[KrisBlueNZ]

If you want, you can try replacing the small transistors in the +48V regulator - Q5~8 - just as you replaced Q1~4. That is:
Q5 and Q6 same as Q1 and Q2, which are MPSA92 or NTE288;
Q7 same as Q3, which is 2N3904;
Q8 same as Q4, which is MPSA42 or NTE287.

I had that thought too, but it also occurred to me that when I took the readings straight from the transformer output plug it was low there as well. What are the odds that the transformer itself is faulty?

Pretty low, unless it is getting hot. But the AC voltage from that secondary isn't low... In post #24 you said "Pins 3 and 4 (which feed a 48V line) has 58.8V". 58.8V AC is plenty high enough.

Did you disconnect the wire from the HV MULT unit? Did that have any effect on the smell?

No, I never located that wire.

Well, you should try again. I didn't suggest it for fun! Find the wire from the ultor cap on the side of the CRT and trace it back to the HV MULT unit (aka "tripler"). There should be one wire, probably a well-insulated one, from the tripler to the HVPS board.

No sizzling or crackling sounds, or if they are present they are drowned out by the buzzing of the neons. I'll look more closely at the transformer this time to see if anything is happening beneath the board where it's harder to see.

Good idea. The corona may be so slight that you can't see it, but it can still produce that smell. Also, it could be inside something, I suppose.

 

[tryppyr]

I checked the schematic and if I read it correctly the cable should terminate at the front side of the CRT. The most heavily insulated cable I see going that way is the pink one with the thick white connector seen in this image. Nothing I can see from this vantage point declares itself to be A6 or HV MULT, but that's probably because it is under A15. Is that the cable I should disconnect?

Update: Disconnecting that cable had no effect on the smell.

The smell SEEMS to be coming from the transformer on the HVPS (or something beneath it).

Looked at it from the underside in the dark... still no visible corona.

 

[tryppyr]

I removed the HVPS to see if I could locate the HV MULT (A6). I noticed a well insulated black cable with a yellow insulated connector. That connector plugs directly into the transformer board. I removed the housing and revealed the module beneath it.

Update: Checking the parts list this is the HV MULT assembly. No wonder I couldn't find the cable to disconnect.

 

[tryppyr]

With HV MULT disconnected the ozone smell is still apparent, still seems to be coming from the transformer, and I still see no corona.

 

[KrisBlueNZ]

OK. It sounds like it might be the transformer. I don't know what would cause that problem in a transformer. I expect that dampness getting into it might do it.

The other possibility is a problem with the regulation. The HVPS circuit may be generating too much voltage. That would explain why the neons are constantly lit. I think that would be the first suspect.

In that case, you should avoid running the oscilloscope for any length of time. When you need to do a test, set up the multimeter in advance, turn the scope ON, wait until you get a measurement, then turn it OFF.

The first thing to check is the regulated negative output voltage, which comes out on the right end (on the schematic) of CR3, and feeds through R11 to TP4. The voltage on TP4 is what's important, but it's too high to measure with a normal voltmeter, so keep away from it!

Instead, you need to check the voltage on the junction of R13, R23, R37 and C14. This should be around 0V. If it's significantly negative, this points to A15U1 as the fault. This is a CA3094A "operational transconductance amplifier", which is now obsolete. Can you measure that voltage - carefully - and let me know?

 

[tryppyr]

Okay, before I run the test, should I reconnect the HV MULT or leave it out?

Regarding the instructions on testing for very short durations, that's been the M-O. Something tells me if I leave it running more than a minute or so something will blow again.

As an aside, I do have a voltmeter that caps out at 1000V. Not that I'm particularly anxious to try it at high voltage...

 

[KrisBlueNZ]

Leave the HV MULT disconnected. It won't affect the reading, and it's safer that way.

Standard voltmeters go up to 1000V. This test point is supposed to be more than twice that!

 

[tryppyr]

I tested the voltage at the place marked by the green oval. It was jumping all over the place. Never got a good reading.

While examining the area, I noticed that the resistor I replaced when all this started has already burned up (see the red oval). Clearly something is cooking that junction. It made the old pair of resistors pop off, and cooked this one.

 

[tryppyr]

I just removed the cooked resistor. Just as a data point I checked it and found that it had been reduced to a 49K resistor. I have a few other 68K resistors, but none small enough for that slot. I also have tons of new 62K resistors (blue ones). That's as close as I get with existing stock and all the stores are closed today.

 

[KrisBlueNZ]

OK. It's difficult to diagnose this circuit. U1 is not easy to test, so let's assume it's OK. Another component that's likely to fail is R13. I think it's the big cylindrical red thing directly left from the top of the green oval in that picture in post #88. Can you lift one end out, confirm that it's marked R13, and measure its resistance. Also, can you tell me the markings on it. I can't read the value on the schematic.

As for the 68k resistor, there is obviously a problem there. If it's not the CRT, it's most likely to be the transformer. To find out, you need to disconnect transformer pins 9 and 10 from the rest of the board. If that requires cutting tracks, you should do it. Be sure you check both sides of the board. You can replace the 68k resistor with a 62k resistor; the value is not critical. If the transformer is the problem, cutting the tracks should mean that the new resistor doesn't overheat.

 

[tryppyr]

I did go ahead and install the 62K resistor in the R14 spot, then took the voltage measurements at R13. Before the neons started buzzing it was at or near zero. Once they started buzzing it fluctuated between +40 and +60 (much narrower range than before I switched out the resistor).

Yes, the long red component is the R13 resistor. The markings are:

PFR8050
PVC175
30M 10%

I'll lift a leg out a bit later. We're preparing Thanksgiving dinner.

 

[KrisBlueNZ]

OK. You won't be able to measure that resistor; normal multimeters can't measure resistances higher than 20 megohms. You can measure it using the 120V rail but you'll have to do a bit of work.

First, check that your multimeter has an input resistance of 10 megohms when measuring voltage. This is normal and standard for modern digital multimeters.

1. Remove R13 from the board.
2. Disconnect off-board Q1 from the HVPS board, to disable the HVPS circuit completely (very important).
3. Using alligator clip leads, connect one end of the R13 resistor to the +120V test point, and connect the other end of the resistor to the positive probe of your multimeter. Connect the negative probe of the multimeter to the 0V rail or chassis of the oscilloscope.
4. Set the multimeter to a 50V or higher DC voltage range.
5. Power up the scope and look at the multimeter reading.

If the R13 resistor is OK, you should measure about 30V and the measurement should be fairly stable (assuming the 120V rail is stable).

If you're interested, here's how I calculated that voltage.

When you connect two resistances (the R13 resistor and the multimeter's input) in series, their resistances add. The total resistance will be 30 megohms plus 10 megohms, which is 40 megohms.

Ohm's Law says I = V / R, where I is current in amps; V is voltage in volts, and R is resistance in ohms. With 120V across a 40 megohm resistance, the current should be 120 / 40,000,000 which is 0.000003 amps, or 3 µA (microamps).

Ohm's Law can be rearranged to V = I R. When that's applied to the multimeter only, V (voltage) is 0.000003 amps multiplied by 10,000,000 (the input resistance of the multimeter), which is 30 volts.

Therefore there should be 30 volts across the multimeter and that is what you should measure.

If the voltage is badly wrong, especially if it's too low, or if it jumps around constantly, you can replace it with nine 3.3 megohm resistors in series. Solder them together in a long string, and mount them off the board. Use metal film resistors.

 

[tryppyr]

Honestly, it seems like it would be easier to just replacement resistors and try that out than to test the current resistor. Vetco has 10M resistors for only $1.35 apiece... or 15M resistors for the same price, and I could pick them up tomorrow. If they solve the problem, great. If not, I have a few more resistors in my supply box.

One thing. I note that the original resistor is a 3W and these metal film ones are only 2 watt. Is that likely to be a problem?

 

[(*steve*)]

Check the voltage rating for the replacement resistors.

The voltage rating is probably the most important rating (aside from the resistance) in this application

For 2W, a 10M resistor would require 4400 volts across it.

 

[KrisBlueNZ]

Right. As Steve said, the voltage is the problem. The top of that resistor has about 2200V on it, and that's far too much for a normal resistor, which is why they used that big thing originally. That's why I suggested nine 3.3M resistors - they will have about 250V across each one. Normal 1/4W resistors will be fine.

 

[tryppyr]

Okay. Thanks!

 

[tryppyr]

I'm constructing the off-card resistor block and now it occurs to me to wonder whether I need a particular gauge of wire running to/from that block. I have 20 and 22 gauge mono-filament. Will that be adequate?

 

[KrisBlueNZ]

There's a very high voltage across that resistance. If you mount all the resistors on a single board, that voltage will appear across that board, so make sure the board is rated for that voltage. If you use stripboard, or any other board with a copper pattern on it, remove all the copper apart from what you need to connect the resistors together. The wire gauge is not important because the current is very low.

While you're at it, you might want to clean the HVPS board of any impurities, by scrubbing all accessible parts using a toothbrush and isopropyl alcohol and rubbing it off with a cotton swab.

 

[tryppyr]

Damn. Sounds like I need to start over. (sigh) Got ahead of myself.

 

[tryppyr]

Based on the previous reply, I cut a piece of old phenolic micarta and made an old school board. No copper traces to worry about. The ends of the resistors are just soldered together like the old tube type radio circuits. Then I coated them with liquid tape to insulate them (though that was superfluous, given that this board will be bolted to the frame at the empty end). Ugly, but serviceable.