EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: c4757p on January 06, 2014, 07:16:25 pm
-
I've got another V-665 and it's a sick puppy. Can anybody help me figure out what its problem is? I really have no idea where to even begin.
The horizontal drive amplifier is oscillating, at about 25 kHz, and it is being clamped at both ends. The oscillation appears to worsen when it clamps, and the clamp levels expand as it warms up. Once it's been on for 20 minutes, the picture finally fills the entire display. I've taken scope grabs:
- Its own screen about a minute after power-on.
- The horizontal drive output to the CRT.
- The horizontal sweep signal entering the amplifier.
- The signal at the emitter of TR852 (see schematic). It doesn't seem to be a particularly interesting point, but it does have an interesting signal...
And a schematic. Ignore the cursors - I just forgot to shut them off.
No components appear to be bad! :-// There is noise on the power supply, but it's clearly from the amplifier itself, because it doesn't appear at the PSU end of the power cable. The amplifier is very difficult to probe inside, as the oscillation is leaking into damn near everything, and the signals are absolutely tiny at some points. It's hard to tell what's on the signal and what's coupling into the probe... I have no idea why a working amplifier would suddenly start to oscillate. Anybody?
Edit - the signals appear to be symmetric. Both halves are oscillating...
-
Can you disconnect the input and measure the dc voltages round the transistors?
-
I didn't disconnect the input, because it obviously contributes to biasing the common-base input stage, so I switched to X-Y mode, centered the spot and shut off the readout. There was a small bit of oscillation still, but it didn't interfere with measurements. Here you go. It all looks good to me :-//
TR881 and 882 both had about 11.4V on their collectors - forgot to add that one
Update: OK, sorry, it's not the H amp. I really should have fully disconnected the input - not only is the oscillation gone, but it is present in the input cable. It's clearly coming from the sweep circuit, not the amplifier...
There's still about 100mVp-p of noise/oscillation/something on the H drive output, but that's 100mV on top of 45-50V or so, and it's not enough to fuzz the display even a bit.
Though I may find myself at a dead end there as well, and back soon...
-
Update: OK, sorry, it's not the H amp. I really should have fully disconnected the input - not only is the oscillation gone, but it is present in the input cable. It's clearly coming from the sweep circuit, not the amplifier...
I was just about to post that, as there is a very small ripple/oscillation on your input image.
So the ripple is present in other/all low voltage circuity?
Can you post the sweep schematic?
-
OK, it looks like I don't know how to operate a probe. :palm:
The signal entering the output drive amp is perfectly clean. Don't know where I picked up noise... but the output amplifier is oscillating at 29 kHz, 760mVp-p with no input cable connected.
Here's the full schematic for that page. The cable I disconnected joins the left and right halves.
Further explanation:
It's actually PSU noise, and I think it's normal. That's the SMPS frequency. It is too small to be visible on the display, and I think that's the best filtering they cared to have (all the filter caps are brand new and excellent quality, so it's not a failing cap). Clearly, what is happening is that the amplifier has decent PSRR normally, but it is saturating, and at that point the whole rail noise is dumped to the output, which is why the trace is "squiggly" at the end.
So why would it saturate like that?
I'm going to plug the input back in and play with it a bit.
-
The one thing that strikes me is that the output should be 45V +/-40V, but you've measured +30V.
Is what's connected to the input changing the biasing differently to how it should.
-
I think you and I are onto the same thing. Look at this.
The scales are all 20V/div (my channel 1 has a faulty 10x sense). The top is the noninverted output. The bottom is the inverted output, inverted again. (Both referenced properly to the center graticule line.) The center is the difference.
In the middle, the noise is common-mode, so it doesn't appear on the display. But then the traces slam against the bottom, at about 9V, apparently saturating. The noise disappears from that side, it's no longer common-mode, and it shows up on the display.
So first, the signal should centered higher. But it also shouldn't be appearing to saturate at 9V; 45V +/- 40V implies it can hit 5V.
:-// :-//
-
Can you now go back to x-t mode, set it at around 9V output and see where the saturation is by looking at the dc voltages round the transistors.
I've got a pile of V-552s in the corner with a similar fault and I think that the circuitry is very similar, so I'm interested in the results.
-
Yup. I had to go make and eat dinner, or else I'd have probably made more progress. I'll update soon.
It is tempting to just shotgun all the transistors in the amp. This scope had a whole bunch of failed semiconductors in the PSU, so I really don't trust whatever else is in it. I don't know what happened to it - bad batch of components, power surge? No idea. This was not the first fault it had...
I must say I am quite pleased that the fault is in the output circuit. I have no idea how you are supposed to work on the H sweep board. It's sandwiched between the frontend and the PSU, with very little space between (and menacing voltages hanging out on multiple sides).
-
Yup. I had to go make and eat dinner, or else I'd have probably made more progress. I'll update soon.
Just like fancy analog, good food is an art, I had homebrew pizza for dinner?
-
Interesting. As I move the spot across the display, the voltage on TR854's collector varies from 1.25 to 1.10 apparently linearly. The voltage on TR853's collector does not - it goes from 1.10 to 1.21 right before the trace craps out, then spikes up to 1.64.
But I didn't find the problem... because I already replaced that transistor on a hunch. Something else must be causing it to do that...
If it's TR851, that's a 2SC535C: 450 MHz minimum, 940 MHz typical, 100-200 hFE. The closest I have is MPSH10, which has a higher minimum GBW product of 650 MHz, but no typical. Hmm... do you think they binned them for bandwidth? I suppose I could just calculate minimum GBW for the circuit.
- Nope, it's not TR851.
So it's not TR853 or TR851. What the hell is it then?? All I can think of is D873/D874. What are they there for, anyway? Clamping? I don't really suspect them because I think they'd affect both sides equally, wouldn't they? But it's only TR853/1 that have a problem.
-
Diodes?
-
Yes and no. I just lifted both D873 and D874. The trace can now reach both edges of the screen fully... but it still distorts. Just, instead of hitting a wall at that point, it pushes (sluggishly) past it now.
And with TR851 changed, the voltage on TR853 still explodes upward at the right edge of the screen. Hmm.... maybe I will replace 853 again. I wonder what are the chances that I installed a replacement with the same fault...
Nope, that doesn't help either.
-
Every transistor, resistor, diode and capacitor in the signal path from where the signal is good to where the signal is bad tests OK.
You know, I've never Widlarized an entire oscilloscope before... :-BROKE
Oh, new suspect: TR881. Looks like it has a leaky C-B.
Or not. It's just fine. But something is making it look like that in circuit, when TR882 on the other side does not.
And that's enough for tonight. >:( This oscilloscope has got to be one of the most irritating ones I've ever worked on. The system layout philosophy seems to be something along the lines of "eh, it'll fit there" and "nobody will ever have to repair this, right?"
-
Well hey, at least it's a constantly occurring fault. I've got two scopes I'm still working on that only fault if the thermometer ends in an odd digit and the time is anytime I am not ready to do work on it...
Good luck :)
-
It's a pity you're not in the UK - I have a 665 which someone has remove the HV transformer from - while I assume that's because it had other problems the two 'scopes might have been a better starting point to get one working one at the end :)
-
I am going to just shotgun the whole amplifier. All transistors, all diodes, all high(er) voltage capacitors. This damn thing is just not worth my time.
OK.... and I just did. Except for the caps, I didn't have any that matched, but I tested the ones it had at full working voltage.
Bloody thing still doesn't work. It must be something external to the amplifier........ so I have no idea how it can have a good signal going in and a bad signal coming out.
All the resistors check out too.
This oscilloscope is going to spend a nice vacation on the "can't be bothered" pile, a.k.a. the "sometimes I hate electronics" pile.
-
Wow! January! Forgot I'd had this for so long. I was thinking April, or maybe March...
Figured I'd come tie up the loose thread, as I dragged this out one last time yesterday and got it fixed (well, sort of...) over the two days.
The amplifiers in this thing are biased with precision resistors for a reason. They're DC-coupled (obviously) and have quite a bit of gain (something like AV=400?), and made from discretes - so if those discretes drift even a bit towards the input end of the amplifier, the drifting bias point is amplified significantly.
The amount of drift that causes this is quite small, so I never did fully track down which part it is. It's on the microprocessor board, which is sandwiched in the most awkward way between two other boards, and I couldn't find it by probing around with the power off. The signal coming out of this board was perfectly fine, it was just shifted some 40mV.
Rather than replacing parts on a board I could hardly test, I simply injected some current into the horizontal drive amplifier to compensate for the drifting bias. The annotated schematic shows that I measured a center point of about +30V on the negative output, and a 22k resistor added between the input and the -12V rail pulled enough current through that amplifier to push the output waveform up to around 44V, preventing it from going into cutoff.
Unfortunately, this took away a significant portion of the horizontal adjustment range system-wide, because the other side of the amplifier was also debiased, though not as badly. A 33k resistor in the same place fixed this, and now the oscilloscope performs perfectly.
Calibrated once, and then the crusty old trimmers drifted enough in a few hours to need a second calibration. I'll have another go at testing it out tomorrow, and then I've got another good scope! :-+
-
It's good to hear you've finally found a solution.
-
I've got a pile of V-552s in the corner with a similar fault and I think that the circuitry is very similar, so I'm interested in the results.
Did you ever fix yours?
-
I've got a pile of V-552s in the corner with a similar fault and I think that the circuitry is very similar, so I'm interested in the results.
Did you ever fix yours?
I suspect I'll make a start next week. I did drag five out of the storage unit this morning.