Tempst video, left third of screen 'keystoned'

cwilbar

Well-known member
Joined
Sep 8, 2015
Messages
1,209
Reaction score
31
Location
Massachusetts
So, I have this Tempst PCB set I'm working on. At least one pokey bad, one of the TL082 for the X output was bad, resistor for y size snapped in half. I've replaced the MC1495Ls for XCOR and the Y output, but I don't think they were actually bad.

What is happening is the left third of the screen is keystoned (vertical height squished on the left margin, and by about 1/3rd of the screen, is normal).

I have only worked on a few Tempest analog outputs previously and have not encountered this issue.

Best I could tell with my scope (which due to the smaller screen and age of the scope does not show the problem as clearly) was that the output of the 2 op-amps (in the same chip) after the MC1495L seem to also show the distortion on the scope (as best I can tell).

Could this be from that op-amp ?

Hoping someone has seen this before, as I don't like replacing good parts (shotgunning).

Thanks for any hints/tips/solutions/etc you can provide.
 
Here is a pic of what it is doing....

I've replaced 2 of the MC1495L, and I'm wondering if I should be replacing the other 2.

I don't quite understand the XCOR and YCOR signals from the Y and X sources respectively.

But, I feel there might be something to that, as the Y size is the problem based on the X position.

Tempest gurus, am I on the right track ?
 

Attachments

  • tempest- partial- keystone.jpg
    tempest- partial- keystone.jpg
    485.6 KB · Views: 59
Could be digital or analog. I'd suspect analog first.

The 1495's make up the pincushion correction circuit. Two of them each take a slight bit of info from one axis, and use it to modulate the other axis (thus the Y will be modulated according to the position of the X signal), which is what corrects the pincushioning.

When you look at the schematics, you'll see that they cross axes. (X feeds into Y, and Y into X). Thus, when you are troubleshooting issues related to them, you sometimes have to socket all 4 of them, as you can't always tell which amp is bad (i.e., the one doing the feeding, or the one being fed.)

If it's a digital issue, you should be able to tell by probing the respective X and Y signals from the first TL082 after the DAC. These will basically just be a scaled version of the output signal. So just clip on to those signals to display your image. If you are still seeing the distortion, then your issue is somewhere from that point back.
 
I believe this is an analog issue. I used the outputs of the op amps that feed the MC1495s, and I do not get the screen distortion.

I've repalce all MC1495s, so it isn't a bad one of those. I can't see how it could be an op-amp, as a Y deflection op-amp is not doing anything more than amplifying what the MC1495 chips are putting out. Plus the same op amps deflect fine on the right side of the screen.

Seems it still must be someng in the mc1495 area.

Wondereid if pulling A/B12 and A/B13 (that generate XCOR and YCOR) would result in different behavior.... and if so, if the issue goes away, it indicates something around XCOR or YCOR signal generation or consumption.

Any further ideas on possible components/things to check is appreciated.

I've never claims to be an analog electronics guru... I'm much more at home with the digital side.
 
Socket and replace the final TL082's, just to rule them out. They are generally good to have socketed anyway, as they are the outward facing chips to the monitor, and are common failures, so you're futureproofing the board anyway. Also, amplifiers can fail in weird ways, and can go non-linear, showing all sorts of distortion, so anything is possible with analog parts.

Also note that those TL082's are dual op-amps, so there is one stage that outputs on pin 7, then the next stage outputs on pin 1. You can probe both to trace further.

You can start looking for cracked resistors, etc in the 1495 area, but I've never seen anything there fail naturally (i.e., outside of physical damage.)

Tone every connection out as well, as you might just have a broken joint somewhere, on or around the 1495's.

Also, make sure your pots are ok. (Cleaned with DeOxit at least, if not replaced with sealed pots.)
 
Do all the adjustment pots work? If you find one that doesn't, that would be a clue. X/Y size control one of the inputs to the MC1495s and if it's open circuit the MC1495 will be doing its job with pin 3 floating... garbage in, garbage out. Similar deal with X/Y center. I'd also check your +6.8v.

Note that it's not just the outer edge... looking at the 0000s on your picture they seem keystones proportionally.
 
It is not just the outer edge, the whole left half squishes proportionally in Y based on the X position.

The issue is in the analog section. I haven't had any more time on it yet, but I've got some things to check and try.
 
Well, this gets even more interesting.

If I pull A/B12 the 'keystone' goes away. This confirms that either XCOR is incorrect, or it's effect in the Y section is incorrect.

I was looking at the schematics, and there are class caps (.1 uF if my memory is correct) at each MC1495L power rail (+15V and -15V). Given that this distortion is only on one half of the X where the Y is affected by XCOR, I'm wondering if one of the glass caps on A/B12 is bad ? I wouldn't think the effect of these is very large.... but the resistors all measure in spec, and I can see the XCOR is connected into the first Y MC1495L, and that the X feed into the MC1495 that generates XCOR is connected.

So I could not find a broken connection, resistors looked fine, all MC1495Ls have been replaced (with multiple replacements for verification). Could one of these glass caps be responsible ?

If so, what type of cap is best to replace one of these ? polymer, disc, etc ?
 
I wouldn't suspect decoupling caps on the power supply. They are just for shunting HF noise from the power supply.

What is your +6.8V measuring?
 
I can re-verify that again, but last I checked, that voltage was fine, as was +15V and -15V.

Seems that for some reason, XCOR in one direction (-V?) is incorrect while it is fine in the other (+V?).

I'm not sure what measured by a DVM should yield, but the XCOR value and YCOR values are different when measured to ground. (I don't have my notes handy, but 6.X for one and 11 for the other). Probably not very useful, as this signal is likely rapidly changing based on X vector position. But I did use that just to see if it would spot something way off.
 
Verified all voltages. spot on.

Verified all resistances around the *COR MC1495Ls. Verified resistances between 'main' MC1495Ls and the *COR MC1495Ls.

IIRC I couldn't get an image from the output of the primary MC1495Ls (a + and - feed from the MC1495Ls to the first op-amp, and not a feedback of output to - like on the other op-amps, maybe that was why). So I was wondering if the first opamp for Y after the MC1495L, could be the culprit. However, I don't hold much hope for that, since if I remove A/B13, taking the XCOR signal away from from the Y circuit MC1495L I get a good non distorted picture on my scope (which means the op amp should be working ok).

Any more ideas of what to check. It seems like a fairly small circuit that this has been narrowed down to, but I'm at a loss to explain the behavior.
 
*bump*

I haven't gotten much more time in on this, but still stumped at the moment.

I was thinking of using the dual trace scope to look at X compared to XCOR and Y compared to YCOR (assuming they operate similarly), to see if the generation of the signal looks correct.

The issue must be with the generation or consumption of the signal. Given that consumption of it is just on one input pin of an MC1495L, and I've swapped around multiple MC1495Ls with no change, I'm assuming it is in the generation of XCOR.
 
I haven't replaced any further TLO82 chips yet.... Not sure that it could be that, as when the 1495 that produces XCOR is pulled (and therefore no XCOR signal going into the Y 1495), the problem goes away, which would seem to show that the op-amps are not amplifying as expected.

My hunch is still that the XCOR signal is not generating properly.... but I'm unsure as to why, or how to verify if that is the issue (short of examining using a dual trace the generation of XCOR from the input to the 1495 that generates it, and comparing that to YCOR and the 1495 and its input to see if any difference can be determined. My scope is not a digital one, so getting an idea of the form of these signals based on a trigger is undetermined as to how that will work.

Probably faster and easier to start swapping components ? (not the approach I prefer)

besides the 1495 there are no other 'active' components in the circuit that generates XCOR. There are only the resistors and the small caps on the power rails. I can't recall if I checked the resistors around the 1495 that takes the XCOR signal .... but I suppose that if it isn't in the generation of XCOR it could be in the 1495 that consumes it.

These are the times I wish I was an analog guru so I could understand the operation of the 1495s and what things could cause them to misbehave in that fashion. They appear to be very versatile chips.
 
Are all of the 1495's socketed? If so, swap them around, and see if any of the behavior changes.

The analog parts fail more often than most other stuff on the board, so I always see socketing them as future-proofing, as they'll fail again, even if it's 30 years from now. As long as you do it well, you're doing the next guy a favor.

Also, because the analog parts are so dependent on each other, you can't always know for sure if a given part is bad. If you see a dead signal at the input to a chip, it could be the input section of that chip is damaged and shorting the signal, or it can mean the chip before it is giving it a bad signal. Or maybe a 3rd chip that interfaces to both is bad. And the same thing goes for the outputs. Or maybe the power supply for that chip is having issues, etc.

So there are a lot of factors, and sometimes you can't nail it down the way you can with digital chips, without socketing some stuff. Also, pulling the chips lets you inspect the traces underneath it, where you can find other interesting things. So with respect to the analog section, if you're experienced at soldering (which you are, i.e., I wouldn't recommend the same to a noob), socket away.
 
Yes, I have swapped around the 1495Ls, including 4 new (Chinese IIRC) ones. No difference between any combo I've tried, except for one verified original one from the board that was bad. I've tried multiple combinations of new/original, all new, all original (minus the one bad one, so technically 3 original 1 new), etc.

I feel relatively safe saying that I don't think it is the 1495Ls, but somehow how they are acting in circuit due to the components around/before/after them.

And socketing is a must. If I remove a chip, 99.9% of the time I socket it. Even if I'm out of the sockets, I'll use pin strips if need be. Too much soldering/desoldering would eventually lift traces.
 
Back
Top Bottom