Problem with LM317T.
40 volt limit or the chip blows up.

I'm not too "proper" to redesign the 30 volt power supply, I just needed to point out the voltage limit on the chip.
I believe a 60 volt version is available.

CR841 will not "rectify" in the manner you suggested, however, if it is shorted, that would be a way to explain the whole problem. Lift a leg and check.

ESR is not a likely candidate for this problem. The capacitor would have to look like it's open before this kind of fault would cause this much error. (Don't worry about not having an ESR meter.)

Referring to post#4, I can see why you can't get zero volts on the base of Q850. As long as the +30 and -30 aren't balanced, your adjustment range will not center. The problem is that this adjustment should cause the balancing to happen, and it isn't. Q850 is responsible for pulling the down base of Q840 and it isn't. This makes Q850 a prime suspect. Either it's open or it isn't being fed properly.

LM317HV

Thanks for the input, I previously tried replacing Q850 with a new 2n3904 transistor, and the old one tested fine after I had it out. So.. If Q850 is indeed the problem then it must be something about how it's getting fed.

I've checked most of the components that I think would have an impact on the current flow through Q850 and they check "OK" with my relatively primitive methods.

 

I totally agreed with #12. I think the fault should be around Q850 and dont give up too early!

Where does the +200V come from? Was there any ripple on it? I'd suggest putting a 10uF 50V cap on the collector of Q850 and GND and see if situation improves. BTW what part no is Q850?

Allen

The 200V is rectified off the main transformer through 4 separate diodes. Each of those diodes check OK. I Don't have a 100x probe so I can't check for ripple, but I imagine it would have some ripple at the 200v line as it is unregulated in normal operation.

 

Diagnostic procedure: Connect a 10k resistor from the collector of Q850 to ground...NOT to negative 30 volts, Not to emitter, to ground. If the +30 line colapses to about 15 volts, you have found the part that is not dumping the "go" signal and it is Q850.

Edit: Q850 can be a TO-92 (small) as it is only dumping about 2ma.

I tried something similar to this to reduce the collector resistance to -30V, but didn't try it to ground. I will try this tonight and report back.

At this point I don't suspect Q850 as I already replaced it with a known good 2n3904 transistor, which matches what is specified in the parts list. But this test may give interesting information to help zero in on what's going on.

 

I jumpered the collector of Q850 to ground through a 10k resistor, and the 30v rail dropped from 36.2 to 35.6 volts.

After being on for a few minutes, it drifted up to 38.20 volts, and the 10k resistor to ground dropped it to 37.5 volts. Any significance to the 0.7v drop? Maybe a diode drop or just coincidence?

A 3.3k jumper dropped the 30v rail to 36.4 volts. A 1k jumper dropped it to 33v and got toasty pretty quick... Too hot to hold in my fingers after about 1 second and it was a 1/4 watt type.

 

Looking at the schematic, the only way that my 1k jumper could ever pull enough current to get hot would be if Q840 or Q845 is feeding a voltage onto Q850's collector. So.. that may answer the question for me, it looks like Q840 is probably bad and managing to leak current. I don't have a replacement handy for Q840, but I'm going to remove it and see what happens.

 

I lifted a leg of R846 (390 ohm feeding Q845), and the voltages on Q850's collector didn't budge, so I'm assuming at this point Q840 is gotta be the problem.

I should have suspected it first, since Murphy's law dictates that the most difficult to access (and cross-reference) component is the most likely to fail.

 

Don't worry about the 390 and Q845 right now. They are an overcurrent protection circuit. You could take them out and the circuit would still work.

When you do that 10k jumper to ground, the voltage at the junction of the 82k, collectorQ850 ans baseQ845 should drop like a rock to about 16 volts. It didn't. Where's the excess current coming from? The 82k resistor (not likely) or Q840 leaking backwards from collector to base. Likely.

and don't worry about a cross reference. Power supplies are not that picky about things like high frequency response and inter-electrode capacitance. Any generic Darlington with the right voltage rating and enough case to get rid of the heat will work.

Pull Q840 out and read its pedigree (part number).

 

I agree with your analysis, thanks for the tip on the diagnostic with the 10k resistor, it really showed what was going on.

Q840 is a General Electric D40K2 NPN in a TO-202 case heat-sinked to the oscope frame. From what I've found I should be able to replace it with a modern MJE802 type Darlington for about $0.50.

To get this fella out requires removal of about 8 screws, 12 hard-soldered wires, and two rail frames on the scope. The MJE802 will take a couple days to get here, but it might sit around waiting for a disassembly!

In your experience is this a common mode for darlingtons to fail? I checked resistances across it (in circuit only) and it wasn't a short circuit at all, but due to being in circuit with that 150-ohm power resistor I couldn't get an accurate test for BE / BC drops.

 

Personally, I've never met a failed Darlington. I couldn't guess what is common for their failure mode.

Do not disassemble until the parts arrive. You will forget little details in just a day or two. And make a drawing of every wire you disconnect! Line up some shot glasses (or similar) for the different types of screws. Be careful about wrecking the insulator between the transistor and the metal of the heat sink. Have some heat sink compound ready to grease the new transistor. Hold the legs with skinny pliers next to the body of the new transistor then bend the rest of the lead (if needed).

ps, the 150 ohm resistor isn't connected to the base, only the collector.