Tangent Young-Jung power supply - Page 6

Thanks error401!

Interesting... I'll definitely be up for a couple boards. Would this thing be good for running down in the DAC power supply range? 1-5V? I've got a TREAD and a protoboard'ed STEPS ready for the DAC I'm building, but I wouldn't mind trying something a little nicer...

As a side note, would an RC filter (something like 10-20ohms) on the output of a TREAD/STEPS get a bit more ripple rejection? Would it be worth it, or a waste of time?

Getting down to 2.5V isn't a problem. Below that, you might have to get tricky. Basically, you replace the voltage reference with a lower voltage one, and there are readily available ones in this package and pinout down to 2.5V. There are lower voltage references, and alternatives to this sort of reference that go below 2.5V, but I don't know that they fit the footprint.

EDIT: You'll probably also have to change the AD825 out for a chip that's happier on lower supply voltages.

EDIT2: Ditto D3. I've modified the YJPS part selection guide to discuss the D3 and U2 changes, and to cover the effect of changing D4 more clearly.

This should be especially good for a precision mixed signal circuit, due to the wide regulation bandwidth and the AC line filter.

Yes, if your only goal is to power a multimeter's measurement input terminals. Or *maybe* a true class A device. As soon as you put any kind of varying load on it, you get varying V drop across the R, giving ripple, probably far more than you are trying to fix.

This is why we take so many pains to lower the output impedance. The ideal power supply has 0 ohms output impedance from DC to daylight.

Problem with YJPS

My YJPS went together easily, but the regulator isn't working - I'm getting around 48-49 volts before and after the regulator. Voltages up to the regulator are fine. Can anyone help? I'm relatively new at this, and this is the first time I've really been stumped. More details:

1. D4 gets hot (I'm not sure about the op-amp - I don't want to stick my finger down there! I'll use a digital thermometer when I get back to work)
2. Other voltages to ground (VSET at 482 ohms, no load):

a. between R8, R9 and D4: 7.0V
b. between R8 and opamp pin 3: 12.0V
c. between R10 and R11 (opamp pin 2): 12.8V
d. between R11 and VSET: 6.4V
e. TP6: 48.9V
f. both before and after R7: 1.0V
g. between C8 and Q1/Q2: 49.2V

3. Adjusting VSET does not affect the output voltage appreciably.
4. D2 lights momentarily when the unit is powered and then shuts off. With no load I thought this would be normal but I wasn't sure.

Thanks in advance to anyone who can help!

Here are some pics:
Attachment 12113

Attachment 12115

Attachment 12117

Attachment 12119

Attachment 12121

First off, that's going to be a very inefficient power supply even once you do get it working. Did you not use the power supply parameter estimator to choose the transformer? It would have predicted this. It doesn't yet go out and tell you that this is hugely wasteful, it just tells you what the voltage drop is, and assumes you know this is bad.

If the estimator says the configuration is "sane," that means it can be made to work, not necessarily easily, but it's at least possible.

If you're willing to accept 18-20 V regulated out of this, you can fix the efficiency problem by paralleling the transformer secondaries instead of putting them in series. Your alternatives are to either cope with the heat it generates (possibly requiring forced-air cooling when enclosed) or swap the transformer, which would be difficult.

It's almost certainly dead, due to having 49 V across its rails. If not dead, at least damaged.

The only question is whether this is whether this is the chicken or the egg. If the op-amp died while the regulator was still working, the resulting high voltage came afterward. If the high voltage is no fault of the op-amp, that killed it, so it could no longer force the regulator back to a normal output voltage.

You're going to have to stick something down there. It's going to have to be replaced.

Before you do that, check to be sure it wasn't installed backwards, and that you used the right sort of chip. Either error would explain the symptoms.

Also, a useful test will be to just remove the op-amp and re-test. If the op-amp was the source of the problem, your build will turn into an overly expensive unregulated supply: you'll see the expected ~2.5 V drop across the preregulator, and D2 will stay lit up. It should be possible to power at least a light load with it, though unless you also parallel the secondaries, you may have some trouble finding a circuit it won't fry due to the high voltage it's putting out.

If you try this and either the preregulator doesn't appear to be working or D2 won't stay lit, you have a different problem. You need to find and fix that before replacing the op-amp, or it'll get killed again as soon as you apply power.

D4 is fine. You should be seeing about a 55°F rise over ambient. It's enough to make you take your finger back off it PDQ even when ambient is room temp, but not enough to hurt the IC, unless you put it into a hot enclosure. (See above.)

The D4 temp will drop a lot when the regulator starts working.

That's probably just Q3/Q4 doing their jobs, trying to clamp the +/-IN difference to a sane value. They should be getting warm, which is expected and harmless.

That's very weird. I would expect about 9V across VSET in your current situation. The voltage across R10+R11+VSET is 49 V, which puts about 19 mA through here. Another 9V or so should be across R11, and the rest -- around 30 V -- should be across R10.

I can't see JRSV in your pictures: is it jumpered? If not, that would explain a lot of problems.

Is there a ~2.5 Vdc drop across the preregulator? (TP4 to TP5) If not, the preregulator isn't working, which could have fried the error amp at least, disabling the regulator, so would be the actual problem.

Can you post a picture of the solder side?

Speaking of pictures, the biggest thing you can do to improve them is turn on more lights, open the blinds, or take it outside to use the big bright day-ball for illumination. In the one set of EXIF data I looked at, the flash fired, the aperture was wide open, and it still managed only a bare-minimum 1/60s exposure. All of this says you didn't give the camera enough light to work with, which is why depth of field is so shallow.

Well now I'm completely embarrassed. I did not use the power supply parameter estimator - it was neglect and lack of thoroughness on my part. I'll make use of it once I figure it out and try to find a more appropriate transformer. I'll also have to assume that when you say that it is "nearly impossible to remove a PC mount transformer from the board once it’s soldered down without damaging the board" that you don't mean that it is completely impossible!

To clarify, I'm actually after a 30V supply to use with a PPA v.2. I realize VSET is set wrong for this (I had been adjusting it to see if I could change the output voltage) - 482 ohms is just where it happened to be when I measured the voltages. If I understand your reply correctly, the transformer I have is still not ideal.

Let me see if I have this right - I should look for a transformer that will give me around 32.3V DC between TP3 and TP5 instead of the ~51V I get with this one. Roughly speaking, that means a ±10V transformer wired in series (or ±20V in parallel) - does that sound right? I'll figure out the estimator and use it before I make a final decision.

Other details:
1. JRSV is jumpered
2. The opamp (AD825) is installed in the right direction (I double and triple checked before I installed it - the little circle by pin 1 is over the small silver dot on the board), all the connections are good, and there are no solder bridges.
3. Voltage between TP4 and TP5 is 1.6V - should it be 2.5 volts now, or just when the opamp is removed?
4. I remeasured the voltages around R10, R11 and VSET. They're a little different because I'm now in a different location, so the wall voltage is probably a little different.

TP3 to TP6: 47.9V

TP3 to opamp pin 2 (between R10 and R11): 12.6V

TP3 to between R11 and VSET: 6.3V

TP6 to pamp pin 2: 35.1V

That gives about 9mA heading toward the opamp (?!)
5. Pic of solder side (sorry about the quality of the other pics - I hope this is a little better)
Attachment 12245
6. D2 lights up and turns off abruptly every time I plug in the unit - is it possible there is protection circuitry in the AD825 and it's just shutting down? If it was fried, wouldn't D2 cease to light up at all? I'm hesitant to remove the opamp until I'm sure I have to, since it's pretty cramped in there now. (I was only reluctant to stick my finger down there when it was powered up!)

Thanks for your help - it is much appreciated.

Well, the bright side is that it's better now than before I wrote the previous post. I'd been meaning to add more warnings to it about obviously-wrong configurations, because experience has proven them to be non-obvious.

True. I've done it once, and it's not fun at all.

Ideally, what you'd want is to get four DIY friends to come over, set the YJPS upside down on a table, build up a solder blob to connect each pair of pins in the corners of the transformer so one iron can heat both pins at once through the blob, then have everyone apply their irons while one of you gently wiggles the transformer from below to get it to drop out of the board once all 8 joints turn liquid. Then you owe them beer.

When you have to do it solo, you do much the same thing, only you dip your iron rapidly into the solder puddle, then go to another corner on the same edge, then back until you can get those two heated up. Meanwhile, you've got a small screwdriver under that same edge of the transformer gently prying up. Once you raise it by about half a millimeter, you transfer your efforts to the opposite edge. Half mm by half mm, you eventually lever the thing up out of the board. Then you drink lots of beer.

There's nothing magic about the 24 V and 30 V numbers you see tossed around here so frequently. A PPA will run on 20 V just as well. There may be some slight loss of sound quality, but that may be a tradeoff worth taking to avoid the desoldering challenge, and corresponding risk of cracking or removing PCB traces.

No, it's just that while the circuit is damaged like this, it can't affect the output voltage. It's useless.

Sort of. You have two problems. One is a broken regulator or preregulator circuit. (Exact fault and fix still to be determined.)
The other is an inefficient transformer configuration, which can be fixed two ways: replace it with a more appropriate transformer, or parallel its outputs and accept a lower output voltage, in the 20 V neighborhood.

Just let the estimator guide you. Transformers and unregulated power supplies have enough weirdness in their behaviors that you don't want to try to figure out what they're going to do on the back of a napkin. That's why the estimator exists: so you don't have to try and work it out on paper.

Just checking...pin 1 is indicated on the AD825 by....?

It should be ~2.3 V all the time. This suggests the LM317 may be dead. That would explain everything.

Removing the op-amp before making that determination is still a good thing to do, however, because it's possible that a broken op-amp could jerk the LM317 around enough to cause problems here.

It occurs to me that it probably makes no sense to make decisions about what you see here while Q3 and Q4 are conducting. If you did want to check the sanity here, you'd remove the op-amp and temporarily lift one end of either R8 or R9. That will get the transistors out of the circuit so the divider just acts as a divider.

I don't see anything seriously wrong with the soldering, though I could wish for better resolution and less JPEG compression... There's one blobby joint over near the right, but it's probably harmless.

The protection circuitry in an op-amp is connected to the input and output pins, not the power pins. The datasheet says the absolute maximum voltage you should put across this chip's power pins is 36 V. You've exceeded that maximum by 36%, repeatedly, and probably for a considerable amount of time each time.

Is it possible for the chip to live through this? Sure. Sometimes people who get hit by lightning live to tell about it, too.

If you're looking for something easy to remove as a first step, you can replace the LM317, too. But, if you suddenly start getting ~2.3V from TP4 to TP5, you still have a trust problem with that op-amp: why would you believe it's unharmed? It's not like the op-amp is anything other than the single most important component on the board...

I'd remove C11 and R12 so you can get at it.

Then when replacing R12, put a properly-sized resistor in there. This is just the LED current-limiting resistor...there's no call to use a Vishay-Dale here. Any old carbon resistor will suffice.

As you should be. 49 V is enough to kill humans.

Finally got around to casing up my YJPS which will power an M3.

I used the estimator and found a couple of 2x12V transformers that will work, so I'll have a crack at taking this one out (hopefully not cracking the board in the process!). I'll order a new opamp and LM317 at the same time.

I'll probably replace the transformer before I remove the opamp, since I don't really want to make do with the current transformer and I am clinging (irrationally?) to the hope that the opamp may be OK after all. If I get the new transformer in there and the numbers still aren't right, then I'll remove the opamp and test the voltage drop over the LM317 and replace it if it isn't 2.3V or so.

Determining pin 1: As far as I can determine from the data sheet, the AD825 has a small circle of a different texture on the plastic case near pin 1 - that's right, isn't it? I think all of the opamps I've used so far are that way.

I just used the Vishay-Dale for the current limiting resistor because I happened to have an extra one, that's all. I've used Radio Shack carbon resistors in the past but I ran out.

I'd rather have a 30V supply, because that's what I was using previously with the PPA and I liked it. I'll probably be able to reuse the 70053 transformer in something else - in fact, do you have any STEPS boards left? I wouldn't mind building another one of those. I used 70053 in the one I built earlier and it works nicely.

Thanks again for your help, Tangent. I'll post again once I get the new transformer mounted.

The transformer is off, and it wasn't that difficult. I used two soldering irons, one screwdriver and a college-age son to do the job. After sucking all the solder off the joints I gently pressed on the two pins at each corner with the two irons while my son pried on the other side with the screwdriver. It took about six times around to ease the pins out, but now it's done and the board looks good!

I suppose that's why they let them home over the holidays...

It's a beautiful build, Nate!

Which 1455 is it?

That counts as difficult. You owe yourself and your son a beer.

Quite irrational, I'm afraid. It's possible it "works" in some limited sense, but complete lack of damage is more than I would hope for. You might as well replace the op-amp with a 741 if you're not going to worry about running a high-end supply like this with a damaged error amp.

Nope.

You could pair the transformer with a TREAD, though.

The YJPS is less forgiving of excess voltage appearing on its output, since there's a plain old op-amp there. Also, it splits its heat load between two smaller heat sinks instead of using one really big one like the STEPS did. There is room on the YJPS for a bigger heat sink on Q2, but then that pushes parts of the regulator closer to the transformer, probably not a good idea.

Schubert, I was thinking more about this, and it might be a good idea to choose a different op-amp, one made for greater than +/-15V supplies. There aren't a great many to choose from, particularly if you restrict your choices to JFET-input chips, which are reportedly more stable in the Jung regulator. (It's not a DC offset issue, it's an RFI susceptibility one.)

I might try an OPA604 myself, as it's actually rated for +/-24V. There are a fair number of +/-18V supply chips out there, too.

It's the T2201 and was built to match a friend's M³. Using short standoffs (3/8") there was about 1/16" clearance between the heat sinks and the top of the case.

Schubert: I once removed a trafo, pretty much the way Tang described, also cause I used one with too high of an output voltage.. went pretty smoothly though.

Nate, Got any comparisons to a STEPS with the M^3? I need to get myself some sort of drill press too.. I strive to have holes that aligned