JHL Zener Diodes
Kroto,
Good idea.
I have 3 x 1N4148 in the right channel and 4 x 1N4148 in the left channel and I can't hear any difference, so maybe it really doesn't matter !
It would be interesting to know how the 'JHL' pcb designers got from John Linsley Hood's 2 unspecified diodes D1 and D2 in series to a 2.7V Zener diode.
See attachment :
Mark
Kroto,
Good idea.
I have 3 x 1N4148 in the right channel and 4 x 1N4148 in the left channel and I can't hear any difference, so maybe it really doesn't matter !
It would be interesting to know how the 'JHL' pcb designers got from John Linsley Hood's 2 unspecified diodes D1 and D2 in series to a 2.7V Zener diode.
See attachment :
Mark
I'm not sure that the voltage value at the bottom of the resistor is that important.
What is important is that the voltage varies less than any variation that will happen on the -ve supply rail.
The two diodes D1 & D2 act like a shunt regulator to give a near constant -ve voltage at the bottom of R1.
The current flows from Q4 emitter > Q4 base > through R2 (4k7) > through R1 (47k) > to regulated voltage ~-1.3Volts to -2.7Volts.
After Q4 warms up to operating temperature, we have a near constant quiescent current flow and you choose resistor values to minimise the input offset voltage. This current flow will vary if the output offset varies. The output offset should be pretty low for this dual polarity supply. The voltage at Q4 base will be -ve and sit >0.6V below the output voltage.
This means the voltage across R1+R2 = regulated Voltage = Base voltage.
With only two diodes, the V(r1+r2) is <0.6V
increasing to three diodes increase this to <1.2V and substituting a 2.7V Zener increases this still further to <2.1V
You can see that the current variation for the 0.6V diff will be much larger than for a 1.2Vdiff or 2.1V diff.
It could be that output offset variation improves as the regulated voltage is increased.
Diodes, or LEDs, or Zeners could all be used.
Worth experimenting to see if any combination is better or worse than others.
What is important is that the voltage varies less than any variation that will happen on the -ve supply rail.
The two diodes D1 & D2 act like a shunt regulator to give a near constant -ve voltage at the bottom of R1.
The current flows from Q4 emitter > Q4 base > through R2 (4k7) > through R1 (47k) > to regulated voltage ~-1.3Volts to -2.7Volts.
After Q4 warms up to operating temperature, we have a near constant quiescent current flow and you choose resistor values to minimise the input offset voltage. This current flow will vary if the output offset varies. The output offset should be pretty low for this dual polarity supply. The voltage at Q4 base will be -ve and sit >0.6V below the output voltage.
This means the voltage across R1+R2 = regulated Voltage = Base voltage.
With only two diodes, the V(r1+r2) is <0.6V
increasing to three diodes increase this to <1.2V and substituting a 2.7V Zener increases this still further to <2.1V
You can see that the current variation for the 0.6V diff will be much larger than for a 1.2Vdiff or 2.1V diff.
It could be that output offset variation improves as the regulated voltage is increased.
Diodes, or LEDs, or Zeners could all be used.
Worth experimenting to see if any combination is better or worse than others.
Last edited:
JHL Diodes
Andrew,
Thanks for the detailed response. If I understand you correctly the diodes are there to reduce the variation in the voltage to Q4, and that the voltage variation will reduce as you increase the voltage drop across the diodes i.e. more diodes = more voltage drop = less current variation to Q4 and potentially less (DC ?) output offset. Have I understood you correctly ?
If this is correct then I can presumably measure the effect of adding more/different diodes on the output offset by checking the DC voltage between the output and the star earth. However, how/where would I measure the effect of adding more/different diodes on the variation in voltage using a multimeter ?
Regards
Mark
Andrew,
Thanks for the detailed response. If I understand you correctly the diodes are there to reduce the variation in the voltage to Q4, and that the voltage variation will reduce as you increase the voltage drop across the diodes i.e. more diodes = more voltage drop = less current variation to Q4 and potentially less (DC ?) output offset. Have I understood you correctly ?
If this is correct then I can presumably measure the effect of adding more/different diodes on the output offset by checking the DC voltage between the output and the star earth. However, how/where would I measure the effect of adding more/different diodes on the variation in voltage using a multimeter ?
Regards
Mark
Yes, I am fairly sure you will find this to be true.
You can measure output offset at the output when no speaker is attached and when the input is shorted, or fitted with a dummy load to mimic your source.
You can measure the variation in output offset as the amp warms up. You can also warm up the heatsink artificially using a hair drier or similar.
I tried simulating the JLH circuit 1 in LTspice. This is what I got for the input and output signal.
Why does it look so weird ? The output signal is not centered at 0V line but is centered at 0.8V. I am still learning to use LTspice which means I must have done something wrong. Anyone knows what I did wrong making the signal output look weird ?
This is what I simulated.
An externally hosted image should be here but it was not working when we last tested it.
Why does it look so weird ? The output signal is not centered at 0V line but is centered at 0.8V. I am still learning to use LTspice which means I must have done something wrong. Anyone knows what I did wrong making the signal output look weird ?
This is what I simulated.
An externally hosted image should be here but it was not working when we last tested it.
The LTspice circuit looks OK and the plots look pretty good, apart from the output not being centered around 0V. I am not familiar with the JLH circuit 1 but did build the "JHL" clone amp based on the same design, which was further modified as per my posts much earlier in this thread. The problem is the output DC bias operating point is not set properly. For a quick and easy resolution, just try 2 simple things in simulation to see if things can be improved:
1) Remove or reduce R3 (prevents Q1/Q3/Q7/Q9 from biasing properly)
2) Add extra diodes to D3,D4 to see the effect on the output DC operating point
Finally, simulate a potentiometer voltage divider (two resistors) fed from the diodes reference to R6 to see if you can get close to 0V DC output. You will need to implement this component if you want to adjust the output DC operating point of a real circuit.
1) Remove or reduce R3 (prevents Q1/Q3/Q7/Q9 from biasing properly)
2) Add extra diodes to D3,D4 to see the effect on the output DC operating point
Finally, simulate a potentiometer voltage divider (two resistors) fed from the diodes reference to R6 to see if you can get close to 0V DC output. You will need to implement this component if you want to adjust the output DC operating point of a real circuit.
When you say linear does it mean that the transistor's gain is linear over a range of Ic values (by looking at the HFE vs Ic graph) ? Is this what is meant by linear ?
I have been reading online on how to bias transistor and how it works as an amplifier. But most of the stuff I read does not show how to use datasheets and what to look for when trying to use the BJT in an amplifier.
Incidentally, people only talked about this JLH 1979 HPA as if it were the one and only HPA design from John Linsley Hood.
It is partly because of the availability of low-cost (sometimes also "cheap") kits and PCBs on the internet.
And it is essentially a low power version of the JLH1969 Power Amp.
But there are other later designs from Linsley Hood that were specifically designed as Headphone amps.
For example the complementary BJT design from 1985 (circuit 2 from the first link).
http://www.startfetch.com/jlh/jlh-1985-headphone.pdf
A Paul Kemble web page - John Linsley Hood preamp designs.
Complementary BJTs are so good these days that there is no real reason for doing quasi push-pull, except for fun.
Patrick
It is partly because of the availability of low-cost (sometimes also "cheap") kits and PCBs on the internet.
And it is essentially a low power version of the JLH1969 Power Amp.
But there are other later designs from Linsley Hood that were specifically designed as Headphone amps.
For example the complementary BJT design from 1985 (circuit 2 from the first link).
http://www.startfetch.com/jlh/jlh-1985-headphone.pdf
A Paul Kemble web page - John Linsley Hood preamp designs.
Complementary BJTs are so good these days that there is no real reason for doing quasi push-pull, except for fun.
Patrick
Just stumbled across this thread and bought one of these. So small and inexpensive, i couldn't resist. I don't understand a word of the instructions though.
Finally arrived. I took another look at the ebay listing and whoo boy there are some 'instructions':
Er, anyone willing to translate that?
Why would you even buy that particular kit that uses a large output coupling cap and comes with such goofy "instructions"??
Most people, including myself, go with this kit or something similar:
GZLOZONE JLH HOOD1969 Class A Headphone amplifier Kit / Pre-amplifier Kit | eBay
Best $20 I've spent.
Most people, including myself, go with this kit or something similar:
GZLOZONE JLH HOOD1969 Class A Headphone amplifier Kit / Pre-amplifier Kit | eBay
Best $20 I've spent.