Modifications to the JLH 15-20W Class AB amplifier (1970)

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Member
Joined 2010
Paid Member
When I ran the simulation the maximum unclipped output voltage into 4R test load was 25 peak.

The static current drawn by the output stage was 250 m.A. per pair while that drawn by the driver stage was 120 m.A - so things are warming up.

Transistor current gain in transistor datasheets is specified at room temperature 25 degrees C but it increases in proportion to degrees K where your 25 degrees translates to 298.15.

The output transistors are likely to get warm in which case you should be able to reduce the standing current of the driver stage a little.

MJL3281 and MJL1302 have higher current gain than MJL21194/21193 and for the same purpose these are a viable alternative.

There is quite a lot of diffusion capacitance in the base-emitter junction of a transistor which has some low pass filtering implications.

I would use lower value stopper resistors and try to run the output stage at low dissipation as well as that you have so designed.

Yes, you are correct, the total dissipation per channel with no signal to the amplifier is about 28 Watt.
I think I will decrease the output bias current to 150 mA or lower per pair and also the supply from +/- 28 V to +/- 24Vdc.

Your proposal for MJL3281 and MJL1302 is also very good. I will check with these transistor and probably I will use them.

I will decrease also the the base resistors to 4.7 Ω.

regards
George
 
Member
Joined 2010
Paid Member
Only two real issues found.
4. I messed with the bootstrap resistors. I can't think of a good reason to portion them the way they were. The main difference is that the bootstrap cap doesn't need to be so big.

The portion of R5 and R7 adjusts the open loop gain and the open loop bandwidth of the amplifier.

I think I will reverse the portion and I will use R5=5.6 k and R7= 1.5k
Doing so, the open loop gain drops from 81 dB to 76 dB but the open loop bandwith increases from 3.4kHz to 6 kHz.
The distortion is not very different, but the strange thing here is that with the lower open loop gain the distortion is slightly better.
From 0.015% drops to 0.012% when the output is +/- 22V at 4 Ω.
 
Can you explain what you mean by "Negative rail sticking". I am not sure that I understand what you mean.

Yes, you are correct for the Vbe multiplier. I will check it.

regards
George

Do the simulation with the amp driven into clipping and look closely at the trailing waveform after clipping. This is a common issue because the early stages of an amp go into overdrive when the output fails to follow and there is no feedback. Then when the input returns within range, there is a delay while the input stage comes back into normal range.
This is usually the VAS driven into saturation/ over-current. The standard solution is a (~Schottky) diode across the VAS collector-base that prevents it from saturating. But the wrong diode can cause some distortion due to leakage. This is why "Schottky TTL" logic gates were faster and used less power than standard TTL logic.
Normally rail sticking only shows up at 10KHz where the recovery delay time is significant, but this is bad enough to be seen at 1KHz.
There are lots of stuff on the subject if you do a little searching:
http://hifisonix.com/wordpress/wp-content/uploads/2019/02/Anti-Saturation-Diodes.pdf
Occasional clipped peaks are hardly noticeable on an amp that has no rail sticking.
 
The portion of R5 and R7 adjusts the open loop gain and the open loop bandwidth of the amplifier.

An ideal bootstrap circuit presents a current source (~infinite Z) load on the VAS. But of course the EF has some loss so the actual VAS load is the lower resistor plus the upper resistor times 1/(1-EFgain), note: 1/(1-1)~=infinity. The base of the driver transistor may be the most significant part of the load, so the bootstrap resistors are actually about these things:
1. Setting the VAS idle current.
2. A 50/50 split allows the use of the smallest bootstrap capacitor for a given low frequency limit.
3. In the case of positive rail sticking (assuming a positive bootstrap) , the lower resistor portion can be reduced to limit the clipping overdrive, ie reduce the bootstrap peak voltage. This becomes important with MOSFET amps because the MOS gate does not clamp the VAS like a BJT.
 
Member
Joined 2010
Paid Member
Here is a new update of the amplifier:

I changed the output transistors to MJL3281/MJL1302 pair, I reduced the resistors of the Vbe multiplier, I reduced the base resistors of the output transistors to 4.7 Ω, and I reduced the supply to +/-24V.

What I found is that by replacing the VAS transistor to a 2N5551 type, no negative rail sticking is shown at 10 kHz and 3 V peak input. The peak current in the VAS transistor is about 19mA in this overload condition. So maybe a current limiter is not necessary.
What do you think?

Here are the updated schematic and the LTspice file.
 

Attachments

  • JLH ClassAB mod.22.09.2019.asc
    11 KB · Views: 47
  • JLH.png
    JLH.png
    86.6 KB · Views: 241
Here is a new update of the amplifier:

I changed the output transistors to MJL3281/MJL1302 pair, I reduced the resistors of the Vbe multiplier, I reduced the base resistors of the output transistors to 4.7 Ω, and I reduced the supply to +/-24V.

What I found is that by replacing the VAS transistor to a 2N5551 type, no negative rail sticking is shown at 10 kHz and 3 V peak input. The peak current in the VAS transistor is about 19mA in this overload condition. So maybe a current limiter is not necessary.
What do you think?

Here are the updated schematic and the LTspice file.

It seems to work fine.
1. The VBE base current is still 50% of the divider current. I would have reduced the resistors more but this may make use the the increased beta at high temperatures for more temperature compensation.
2. The 22 Ohm VAS emitter resistor reduces open loop gain but the THD seems to be fine anyway.
3. The bootstrap cap can now be a lot smaller, say 220uF.
4. The bias levels are still pretty high, but if you are comparing it to class-A and you have the heat sink then it's fine.
 
Cooler operating options

Yes, you are correct, the total dissipation per channel with no signal to the amplifier is about 28 Watt.
I think I will decrease the output bias current to 150 mA or lower per pair and also the supply from +/- 28 V to +/- 24Vdc.

Your proposal for MJL3281 and MJL1302 is also very good. I will check with these transistor and probably I will use them.

I will decrease also the the base resistors to 4.7 Ω.

regards
George

Good that you have gone with those devices and lowered the supply rails.

I assume you will incorporate supply rail fuse protection in your build. I have concerns for output stage survival prospects if the output terminals are shorted.

You can expect low distortion results with resistive loads down to 4R with peak output voltages of 18 V (9V with 2R and half the input voltage).

That said without the need for any heating, the current gain of the substitute MJL3281/1302 pairs allows a reduction in the driver current.

R8 could be increased in value from 10R to anywhere up to say 22R reducing emitter Q4 current down to as little as 57 mA.

JLH intended his amplifier to be Class AB with some output in the Class A region up to some mW level.

He determined this was possible with 200 m.A. of output standing current. On that basis you can run your output sets at 100 m.A. a pair.

With 0.22R emitter resistors this standing current level would also align with a paper by Oliver.

Thermal stability is something to think about.
 
Member
Joined 2010
Paid Member
Thank you both, steveu, mjona, for your comments and suggestions.

Here is another update of the amplifier with the following changes:

- I decreased the current of Q6,
- I reduced the capacitors C3 and C4,
- I included Q11 to protect the VAS transistor in case of hard overload
- I changed the bias circuit of the input transistor (with Q10) to reduce the output offset variation due to temperature changes.

I think it is a nice amplifier with very low distortion, mainly second and less third harmonic.

I will proceed to design a nice PCB for this amplifier.

Regards
George
 

Attachments

  • JLH Class AB Ver.24.9.2019.png
    JLH Class AB Ver.24.9.2019.png
    94.9 KB · Views: 188
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.