I should have written Christmas 1967 and a few months later not 1969.
Article scanned here: https://deramp.com/swtpc.com/PopularElectronics/Dec1967/PE_Dec1967.htm
Article scanned here: https://deramp.com/swtpc.com/PopularElectronics/Dec1967/PE_Dec1967.htm
Last edited:
Interesting video. Not sure why anyone would be surprised by poor performance at 20kHz using 2N3773. The original spec. was 200kHz fhfe, so it will not maintain whatever gain it had at 1kHz at 20kHz. The ON semi datasheet does not change that, although it is very likely to be built on a higher performance process today. The slew rate is best when modern high speed devices are used, as I have suggested on more than one occasion in this thread.
Of course the transistors might not be ON originals. Or ON might be using an older process for these devices. I would recommend at least trying MJ150xx types at 2MHz Ft. (MJ15003 etc). if people do not like the high speed types, but small correction capacitors can be used to fix HF oscillations.
Of course the transistors might not be ON originals. Or ON might be using an older process for these devices. I would recommend at least trying MJ150xx types at 2MHz Ft. (MJ15003 etc). if people do not like the high speed types, but small correction capacitors can be used to fix HF oscillations.
I can't find what/who you are referencing but the JLH circuit drives the OPs with an impedance about 2K, so ya, that's going to be slow. 100 Ohms in a typical AB amp is marginal for those old BJTs and Crown had it right with about 10 Ohms. Slow OPs generally make GFB less stable because they add a second "dominant pole", but real fast devices may be unstable due to wiring coupling. So "whatever works".
Greetings to the respected forum and all lovers of high-quality sound!
Problem with MJ15001G power transistors on JLH-1969 amplifier.
Faced with the problem of replacing power transistors on the JLH-1969 amplifier. Perhaps someone can help... The amplifier circuit is classic for an 8 Ohm load. (Power supply 27 Volts, quiescent current - about 1.2 A, the ratings of all elements are as in the classic circuit). To test the board, I installed Chinese counterfeit transistors MJ15003G. I tuned the amplifier and got very good results: the sine wave and square wave showed themselves well in the audio frequency spectrum. After that, I replaced the counterfeit Chinese transistors with original MJ15001G from OnSemi, adjusted the midpoint and quiescent current on a sine wave signal at a frequency of 1 kHz. Then I began to change the frequency and shape of the signal. The results were very bad: the waveforms (sine and meandr) started to distort literally from 3 kHz. At 7 kHz the waveform was just awful, and at 15 kHz the waveform just became sawtooth. In order to eliminate the possibility of an accident, I installed another pair of MJ15001G - the result was the same. There was a feeling that something was preventing the original transistors from entering the normal operating mode. But what? My input PNP transistor is BC559C - I did not touch it. I decided to replace the driver NPN transistor BD139-16 with a more powerful 2SC5707. I hoped that this would make the output power transistors work better, but this did not happen. I have read many times about directly replacing the popular 2N3055 with MJ15003, but I have never heard of such a problem as mine. The signal amplitude from the generator at the amplifier input was a maximum of 0.66 volts, as Hood recommended for an 8 ohm load. The amplifier power was almost 12 watts. When the input signal amplitude was reduced, the shape on the oscilloscope became a little better, but the power dropped significantly. I decided to install Chinese counterfeit transistors again and everything was fine again.
If anyone has encountered something similar or just knows what to do in such a situation, please let me know. How to make the original MJ15001G work normally?
Perhaps the problem is the high capacitance of the transistor collector? The original transistors that Hood himself worked with (Motorola MJ480) have a capacitance of 200 pF, the popular 2N3055 transistor has a capacitance of about 600 pF, and the MJ15003/1G has a capacitance of 1000 pF. Perhaps because of this some elements of the circuit (resistors and capacitors) require changing the ratings?
P.S. MJ15001G is a "light version" of MJ15003G: the only difference is in the maximum collector current: 15 A for MJ15001G (power 200 watts) and 20 A for MJ15003G (power 250 watts). I have no doubts about the originality of the transistors - I bought them on the Mouser website: https://eu.mouser.com/ProductDetail/onsemi/MJ15001G?qs=HVbQlW5zcXWBmwlm1eiLzQ== Several pieces were bought. I matched the transistors in pairs by hFe on a power bench. (Voltage 24V, current 1.5A, load power resistor 8 Ohm, 25W). I got very good results for the current gain. The average hFe was about 195, and the spread was small: from 189 to 204.
P.P.S. Sorry for my English.
Problem with MJ15001G power transistors on JLH-1969 amplifier.
Faced with the problem of replacing power transistors on the JLH-1969 amplifier. Perhaps someone can help... The amplifier circuit is classic for an 8 Ohm load. (Power supply 27 Volts, quiescent current - about 1.2 A, the ratings of all elements are as in the classic circuit). To test the board, I installed Chinese counterfeit transistors MJ15003G. I tuned the amplifier and got very good results: the sine wave and square wave showed themselves well in the audio frequency spectrum. After that, I replaced the counterfeit Chinese transistors with original MJ15001G from OnSemi, adjusted the midpoint and quiescent current on a sine wave signal at a frequency of 1 kHz. Then I began to change the frequency and shape of the signal. The results were very bad: the waveforms (sine and meandr) started to distort literally from 3 kHz. At 7 kHz the waveform was just awful, and at 15 kHz the waveform just became sawtooth. In order to eliminate the possibility of an accident, I installed another pair of MJ15001G - the result was the same. There was a feeling that something was preventing the original transistors from entering the normal operating mode. But what? My input PNP transistor is BC559C - I did not touch it. I decided to replace the driver NPN transistor BD139-16 with a more powerful 2SC5707. I hoped that this would make the output power transistors work better, but this did not happen. I have read many times about directly replacing the popular 2N3055 with MJ15003, but I have never heard of such a problem as mine. The signal amplitude from the generator at the amplifier input was a maximum of 0.66 volts, as Hood recommended for an 8 ohm load. The amplifier power was almost 12 watts. When the input signal amplitude was reduced, the shape on the oscilloscope became a little better, but the power dropped significantly. I decided to install Chinese counterfeit transistors again and everything was fine again.
If anyone has encountered something similar or just knows what to do in such a situation, please let me know. How to make the original MJ15001G work normally?
Perhaps the problem is the high capacitance of the transistor collector? The original transistors that Hood himself worked with (Motorola MJ480) have a capacitance of 200 pF, the popular 2N3055 transistor has a capacitance of about 600 pF, and the MJ15003/1G has a capacitance of 1000 pF. Perhaps because of this some elements of the circuit (resistors and capacitors) require changing the ratings?
P.S. MJ15001G is a "light version" of MJ15003G: the only difference is in the maximum collector current: 15 A for MJ15001G (power 200 watts) and 20 A for MJ15003G (power 250 watts). I have no doubts about the originality of the transistors - I bought them on the Mouser website: https://eu.mouser.com/ProductDetail/onsemi/MJ15001G?qs=HVbQlW5zcXWBmwlm1eiLzQ== Several pieces were bought. I matched the transistors in pairs by hFe on a power bench. (Voltage 24V, current 1.5A, load power resistor 8 Ohm, 25W). I got very good results for the current gain. The average hFe was about 195, and the spread was small: from 189 to 204.
P.P.S. Sorry for my English.
This doesn't make sense to me either. Differences in junction capacitance should have zero effect in a design like this but I can't just explain what might be going on.
Are you seeing this distortion with and/or without a load?
You have scoped the output to see this issue but have you also put the scope on the 27 volt rail to see if anything is going on there?
Are you seeing this distortion with and/or without a load?
You have scoped the output to see this issue but have you also put the scope on the 27 volt rail to see if anything is going on there?
Yes, of course, in both cases the load was a 25 Watt, 8 Ohm power resistor and the oscilloscope was connected to its terminals.
With a 27 Volt supply, everything was fine in both cases.
In both cases I monitored the input signal shape: with the fake MJ15003 the output shape followed the input signal, while with the original MJ15001G transistors the output shape was bad, although the input signal remained good.
That is, both types of transistors worked under the same conditions. The counterfeit transistors showed good results, and the original ones - bad. Of course, after the replacement I adjusted the settings, but this did not lead to noticeable changes - the result of working with MJ15001G was terrible. Why is this so - I do not know ...
With a 27 Volt supply, everything was fine in both cases.
In both cases I monitored the input signal shape: with the fake MJ15003 the output shape followed the input signal, while with the original MJ15001G transistors the output shape was bad, although the input signal remained good.
That is, both types of transistors worked under the same conditions. The counterfeit transistors showed good results, and the original ones - bad. Of course, after the replacement I adjusted the settings, but this did not lead to noticeable changes - the result of working with MJ15001G was terrible. Why is this so - I do not know ...
Last edited:
AFAIK the JLH 1996 version has symmetrical power supplies and does not need a capacitor but you included one that introduced an issue. This is a consequence of adding superfluous parts. The capacitor will charge when there is a voltage and a loudspeaker is connected. The charging goes of course via the capacitor and the loudspeaker. Depending on the level of that DC voltage sometimes pretty high currents may flow. DC current. Through the loudspeakers. Now assume (worst case) that to be either one of the full supply voltages. Calculate.
So if you really want to keep that cap and don't want the thump you could include a loudspeaker relay and have the NC contact connected to GND with let's say a 22 Ohm resistor. The cap will charge via the resistor at power on and when the relay switches to loudspeaker after a few seconds there will be silence.
Power on thumps are a nuisance and a design imperfection that is sometimes accepted. Completely unnecessary. Now we are audio enthusiasts that can put much time in debating tiny details and we hear everything that is detrimental so that relay of course is suspect. If one likes to think about relay drawbacks and prefers to have DC current through the precious loudspeakers look no further. If not then search for relays with silver based contacts or ones that have both tungsten contacts AND silver contacts. Or use an extra switch with gold contacts labelled "Speaker direct" to bridge the relays contacts 🙂
Or remove the cap.
Last edited:
It would probably be logical if I posted here the oscillograms of the problematic transistors. Perhaps this will help identify the problem.
I didn't save the oscillograms of the fake transistors, but there were very good oscillograms up to 50 kHz. (Both sine and meandr).
Last edited:
That is very strange indeed and I have absolutely no idea at a distance what is going on.
For curiosity I tried adding some base/collector capacitance and tbh it is more susceptible to this than I would have thought given the high currents floating around in all the stages.
This is 50kHz and 8000pF added. I don't know what to say beyond suggesting you try something different for the outputs. They are the only variable you are swapping:
For curiosity I tried adding some base/collector capacitance and tbh it is more susceptible to this than I would have thought given the high currents floating around in all the stages.
This is 50kHz and 8000pF added. I don't know what to say beyond suggesting you try something different for the outputs. They are the only variable you are swapping:
One other thing.
Can you check the values of R1 and R2 in your build of the amp. If one is a preset what value is it set to? The value of the two together will give an idea of the current flow in the voltage amplifier stage. High gain outputs need less base current and increasing the value of these resistors will limit the HF performance of the amp.
Can you check the values of R1 and R2 in your build of the amp. If one is a preset what value is it set to? The value of the two together will give an idea of the current flow in the voltage amplifier stage. High gain outputs need less base current and increasing the value of these resistors will limit the HF performance of the amp.
My question would be: Why you felt the need to replace components when you did not have a problem in the first place? Just the fact that Mooly had to add so much base collector capacitance to simulate that problem is a sure defect in the transistors you have replaced the originals with. I always (and others) Say, if something does not itch, don't scratch it.
Comparing data sheets (On Semi) of the original transistors and the replaced ones show the GBP to only be 2MHz while the original is 30Mhz.
Comparing data sheets (On Semi) of the original transistors and the replaced ones show the GBP to only be 2MHz while the original is 30Mhz.
R1 is a constant 100 Ohm. R2 is a variable 2 kOhm (currently - about 600 Ohm). That is, everything is according to the original scheme. But there is one nuance that I remembered: resistor R1 heats up very much. I installed it with a power of 3 Watts and it is still quite hot. This is not the first time I have assembled the JLH-1969 amplifier, but this is the first time I have encountered such a large heating - usually 0.6 Watts is enough.
Even the original MJ480 from Motorola has FT=4MHz twice that of your newly acquired transistors. I think that is where your problem lies. Put the old ones back and be happy.
While Mooly's simulation adds 8nF to the OPs, he used 30MHz MJE15032 and not 2Mhz MJ15003. Note that with MJ15003, the upper resistor (~R1) should be about 1K.
Thanks.
In the simulations those values only give around 600ma output stage current and around 18ma in the resistor chain with around 10ma going to TR3, the voltage amplifier stage.
The assumed voltages tell us that the midpoint voltage if correct (say 13 volts give or take) will mean a base voltage on the upper output of lets say 14 for easy reckoning. Two series resistors, 100 ohm and 600 ohm are dropping 27-14 which is 13 volts. The 100 ohm should only be seeing a couple of volts at most across it and the dissipation should be milliwatts.
So something is wrong somewhere. Its worth checking the actual voltages and also make sure R1 reads 100 ohm although even if it were say a 1k by mistake it would still only be milliwatts dissipation.
All that needs checking.
The weird thing though is that zillions of JLH's have been built using every type of transistor you can think of and this issue has never been seen (as far as we know) before. It is odd.
Thanks for that 🙂 I did have play with altering the transistor models to simulate higher capacitance but its all a bit inconclusive.
Original values:
And just playing. I've no experience altering actual models or creating them. Are we in nano farads here xxE-9 and all that: