Hello, I have build the 1969 model amplifier from JL, I went with bc212, 2n1711 and 2x 2n3055 ( old ones ) I find the sound very warm , pleasant but a bit of distortion in bass at high levels.
how can I fix that? ( maybe parralell two 2n3055's will do the job
and also
Can i mod this amplifier to work in class ab or Class a but with 100-200mA bias or so, but keeping it's warm sound
Someone any ideas ?
Thanks
Also Im thinking to pair it with Rod Elliott's project37a Pre-amplifier
how can I fix that? ( maybe parralell two 2n3055's will do the job
and also
Can i mod this amplifier to work in class ab or Class a but with 100-200mA bias or so, but keeping it's warm sound
Someone any ideas ?
Thanks
Also Im thinking to pair it with Rod Elliott's project37a Pre-amplifier
Baising classic 2n3055 output transistors with bootstrap
creates a situation that overbiases beyond class A in the
middle, yet runs out of bias before reaching either rail.
Sorry, but the beta of these devices is just plain curved,
and cannot be made linear by steering a constant current.
Easiest fix would be modern output devices that aren't as
curvy to begin with...
Second, and maybe easier if you've already mounted TO3's.
Is just deliberately overbias by selecting R1 and R2 values
that are purposely too low. Then discard the excess bias
with a shunt regulator, I'll describe how...
Place a pair of current monitoring resistors in the output
totem, with your load halfway inbetween. Choose values
such that sum voltage drop across the pair is 0.65V when
your output is at the linear class A bias that you want to
maintain.
Now place a new NPN TR5 such that TR5 base watches
the emitter of TR2. TR5 Emitter watches the collector
of TR1. TR5 collector shunting away bias from the base
of TR2. Steering excess bias into the collector of TR1,
rather than into TR1's base as TR3 normally does). VBE
TR5 expects to see 0.65V when sufficient overbias has
been discarded.
Now you will have the correct non-linear bias to get
a perfectly linear class A output, regardless how old
and curvy your classic output transistors might be.
TR5 holding the voltage drop across the monitoring
resistors, locking the behavior into class A.
Now, one can theoretically make this AB by abuse of
Schottky diodes, rather than resistors for the senses.
But this solution has not proven stable. Might sim OK
in LTSpice, but usually makes an oscillator. Stay with
linear resistors, and there will be no problem.
Search these forums for SRJLH, and you should be
able to find a few drawings of the shunt regulator.
Again, ignore anything I might have said, drawn, or
simulated to do with a Schottky. Resistors only...
creates a situation that overbiases beyond class A in the
middle, yet runs out of bias before reaching either rail.
Sorry, but the beta of these devices is just plain curved,
and cannot be made linear by steering a constant current.
Easiest fix would be modern output devices that aren't as
curvy to begin with...
Second, and maybe easier if you've already mounted TO3's.
Is just deliberately overbias by selecting R1 and R2 values
that are purposely too low. Then discard the excess bias
with a shunt regulator, I'll describe how...
Place a pair of current monitoring resistors in the output
totem, with your load halfway inbetween. Choose values
such that sum voltage drop across the pair is 0.65V when
your output is at the linear class A bias that you want to
maintain.
Now place a new NPN TR5 such that TR5 base watches
the emitter of TR2. TR5 Emitter watches the collector
of TR1. TR5 collector shunting away bias from the base
of TR2. Steering excess bias into the collector of TR1,
rather than into TR1's base as TR3 normally does). VBE
TR5 expects to see 0.65V when sufficient overbias has
been discarded.
Now you will have the correct non-linear bias to get
a perfectly linear class A output, regardless how old
and curvy your classic output transistors might be.
TR5 holding the voltage drop across the monitoring
resistors, locking the behavior into class A.
Now, one can theoretically make this AB by abuse of
Schottky diodes, rather than resistors for the senses.
But this solution has not proven stable. Might sim OK
in LTSpice, but usually makes an oscillator. Stay with
linear resistors, and there will be no problem.
Search these forums for SRJLH, and you should be
able to find a few drawings of the shunt regulator.
Again, ignore anything I might have said, drawn, or
simulated to do with a Schottky. Resistors only...
Last edited:
Hi,
No, it won't work as Class AB at reduced current. For
maximum power it must be carefully load matched.
See Rod Elliott's Death of Zen article for updates.
rgds, sreten.
Same problem I had with a Kelvin Labs class A, great
sound but the bass gave up the ghost far too much.
D. Self
No, it won't work as Class AB at reduced current. For
maximum power it must be carefully load matched.
See Rod Elliott's Death of Zen article for updates.
rgds, sreten.
Same problem I had with a Kelvin Labs class A, great
sound but the bass gave up the ghost far too much.
D. Self
Hi xXBrunoXx, Very interesting post. You are obviously aware of Rod Elliot's site. He now hosts the JLH Class A amp site developed by Geoff MOSS in the UK.
Before you do anything else can I suggest you go there and check out the 15-20 Watt Class AB design that JLH produced just after the Class A 10 "Watter".
Google "Class A amp site".
Then go to "Other JLH Amplifier designs". (second option) and then the 15-20 watt Class-AB entry.....again, second option.
The preceding article is worth a read too.
It was response to your exact point about the same sound with lower bias.
That should address many of the issues you raise.
Cheers,
Jonathan
Before you do anything else can I suggest you go there and check out the 15-20 Watt Class AB design that JLH produced just after the Class A 10 "Watter".
Google "Class A amp site".
Then go to "Other JLH Amplifier designs". (second option) and then the 15-20 watt Class-AB entry.....again, second option.
The preceding article is worth a read too.
It was response to your exact point about the same sound with lower bias.
That should address many of the issues you raise.
Cheers,
Jonathan
Working properly, the classA JLH is a very good amp. I built one as my first ever hifi amp and had the issue you did, basically, as mentioned earlier, old transistors suffered from bi-polar gain droop and also low gain, some power transistors had a gain as low as five!. I would measure the gain of the devices first, if its low and I suspect it will be, then change to a more modern type, or if you are feeling very brave you can use the SiC design I posted a few days ago.
Member
Joined 2009
Paid Member
You can parallel devices, I used 16 of them for one channel. But you can't run in Class AB without changing the topology and even with parallel devices the sound is still ultimately limited by these old output devices. In my case the power devices were all riveted to heatsinks, pulled out of two old power supples but I have been tempted to drill them out and replace with something modern. Nevertheless, built as-is the design is a 'classic' and the sound no doubt close that which JLH enjoyed with the original.
http://www.diyaudio.com/forums/solid-state/268846-tgm9-my-version-jlh-69-class-amplifier.html
Thank you very much ! , this is the schematic ?
A few questions :
In the original 1969 version of JLH can I mod it to use N channel mosfet's ( i have some lying around )
what can i change to make it work well with 2n3055's ( i know, i know - they are old and poor , but I found the sound very pleasant.
ALSO in the schemathic that you recommended Class AB 15-20w
What transistors i can use instead of:
MJ481/MJ491
2n3054
Tr2 in picture is 2n1711 i think ? ( i have some old/original ones 😀 )
Thanks again
A few questions :
In the original 1969 version of JLH can I mod it to use N channel mosfet's ( i have some lying around )
what can i change to make it work well with 2n3055's ( i know, i know - they are old and poor , but I found the sound very pleasant.
ALSO in the schemathic that you recommended Class AB 15-20w
What transistors i can use instead of:
MJ481/MJ491
2n3054
Tr2 in picture is 2n1711 i think ? ( i have some old/original ones 😀 )
Thanks again
I have found some posts here about JLH shunt regulator, but over my knowledge.
I might build the one you recommended
I might build the one you recommended
Thank you for taking your time to explain
i searched for shunt regulated JLH and it's a bit above my knowledge
still have a few questions:
If i change the 2n3055's with some modern ( better transistors ) would I still need to " shunt regulate " it ?
Also i have some KD503 ( old ones, similar to 2n3055's maybe they are better ? )
Also , if you have time to help a newbie
can you draw the modifications you sugested on the " second solution " you described .
Thank you !
i searched for shunt regulated JLH and it's a bit above my knowledge
still have a few questions:
If i change the 2n3055's with some modern ( better transistors ) would I still need to " shunt regulate " it ?
Also i have some KD503 ( old ones, similar to 2n3055's maybe they are better ? )
Also , if you have time to help a newbie
can you draw the modifications you sugested on the " second solution " you described .
Thank you !
While you could certainly 'get it to work', I think this is one amplifier where doing so would be sub optimal in pretty much all respects. They would not be a drop in replacement (whether laterals or HexFets) and would require alteration of several of the resistor values.
Member
Joined 2009
Paid Member
I wouldn't use FETs for another reason - the amplifier was designed around there being base current into the output devices. Nelson Pass did a FET version but it was never popular. The bipolar JLH on the other hand has been extremely popular.
For me, I would not add the 'shunt' thing either, you want it to be a JLH classic and operating in Class A - that's the whole point. If you want Class AB then build a different amp. Or look at Elvee's Circlophone, rich-biassed Class AB.
One change you could consider is to improve the PSRR with an RC filter in the power rail to the front end of the amplifier - see attached.
For me, I would not add the 'shunt' thing either, you want it to be a JLH classic and operating in Class A - that's the whole point. If you want Class AB then build a different amp. Or look at Elvee's Circlophone, rich-biassed Class AB.
One change you could consider is to improve the PSRR with an RC filter in the power rail to the front end of the amplifier - see attached.
Attachments
Hi Bruno, YES, that is the circuit I had in mind. It tries to get "Class A sound with low bias".
As Bigun says Nelson PASS has developed a MOSFET version. It is in an article he wrote try. "PLH" on his "passdiy" site.
You will not get MJ481/MJ491 transistors these days. The 2N3055 and 2N2955 are a rough substitute but with lower Ft and lower gain (but higher power ratings).
You can just google on the web to find substitutes.
I still think that a visit to the "Class A Amp site" will address a lot of your questions and suggest modern substitutes etc.
I think you will benefit from reading around this area on that site.
There is also a very long thread on this forum devoted to the JLH Class-a amp. It has several thousand posts!
Cheers,
Jonathan
As Bigun says Nelson PASS has developed a MOSFET version. It is in an article he wrote try. "PLH" on his "passdiy" site.
You will not get MJ481/MJ491 transistors these days. The 2N3055 and 2N2955 are a rough substitute but with lower Ft and lower gain (but higher power ratings).
You can just google on the web to find substitutes.
I still think that a visit to the "Class A Amp site" will address a lot of your questions and suggest modern substitutes etc.
I think you will benefit from reading around this area on that site.
There is also a very long thread on this forum devoted to the JLH Class-a amp. It has several thousand posts!
Cheers,
Jonathan
Last edited:
For the Class AB circuit I would consider MJL-21194/21193.
These are rated for their linearity without negative feedback ( less than 1% as push-pull emitter followers up to 20 kHz) and are a good enough match for hFE and fT of MJ481/MJ491 at the lower power range of this amplifier.
They will also be more reliable since the power rating for the MJL pair is more than double the obsolete devices.
Hi,
For Class A see Rod Elliots :
DoZ power amplifier
DoZ preamplifier
DoZ capacitor multiplier supply
rgds, sreten.
For Class A see Rod Elliots :
DoZ power amplifier
DoZ preamplifier
DoZ capacitor multiplier supply
rgds, sreten.
- Status
- Not open for further replies.