I guess you're right again..
,Geoff said:
take your time, geoff. we are in a holiday mode as well here.
to tschrama's point, the only requirement for a push-pull output stage to work is that when one output device blows, the other sucks (sounds a little weird,
).Of course, they should be in complete sync. otherwise, you don't get the right output. What "triggers" the output devices is 2ndary.
What the driver does here is to generate two outputs, in perfect phase reversal so that it drives the push-pull output stage.
The bootstrap does is to enhance the gain of the upper device and enhance power output. You can prove this by letting the bottom device work like a current source (current sink more precisely).
a good way to understand this is to replace the output devices with controlled current sources (current controlled for BJTs and voltage controlled for MOSFETs).
This current drive / voltage drive point seems just like a viewpoint thing, different ways of understanding the same situation.
My understanding is that generally a change in current is caused by a change in voltage
The idea to bootstrap the i/p tr collecter resister is neat.
I did not do exhaustive research on it but I think you may find that if you make the driver transistor a small mosfet and use a CCS as the i/p tr collector load you will notice that the distortion goes to much lower levels.
I tried it and like the sound
it may create slight stability issues
mike
My understanding is that generally a change in current is caused by a change in voltage
The idea to bootstrap the i/p tr collecter resister is neat.
I did not do exhaustive research on it but I think you may find that if you make the driver transistor a small mosfet and use a CCS as the i/p tr collector load you will notice that the distortion goes to much lower levels.
I tried it and like the sound
it may create slight stability issues
mike
PSU Voltage
Hi!
Some other kind of question:
What is the max PSU voltage for the JLH 1996 version?
I have a nice 300VA 2X24 V Hipersil transformator and I would like to use for the JLH updated amp, in a non-regulated PSU, but I don't know the 30V isn't too high?
(Somewhere I read that 60 V is the max and 40 V is the optimum for the 1969 version.)
Thanks
Hi!
Some other kind of question:
What is the max PSU voltage for the JLH 1996 version?
I have a nice 300VA 2X24 V Hipersil transformator and I would like to use for the JLH updated amp, in a non-regulated PSU, but I don't know the 30V isn't too high?
(Somewhere I read that 60 V is the max and 40 V is the optimum for the 1969 version.)
Thanks
I am running 36v on my mosfet 1969 version. it works fine.
I don't consider rail rail voltage being a big factor, especially with BJTs as they are a lot more efficient than my mosfets. the thing has a gain of 10x, so it is about 5v each half wave. Plus may be 5 - 6v for the output devices, you should have at least 2x(5+6)=20v in rail voltage for it to work.
a higher voltage will limit your ability to use high Iq on the output devices as you may be approaching the SOA of your output devices.
I am sure you can go through similar calculations with the rual rail version.
I don't consider rail rail voltage being a big factor, especially with BJTs as they are a lot more efficient than my mosfets. the thing has a gain of 10x, so it is about 5v each half wave. Plus may be 5 - 6v for the output devices, you should have at least 2x(5+6)=20v in rail voltage for it to work.
a higher voltage will limit your ability to use high Iq on the output devices as you may be approaching the SOA of your output devices.
I am sure you can go through similar calculations with the rual rail version.
Re: PSU Voltage
on a 2nd thought, if you ran the 1969 at 1.5amp iq. it would dissipate about 90w total under 60v rails. That means about 45w each output device. so you probably need transistors rated at 150w.
for 40v, you are talking about 60w total and 30w each. in the ball park for the IRFs I am using (I think they are rated at around 100w).
Tyimo said:
on a 2nd thought, if you ran the 1969 at 1.5amp iq. it would dissipate about 90w total under 60v rails. That means about 45w each output device. so you probably need transistors rated at 150w.
for 40v, you are talking about 60w total and 30w each. in the ball park for the IRFs I am using (I think they are rated at around 100w).
So I'm back...
I just build MOSFET version to see if it would work,how it would sound and how it would behave...
The IRF540's that I bought were tested for Vgs threshold: both around 3.7Vfor 500mA,give or take 50 to 100mV because the heatsink get warm while testing and this makes the Vgs shift.
I build Millwoods versionwith slightlydifferent component values:
input bias : 2 x 100K, 15K/22uF for Vsupply-smoothing
first gain stage : 2 x4.7K, BC560C, 100R/220uF
Bootstrap : 776Ohm / 220uF
second stage: 470R, BD139,470R
Feedback: 1K
MOSFET: 100R gate snoopers, 0.5R source resistors
Vsupply: 30V regulated lab supply
I behaved very nicly.. looked stable on my scope when idle, so I connectedmy speakers...: sounds nice..but just one channel build so nothing definite..
But:
The Ibias is not proportional to Vsupply, but suddely starts and rises very sharply after that. This design is very, very depended on Vsupply.I think 1volt makes 300mA difference... This is offcourse due to the transductance charateristics of the MOSFETs.
I think this design needs a regulated supply,which I has at hand.
I had to set Ibias to 600mA, because of the small heat sink. I did this by putting 2k2 parralel to my 1k2 resistor (too low bias): makeing 776R (about right)
Then I tested some test signal... everything looks ver clean. But on a 20KHz square wave there was servere ringing. So I tried to stop that by putting 150pF across the 1K feedback resistor...
OOOOOOOoopps
suddely I bias wentuptot 3.1A and it started to smell bad 
I'm sorry,I should have been more carefull,but I was so curious..hmmm after inspection it looks like the driver is blown.
Do any of you have a better suggestion how to stabilize the amp in a simple way,savely?
That's it for now.. butI'll be back...
best regards,
Thijs
I just build MOSFET version to see if it would work,how it would sound and how it would behave...
The IRF540's that I bought were tested for Vgs threshold: both around 3.7Vfor 500mA,give or take 50 to 100mV because the heatsink get warm while testing and this makes the Vgs shift.
I build Millwoods versionwith slightlydifferent component values:
input bias : 2 x 100K, 15K/22uF for Vsupply-smoothing
first gain stage : 2 x4.7K, BC560C, 100R/220uF
Bootstrap : 776Ohm / 220uF
second stage: 470R, BD139,470R
Feedback: 1K
MOSFET: 100R gate snoopers, 0.5R source resistors
Vsupply: 30V regulated lab supply
I behaved very nicly.. looked stable on my scope when idle, so I connectedmy speakers...: sounds nice..but just one channel build so nothing definite..
But:
The Ibias is not proportional to Vsupply, but suddely starts and rises very sharply after that. This design is very, very depended on Vsupply.I think 1volt makes 300mA difference... This is offcourse due to the transductance charateristics of the MOSFETs.
I think this design needs a regulated supply,which I has at hand.
I had to set Ibias to 600mA, because of the small heat sink. I did this by putting 2k2 parralel to my 1k2 resistor (too low bias): makeing 776R (about right)
Then I tested some test signal... everything looks ver clean. But on a 20KHz square wave there was servere ringing. So I tried to stop that by putting 150pF across the 1K feedback resistor...
OOOOOOOoopps
suddely I bias wentuptot 3.1A and it started to smell bad I'm sorry,I should have been more carefull,but I was so curious..hmmm after inspection it looks like the driver is blown.
Do any of you have a better suggestion how to stabilize the amp in a simple way,savely?
That's it for now.. butI'll be back...
best regards,
Thijs
tschrama said:
that may be a little on the low side. I would go to 1amp.
tschrama said:That's it for now.. butI'll be back...
Thijs
sounds like an interesting experience. sorry that I cannot help you on the ringing thing.
did you get turn-on thump? I did with mine and I am struggling to figure out how to cure it.
Yeh,your right about the bias, but I don't have the money to buy a larger heatsink actually,I don't have any at the moment, jsut enough to get me some MOSFETs.. I made the thing out of parts the were stolen from other equipment,I only had to buy the MOSFET's.. (still 2,20Euro/pcs)
BTW.. I found out that the MOSFET version will do over 20V/uS slwrate with 100pF acroos the 1K FD resistor.. that not bad.. not bad at all!!
but I'm still a bit and 
about blowing up my version.. this is actualle the first time I blow a power amp...I soooooo hope the MOSFET are OK... we'll see
I do have a turn-on bump.. I thing this is because the 15K/470uF bias stabilization... it makes the amp draw twice the current for a moment or two...actuallymore than that
Geoff are you still there?? Any wise words for me?
Regards,
Thijs
actually,I don't have any at the moment, jsut enough to get me some MOSFETs.. I made the thing out of parts the were stolen from other equipment,I only had to buy the MOSFET's.. (still 2,20Euro/pcs) BTW.. I found out that the MOSFET version will do over 20V/uS slwrate with 100pF acroos the 1K FD resistor.. that not bad.. not bad at all!!
but I'm still a bit
and about blowing up my version.. this is actualle the first time I blow a power amp...I soooooo hope the MOSFET are OK... we'll see I do have a turn-on bump.. I thing this is because the 15K/470uF bias stabilization... it makes the amp draw twice the current for a moment or two...actuallymore than that
Geoff are you still there?? Any wise words for me?
Regards,
Thijs
millwood said:
Sorry Millwood, but could you please explain parts of this paragraph. I don't understand what you are saying.
"the thing has a gain of 10x, so it is about 5v each half wave."
The standard JLH has a closed-loop gain of approximately 13 but this has no bearing on the selection of supply rail voltage unless one is designing on the basis of input sensitivity rather than output power (which is not the usual way). For example, for a 1Vrms input and a gain of 13, the maximum output voltage will be 13Vrms, or 36.8Vp-p. The supply rail will need to be this voltage plus an allowance for the voltdrop across the output transistor and any allowance necessary for the preceeding Iq control circuitry (nothing for the bootstrap circuit, but around 3V for the 1996 TR5 arrangement). So that gives a single 40V rail or a split supply of +/-22V.
"Plus may be 5 - 6v for the output devices, "
I have yet to find a BJT that requires more than 1V at the current we are talking about (2-3A).
Re: PSU Voltage
The answer to your question depends on what load impedance you intend to drive. If your speakers are 15ohm (very rare these days) and their impedance does not drop below this figure, then +/-30V supply rails and a quiescent current of 1.4A would be ideal. However most modern speakers use 4 ohm bass units to 'improve' the sensitivity specification and so the optimum balance is for lower supply rail voltages and a higher quiescent current. You could use the higher rail voltage, but this would lead to a lower peak power availability and greater inefficiency (which is not high to start off with
I think you have two options. One is to use the +/-30V rails and parallel output transistors so that the quiescent current can be increased and some efficiency restored. However, there will be a limit on how far the current can be raised due to the VA rating of the transformer (particularly if you intend to use a single transformer to feed both channels). Also, this may not be cost effective (the extra heatsinks etc required will not be be cheap), particularly if you do not need the extra power. It would probably be cheaper to by a new transformer with the correct secondary voltage
The other is to parallel the transformer secondaries and have a single 30V rail. This will require the use of an output capacitor, a device not favoured by many people. However, you would be able to set the quiescent current to 2.7A (with only single output transistors) which would give an output power of nearly 25W (50Wpeak) into 4ohm.
Geoff
Tyimo said:
The answer to your question depends on what load impedance you intend to drive. If your speakers are 15ohm (very rare these days) and their impedance does not drop below this figure, then +/-30V supply rails and a quiescent current of 1.4A would be ideal. However most modern speakers use 4 ohm bass units to 'improve' the sensitivity specification and so the optimum balance is for lower supply rail voltages and a higher quiescent current. You could use the higher rail voltage, but this would lead to a lower peak power availability and greater inefficiency (which is not high to start off with
I think you have two options. One is to use the +/-30V rails and parallel output transistors so that the quiescent current can be increased and some efficiency restored. However, there will be a limit on how far the current can be raised due to the VA rating of the transformer (particularly if you intend to use a single transformer to feed both channels). Also, this may not be cost effective (the extra heatsinks etc required will not be be cheap), particularly if you do not need the extra power. It would probably be cheaper to by a new transformer with the correct secondary voltage
The other is to parallel the transformer secondaries and have a single 30V rail. This will require the use of an output capacitor, a device not favoured by many people. However, you would be able to set the quiescent current to 2.7A (with only single output transistors) which would give an output power of nearly 25W (50Wpeak) into 4ohm.
Geoff
Re: Re: PSU Voltage
I am not sure how much of a voltage drop a typical BJT has, but mosfets usually drop 5 - 6v and I know that BJTs are more efficient so I wouldn't go wrong with the 5-6v figure.
what I did is to until the center tap of my transformer and form two windings of 28v. From there, I set up two independent sets of rectifier / filter cap circuitries. They are set up such that if needed, i can use one jumper wire to form +/- 36v PS. or if I so desire, I can have two completely independent 30v single rail PS.
worked well for me.
Geoff said:
I am not sure how much of a voltage drop a typical BJT has, but mosfets usually drop 5 - 6v and I know that BJTs are more efficient so I wouldn't go wrong with the 5-6v figure.
Geoff said:
what I did is to until the center tap of my transformer and form two windings of 28v. From there, I set up two independent sets of rectifier / filter cap circuitries. They are set up such that if needed, i can use one jumper wire to form +/- 36v PS. or if I so desire, I can have two completely independent 30v single rail PS.
worked well for me.