I have built the original circuit with split power supply and the DC offset at the output is very stable and I use the amplifier every day for some years now.
The emitter resistor is 0.47Ω in my new modified circuit. Maybe I have to decrease this to 0.22 Ω. I will check that.
Nice idea! One green LED is enough.
I run the simulation with only one LED and decreasing the resistor to set the same current. This gave only 0.2 more Volt at the negative side.
But I will keep your proposal.
I have also built a lower power version of the JLH 1996 Class A amplifier and the only problem is than runs very-very hot. Especially here in Greece it is not possible to operate it during the hot days of summer.
So I was looking for another solution and I built the original JLH Class AB (1970 circuit). It runs without much hot and I cannot hear any difference with the Class A (1996) circuit.
regards
George
As I mentioned in post #1, I used split power supply even for the original circuit of JLH 1970 amplifier.
Also I want to keep the original amplifier, and I want to built a new one with the suggested modifications.
Yes, I agree it is a good idea to decrease the voltage of the power supply, for a 4 Ω load. But I have found that the maximum peak voltage that I need for my loudspeaker is about 25-26 Volt peak during loud passages of music. So that is why I use a voltage of +/- 34 Volt and accept the greater heat consumption.
I will keep only one LED for the current source.
thanks
George
I will design a PCB when the amplifier design is finalised. But this it will take some time I think!
I will post also the LTspice files later.
You can read the original article of JLH here :
https://americanradiohistory.com/Archive-Wireless-World/70s/Wireless-World-1970-07.pdf
See page 321 of the magazine.
I see some similarities but the Quad is much more complicated.
regards
George
Another option is to add an pair or large power transistors to the existing circuit. 25 years ago, I did this to a Sansui G-4700 receiver that I was given because it had blown up. You change the existing emitter resistors to about 2 Ohms and drive the added transistors (c/w heat sink) from there. I used a 2n3773's and 2.2 Ohms. Note this was quasi addition to a complementary circuit so the lower 2n3773 was driven from the PNP collector. Depending on the bias, this gives only fair THD but indestructible output. My old receiver has since been shorted and abused with no failure. The idea is a bit like current dumping. It would be a great way to handle low-Z loads.
It turns out that D1 is a big part of the 4 Ohm problem, ie it is a current limit that kicks in even before the PNP runs out of gain. But you still need more gain on the PNP side so the best solution is to add a pnp driver and replace your big PNP with a NPN, and the emitter resistors need to be reduced.
I should note that using the big PNP as a driver with a quasi NPN added like this leads to stability problems because the big PNP is slow. Drivers need to be fast. You could also drive the big PNP with a pnp driver but then the bias needs to change etc. Lots of possibilities but this is probably the best bang-for-the-buck.
And the bias should be changed to a VBE multiplier but I didn't show that here.
And I didn't fine tune the current limit. This is related to the emitter resistor values.
Last edited:
How you came to this conclusion?
If I remove completely the zener D1 from the circuit, the simulation shows that the maximum voltage at the negative side does not increase in level.
Yes.
As JLH states at the article, the 1970 class AB design was a try to achieve the good listening performance of the 1969 class A design. And according to his words from the 1970 article:
"However, the most careful comparative listening trials, with several of the author's long suffering friends, have failed to uncover any audible difference between these two designs, both of which will almost certainly surpass in performance the best available valve-operated, transformer-coupled units."
And the big advantage of the 1970 design is that it does so without the heat.
regards
George
Jan, it's always a pleasure to read kind words from you! Thank you!
Actually, I was thinking to use this amplifier to drive some of the channels of my AXL active loudspeaker.
What do you think about this? Any suggestions ? I appreciate your knowledge about amplifiers.
regards
George
I appreciate this is has not been subject to a wide battery of tests.
I have tried a single THD test which shows the maximum peak output voltage to be at little below 16 Volts into 4R with 20kHz sine wave input. The THD is about 0.05% provided C11 and R18 are left out -if not the figure rises to 2.6%
without D1
My guess is that you didn't change the load to 4 Ohms? Attached is two simulations with and without D1 into 4 Ohms.
My guess is that you didn't change the load to 4 Ohms? Attached is two simulations with and without D1 into 4 Ohms.
Attachments
The piece you have quoted attracted my interest too. One might expect if such is the case then the FFT of the Class AB design would resemble that for the Class A.
The closest I have been able to get to this is with any simulation is for the second to fourth harmonics to decline in a uniform pattern.
For this result the output Iq has somewhat of a sweet spot which is more in line with the expectations of some contemporary 1970 critics at around 50 m.A. At that level the outputs should run cool.
According to my simulation a minor shift in output Iq level can cause odd order harmonics to dominate.
On changes made the output is direct coupled as in the 1996 Class A design. I replaced the bootstrap load on the voltage gain amplifier which runs at 2 m.A. in favour of a constant current source running 6 m.A.
The aim there is to increase the gm enough to remove the 47R collector load on the driver stage which is otherwise outside the compensation stability loop - Also 2m.A. seems too low to exclude the possibility of Early Effect. Increasing the current gives lower THD but in this instance the harmonic structure suffers thus 6m.A. seems to be another CCS sweet spot.
The driver stage based on MJE340 runs in Class A around a median current of 50 m.A. while there are faster transistors around this one looks good enough for speed and the power rating is good.
When writing about a headphones Class A amplifier published in HiFi News & Record Review based on the Simple Class A amplifier in January 1979 JLH outlined his preferences.
This article is an interesting read however among these preferences were the absence of distortion residues of significant amount of high order 7th, 9th, 11th etc harmonics and the absence of inter-modulation products due to these.
To recast JLH's words a little he thought that 0.1 % THD at 20kHz was for the the impressionable as he knew no-one who could hear the second harmonic let alone the third.
The 1970 Class AB design was early First Watt thinking so I have copied the FFT from my simulation at 4 Volts output into 8R to see today what people may think.
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.