JLH 10 Watt class A amplifier

The only way out would be for someone to extract the points made into a summary document.
Not that is really necessary - just reading the original article and follow-ups covers most of them.
Here (see attachment) is the most comfortable use of this topology with measurements.
They write that someone there set it up for a long time to produce this circuit - in fact, simple mathematics, the next cascade limits the frequency to the root of 2 from the previous cascade.
1 input filter and 2 gain stages
Class A, efficiency - 30%, other characteristics were measured on my version with an output power of 9 watts, everything is shown in the diagram.
at an output power of 40 watts distortion (1 kHz) 1%.
 

Attachments

  • JLH 2003  Hennady DIY.jpg
    JLH 2003 Hennady DIY.jpg
    87.9 KB · Views: 221
  • Like
Reactions: 1 user
Hi.I need them for 4 ohms speakers.So lowering the voltage and increasing the current to 1,7A each transistor would help IMO
for example, we lower the voltage to +/-18 volts, the current of the output transistors is 1.7A per pair, the total current will be 3.4A
It will be necessary to reduce the sensitivity of the circuit; to do this, increase resistor R4 from 220 ohms to 390 ohms.
The maximum power will drop to 28 watts into a 4 ohm load.
Otherwise, everything is the same, except perhaps for the distortion spectrum, which the closer it is to 2/3 of the maximum power, the more odd distortions will appear in the spectrum.
In the version with an 8 ohm load on the circuit of post # 9586, the even harmonics dominate almost up to maximum power.
 
  • Like
Reactions: 1 user
Thank you for the information. So if build for 4 ohm speakers distortion is mostly odd harmonics( I think I prefer second harmonic because it is more natural to ear )
4 ohm speakers are problemic for amplifiers. Most good amps are designed for 8 ohms. Like Jlh , aleph j etc. Maybe it is better to buy 8 ohm speakers.
 
Thank you for the information. So if build for 4 ohm speakers distortion is mostly odd harmonics( I think I prefer second harmonic because it is more natural to ear )
output power close to maximum forces the current through the upper and lower transistors to modulate slightly, this leads to the appearance of odd harmonics. With an 8 ohm load, the maximum power threshold is higher and the amplifier operates in Class A linear mode over a larger range of volume levels for the listener.
4 ohm speakers are problemic for amplifiers.
Such acoustics require powerful amplifiers with low output impedance. It's not for nothing that amplifier manufacturers write the recommended load impedance.
Most good amps are designed for 8 ohms. Like Jlh , aleph j etc.
These amplifiers will work well as part of an audio complex, where a separate powerful amplifier operating at frequencies below 120 Hz will be added for the woofer or subwoofer.
Maybe it is better to buy 8 ohm speakers.
High quality reference acoustics, usually with an impedance of 8 ohms.
High-quality acoustics with an impedance of 4 ohms can be very demanding on the power parameters of the amplifier.
 
The problem with increasing R4 from 220 to 390 ohms is somewhat self-defeating. The open loop gain will be reduced, so distortion will not be reduced as much as might have been expected.
For better results with 4 ohm loads, R4 needs to be reduced, and the feedback resistor reduced too. But preferably the input stage transistor current needs to be increased as well. That could be arranged if an additional transistor is added to make a current mirror before the VAS/driver stage.
You might also consider using linear gain transistors in the output. The MJ15024 suggested by Hennady is an option with a pair for 4 ohms even running at a lower voltage if you do not need 40W. 2SC5200 and MJL3281A have been used successfully, but at 8 ohms in my tests.
One of the problems I have pointed out before is that in many of the older transistors the current supplied by the output transistor which is conducting more is falling in gain. So the driver transistor current has to conduct enough to ensure that it can supply the required current at peak output. The transistor being turned off may not turn off as much, so the current division becomes asymmetrical and second harmonic distortion is less cancelled. That is the problem for lower impedance speakers.
While it is better to aim for 8 ohm speakers it is worth noting that so-called 8 ohm speakers may have a lower impedance at some frequencies.
 
For better results with 4 ohm loads, R4 needs to be reduced, and the feedback resistor reduced too.
this contradicts the optimal DC mode first cascade, which is equal to R3/R5= 8.2/2.7=3, it will also increase the gain of the open loop of this stage, which will require additional frequency compensation, because unbalances the correspondence of the cutoff frequencies of the cascades F2=F1* Root of 2
 
One of the problems I have pointed out before is that in many of the older transistors the current supplied by the output transistor which is conducting more is falling in gain. So the driver transistor current has to conduct enough to ensure that it can supply the required current at peak output. The transistor being turned off may not turn off as much, so the current division becomes asymmetrical and second harmonic distortion is less cancelled. That is the problem for lower impedance speakers.
While it is better to aim for 8 ohm speakers it is worth noting that so-called 8 ohm speakers may have a lower impedance at some frequencies.
The input impedance of a single speaker is purely active at only two frequencies: mechanical and electromechanical resonances. At all other frequencies, the total resistance has a reactive component, and its module near the main resonance and at the upper limit of the operating range can be several times greater than the active resistance.
Let's calculate the required output current of the amplifier for the peak value of the harmonic signal, which depends on the output power of the amplifier. Therefore, for the active resistance of an acoustic speaker, the current with an effective output power of 26 watts(effective = 2/3 max Pout) and an active resistance of 8 ohms will be the square root of (2 * 26/8) = 2.55 A. For the complex impedance (~4*Ract - this is 12dB of frequency response unevenness) of the speaker system, the peak current will be the square root of (75*26/32) = 7.8 A!!!
This is the reason for the appearance of nonlinear distortions. and precisely the odd harmonics, because with such a peak current the shoulders, top and bottom, begin to modulate as in a quasi push-pull mode...
In a push-pull amplifier at complex load, short-term clipping occurs, which is not even audible at high volumes.
 
Last edited:
  • Like
Reactions: 1 user
this contradicts the optimal DC mode first cascade, which is equal to R3/R5= 8.2/2.7=3, it will also increase the gain of the open loop of this stage, which will require additional frequency compensation, because unbalances the correspondence of the cutoff frequencies of the cascades F2=F1* Root of 2
That is why I suggested that a current mirror is added, as a higher current in the input transistor would have needed a lower resistor than 8.2k with increased nonlinearity on the second stage.
I agree that the effect will be to put the open loop gain up higher, which is to bring the distortion down. It may well need some modification to the frequency compensation as you say. The point being whether the overall closed loop distortion can be reduced particularly for heavier loads, which simulations suggest it can be.
Regarding your loudspeaker calculation, if a speaker has an impedance 4 times the nominal impedance then the output current will be surely approximately quarter of the nominal output current (if I understand you suggesting z=8+32j) for a given output voltage. The current will be out of phase, which makes it more of a challenge for the amplifier.
 
  • Like
Reactions: 1 user
The point being whether the overall closed loop distortion can be reduced particularly for heavier loads, which simulations suggest it can be.

you are talking about the overall coefficient distorcion of the amplifier, I am trying to convey to you that you need to understand where the node is where the distortion is maximum and modify it.
This is the node - is the base of the output transistors.
Regarding your loudspeaker calculation, if a speaker has an impedance 4 times the nominal impedance then the output current will be surely approximately quarter of the nominal output current (if I understand you suggesting z=8+32j) for a given output voltage. The current will be out of phase, which makes it more of a challenge for the amplifier.
Yes, this is a calculation for the inductive nature of the load for “pink noise”, this is only needed to calculate the maximum current of the output transistors, and since this current for a real signal is 4 times less, there is nothing to worry about, because this current has the opposite sign, so part of it will close the Zobel circuit and part will go through the output transistor and close to the supply voltage bus without a blocking capacitor.
As for the capacitive nature of the load, the current is not subtracted but added to the current through the output transistors, given that in a wideband speaker it is located near the resonance frequency and has a small segment, it only affects the transient response of the amplifier, which is more important for amplifiers operating in the class AB.
 
Last edited:
  • Like
Reactions: 1 user
Yesterday my jlh mono arrived, and I have a few questions.
I just bought them without asking any questions, they were that cheap I couldn't pass.
Seller said it was jlh 1969, which I believe, because its single rail, and has output capacitor, the home etched pcb also reads jlh class a, but this one has 4 to3, never seen that , current sharing perhaps?, I will not dissect them just yet, but slowly power them up and listen to them.
Ps I bought 20 Motorola mj15003 because I saw the used ones aren't the same some toshiba, some rca
20240314_180350.jpg
20240314_181255.jpg
20240314_180027.jpg
20240314_180058.jpg