P3A now running at 225mA bias. The sound is a little smoother, less grainy and soundstage seems to have opened up. Bass seems to have more control and impact as well.
I played "My Parties" from Dire Strait's album - On Every Street. The triangle, hi-hat, saxophone and bass/rhythm are etched precisely in the 3D soundstage, vocals dead on centre. The crack of the snare drum is reproduced with so much transient attack that it's frightening!
I played "My Parties" from Dire Strait's album - On Every Street. The triangle, hi-hat, saxophone and bass/rhythm are etched precisely in the 3D soundstage, vocals dead on centre. The crack of the snare drum is reproduced with so much transient attack that it's frightening!
Hello Sakis,
why dont you share these changes with the forum, do you consider it IP you wont share or because Rod may object?
well no ... since the circuit is free on the internet any one can do anything with it ...
through the post though all the things i ve done and changed have been reported more than once but let me refresh for you ...
--- i use my own pcb ( i ve been trying a few and setled in a specific one )
--- yet never heard an original version with an original pcb hell it could be by far better than mine but still dont know
----i use very carefully matched ltp stage
---- my ltp has thermal junction
---- my pcb features 4 R68 5W vitrohm resistors (paired in two per transitor )
---- been working so far with BD139-140 pairs but i found BD829-830 to sound better and with even less bias
---- been playing with 30-120ma of bias .... found that with slow drivers more is needed while faster ones will happily work with less
---- Drivers 829-830 are common heatsinked and feature also the VBE multiplier on the small heatsink
---- Been using mostly styroflex caps and i was happy and stable enough with 68of on the VAS next to the 100 that was placed originally
---- i use silver mika input filter capacitor
---- my pcb features rail capacitors on board ( 10.000 mfd X2 )
---- and finally on most of the P3A amplifiers i make i use the impossible to find 2SA1302
2SC3281 pairs since i have about 100 more original pairs available ....
be sure that very against to what Rodd says about the "amplifier being very tollerant to change or choise of parts" which is partially true since the amp will happily work very well with a bad choise of ltp but as i said be very very sure that small changes like the above will of course alter the performance of the amplifier ....
working with this amplifier for more than 4 years now and playing arround from time to time the above is my results in a nutshell ...
For the rekord i will state that eventhough latest times and with the economics world wide i am way to busy to update the thread too often but for you information rates remain as was in the first post and none of the comercial hifi i repair ( now days about 250 amplifiers per year ) has been able to beat a well made P3A from any aspect at the same power class ...
Finally i will have to state once more that i think the DX amplifier is technically a better amplifier than the P3A still the P3A remains more warm and musical ( benefit of the sziklai topology i think ) for my ears
happy regards
Sakis
through the post though all the things i ve done and changed have been reported more than once but let me refresh for you ...
--- i use my own pcb ( i ve been trying a few and setled in a specific one )
--- yet never heard an original version with an original pcb hell it could be by far better than mine but still dont know
----i use very carefully matched ltp stage
---- my ltp has thermal junction
---- my pcb features 4 R68 5W vitrohm resistors (paired in two per transitor )
---- been working so far with BD139-140 pairs but i found BD829-830 to sound better and with even less bias
---- been playing with 30-120ma of bias .... found that with slow drivers more is needed while faster ones will happily work with less
---- Drivers 829-830 are common heatsinked and feature also the VBE multiplier on the small heatsink
---- Been using mostly styroflex caps and i was happy and stable enough with 68of on the VAS next to the 100 that was placed originally
---- i use silver mika input filter capacitor
---- my pcb features rail capacitors on board ( 10.000 mfd X2 )
---- and finally on most of the P3A amplifiers i make i use the impossible to find 2SA1302
2SC3281 pairs since i have about 100 more original pairs available ....
be sure that very against to what Rodd says about the "amplifier being very tollerant to change or choise of parts" which is partially true since the amp will happily work very well with a bad choise of ltp but as i said be very very sure that small changes like the above will of course alter the performance of the amplifier ....
working with this amplifier for more than 4 years now and playing arround from time to time the above is my results in a nutshell ...
For the rekord i will state that eventhough latest times and with the economics world wide i am way to busy to update the thread too often but for you information rates remain as was in the first post and none of the comercial hifi i repair ( now days about 250 amplifiers per year ) has been able to beat a well made P3A from any aspect at the same power class ...
Finally i will have to state once more that i think the DX amplifier is technically a better amplifier than the P3A still the P3A remains more warm and musical ( benefit of the sziklai topology i think ) for my ears
happy regards
Sakis
Last edited:
Surely the most interessant feature in the list...
Yet, P3A is better with a classic double emitter follower and can be improved a little step further by using an adequate TMC compensation network.
small thing to to add and reverify the above is like that :
I made for a costumer a mod and installed two of my pcb in a specific very small size box the power supply ( existing ) could produce 24+24 volts after rectification ...obviously low ...specific board also was made with BD139-140 pairs and fairchild 1943-5200 and after carefull listening low bias is not the best for these pairs
higer somethink like 100 ma and everything is a bit warmer but sounds really fine
Kind regards sakis
I made for a costumer a mod and installed two of my pcb in a specific very small size box the power supply ( existing ) could produce 24+24 volts after rectification ...obviously low ...specific board also was made with BD139-140 pairs and fairchild 1943-5200 and after carefull listening low bias is not the best for these pairs
higer somethink like 100 ma and everything is a bit warmer but sounds really fine
Kind regards sakis
Then, perhaps, it is no more a P3A.
Well, the output stage would still use the same components
while TMC would require adding a little cap and a resistor.
Hardly a big change in the topology..
I also believe that P3A will sound differently with a standard (boring) EF.Well, the output stage would still use the same components
while TMC would require adding a little cap and a resistor.
Hardly a big change in the topology..
The crossover behaiviour should be completely different, at least if low- biased at 30-50. The graphs and measures on D. Self book are very esplicative on that subject.
Then
I have read here in similar amplifiers threads that (OS's BX ?, JAS ?) that TMC won't work without a high open loop gain (say 120 dB). I do not believe that P3A has so much. But I am not expert on this more "refined" compensation to say more.
effebi
Boring??...
An EF , according to Self s measures produce significantly
less distorsion at low levels.
Besides, he use very old slow devices in theses measurement.
Would you use MJ802/MJ4502 pair in a modern design unless
it is for a bass box ?..
To the contrary, TMC works better with high OLG amps provided
the VAS has very high gain.
Of course , it works not as well in amps such as P3A but it bring
significant improvement that is worth taking account of since
this amp s VAS has a gain on the order of 60db.
Yes, boring, very boring . When something is used by everybody becomes boring. I am not saying it is not good or that has not some adavantages. Just (highly subjectively)boring.
I am sure to remeber that Self (on the 3rd ed. I believe) demonstrates that as OP stage CFP has LESS (CROSSOVER) distorsion, with numbers and graphics.
Many CFP also work well with "relatively" modern devices like the ones quoted by Sakis (or the omni present 2500/1943), so I do not see your point so easily criticizing the Self's choice. Maybe I am misunderstanding you, sorry for that.
BTW I have around a PA3 with (slow!) 15003/4 OP that sounds very good (to some people with good ears, not mine) but I do not have instrumental THD measuments to support my thesis, so it's up to you . And no, it is not for a bass box and yes is HF stable with TO3 mounted on the same PCB.
About the TMC , seems we agree that the OLG should be high, just some members here had experimentally quite bad results with smaller numbers than 120. I agree also that the tecnique should bring neat improvements, if made to work. I would give it a try as soon I have P3A on hand again, I am courious now.
effebi
Distortion In Power Amplifiers
P3A works better , in sims at least, with EF than CP.
TMC will not improve a CP version by much but in conjunction
with an EF OS , it will improve the THD ratios by some margins,
although the result will surely be unoticeable in listening tests,
but who knows..
Funny that i elected to use the ones on the third curve ,
2SJ50/2SK135 as definitive devices in almost all the amps
i built since late 80s...
Indeed, they allow very good stability and very wide OLG.
As pointed by Tekko, CFP are less stable when used in amps
that have global NFB , and this show in simulations by wider
bandwith of THD residuals.
Attachments
