P3A does work well if built well, but that CFP output can be a pain. My suggestion of adding local power supply decoupling of 100uF in addition to the 100nF on the schematic often does the trick.
Likewise if you use fast output transistors, you should use fast drivers too - such as the MJE15032/3 devices. I am not a fan of using the BD139/140 there, as the modern ones are not as good as the old Philips devices were. MJE243/253 also work well.
Likewise if you use fast output transistors, you should use fast drivers too - such as the MJE15032/3 devices. I am not a fan of using the BD139/140 there, as the modern ones are not as good as the old Philips devices were. MJE243/253 also work well.
Hi Sakis, nices to hear from you again...
but,
I do not see anything difficult to underderstand:
imput: diff. pair with LED current source
VAS: single BJT with bootstrap.
Output sigle CFP pair.
Compensation: Cdom on the VAS
Simple and beautiful.
And about the problem: seen on scope on square wave ,
changed CB capacitor, problem disappeared.
What am I missing?
your case is a bit different since if i remember well you are using cans in the outputs am i wrong ?
Change a pair of drivers to any other model ( given as a fact that miller cap remains 100pf ) and observe the problem to see if its better or worst ....but change the drivers with anything available but different to see what will happen and get back to me ...
Kind regards
sakis ( good to hear from you too !! )
Change a pair of drivers to any other model ( given as a fact that miller cap remains 100pf ) and observe the problem to see if its better or worst ....but change the drivers with anything available but different to see what will happen and get back to me ...
Kind regards
sakis ( good to hear from you too !! )
@ jay cee small joke if i am allowed :
When i worked for the local Nak-Adcom dealer in my very young days that company also represented JRC RAtheon ship radars , so inside the bench was a tech named Jim Pappas repairing radars next to me and every time there was a super weird Japanese transistor burned beyond recognition in some weird also ( for me ) stage of the radar Jim come over to my bench virtually still a BC 547 from my parts a couple of resistors and a couple of IN4148 installed those on the fly and radar was working all over again .
Through the years it actually become an internal joke and in any case any of us 3 techs inside the bench was into any possible problem the solution of course was to """ADD A BC 547 AND A COUPLE OF RESISTORS "" and you will be just fine ..
Similar to the above we both follow close P3A threads and every now and then you talk about those two capacitors of 100uf ...fair ..and helpful but this circuit has a bit more juice than that ... you need to look a bit more carefully to stop thinking EFP when you look at the schematic ....
Add 100uf decoupling and you will be just fine !!!
Kind regards
Sakis
When i worked for the local Nak-Adcom dealer in my very young days that company also represented JRC RAtheon ship radars , so inside the bench was a tech named Jim Pappas repairing radars next to me and every time there was a super weird Japanese transistor burned beyond recognition in some weird also ( for me ) stage of the radar Jim come over to my bench virtually still a BC 547 from my parts a couple of resistors and a couple of IN4148 installed those on the fly and radar was working all over again .
Through the years it actually become an internal joke and in any case any of us 3 techs inside the bench was into any possible problem the solution of course was to """ADD A BC 547 AND A COUPLE OF RESISTORS "" and you will be just fine ..
Similar to the above we both follow close P3A threads and every now and then you talk about those two capacitors of 100uf ...fair ..and helpful but this circuit has a bit more juice than that ... you need to look a bit more carefully to stop thinking EFP when you look at the schematic ....
Add 100uf decoupling and you will be just fine !!!
Kind regards
Sakis
P3A is obviously inspired by Motorola App. Note 485 circuit, but without complicated output protection, and bootstrap instead of transistor CS for VAS. Original amp designed by Motorola applications engineer Richard Ruehs used 50pF compensation in VAS and 50pF for lower driver. Obviously even with low values of these caps Sziklay outpuit stage can be stable, depending on drivers and output transistors used. If one decide to use ceramic caps for these positions depending on type of ceramic and temperature inside amp, 100pF value specified in Rod's P3A circuit will go down due to negative tempco of ceramics.
Has enyone tried 2SB649/2SD669 or 2SA1837/2SC4793 as drivers in P2A?
Has enyone tried 2SB649/2SD669 or 2SA1837/2SC4793 as drivers in P2A?
Also, in CFP output stages 0,33R wirewound emmiter resistors may cause instability or oscilations in the range 5 to 50 MHz due to their inductance. I found 2W 0,33R MOX resisitors in the local shop and shall use them in P3A that I intend to build soon. Since CFP output stages are more prone to instability than other output topologies it is better to prevent any possibility of oscilations by using non-inductive metal oxide emmiter resistors.
Ceramic capacitors used for compensation should be of COG/NPO type, and that one has temperature stable capacitance.
Indeed. I used this type of ceramic capacitor when I built a P3A (on Veroboard too!) and it was stable. Ordinary ceramics can go as much as 50% off value, usually lower, in use.
Metal oxide resistors should have less inductance than wirewound, and they are often recommended for passive loudspeaker crossovers (L-pad) for that reason. How much less I can not say.
But it is possible that carbon resistors are also non inductive depending on fabrication method. Metal film resistors are inductive since resistive element usually have the form of spiral. In Japanese amps one can see emmiter metal film resistors in the form of cube and it is possible that these are made in special way to make them non-inductive. But it is impossible to buy them anywhere.
I think that Rod recommended carbon emmiter resitors for his P101 Mosfet amp because these are non-inductive.
But it is possible that carbon resistors are also non inductive depending on fabrication method. Metal film resistors are inductive since resistive element usually have the form of spiral. In Japanese amps one can see emmiter metal film resistors in the form of cube and it is possible that these are made in special way to make them non-inductive. But it is impossible to buy them anywhere.
I think that Rod recommended carbon emmiter resitors for his P101 Mosfet amp because these are non-inductive.
What is the temp coeff for 2KV low pF caps ( blue in colour) ?
I had some 100pf /2 kV which I measured at room temperature 28 deg C and dipped in hot water ( about 80 deg C). The value didn't seem to change drastically. Certainly not 50%.
Maybe I should do it again now to make sure I didn't goof the measurement !
I couldn't find any COG/NPO in the local retail market. I have a few which I bought in Singapore.
I had some 100pf /2 kV which I measured at room temperature 28 deg C and dipped in hot water ( about 80 deg C). The value didn't seem to change drastically. Certainly not 50%.
Maybe I should do it again now to make sure I didn't goof the measurement !
I couldn't find any COG/NPO in the local retail market. I have a few which I bought in Singapore.
Carbon film , metal film and metal oxide resistors are all fabricated similarly and spiral grooving, used in production to economically trim the values of film types, creates an inductive component to them all. It will vary somewhat, according to the length of spiral path necessary for each value in the individual manufacturer's formulation/decade range. Low value 5W box wirewounds (at least) are wound for low inductance (bifilar) below 100R. That is open to question but the specs of common brands like NOBLE, XICON seem correct, at <10% of the nominal single layer solenoid wound value, using an LCR bridge.
In any case, the original P3a design is quite stable with 30MHz Ft output and 150 MHz Ft driver transistors - as is. There are tens of thousands of these in wide application around the globe. It is probably the most tested and proven of all DIY amps! The main threat to its stability is ill-conceived PCB layout and wiring as DIYs are wont to experiment and try for themselves. In such a simple design, it's inevitable that someone will find a limit to how far it can be pushed before stability is stupidly compromised. In that case, I wish them lots of toasted parts for their trouble!
The reason for using COG/NPO ceramics is not for thermal stability of compensation but for distortion reasons, which these types don't measurably incur whist other grades can be disastrous due to their non-zeroing behaviour with AC signals. John Curl has much to say on the subject as does D. Self.
I can't find a direct reference to the age of AN485 - It could well have been the inspiration of this and many other audio designs when the audio knowledge base was still dominated by vacuum tube technology. ESP P3 design goes back to about 1970 according to Rode and I think he should know.
BTW Ashok, I can't guarantee it, but most quality sources of ceramics <100pF are actually NPO. It's not hard to test with a little air heating in any case. You only find COG in SMD parts, AFAIK.
In any case, the original P3a design is quite stable with 30MHz Ft output and 150 MHz Ft driver transistors - as is. There are tens of thousands of these in wide application around the globe. It is probably the most tested and proven of all DIY amps! The main threat to its stability is ill-conceived PCB layout and wiring as DIYs are wont to experiment and try for themselves. In such a simple design, it's inevitable that someone will find a limit to how far it can be pushed before stability is stupidly compromised. In that case, I wish them lots of toasted parts for their trouble!
The reason for using COG/NPO ceramics is not for thermal stability of compensation but for distortion reasons, which these types don't measurably incur whist other grades can be disastrous due to their non-zeroing behaviour with AC signals. John Curl has much to say on the subject as does D. Self.
I can't find a direct reference to the age of AN485 - It could well have been the inspiration of this and many other audio designs when the audio knowledge base was still dominated by vacuum tube technology. ESP P3 design goes back to about 1970 according to Rode and I think he should know.
BTW Ashok, I can't guarantee it, but most quality sources of ceramics <100pF are actually NPO. It's not hard to test with a little air heating in any case. You only find COG in SMD parts, AFAIK.
Hi Sakis,
You have a good memory about metal cans!
but it was not that implementation.
It was the rather standard one that I mounted the Rod's board
http://www.diyaudio.com/forums/solid-state/191055-tecnics-su-z1-refurbish-2.html#post2981132.
To stay on the DIY side I had used my mix of BJTs, in particular a Sanyo 2sA1209 as a VAS ...
Thanks also the valuable Sakis help with the Technics schematics, that build was quite a success at the end and still today it plays on a classic Dance Studio like a mini-PA system every day!
BTW, for the Cdom I used a 500V 100pf mica and the board HAS the 100 uF filters installed as standard (PCB is a"second version").
You have a good memory about metal cans!
but it was not that implementation.
It was the rather standard one that I mounted the Rod's board
http://www.diyaudio.com/forums/solid-state/191055-tecnics-su-z1-refurbish-2.html#post2981132.
To stay on the DIY side I had used my mix of BJTs, in particular a Sanyo 2sA1209 as a VAS ...
Thanks also the valuable Sakis help with the Technics schematics, that build was quite a success at the end and still today it plays on a classic Dance Studio like a mini-PA system every day!
BTW, for the Cdom I used a 500V 100pf mica and the board HAS the 100 uF filters installed as standard (PCB is a"second version").
Last edited:
Guessing at the brand but since those blue 2kV are cheap and everywhere, they are Chinese and marketed in the US as XICON.
They are branded variously elsewhere. I think then, they are likely Y5P coefficient but will be unsuitable for the comp. cap by their
formulation anyway.
Is this appearance similar?http://www.mouser.com/catalog/specsheets/XC-600138.pdf
or perhaps TDK's more expensive one http://www.mouser.com/ds/2/400/e413_ck45-6483.pdf
These have a -20/+50% temperature characteristic and I think that also implies a no-go
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.