Did you mean 8 mV across LME49810 outputs (source, sink)?
I guess your supply is a conventional linear one with transformer, rectifiers and caps. Did you observe the rail drop after the in series "protection resistors"?
Do you have the amp with x2 output devices running with +/- 30 V rail?
Last edited:
Panson: 8mV is between collector of Q5 and collector of Q6. The supply is conventional 500VA 2x40V. I do not have protection resistors but I use a variac and a light bulb. The DC voltage drop is progressive from +/- 20AC onward. The light bulb is not on.
The amp with +/- 30V rail only has one pair of output transistors on a similar but different pcb layout, but the schematic is otherwise identical to post #318.
The amp with +/- 30V rail only has one pair of output transistors on a similar but different pcb layout, but the schematic is otherwise identical to post #318.
It's easily possible that this is a problem because you didn't put balancing resistors on your Q3-Q6 output transistors (~0R22 in series with emitters)...turn off one output pair and see if the problem disappears.
Update
I found the fault. I used a 7.5k for R7 since I could not find a 15k resistor. That laziness has done me in. A lot of current is being drained into the chip via the mute pin causing R7 to overheat over time. Because there is no smoke I wasn't paying attention to this pin while looking for other more probable places. I have checked with the voltage across R13 and R23. They are equal so spreading the load. So are R14 and R24. After the switch of R7 back to 15k, I now have 56V of juice running on each rail and the chip is only lurk warm after an hour. The final value of R9 is 510, R10 600, VR1 200, R16 2.2k. I also have a 100pF cap across the base and collector of Q1 for good measure. I must admit I don't know what D4 is for. I just found that arrangement in a lot of circuits. Guilty as charged. Thank you all for the suggestions.
I found the fault. I used a 7.5k for R7 since I could not find a 15k resistor. That laziness has done me in. A lot of current is being drained into the chip via the mute pin causing R7 to overheat over time. Because there is no smoke I wasn't paying attention to this pin while looking for other more probable places. I have checked with the voltage across R13 and R23. They are equal so spreading the load. So are R14 and R24. After the switch of R7 back to 15k, I now have 56V of juice running on each rail and the chip is only lurk warm after an hour. The final value of R9 is 510, R10 600, VR1 200, R16 2.2k. I also have a 100pF cap across the base and collector of Q1 for good measure. I must admit I don't know what D4 is for. I just found that arrangement in a lot of circuits. Guilty as charged. Thank you all for the suggestions.
To my knowledge, D4 is used to "protect" the single polarity e-cap. Avoid reversed bias, For safety. Electrolytic Capacitor >> Capacitor Guide
Old thread I know..... But here goes.
Over the last week or so I got to revisit this (compact) amp from years ago. Put in a new transformer, higher voltage supply and better PSU (cap-mx). Rails are now at 44v. I have two questions:
Bias measured across the two thermaltrak emitter resistors (0r22) is 20mV - which should equate to 22mA per side, the recommended amount for minimal distortion. The amp takes a while to come up to that bias, which was set with inputs shorted and outputs open after an hour or more and was stable for several hours in this config. After the amp has been playing for two or three hours into a fairly tough speaker load, the heatsinks are quite warm (55degC) and the thermaltraks measure at 62degC. Both sine and square waves are clean, measured power into 8r is just under 80w. I'm a little concerned that the heat is a bit high which must mean more current through the thermaltraks and particularly that I'm missing something obvious... But thought I would appeal to the wise heads here. I'm not at all used to this type of amp.
The second question is about whether to use a zobel network on the output (there isn't one on board). If I should, what circuit to use? 10r-100nf output to ground and small uH inductor with 10r in series with the output?
I think it's a shame these parts were discontinued, the sound is very good but I suppose there are newer components out now too.
Thanks in advance
Over the last week or so I got to revisit this (compact) amp from years ago. Put in a new transformer, higher voltage supply and better PSU (cap-mx). Rails are now at 44v. I have two questions:
Bias measured across the two thermaltrak emitter resistors (0r22) is 20mV - which should equate to 22mA per side, the recommended amount for minimal distortion. The amp takes a while to come up to that bias, which was set with inputs shorted and outputs open after an hour or more and was stable for several hours in this config. After the amp has been playing for two or three hours into a fairly tough speaker load, the heatsinks are quite warm (55degC) and the thermaltraks measure at 62degC. Both sine and square waves are clean, measured power into 8r is just under 80w. I'm a little concerned that the heat is a bit high which must mean more current through the thermaltraks and particularly that I'm missing something obvious... But thought I would appeal to the wise heads here. I'm not at all used to this type of amp.
The second question is about whether to use a zobel network on the output (there isn't one on board). If I should, what circuit to use? 10r-100nf output to ground and small uH inductor with 10r in series with the output?
I think it's a shame these parts were discontinued, the sound is very good but I suppose there are newer components out now too.
Thanks in advance
62C is not dangerous to output transistors but if you add more heatsink your transistors might last longer.
I=V/R so .02v/.22ohms=.09 A or 91 ma. That is a bit high for idle current. Try for 4.4 mv to get 20 ma.
Zobels prevent oscillation into capacitive loads. 10 ohm series .1 uf sounds normal. Dynaco mounted them on the back of the speaker banana jacks in the model I own.
The inductor keeps police/fire/CB/AM radio from coming in from the speaker wires and feeding back into the feedback transistor of the differential pair. 11-14 turns of 20 ga wire (for a 100 w amp) wound around a AA battery or china marker is normal. You parallel that with a 10 ohm wirewound resistor, then series the pair with the speaker hot jack. The 10 ohm resistor can be 3 watts on a 100 watt amp, more like 5 watts at double the wattage. These should go as close to the speaker jack as you can mount them. Longer wires from jack TO the inductor make little antennas.
These projects tend to assume you live in an ivory tower in neverneverland. I live on a street 4 miles from an AM radio station and there is a yellow truck that drives by that emits CB radio of dogs barking out the tune Dixie. Probably a bootleg 500 w CB radio. I like zobels & speaker inductors. I had to boost the input cap of one amp from 68 pf to 150 pf to get rid of the barking CB dogs.
Great experiment.
I=V/R so .02v/.22ohms=.09 A or 91 ma. That is a bit high for idle current. Try for 4.4 mv to get 20 ma.
Zobels prevent oscillation into capacitive loads. 10 ohm series .1 uf sounds normal. Dynaco mounted them on the back of the speaker banana jacks in the model I own.
The inductor keeps police/fire/CB/AM radio from coming in from the speaker wires and feeding back into the feedback transistor of the differential pair. 11-14 turns of 20 ga wire (for a 100 w amp) wound around a AA battery or china marker is normal. You parallel that with a 10 ohm wirewound resistor, then series the pair with the speaker hot jack. The 10 ohm resistor can be 3 watts on a 100 watt amp, more like 5 watts at double the wattage. These should go as close to the speaker jack as you can mount them. Longer wires from jack TO the inductor make little antennas.
These projects tend to assume you live in an ivory tower in neverneverland. I live on a street 4 miles from an AM radio station and there is a yellow truck that drives by that emits CB radio of dogs barking out the tune Dixie. Probably a bootleg 500 w CB radio. I like zobels & speaker inductors. I had to boost the input cap of one amp from 68 pf to 150 pf to get rid of the barking CB dogs.
Great experiment.
@indianajo - thank you so much for the information.
I should have been clearer - the 20mV is across two emitter resistors, so 20mV across 0r44 should give about 44mA total, or 22mA per thermaltrak transistor. I suppose what I'm really worried about is thermal runaway (once the sound is good!). It doesn't seem to continue to get hotter and hotter, and another good indication is that the power supply doesn't seem to be under pressure (its a cap-mx and the transistors there are holding steady at less than 50degC when the inside of the case is in the 30s). I'm going to do an extended test today, leave it running for hours with inputs shorted, outputs open and see does the bias drift. I need to read more examples on the thermaltrak as well, I am pretty ignorant of their typical use situations, others builders experience and so on.
I'll put together a little board with the zobel on it and fit it at the terminals of the amp.
I should have been clearer - the 20mV is across two emitter resistors, so 20mV across 0r44 should give about 44mA total, or 22mA per thermaltrak transistor. I suppose what I'm really worried about is thermal runaway (once the sound is good!). It doesn't seem to continue to get hotter and hotter, and another good indication is that the power supply doesn't seem to be under pressure (its a cap-mx and the transistors there are holding steady at less than 50degC when the inside of the case is in the 30s). I'm going to do an extended test today, leave it running for hours with inputs shorted, outputs open and see does the bias drift. I need to read more examples on the thermaltrak as well, I am pretty ignorant of their typical use situations, others builders experience and so on.
I'll put together a little board with the zobel on it and fit it at the terminals of the amp.
If you are measuring from the emitter of an npn push output transistor, to the emitter of a pnp pull output transistor, across 2 emitter resistors, no current should be going to the speaker. The output terminal should be 0 v to the speaker ground, or pretty close. In that case the idle bias current coming out of the npn goes into the pnp. There is no reason to halve the current you measure. A 20 mv you're running idle bias current of 45 ma. For 20 ma idle bias current you should measure 8.8 mv across the two emitter resistors. Lower idle bias current will reduce some of your heat.
It is really better to measure voltage of the two resistors of the push pull pair separately, in case output voltage is not zero at idle. Or measure the output voltage on speaker at near zero, then measure voltage across one emitter resistor.
It is really better to measure voltage of the two resistors of the push pull pair separately, in case output voltage is not zero at idle. Or measure the output voltage on speaker at near zero, then measure voltage across one emitter resistor.
I was just about to come back and post my findings from today when you replied @indianajo. I wanted to experiment a bit with bias, and look at stability of bias over time.
Panson set up the board with two test points to be able to measure the transistor bias relatively easily (see tst1 and 2 in the schematic attached). That is where I am measuring - which assumes that both halves are working equally.
I'm puzzled by setting the bias here. So I turn the pot down to zero ohms (as per Panson's past instructions here in the thread), and in this case measuring from the two test points, you get no voltage. Then bringing up the bias using the pot, you get to some level of bias - say for now that is 15mV. When I leave that settle, with the amplifier inside at room temperature, with the lid on the amp and the test leads connected etc, the bias slowly creeps up as the amp warms up. This happens pretty slowly, over a few hours at idle. So say after 2 hours or so, I've adjusted it so that its now sitting at 20mV. I bring the amp outside and it cools down pretty quickly to say 10degC. Bring the amp back inside, turn on, and measured bias is down at 0 to 1mV. After an hour this has risen to 8mV, and I suspect leaving it longer it will rise further.
That's a long winded way of saying that setting the bias seems incredibly sensitive to temperature within the amp, and on top of that, the warmer it gets, the more it rises. surely that was what thermatrak was meant to avoid?
I have the simple zobel inserted at the speaker terminals, but haven't listened to it yet.
I'm obviously missing something here!!
Panson set up the board with two test points to be able to measure the transistor bias relatively easily (see tst1 and 2 in the schematic attached). That is where I am measuring - which assumes that both halves are working equally.
I'm puzzled by setting the bias here. So I turn the pot down to zero ohms (as per Panson's past instructions here in the thread), and in this case measuring from the two test points, you get no voltage. Then bringing up the bias using the pot, you get to some level of bias - say for now that is 15mV. When I leave that settle, with the amplifier inside at room temperature, with the lid on the amp and the test leads connected etc, the bias slowly creeps up as the amp warms up. This happens pretty slowly, over a few hours at idle. So say after 2 hours or so, I've adjusted it so that its now sitting at 20mV. I bring the amp outside and it cools down pretty quickly to say 10degC. Bring the amp back inside, turn on, and measured bias is down at 0 to 1mV. After an hour this has risen to 8mV, and I suspect leaving it longer it will rise further.
That's a long winded way of saying that setting the bias seems incredibly sensitive to temperature within the amp, and on top of that, the warmer it gets, the more it rises. surely that was what thermatrak was meant to avoid?
I have the simple zobel inserted at the speaker terminals, but haven't listened to it yet.
I'm obviously missing something here!!
Both channels measure identically - one side is 11.4mV across the 0r22 emitter resistor, and the other is 9.1mV (measuring from test points to ground)- so one half is running a little harder than the other at either (41mA vs 51ma). Not sure how big an issue this is, maybe its a big deal, I don't know. Rail voltages are equal.
I thought I was doing something wrong but maybe not, anyway, I'll see how it works out for a while. Its not that I'm worried about 60degC, more that would it run away catastrophically if left unattended!
I thought I was doing something wrong but maybe not, anyway, I'll see how it works out for a while. Its not that I'm worried about 60degC, more that would it run away catastrophically if left unattended!
Still harping on about this...... like a dog with a bone I know.
So Panson originally left some an option on that board to add another transistor which I understand might be for active bias control (this is Q5 in the schematic above in post 333 above). Do you think that would give me more accurate control? On. the parts list (attached below) Q5, R12 and R22 are listed as optional for Vbe multiplier - any ideas as to what would be suitable here? Panson mentions in post 269:
But there is no mention of what R12 and R22 might be. Also - while I have a good number of common transistors, I'm pretty sure I don't have MJE15030 in the parts bin. What else could be used?
You can see I am well out of my depth! If anyone has any ideas, I'd be most grateful.
So Panson originally left some an option on that board to add another transistor which I understand might be for active bias control (this is Q5 in the schematic above in post 333 above). Do you think that would give me more accurate control? On. the parts list (attached below) Q5, R12 and R22 are listed as optional for Vbe multiplier - any ideas as to what would be suitable here? Panson mentions in post 269:
But there is no mention of what R12 and R22 might be. Also - while I have a good number of common transistors, I'm pretty sure I don't have MJE15030 in the parts bin. What else could be used?
You can see I am well out of my depth! If anyone has any ideas, I'd be most grateful.
There is not much voltage or current across a Vbe multlplier. Toughness of Vceo 150 of MJE15030 not necessary. I'm not downloading a datasheet of a transistor I can't buy like 2sc1815 but if it is silicon, npn, TO220 or TO202 package, Vceo >30 it should work. Resistors are determined experimentally if you don't want to run spice simulation. Mounts on the heat sink between the output transistors obviously. Use a mica or silicon heat insulator.
Thermaltrak should be better, quicker response to heat. I'd increase size of R21 R23 diode clamps first.
Thermaltrak should be better, quicker response to heat. I'd increase size of R21 R23 diode clamps first.