Yessir. One channel works fine. It's sitting steady at ~25mv across TP1/TP2. I've listened to music on it for an hour like that. Both channels were built at the same time using the same lot of parts. I even hand-matched resistors between the two.
Next, I'm going to:
- Re-check cap values. Maybe I have a wrong value causing some oscillation or something.
- Measure emitter resistors on the outputs. Maybe one of them has been compromised.
Next, I'm going to:
- Re-check cap values. Maybe I have a wrong value causing some oscillation or something.
- Measure emitter resistors on the outputs. Maybe one of them has been compromised.
Beryllium Copper/Silver Fuse Clips
I tried Beryllium Copper/Silver Fuse Clips from Ebay. (01220083H -Circuit Board Mount Fuse Clips for 0.25 in Diameter Fuses -Littelfuse}. The through-hole solder tab spacing is too wide for board and the BC is far too brittle to adjust without snapping off the little solder tabs. Stick with the recommended part. Not the correct path to "fancy". Just simple info, not fodder for big discussion, I hope.
I tried Beryllium Copper/Silver Fuse Clips from Ebay. (01220083H -Circuit Board Mount Fuse Clips for 0.25 in Diameter Fuses -Littelfuse}. The through-hole solder tab spacing is too wide for board and the BC is far too brittle to adjust without snapping off the little solder tabs. Stick with the recommended part. Not the correct path to "fancy". Just simple info, not fodder for big discussion, I hope.
Thinking about my problem a little more, I wonder if I have a bad emitter resistor. This might explain higher current than expected being drawn by the output sage transistors and I am thinking it would also explain the drifting voltage across TP1/TP2 that I saw. If I have a shorted (or very small valued) emitter resistor, I'd expect the circuit to be less stable as temp changes.
I also point out that I'm not able to measure those resistors very accurately with my cheapo ratshack DMM. While I'd be surprised to see a resistor fail short, maybe I got a bad part and never noticed since I couldn't measure it? I think I'll try replacing them this go-around.
What do you think? Is my reasoning sound?
I also point out that I'm not able to measure those resistors very accurately with my cheapo ratshack DMM. While I'd be surprised to see a resistor fail short, maybe I got a bad part and never noticed since I couldn't measure it? I think I'll try replacing them this go-around.
What do you think? Is my reasoning sound?
make up a two transistor CCS using a power transistor.
Set the CCS current to 100mA.
Connect a dozen, or so, emitter resistors together in series.
Say you have 0r33 +-10%
12 off 0r33 passing 100mA will need 0.396 +-10% of voltage drop.
The CCS will need at least 2V
find a 3V to 12V DC supply.
Connect up.
Measure the voltage drop across each resistor.
That gives you a 33.xmVdc+-10% reading on the 199.9mVdc voltage scale of a dmm.
If you require more accuracy use a higher current. 570mAdc will give a reading across each resistor of ~188.1mVdc.
That allows you a comparison resolution of 1 part in 1800 to select your matching resistor values.
Use your DMM for comparison measurements when you need tight tolerances.
That does not help when you don't know how close the resistor values are.
+-1% or +-10% makes a big difference when trying to find a problem.
Here is an update for those interested. I was testing a new IPS with my Slewmaster OPS with vertical MOSFET outputs and 0R22 emitter resistors. The IPS had some oscillation and blew the three of the outputs and the fuses. I replaced the bad FETs but when I fired it back up I could not get it to bias properly and the offset was bad. So I started measuring across the emitter resistors and some had current across them and others didn't. So I uses a small 2A power supply set to 3V. I hooked the leads of my DMM to the two leads and began touching either side of each resistor. It was instantly apparent which resistors were bad and which weren't. Five of the ten resistors had opened up. I replaced them and everything instantly came into line.
So, for the above issues where fuses have blown, measure each emitter resistor and make sure you don't have any that are open. These cement resistors don't smoke, they just open.
So, for the above issues where fuses have blown, measure each emitter resistor and make sure you don't have any that are open. These cement resistors don't smoke, they just open.
No it won't.
Take the simplest case: A one pair output stage with one NPN device not quite an exact match for the PNP device. Assume the Re values are different by 3%
With the output offset exactly equal to zero and the speaker NOT connected, the only current escaping from the output node is that passing through the feedback resistor. For a 20k feedback resistor, the escaping current is so small that it can be ignored. Effectively ALL the current coming down through the NPN device passes through the output node to the PNP device.
Measure the Vre of the two resistors.
They will be different.
What does that tell you about the output devices?
Now go to a multi-pair output stage.
All the NPNs are different from each other, All the PNPs are different from each other. All the PNPs are different from all the NPNs.
All the Re vary within the +-5% tolerance.
Measure the Vre after setting the output offset to exactly zero.
All the Vre will be different, although two may be very similar.
What does that tell you about the output devices?
You can make a start at improving the information coming from measuring the Vre by ensuring that the Re values are all matched before assembly.
If you want a particular accuracy from your measurements then the re tolerance should preferably be 10 times better than the accuracy you want. You could get away with only 5 times tighter matching tolerance.
Let's suppose you would like to know the bias currents of the output devices to an accuracy of +-5%. You NEED Re tolerance better than 1% and preferably +-0.5%
If you don't use accurate Re matching, you CANNOT get accurate results in converting Vre to bias current.
BTW, absolute accuracy of Re is NOT required. It is matching accuracy that IS required.
Last edited:
Thanks both of you!
I am hoping that my first failed attempt at biasing this channel caused one or more resistors to blow and subsequent attempts at replacing the output devices would understandably fail thereafter.
Now that we're discussing it, I also note that upon inspection of blown output devices, 2 of them show shorts across all pins and the others do not.
This discussion has given me enough confidence that it's worth trying. It's easy enough to put in new ones. I'm going to pop out those resistors and replace them (once my weekly package from mouser arrives). Before I put the new ones in, I'll make up the CCS and build a 4 wire R tester as suggested. Seems like a handy circuit to have around anyway. I'll let you know how it goes.
I am hoping that my first failed attempt at biasing this channel caused one or more resistors to blow and subsequent attempts at replacing the output devices would understandably fail thereafter.
Now that we're discussing it, I also note that upon inspection of blown output devices, 2 of them show shorts across all pins and the others do not.
This discussion has given me enough confidence that it's worth trying. It's easy enough to put in new ones. I'm going to pop out those resistors and replace them (once my weekly package from mouser arrives
). Before I put the new ones in, I'll make up the CCS and build a 4 wire R tester as suggested. Seems like a handy circuit to have around anyway. I'll let you know how it goes.
Measuring one at a time will depend on how well the CCS holds constant current.
It is better for "comparison" to pass the SAME current through the string of series connected resistors.
Then you measure each one in the string. Recheck to ensure the CCS has not drifted as it warmed up or the voltage source drifted. You will often find that the first two or three measurements have changed due to the CCS warming up.
BTW,
this "comparison" method is so sensitive to small differences, that you can measure the resistance of the lead out wires !!!!!!
You need to measure at the lead out wires such that the SAME LENGTH of lead out wire is included in each measurement.
Last edited: