Hi Fred,
You're getting good advice. Everyone does things a little differently, but the goals are the same.
Pete, I use 100 and 10 ohm resistors for testing sometimes. The 1K resistors may burn out with no faults. For easily measuring current, instead of using a current meter, I use either 1 ohm, or 0.1 ohm resistors in dead fuses. The 0.1 ohm resistors are those "plate" resistors that have very low inductance, and they handle surges easily in a compact package. You must have a really good meter to get meaningful readings with the 0.1 ohm resistance at lower currents, so with a budget meter, use 1 ohm wire-wound or metal oxide resistors (wire-wound preferred here). The same form factor that you'll see in better Japanese designed amplifiers.
For Fred, he might want to increase the AC mains voltage while watching the current draw (one can never have too many meters on the bench).
Hi Nigel,
For those of you who skimmed over the above paragraph, there is a good troubleshooting hint in there.
-Chris
You're getting good advice. Everyone does things a little differently, but the goals are the same.
Pete, I use 100 and 10 ohm resistors for testing sometimes. The 1K resistors may burn out with no faults. For easily measuring current, instead of using a current meter, I use either 1 ohm, or 0.1 ohm resistors in dead fuses. The 0.1 ohm resistors are those "plate" resistors that have very low inductance, and they handle surges easily in a compact package. You must have a really good meter to get meaningful readings with the 0.1 ohm resistance at lower currents, so with a budget meter, use 1 ohm wire-wound or metal oxide resistors (wire-wound preferred here). The same form factor that you'll see in better Japanese designed amplifiers.
For Fred, he might want to increase the AC mains voltage while watching the current draw (one can never have too many meters on the bench).
Hi Nigel,
Yes! I've done that for years, but people starting out are less confident and may mistake how it's wired. You can also simply install a diode across the emitter-base terminals of the removed outputs and driver transistors. This works well, is much easier for inexperienced people to do, and you can test the rest of the circuitry in that area. I'm thinking of current limiting transistors and protection circuit detection components. You also naturally close the feedback loop in the process. Some output configurations require more inventiveness to pull this trick off on.
For those of you who skimmed over the above paragraph, there is a good troubleshooting hint in there.
-Chris
Yep. Checked against the data sheet, against the schematic, and against photos of the original board in it's original state.
Okay, started working on getting the boards in the state that Pete recommended (Post 513). Worked on the input board last night with the following test results:
Removed Q4, tested good, no leakage, 230hfe
Removed Q9, tested good, no leakage, 420hfe
Removed Q10, tested good, no leakage, 420hfe
R16 was burned through. Removed, not yet replaced.
Tested all the following resistors in circuit, all tested good
R1, 2, 5, 6, 7, 8, 9, 11, 12, 13, 14, 15, 17, 18, 19, 20, 21, 23
I believe R3, R22 and R10 need to be out of circuit to test, right? (R4, strangely, does not exist).
Removed Q4, tested good, no leakage, 230hfe
Removed Q9, tested good, no leakage, 420hfe
Removed Q10, tested good, no leakage, 420hfe
R16 was burned through. Removed, not yet replaced.
Tested all the following resistors in circuit, all tested good
R1, 2, 5, 6, 7, 8, 9, 11, 12, 13, 14, 15, 17, 18, 19, 20, 21, 23
I believe R3, R22 and R10 need to be out of circuit to test, right? (R4, strangely, does not exist).
If D5 or D6 are either in backwards, or blown then all of the bias current for those diodes goes to the bases of Q3 and Q4 and are multiplied by the device beta which is not how the current sources are intended to work. A scenario I didn't think of was for Q4 to conduct a lot of current reducing it's own Vce (which saved it) but burning up R16 and the 1K.
R3 and R22 can be tested in circuit, R3 as long as the diff pair are out, and R22 because there is a series cap which is open circuit to DC.
R4 would have been the other diff pair degeneration resistor.
R3 and R22 can be tested in circuit, R3 as long as the diff pair are out, and R22 because there is a series cap which is open circuit to DC.
R4 would have been the other diff pair degeneration resistor.
Okay, with some off-line help from Pete (thanks Pete!) I did the full-power leakage tests on the output transistors.
The amp is in the state described earlier, including 100 ohm current limiters across the fuse holders.
With 82vdc on the rail, I connected C of each transistor to the supply (by reattaching a case screw) and connected the E to the supply through a 21.9k ohm resistor.
Here are the results:
(-) side:
Q13: -1.9mvdc
Q14: -2.0
Q15: -2.2
Q16: -2.0
(+) side:
Q17: 439 mvdc (same transistor that had leakage when tested with the IT-18)
Q18: 4.6
Q19: 0.3
Q20: 0.3
For each measurement, I waited about 10 minutes for the reading to settle.
Pete, correct me if I'm wrong, but I believe we also determined that D5 and D6 were working correctly. Other tests were inconclusive (e.g. Q3).
The amp is in the state described earlier, including 100 ohm current limiters across the fuse holders.
With 82vdc on the rail, I connected C of each transistor to the supply (by reattaching a case screw) and connected the E to the supply through a 21.9k ohm resistor.
Here are the results:
(-) side:
Q13: -1.9mvdc
Q14: -2.0
Q15: -2.2
Q16: -2.0
(+) side:
Q17: 439 mvdc (same transistor that had leakage when tested with the IT-18)
Q18: 4.6
Q19: 0.3
Q20: 0.3
For each measurement, I waited about 10 minutes for the reading to settle.
Pete, correct me if I'm wrong, but I believe we also determined that D5 and D6 were working correctly. Other tests were inconclusive (e.g. Q3).
Hi Fred,
Q17 has been condemned by both methods now, but it hasn't failed catastrophically. That means it is an isolated component failure. Q18 should probably be cleaned and tested again. The high Vce test you are performing can be more sensitive to surface leakage. Also, it's important that each base should have a resistance connection to the emitter so that charges can bleed off. You can easily measure voltage from E-B if there is actual leakage.
What measurements did you get on the terminals of Q3? D5 and D6? My feeling is that they should be okay, but that is not an observation. You have to prove this through your measurements.
-Chris
Q17 has been condemned by both methods now, but it hasn't failed catastrophically. That means it is an isolated component failure. Q18 should probably be cleaned and tested again. The high Vce test you are performing can be more sensitive to surface leakage. Also, it's important that each base should have a resistance connection to the emitter so that charges can bleed off. You can easily measure voltage from E-B if there is actual leakage.
What measurements did you get on the terminals of Q3? D5 and D6? My feeling is that they should be okay, but that is not an observation. You have to prove this through your measurements.
-Chris
Hi Chris,
"Also, it's important that each base should have a resistance connection to the emitter so that charges can bleed off."
What value of resistance?
David.
"Also, it's important that each base should have a resistance connection to the emitter so that charges can bleed off."
What value of resistance?
David.
Hi David,
I would tend to use whatever is in the circuit already. Normally, that would be somewhere between 100 ohms and 470 ohms. I don't think the value is critical, but since you are measuring leakage currents, you don't want a charge to build up on the base. It is high impedance until you start getting close to the rated E-B voltage.
I guess if you looked the part up, you could use an open base connection up to the rated voltage, but I would allow any charges to bleed off as you increased your test voltage to that rated C-E voltage. After that, you could open that connection to allow a very sensitive test for C-B leakage. A small test jig would accomplish that. Things not built in a box don't tend to survive storage until the time you need it next.
-Chris
I would tend to use whatever is in the circuit already. Normally, that would be somewhere between 100 ohms and 470 ohms. I don't think the value is critical, but since you are measuring leakage currents, you don't want a charge to build up on the base. It is high impedance until you start getting close to the rated E-B voltage.
I guess if you looked the part up, you could use an open base connection up to the rated voltage, but I would allow any charges to bleed off as you increased your test voltage to that rated C-E voltage. After that, you could open that connection to allow a very sensitive test for C-B leakage. A small test jig would accomplish that. Things not built in a box don't tend to survive storage until the time you need it next.
-Chris
Hi Fred,
Chris, Vceo is done with the base open, resistance to the emitter will make
the test insensitive to minor CB leakage. There were some very old specs
that used a 100 ohm resistor from B-E but I've not seen those in a long time.
I think we agree that Q17 is bad at this point.
You can compute the leakage current with the simple formula
of I = V/R for them if you are interested.
We did determine that D5 and D6 were working and biased correctly,
but could not explain why R16 burnt up. The best that we could come
up with was that Q4 might be a fake.
We should review what has changed since the last time it was powered up
without R16 overheating. Work was done on the front end board, and it was
rewired into the amp. It would be prudent to check the wiring, however as
I stated before Q4 is a constant current source and should have been able to
have the collector connected to any voltage without a problem. It seems to
me that the only logical reasons might be that Q4 was a fake, or that it was
installed incorrectly.
It is good that the output devices are all tested now. I'd suggest that we slowly
build up the driver board, testing as we proceed. Remove all collector screws
from the outputs and leave out the drivers.
I suggest that you replace R16 and reinstall Q4.
We want to test Q3 and Q4 to determine that they work as current sources at
full voltage.
1. Tack in 1K resistors from Q3 C to ground and from Q4 C to pos supply.
2. Slowly bring up the Variac (25V, 50V, 80V) and note the voltage across
the 1K resistors at each rail voltage. You can compute the current. Watch for
smoke obviously.
Chris, Vceo is done with the base open, resistance to the emitter will make
the test insensitive to minor CB leakage. There were some very old specs
that used a 100 ohm resistor from B-E but I've not seen those in a long time.
I think we agree that Q17 is bad at this point.
You can compute the leakage current with the simple formula
of I = V/R for them if you are interested.
We did determine that D5 and D6 were working and biased correctly,
but could not explain why R16 burnt up. The best that we could come
up with was that Q4 might be a fake.
We should review what has changed since the last time it was powered up
without R16 overheating. Work was done on the front end board, and it was
rewired into the amp. It would be prudent to check the wiring, however as
I stated before Q4 is a constant current source and should have been able to
have the collector connected to any voltage without a problem. It seems to
me that the only logical reasons might be that Q4 was a fake, or that it was
installed incorrectly.
It is good that the output devices are all tested now. I'd suggest that we slowly
build up the driver board, testing as we proceed. Remove all collector screws
from the outputs and leave out the drivers.
I suggest that you replace R16 and reinstall Q4.
We want to test Q3 and Q4 to determine that they work as current sources at
full voltage.
1. Tack in 1K resistors from Q3 C to ground and from Q4 C to pos supply.
2. Slowly bring up the Variac (25V, 50V, 80V) and note the voltage across
the 1K resistors at each rail voltage. You can compute the current. Watch for
smoke obviously.
You could do the same test that we did for the output transistors. See if it will take the full 160V with the base open - Vceo. Voltage collector to emitter, base open with a 10K -22K current limiting resistor. And all the usual tests in your IT-18, beta should be comparable to the original. The other option is to use the original part if you still have it.
I believe Pete has a plan to test the transistor Q4 in circuit.
If it is a fake then it probably won't pass the test. Worst case you might burn the resistor again, but I think everyone would like to know why the resistor burned in the first place.
If the circuit was working properly as a current source then it would be impossible for the resistor to burn. The only way I can see the circuit failing, given that D5 and D6 tested good, is if the CE breakdown voltage of Q4 were exceeded. That's is why there is suspicion that Q4 may be a fake.
David.
If it is a fake then it probably won't pass the test. Worst case you might burn the resistor again, but I think everyone would like to know why the resistor burned in the first place.
If the circuit was working properly as a current source then it would be impossible for the resistor to burn. The only way I can see the circuit failing, given that D5 and D6 tested good, is if the CE breakdown voltage of Q4 were exceeded. That's is why there is suspicion that Q4 may be a fake.
David.
Hi Pete,
Why?
Transistor manufacturers and acceptance tests for end users are dealing with new, clean transistors. Their test jigs are known to be good and the operators trained. As well, the power supplies in use are known to be free from noise on the output. A home experimenter should exercise a little extra caution. unless they know their equipment and procedures align with industry standards.
You are right in that these tests using a resistive link from base to emitter are not often seen these days unless you are looking at SMPS switching transistors (where this is valid). The resistive connection yields a higher withstanding voltage figure. No surprise there.
-Chris
That's right. However, my recommendation was to leave the resistance in place until the test voltage was reached, then open that resistor connection to the base. This would prevent causing the transistor to conduct through the C-B capacitance. Once the test voltage is reached, removing this connection will give a more accurate idea on leakage than if the base was left open the entire time.
Why?
Transistor manufacturers and acceptance tests for end users are dealing with new, clean transistors. Their test jigs are known to be good and the operators trained. As well, the power supplies in use are known to be free from noise on the output. A home experimenter should exercise a little extra caution. unless they know their equipment and procedures align with industry standards.
You are right in that these tests using a resistive link from base to emitter are not often seen these days unless you are looking at SMPS switching transistors (where this is valid). The resistive connection yields a higher withstanding voltage figure. No surprise there.
-Chris
You might check the pin out to confirm it is the same as the original. If it is a fake, they might be different.
Hi Steve,
Excellent point!
I've been nailed that way before and just check all new parts before I install them these days. Takes a while, but it's faster than redoing the job or troubleshooting the problem!
A cynic? Perhaps, but you work in a way that makes the hurting stop.
-Chris
Excellent point!
I've been nailed that way before and just check all new parts before I install them these days. Takes a while, but it's faster than redoing the job or troubleshooting the problem!
A cynic? Perhaps, but you work in a way that makes the hurting stop.
-Chris
Maybe I'm missing something, but if I hook up the TR to the IT-18 and set it to the known type (NPN or PNP), according to spec, then presuming it calibrates and tests correctly, don't I know the pin-outs are correct?
If it tests correctly, then yes. If it doesn't test correctly, is it bad or a different pin out? I ask this because I once received a bag of transistors from MCM that were all the same part number, but had 3 different pin outs. I would have never found the problem if I hadn't owned a tester that would identify the pins for me.