic to the point where the trigger signal makes its way onto the horizontal board.  There are a couple of 7400 logic gates that process that signal - one of them was bad.  As soon as I replaced the offending chip, everything worked fine.

If I remember, it is possible to lift the vertical boards out of the way and probe the horizontal board while the scope is running.

[Smauel]

jdragoset, thanks for the suggestions.

Exercising the switches did not appear to help unfortunately.

Switching the source to external and plugging the vertical input into the external input also did not make any difference. However, switching to X-Y mode does work as expected. That was a helpful suggestion because that tells me that the switch is probably fine since the X-Y mode works, but there is definitely something wrong with the triggering circuit since the external trigger doesn't work properly either.

I assume by "Lf reject" you mean the AC-GND-DC coupling switch. It is in AC mode, but changing to DC mode does not make a difference to the triggering issue, as the signal I'm using is relatively high frequency, so it passes through anyway.


wn1fju, thanks for your input as well.

Glad to hear of someone with a similar issue. I'll take a look at the schematic and see if I can find some datasheets. Then maybe I can get some voltage measurements and such to test those 7400 chips and I guess the rest of the horizontal circuit while I'm at it.

That sounds like really promising advice. I'll update when I get a chance.

Thanks again!

[Smauel]

I took apart the scope a bit, and I've had a long look at the schematic and the voltage levels I'm seeing in the horizontal circuit. I am seeing some strange voltages, some of which are definitely intermittent, which was a bit of a surprise, and makes it more difficult to solve of course.

Unfortunately, the schematic is not clear about most of the voltages I should expect to see, but it states two voltages, both of which are intermittently incorrect. Also, when one is correct, the other is incorrect, but I think this is incidental. The first is -4V at a voltage divider, which is sometimes -7.3V instead, and the other is 2.5V at the emitter of a transistor Q2124, which is sometimes 5.3V instead.
Voltages on power supplies are all good.
From what I can tell, the 5.3V on the 2.5V part are simply a consequence of the -7.3V on the -4V voltage divider.

That voltage divider is between ground and -8V to create -4V at the base (pin 2) of a transistor U2126, part of a five transistor 3086 DIP array. However, the voltage divider sometimes reads -7.3V instead of -4V, meaning additional current is being drawn from R2127, the one with one lead to ground, causing the voltage to drop.

The base of the transistor is also decoupled to ground and to +8V. The collector of U2126 is connected directly to +8V as well, and the emitter is connected to the emitter of a second transistor on the 3086. Those emitters are connected by a 560 ohm resistor to -8V. Thus, the only path from the voltage divider other than the resistors is to +8V, ground, or through the base of the first transistor.

So the suspects have to be the resistors, the transistor, or some short. I poked and scraped at anything that might cause a short, but found nothing. I took the resistors out of circuit and measured them and they were fine, but they were old and crusty, so I replaced them while I was at it. So if the resistors are fine, the current has to be through the transistor. As far as I know, this means the transistor is drawing LOADS of current through its base, about 5 or 6mA by my calculations.

Obviously, that was an immediate red flag, but I took some voltage and diode drop measurements of the transistor in circuit, and they seemed okay. So, I desoldered it from the board and measured again and everything seemed nominal, as far as I could tell.

I don't have a component tester so I couldn't test the hFE or anything, but it still seems very bizarre to me that measurements of the transistor would be reasonable even though it is somehow passing 6mA through its base. I thought 6mA would destroy the base of a transistor. What would it take to destroy a BJT like that?

I assume that 6mA through the base will make it function improperly even if it reads okay out of circuit, so I will definitely replace the chip, but I would still like to know whether it could feasibly be okay for that to happen.

Am I missing something? Is that kind of current reasonable? Could the incorrect -7.3V at the -4V divider be because of something else affecting the emitter of the transistor?

In the meantime, I'll order that part and see if that fixes it, but I'd love to hear anyone's thoughts on this!

[wn1fju]

Don't know definitively whether the 3086 is the problem, but I've had a few of those 5-transistor arrays go bad over the years with somewhat strange faults.  I had one where there must have been some internal short to the substrate and that created havoc for everything connected to all 5 transistors.  I've had one where only one of the transistors tested bad when I pulled the chip out, yet several didn't work correctly when the chip was installed. 

I've also replaced these arrays with 5 discrete transistors (2N3904) soldered to an IC socket and then inserted into the board.  Seemed to work fine!  But I guess it depends on whether they are being used for simple switching or for higher frequency applications.

Good luck.

[Smauel]

Good to know they can tend to test okay even if they aren't working correctly. That is really helpful.

The transistor that I am suspect of is in the trigger amplifier prior to the comparator that creates the trigger signal.

It seems the 3086 is uncommon now, but I found one I can buy from Jameco, which apparently they manufacture, or else I think I can find a comparable replacement. Anyway, I've ordered the Jameco 3086.

Hopefully it works out. I'll update again when I try it.

Thanks for the help again!

[jdragoset]

https://www.ebay.com/itm/380614045665?epid=1483119556&hash=item589e5d6fe1:g:Kf4AAOSwwtdb7cie

[Smauel]

I haven't had much time to work on the issue with classes wrapping up for the semester, and finals coming up in the next couple weeks.

However, I did install the new 3086, and it did not solve the problem. After fiddling some more, I have reason to believe the voltages in the region I suspected were correct all along, even if I don't fully understand them. Basically, switching to a different mode and/or making certain the body of the horizontal section is grounded properly with the screws change those particular voltages. Still could be this section, but I have nothing in particular to suspect, so I want to check other areas before I try to replace every component I can see.

Looking through the circuit more at what could be causing the problem, my suspicion has shifted to two other sections, the 7400 wn1jfu mentioned, and the source of the trigger signal in the vertical section.

The 7400 seems to be working normally, though I'm not certain. The output of one of the NAND gates varies in an analog way depending on the input voltage. I assume this is normal, as the datasheet says the output voltage is not guaranteed when it is between a 0 and 1 value for the digital voltage, however, it seems as though one of the other NAND gates varies only digitally, given a similar input. I'm not sure if this is normal operation.
The schematic of this circuit is attached. Pin 6 of the 7400 NAND gate chip is the one that varies digitally, pin 3 varies analogically.

The other suspect circuit is the trigger input from the vertical board, which might also make sense if the trigger signal amplifier is acting nonlinearly. However, since the display looks correct, the problem has to be after that, and I checked everything there and found no issue. Apparently a 4.3V Zener is a common failure, but it seemed to be okay. All the transistors were biased properly too. None of the voltages seemed wrong.

I've also heard bad solder joints and connectors can be a culprit, but the issue isn't intermittent, its very consistent, just with weirdly specific symptoms mostly to do with the input signal itself. Also, the joints look okay, and the connector pins are clean of corrosion. That said, I may go in and clean and resolder the connector pins between the vertical and horizontal board anyway at some point.

Not sure where to go from here, and I probably won't have much time to work on it for a little while, but as always, if you have any idea about whether what I'm seeing is normal, please tell me.

Thanks!

[wn1fju]

I'm always a bit suspicious when TTL logic gates aren't at TTL levels.  But assuming that 7400 chip is OK, and the trigger signal at pin 6 is OK, the trigger gets processed in the horizontal/sweep section by two more gates, ICs 2212 and 2234.  On my scope, that's where the fault was (I don't remember which chip of the two was bad). 

Not saying that my problems equal your problems, but 7400 chips are pretty cheap and plentiful and you could always simply swap in some new ones.

[Smauel]

I measured again and I'm suddenly uncertain if I really saw non-TTL levels on the output of the logic gates, or if I was just measuring the wrong thing, because I realized I was messing up with reading pin numbers 7 through 14 backwards, 14 to 7 for a couple days. The voltages on the 7400s look good as of checking today, so I think they are all okay, but I will keep an eye on them, and next time I get parts, I will get five or ten 7400s with them.

I am now highly suspect of a 9602 chip, U2224, which is a retriggerable one shot chip, which from my googling, appears to be closely related to a monostable multivibrator. The 15uF electrolytic capacitor, C2226, attached to the circuit seems to definitely have the wrong voltages on it. I think that when the capacitor gets charged for an output pulse it happens too fast for my multimeter to react. (If only I had an oscilloscope to measure the capacitor with ).

A picture of the suspect is attached.

The voltages on the capacitor remain constant at 1.4V and 1.6V, but the 1.4V is on the positive side, and the 1.6V on the negative side of the capacitor. So the capacitor is reverse biased by the one shot, which seems almost certainly wrong, and I think I would expect the voltages either side to be both zero, both 5V, or one 0V and one 5V, as 5V is the VCC of the chip, and the voltage used to charge the capacitor.

Since there is 5V-1.4V = 3.6V across the 20k resistor, R2226, that I believe charges the capacitor, the chip is constantly sinking 180uA of current, (it can't be going through the capacitor, were it leaky, as that also just leads back into the chip, and also is at a higher voltage.

Can anyone confirm that a one shot/multivibrator sinking current like this is unusual, and also that reverse biasing the capacitor at seemingly arbitrary values like 1.4V->1.6V is not normal either?

I suppose I'll get a new 9602 in a couple days along with some more 7400s and a new capacitor, in case that one was damaged by the (small) reverse voltage.

[wn1fju]

I believe Vcc for the chip is supplied at pin 16.  Your measurement at pin 15 does not necessarily mean that the chip is drawing excessive current.  Stuff going on at pins 15 and 14 just set timing parameters and the data sheet seems to indicate the proper voltages are lurking around 1.5V which is what you are seeing.

My guess is that the 9602 is OK.  Can you borrow an oscilloscope for a few days?  It would be a simple matter to see if the one shot is occurring.  Trying to diagnose a fast waveform with a multimeter is not the way to go!  Plus, the service manual is full of pictures of waveforms at various test points.  Usually a good idea to check those.

[Smauel]

I'm not sure where on the datasheet you're seeing that those pins should be around 1.5V. I don't know that it is drawing excessive current, I just don't understand why pin 15 is continuously drawing some current, and I especially don't like that the electrolytic capacitor is reverse biased.

All of the waveforms that are listed to this point I have been able to inspect with just a meter, but many of the ones I am interested in are not listed in the service manual. The waveform for the 9602 is not listed.

As for borrowing a scope, I don't know anybody who owns one, and I'm not currently physically at the university I attend (Covid restrictions), so I can't borrow one in the labs there.

If you think the 9602 is probably okay, I won't get a new one yet. I'll look at more voltages and waveforms in that area of the circuit and see if I can figure out what's going on.

Thanks, as always!

[wn1fju]

You are right, I don't now see from the data sheet how I thought the pins were at 1.5V.

But I assume the capacitor is being charged/discharged fairly rapidly as part of the chip operation.  Could it be that the voltages on the pins are AC voltages, and you are using a DCV meter? If so, you are simply going to get some sort of average value.  So that could explain why you think the cap is reverse-biased.  I'm sort of grasping at straws at this point!  Once again, a scope is your best friend.

I've been fooled many times trying to measure DCV when I really had an ACV signal.  For instance, I had an open electrolytic filter cap in a power supply.  The DCV fooled me into thinking I had a relatively close to normal value at the capacitor.  It wasn't until I scoped it and saw 20 volts of 120 Hz ripple did I know I had a problem.

[Smauel]

Yeah I know what you mean. I'm definitely not getting the full picture with just the multimeter. I also feel like I'm grasping at straws with everything I suspect to be an issue, because nothing is visibly damaged, and all the suspicious voltages are unlabeled and just reasonable enough to make me doubt my sanity. I have discovered a fair bit more, but still nothing conclusive, despite the wall of text.

I measured the capacitor with the AC mode on my multimeter and it reads 0.000V, so I don't think I'm missing anything there, but who knows.

What I understand from reading the datasheets and circuit is that every time the square wave from the 7400 NAND gates that process the trigger signal goes high, a short pulse is produced by the 9602, which does shows up on my multimeter when I turn the trigger level knob.

I can see the pulse on the 1.4V (technically positive side) of the capacitor, which goes up to almost 1.6V before dropping straight back down when the trigger level is changed.
Similarly, when I read the value on the output, Q, pin 10, it is normally close to zero, and very briefly jumps up to a couple volts.

Fun fact: when I twist the trigger level knob faster, the pulses on the output and the capacitor are bigger and longer. Could this be related to the fact that square waves trigger mostly okay, but sine waves don't? Do the sharp edges of a square wave trigger the 9602 successfully, but sine waves can't quite get the speed to trigger it properly? Very suspicious, if you ask me.

The 9602 datasheet says "when input conditions for triggering are met, a new cycle starts and the external capacitor is allowed to rapidly discharge and then charge again."
I would have assumed this means that the capacitor should be charged most of the time, and then get discharged. However, I am seeing the capacitor charged 0.2V in reverse most of the time, and then discharge the low side upwards to match.
Is it normal for a device to operate a polarized capacitor in reverse like that? It seems to produce an output pulse okay (maybe), so it can't be completely broken, if at all, but it still seems extremely odd to me.

Notably, the second one shot in the chip acts slightly differently. The capacitor for the other one is not polarized, but it is also charged such that the pin not connected to the resistor is at 1.6V, however, the pin that is connected to the resistor changes between 1.5 and 1.7V depending on the position of the S/Div switch.

In conclusion, I have no idea whether the 9602 is broken, but its definitely weird, so I'll order a new one with my next batch of parts anyway.

I need to have a closer look at the circuits that come after it to see if there's anything strange there, I should have time for that next week when I'm done with finals.

Thanks again of course.

[wn1fju]

Yes, the 9602 behavior does seem a bit mysterious.  But it is easy enough to replace it.  Sometimes the quickest solution is the one that doesn't require a lot of circuit analysis!

I am not that surprised by your ACV measurement.  One has to be careful when making measurements of either very short pulses or things that are high in frequency.  They can be outside the bandwidth of your ACV meter and you will get an inaccurate measurement.

Hopefully you will get to the bottom of this.  Then you will have a working oscilloscope and will see why a scope is indispensable for repair work.

[Smauel]

I replaced the 9602 and all of the 7400 chips on the horizontal board. I also replaced a few more transistors which I thought may be an issue even though voltages seemed to fine. I measured various resistors and capacitors and all seemed to test fine. I also replaced a diode CR2227 which the manual mentioned may be related to a no trigger issue, which wasn't our issue, but worth a try.

Naturally, the issue persisted as ever.

At this point, I have ordered capacitors to replace all of the electrolytics, because I cannot think of anything else that might be the issue. There are probably only about three that I think might cause this issue, so hopefully one of them is the problem, or else I have no idea what to do to try to fix it next. Any suggestions are greatly appreciated, as always.

[wn1fju]

I wish I had some concrete advice at this point, but I don't unfortunately.

Sometimes you get lucky throwing replacement parts at a malfunctioning unit, and sometimes you don't.  It looks like you are in the latter category.

At the risk of repeating myself, you really need a (second) oscilloscope to hunt this problem down.  Then you could march through the signal flow and compare the waveforms to those in the service manual.  That would isolate the problem circuit very nicely. 

I don't know your financial situation or your interest in acquiring test equipment, but you could buy a cheap scope on eBay and after you use that to fix your present scope, sell one of them.  There are plenty of (advertised as working) scopes on eBay for less than $100.  Something in the Tektronix 2200-series or a generic analog scope like BK Precision or GW Instek would do nicely.

[Smauel]

Great news ! I fixed it!! 

Of course, it was one of the capacitors. I could have sworn I suspected this capacitor and tested it to be sure, but evidently not: When I pulled it out of the circuit it measured less than 1nF (and dropping), when it was supposed to be 100uF, 100000 times smaller than it should have been! I feel foolish for not realizing sooner, because I definitely remember suspecting it, but I don't remember what conclusion I came to.

Also, (with the benefit of hindsight), it makes perfect sense that this capacitor being low caused the issue. The capacitor was at the input of the trigger comparator, and the capacitor was extremely low value, so only really high frequencies were reaching the trigger comparator.
This explains why the issue was worse on sine waves (low slew rate) and better on square waves (high slew rate), but was still able to trigger on both (albeit with a much narrower degree of control, especially on sine waves).
It also perfectly explains why it had a habit of triggering on the wrong edge (rising/falling) depending on the position of the trigger level knob: (the high frequency square wave spikes would rise then fall immediately, or vice versa, causing it to trigger no matter the mode, simply depending on if the trigger level was below or above zero.)

As a result, the trace now looks far cleaner and sharper, because it is triggering much more consistently, and the trigger controls work perfectly.

Thank you for the suggestions along the way wn1fju, and now I finally have a way to troubleshoot my circuits for real

Also, the neon bulb power indicator circuit was broken, (a leaky capacitor, I believe), which caused it to blink incessantly, (it is only meant to blink when the line voltage is incorrect).
So I took the liberty of replacing the very dim, wimpy neon bulb with an LED circuit of my own design that utilizes (and maybe improves, according to my LTspice simulations,) the original blinking circuit.

The results of both repairs/mods are as shown. Arguably the LED is way too bright (it sort of blows out the camera, but its not too bad in person), but anyway, I'm partial to bright LEDs, and I can replace the resistor down the line if I am bothered by it.

[wn1fju]

Good for you!  A defective electrolytic cap wouldn't have been my first guess in a trigger circuit.  A defective electrolytic cap in a power supply, however, is so common that it is always my first guess in that situation. 

[Smauel]

Yes, the power supply caps were my first thought too, I replaced them immediately. I thought I remembered suspecting the electrolytic cap in the trigger circuit and ruling it out, but now thinking harder about it I can't remember what I actually did or if it was remotely thorough.

A good lesson, I suppose, to document every troubleshooting step to be sure I don't miss anything.

Thanks again 

[jdragoset]

"Trigger appears to be in "Lf reject" as is on many newer scopes"
The "LF reject" setting would pass the trigger signal through small capacitance so trigger would only occur with fast rise/fall time.The failing cap from 100 uf to 1 nf would do the same.
Great to solve the problem

[Smauel]

Ah, indeed you did point that out, I thought you just meant it was in AC mode.
Thanks !