I would just rebuild any high voltage power section as a rule of thumb... years ago I used to just replace the faulty part, only to find that something else would fail within a few weeks. Did you check the voltage at the base of Q22? Should be around 193VDC. The zener is a part of the reference circuit. If you want to know the theory of operation, Google series pass transistor regulator.

These circuits can be designed and drawn in hundreds of ways but the concepts are the same. Q21 is the series pass transistor. Q22 forms a Darlington Pair with Q21 to increase current handling. Q23 is the error amplifier. If the error signal changes the transistor amplifies this signal and feeds it back to the base of the series pass transistor to adjust its effective resistance.

If you are only getting about 100 VDC out at TP2, the top of divider network formed by R54/RT1/R56, something is certainly amiss. It might be a component or components, but if not you may also have to check ALL the interconnections between them. That's why I suggest starting with checking all the noted voltage points, but all those should also be present at all the components connected to them (i.e., you should have around 193 VDC at the base of Q22, one side of C27, one side of R51, the collector of Q23, and one side of C28)

Seems like the only part you haven't replaced is the adjustment pot. I would make sure that is working.

Low output voltage is usually bad 140v zener diode in my experience.

1N5275A, 0.5w 140v zener is plenty fine. It was pretty rare they fail, but it does happen. Not sure why.

I am using 1n5382 on my replacement board. I think most 140v zener diodes are obsolete now. That part was actually the cheapest one I could find, ~ $0.04ea. It is bit overkill being a 5w part and the tolerance range seems high, but works fine in this circuit. I am not sure if the drill holes and spacing on the original board is big enough though.

Not exactly you symptoms, but close.

Untitled (resized).png

1N5382 wide tolerance won't matter -- it's an adjustable circuit.

Z4KE140A is a good sub that is still made - 1.5W and is a closer fit.

Chinese transistors are a real crapshoot.

Power transistors under western brands out of china usually big time pass. There are OK Chinese transistors but they usually have a unique part number and official looking datasheet, ie from LCSC and not ebay/aliexpress.

If you are having trouble with finding the original can transistors MJE3439 can be used if you get the pin out correct. TIP50 is close enough too. You don't really even need a heat sink on either of them based on temperature readings I have done, but a small TO220 can be mounted to them with an m3 screw and nut.

https://www.arrow.com/en/products/mje3439g/on-semiconductor

pasted_image (resized).pngpasted_image (resized).png

Based on other parts shown by that ebay seller - he's definitely a counterfeiter. He's got parts listed with date codes that cannot possibly exist. Pretty much puts all his parts as suspect.

VR1. Did you install a new zener here? You previously mentioned getting a zener, but you didn't say if you installed a new one.

Are you sure about the 95V and 98V at the Q22 pins? The voltage on the base should be higher than the emitter, not lower.

SDB_HV1.png

Nice Quench! It is easier to understand the way you have it drawn.

I would like to know how .5 V is on the base of Q23, 80V on the cathode of the zener isn't enough to bias the zener.

If you can't (easily) find a 140V zener it is normal practice to replace it with two zeners in series. IE use a 62V and a 75V in series giving you 137V which will work perfectly in this circuit.

There's a lot of things there that defy logic unless the zener that got installed was actually a 1N5268 (80V zener).... or behaving as an 80V zener.

First - Vbe of Q23 is at 0.5V. A 1N5275 has a max leakage current of 0.1uA. At that current - max voltage at Q23 base would be (0.1u)(1200) = 120uV, nowhere near the 0.5V measured. So where is Q23 base voltage of 0.5V coming from? Only place I can think of is excessively high leakage current through the zener, too low of zener voltage or just a bad zener. Could also be a bad Q23 as that can cause all sorts of erratic measurements even on the base.

Somehow - in theory, enough current is being supplied into the base of Q23 to cause him to turn on and try to pull the voltage down. Again, this points to the zener voltage or a bad Q23.

And as Quench stated, how can the emitter of Q22 be higher than the base voltage of Q22? That's simply not how an NPN transistor behaves. Maybe both Q22 and Q23 are bad?

Q21 behavior looks fine - the output voltage is the base voltage minus the part's internal Vbe voltage drop.

R51 has a 175 volt drop across him -- that would be ((175**2)/22000) or 1.4 watts across a 0.5W resistor! Don't touch that!

this doesn't help you locating the culprit, but at such little cost, replace the components you haven't.

i personally double the wattage on resistors, regardless of what is recommended.

even a resistor, or any component for that matter can 'test' okay out of circuit, but under load can fail.

You should desolder one end of resistors in complex circuits to get accurate measurements.

On the bench and disconnected from anything, when I measure it on a good board from the ground side of R56 to TP2 side of R54, I get 115k ohms immediately even after swapping the meter leads around. There's no capacitance across that resistor ladder and it goes against the reading of your "good" board.

My working board measures 82k ohms across R56 immediately even when swapping the meter leads around.

All resistors measure to spec right away on my board other than R35 but that's because it has the C26 filter capacitor across it.

Is resistor R51 (22k ohm) on the bad board original and still good? You've got 175 volts across it. That's 1.4 watts of power dissipated by a 0.5 watt rated resistor.. She should be burning if she hasn't already gone open circuit.
Measure the long distance resistance across R51 from the collector of Q23 to the collector of Q21 in case there's a connection/trace issue to that resistor.

Correct me if I am wrong , but as long as C26 is still charged when you are taking resistance measurements across various points on the board, they will be affected by it? If C26 is still charged , then it is going to drive the transistors and slowly discharge as you are taking measurements. Even more true if the output is disconnected since there is no more load or anything else added to the circuit. As Quench stated there is nothing to cause capacitance on R56 to R54, his board doesn't measure that , but yours did..

You have the luxury of having a working board to take components and swap them. With so few components in the circuit, it should be a breeze to move the most probable components from one board to the other.

On the bad board picture , the zener hasn't been replaced yet ? Are you 100% sure of the polarity of that zener, verified with it's datasheet ? I know some manufacturers sometimes label the ANODE with a band on zeners.. you know , just to mess up people.

Does RT1 actually vary the voltage but it's just not high enough ? Or does it not vary the voltage at all ?

I would definitely try what wayout440 suggested as well , is that you should try to measure the voltage at all the components involved, like measuring your 95V at the R51 resistor directly , then checking to make sure it's still exactly 95V at the base of Q22 and at the collector of Q23, just to rule out a possible broken or damaged trace that would create resistance, changing the voltages.

Question to the smarter ones out there. If there really is 175V on R51 , then it would be burned open (looks good on the pic). But if it was open because it burned, how would Q22 actually get driven if the output is 97V. Could current leak through C27 and actually drive Q22 ? Like bad filtering on C26 letting AC ripple through , making Q22 act up and open / close very fast , lowering the expected voltage output ?

Your 'good' board appears to have either a bad resistor R35 or a bad connection to it. R35's job is to drain current from C26 after you shut the machine off. Your measurements are showing that the cap is still charged.
This could also be due to a very high capacitance value and R35 is just taking longer to drain the cap. You should wait until the cap is drained before taking any readings. Measuring resistance while there is still voltage present can damage your meter.

Two boards appear to have same but older NOS zeners installed - one on bad board appears to have Bally part number on it, other says "5275" indicating it would be a 1N5275 as expected. Not a good or bad thing... just an observation. Neither of these are new, are you getting these zeners from donor boards?

It's been more than 6 months that I've worked on one of these boards but my rusty memory tells me it takes about a minute to pretty much discharge C26.

Yeah, a minute or two sounds reasonable. Inkochnito still should check to see if his board is discharging or just sitting there with a charge. It won't affect operation of the 'good' board but can be a safety issue.

Good to hear. So you got yet another source of 2N3440 transistors since yesterday and they worked ok?

Can you tell us all the Chinese vendors (with web links) you got the working and non-working transistors from?

For repairs to these boards I use TIP47G (MUST have the 'G') for the drivers and MJE15032 for the output. Never had any issue using these parts and they are very easy to find.

These 2N3440 from Poland might be a better source closer to you and look like good value:
ebay.com link: 20pcs 2N3440 PNP 350V 20mA 1W TO39

Why must it be the "G" suffix -- that only means "RoHS compliant".

I do agree that the TIP50 would be a better choice than the TIP47 to get us into the right voltage range.
And the Wakefield 291-H36AB heatsink should be fine for the voltage setting transistors.

MJE15032 is certainly higher in current passing than the BUX85.
But the MJE15032 (like the 2N3584) was a bit of a lightweight when it came to voltage and cuts it a bit too close for comfort.
I would stick with the BUX85 - he is still at more than 4x the fuse rating and has a far higher voltage rating.

For a heavy duty setup of parts --
I would use:
2x TIP50 with HSS-B20-CP-01 or equivalent heat sinks
1x BUX85 series pass transistor
1x Z4KE140A zener diode

Edit:
Important -- note that the TIP50 and 2N3440 have different pinouts so pay attention to Base, Emitter and Collector pins on transistors at installation.

PinCompares (resized).jpg

So you bought 40 transistors to fix one game? Geez, why not leave some for other folks trying to fix their games?

NTE 384 and 396 parts have worked well for my rebuilds of this circuit.

Yes but the transistor only has 193V across it. The input voltage isn't what is applied across the devices. I have never had a failure using these but I do buy them from reliable places such as RS Components.

... and this is why when one transistor fails, they all do.