30+ minutes of leaving the boards powered on, still no changes to the LED, I am pretty sure I am having an issue with the boards here. Any insight on what to test on the SMT boards when protection circuit is not engaging the relay, yet offset measures 3mV at R15/R16 would be greatly appreciated.
If it worked last night, it cannot be too bad.
You already checked the actual offsets at the output of both channels and found it withing single millivolt range. That means the amp works correctly, so we will be looking at the protection circuit.
First, check the voltage at the output of the comparator - it's pint 1,2,13,14 of the comparator itself. It should be about zero. Alternatively, you can check the voltage across C53 - it should about the same as the negative supply rail. If this checks out, then the problem is in the Schmitt trigger Q51 Q52. If it doesn't, we need to check if the comparator works correctly, If Q53 is open, and if R55 charges C53.
In my testing, the protection didn't work once after I washed the board. The charging current via R55 is small, low tens of μA, and even though I dried the board, the residual water on its surface conducted enough current to prevent C53 from charging.
You already checked the actual offsets at the output of both channels and found it withing single millivolt range. That means the amp works correctly, so we will be looking at the protection circuit.
First, check the voltage at the output of the comparator - it's pint 1,2,13,14 of the comparator itself. It should be about zero. Alternatively, you can check the voltage across C53 - it should about the same as the negative supply rail. If this checks out, then the problem is in the Schmitt trigger Q51 Q52. If it doesn't, we need to check if the comparator works correctly, If Q53 is open, and if R55 charges C53.
In my testing, the protection didn't work once after I washed the board. The charging current via R55 is small, low tens of μA, and even though I dried the board, the residual water on its surface conducted enough current to prevent C53 from charging.
Last edited:
Appreciate the help alexcp, right on the cusp and I hit a snag, such is my luck.
LM339 pins 1, 2, 13, 14, are all -16.9vdc. That seems concerning! COM DMM lead on a ground pin, positive lead on the comparator pins
C53 has 128mV across it.
Both boards show the same thing. Here's some pictures so you can see what I am working with. It seems the problem is with the comparator, with both boards behaving the same way I guess it will be the same problem on both.
Attachments
Sorry about that.
Q53 and "power good" done with Com DMM lead on a Gnd pin, positive lead doing the measurement. Only tested one board this time since both are behaving identically. Despite my previous picture not showing it, I do have a 3pin header shorting the 3 input pins together.
"Power Good" -20.65vdc
Q53 gate -20.70vdc
Q53 source -17.05vdc
Q53 drain -16.95vdc
Edit
Also wanted to clarify on C53's voltage.
DMM leads on either side of C53 I get 128mV. Com lead on ground and putting pos lead on either side of C53 and I get -17vdc or so on either side. Sorry about that, I am but a novice learning as I go.
Q53 and "power good" done with Com DMM lead on a Gnd pin, positive lead doing the measurement. Only tested one board this time since both are behaving identically. Despite my previous picture not showing it, I do have a 3pin header shorting the 3 input pins together.
"Power Good" -20.65vdc
Q53 gate -20.70vdc
Q53 source -17.05vdc
Q53 drain -16.95vdc
Edit
Also wanted to clarify on C53's voltage.
DMM leads on either side of C53 I get 128mV. Com lead on ground and putting pos lead on either side of C53 and I get -17vdc or so on either side. Sorry about that, I am but a novice learning as I go.
Last edited:
No problem, we'll sort this out.
"Actually, my name is Austin Powers. Troubleshooting is my middle name."
Did I get that right?
What we see so far is the amp working correctly, Q53 working correctly, but C53 being close to discharged and not allowing the relay to switch on. Looking at the schematic, there are several potential point of failure. C53 may not be charging through R55, or it may be discharged by the comparator. It may be discharged by faulty Q51, or by something else.
Let test the first two points. First, check if the comparator is happy about the amp's output voltages. For that, check the reference voltage at the point where R51 R53 connect (should be about +80mV vs ground) and at the point where R52 R54 connect (should be about -80mV vs ground). Just in case, check the voltages that the comparator receives from both channels - that's points of connection of R49 C43 and R50 C44, respectively. These two should be close to those measured at the amp's outputs.
"Actually, my name is Austin Powers. Troubleshooting is my middle name."
Did I get that right?
What we see so far is the amp working correctly, Q53 working correctly, but C53 being close to discharged and not allowing the relay to switch on. Looking at the schematic, there are several potential point of failure. C53 may not be charging through R55, or it may be discharged by the comparator. It may be discharged by faulty Q51, or by something else.
Let test the first two points. First, check if the comparator is happy about the amp's output voltages. For that, check the reference voltage at the point where R51 R53 connect (should be about +80mV vs ground) and at the point where R52 R54 connect (should be about -80mV vs ground). Just in case, check the voltages that the comparator receives from both channels - that's points of connection of R49 C43 and R50 C44, respectively. These two should be close to those measured at the amp's outputs.
Great! The inputs to the comparator are what they should be. If the comparator is ok, it should not prevent C53 from charging.
Let's temporarily connect, say, a 10kOhm resistor across R55. No need to solder, just carefully touch both side of R55 with the leads of the extra resistor. The value is not critical - it should be much smaller that 1 megohm, but at least a couple of kOhms; avoid shorting R55. If there is a small leakage somewhere, this resistor should overpower it. Try it and see what happens.
A side question: you mentioned that two boards behave identically. Does it happen with the same power supply?
Let's temporarily connect, say, a 10kOhm resistor across R55. No need to solder, just carefully touch both side of R55 with the leads of the extra resistor. The value is not critical - it should be much smaller that 1 megohm, but at least a couple of kOhms; avoid shorting R55. If there is a small leakage somewhere, this resistor should overpower it. Try it and see what happens.
A side question: you mentioned that two boards behave identically. Does it happen with the same power supply?
10k across R55 - while difficult when you aren't exactly rock steady handed, it appears to blink on the LED, and sounds like a relay clicks. Does that indicate I need to do an even better job washing up the board with isopropyl alcohol? Or something else?
I have another assembled PSU, I haven't tried it yet. I picked up 3 sets of PSU/SMT amp board, assembled 2 sets so far. Tested 1 PSU and 2 amp boards.
I have another assembled PSU, I haven't tried it yet. I picked up 3 sets of PSU/SMT amp board, assembled 2 sets so far. Tested 1 PSU and 2 amp boards.
Aha! The protection almost works!
It indicates a few more things need to be checked. One suspect is leakage from the trace connecting the gate of Q51, the drain of Q53, the output of the comparator, R55 and C53; note it includes a short piece on the bottom of the board. Leakage from that trace to -17V may be preventing C53 from charging. The only place where this trace is physically close to -17V is around Q53. Check that area with a magnifier, wipe it with alcohol, then dry.
Second suspect is R55. If it is not soldered properly, it may be unable to charge C55. Try heating one end of R55 at a time with a soldering iron to make sure it is connected to the trace below.
Let's see if it helps.
It indicates a few more things need to be checked. One suspect is leakage from the trace connecting the gate of Q51, the drain of Q53, the output of the comparator, R55 and C53; note it includes a short piece on the bottom of the board. Leakage from that trace to -17V may be preventing C53 from charging. The only place where this trace is physically close to -17V is around Q53. Check that area with a magnifier, wipe it with alcohol, then dry.
Second suspect is R55. If it is not soldered properly, it may be unable to charge C55. Try heating one end of R55 at a time with a soldering iron to make sure it is connected to the trace below.
Let's see if it helps.
Hey Alexcp and bloqherd
You'll excuse me from the side.
Since the LM339 has PNP inputs, the output will be active LOW if any of the inverting inputs float.
and the relay does not turn ON.
Isn't the 4th pin of U51 LM339 floating?
If pin 4 floats, the non-inverting input, pin 5, will have a bias current that can pull up the 3mV output offset to 8mV.
Looking at the photo of #244, the LM339 looks a little slanted. Also, the size of the footprint resist is for reflow and seems small for hand soldering.
You'll excuse me from the side.
Since the LM339 has PNP inputs, the output will be active LOW if any of the inverting inputs float.
and the relay does not turn ON.
Isn't the 4th pin of U51 LM339 floating?
If pin 4 floats, the non-inverting input, pin 5, will have a bias current that can pull up the 3mV output offset to 8mV.
Looking at the photo of #244, the LM339 looks a little slanted. Also, the size of the footprint resist is for reflow and seems small for hand soldering.
Maybe - I noticed the difference, too, but it is hard to see from the photo taken at an angle. However, if it was the case, connecting 10kOhm across R55 would not switch on the relay, as the gate of Q51 would be driven to -17V by the comparator. So my conclusion is that the comparator is wired, and working, correctly.
Last edited:
Here's as close up of a picture as I could get of the area around Q53 and R55 after cleaning it up.
@mason_f8 the board was soldered with a hot plate and paste as opposed to hand solder. Here's a picture of the comparator close up too. I wet each pin and re-cleaned prior to taking the picture and testing.
@alexcp same behavior following cleanup and a precautionary wetting of R55 and Q53. I was able to find a steadier hand in placing 10k ohms across R55, the LED stays on, but the relay buzzes as if it's opening and closing quickly, if the LED is blinking I can't tell or it's too rapid. Not sure if that is expected when performing that test but thought I would update.
The lower end of C53 (the one pointing to Q51 Q52) on the photo above appears not to be soldered to the pad under it. Same may be the case with C51 and C52. It is sometimes difficult to solder a taller component like a 1uF capacitor, as solder remains on the capacitor only and does not flow to the pad under it.
Perhaps you're right. I did not realize that the relay was buzzing with the 10kOhm resistor. I still have doubts, as the 5mV difference on 102kOhm of the LPF means only 50nA of current, pretty low for two inputs wired together.
I re-wet those, they look better but still the same behavior.
Sadly on the board in pictures, I accidentally shorted pins 1 and 2 of the comparator while probing, an alarm went off and startled me.. sigh, not my day hah. It looks to have cooked Q53, I can replace that, any other parts to worry about on that board?
I put in the other board glad I have multiple to work with here so we can keep on track. I quickly went back over the spots we've tested, the amp DC offset is good, ~3mV each channel.
LM339D pins 1, 2, 13, 14 all at -16.95vdc
"Power Good" -20.65vdc
Q53 gate -20.70vdc
Q53 source -17.05vdc
Q53 drain -16.95vdc
R51-R53 +84.0mV
R52-R54 -84.9mV
R49-C43 8mV
R50-C44 8mV
C53 has 128mV across it and ~17vdc on either side.
10k on either end of R55 had the same result, LED on, relay buzzing.
This is what has stumped me, both boards behaving mostly the same, although this other board has the new behavior or both channels at 8mV where you had me test.
Attempting to follow along as a newb with you and mason, does that point towards the comparator as a problem?
Last edited:
@mason_f8 : good catch, thank you.
But now we have 8mV on both channels.
@bloqhed : not necessarily. Still. let's make sure all comparators' pins are soldered well.
If that doesn't help, I have a plan. Check if it works after each step:
But now we have 8mV on both channels.
@bloqhed : not necessarily. Still. let's make sure all comparators' pins are soldered well.
If that doesn't help, I have a plan. Check if it works after each step:
- Disconnect the PGOOD wire coming from the power supply (e.g. temporarily replace your 5-way cable with four wires) and check if it works without it. I am still concerned that both boards behave identically - errors and faults are common but rarely identical on different boards.
- If you can, remove Q53. If you cannot, just make sure Q53 is closed by shorting R60. An even better idea would be to connect a battery of some other isolated low voltage source between PGOOD (negative) and -V (positive), driving the gate of Q53 below the negative rail with something else than the PGOOD from the power supply.
- Short С43 and С44. This effectively will take the comparator out of the circuit. (In a through hole board, we could just pull it out of its socket, but alas, not here).
- Now we have only R55 and C53 left to prevent the relay from working. Measure the resistance of R55 with your DMM. If it is ok, remove С53.