Member
Joined 2009
Paid Member
Sakis, I was already convinced of the superiority of CFP, what I am discussing is its problematic implementation, which is partially related to the unobtainability of decent transistors for audio use these days. The CFP is not forgiving, on the contrary, the success lies in a meticulous combination of transistor choice and biasing. That capacitor is unacceptable to me.
I firmly reject all technical and mathematical signal measurements, but I always pay close attention to subjective evaluations.
I firmly reject all technical and mathematical signal measurements, but I always pay close attention to subjective evaluations.
One sees a loop in the CFP and doesn't see one in the 2EF, however they are both hight feedback pairs. Both can oscillate because of the feedback.
CFP knowledge, proper compensation and treat, sure I would like to know more.
Hi guys; I don't remember well but it seems to me to have seen a P3A schematic of Rod Elliott in which there was a 2.2uF in parallel with Vbe multiplier transistor but now observing the schematic of P3A this mentioned capacitor does not exist; is this cap very important or not? Thank you very much for your attention
Member
Joined 2009
Paid Member
My 2019 P3A
At last, 2019 is my next visit to P3A circuit. I like to revisit amp projects when I think I have got new knowledge and design secrets that can be used to improve the circuit.
We can make improvement only if we know what to be improved, right? For me CFP designs always sound nice but to be considered as one of the best they usually lack something. They sound unnatural and they lack micro drive (whatever that means). All of them, including the P3A and Circlophone.
I have built more than a thousand amps, so I'm interested only with the best ones. So, when I say that an amp sounds 'bad', I suggest most people to not take it so seriously
First secret that I tried to bring into this circuit is the output transistors. I have researched all transistors sold in US suppliers (thru their parametric search engines) and those sold locally, and I couldn't find anything better than my two pairs. One is Panasonic B1317/D1975 and the other one I forget and I haven't tried yet.
Next secret is the importance of 'stability' to amplifier and how it can be fixed thru compensation techniques. The unnatural sound of CFP is indeed caused by the nature of its instability trait (plus a little from the lack of drive).
To fix the drive issue without changing the drive circuit (to keep it a P3A), I can see that the driver current should be increased. At least the collector resistor should be reduced to below 100 Ohm (original is 220 Ohm). All CFP that I have seen (commercial or not) have this low value collector resistor (especially those with better drive). But after studying and fiddling I can appreciate why Rod chooses 220 here. I don't want to reduce it either. Stability is paramount in CFP to make the amp sound natural. The issue with drive, I will solve with other approach, even if I have to accept that it is not the best amp at that quality parameter.
I choose 30V for supply because I want to be able to use devices with 50V Vceo. The thing that I disagree with Rod and Sakis and most people here is that I need high bias current for the amp to sound good. I'm now using 0.47 output resistors, and I tried to keep 60mv across it, so 128mA of bias current!
First setup that is considered as 'worth to keep' has these: bc546 input, b649 vas, c5171 driver, c2632-r bias, B1317/D1975 output. 100n across drivers' bases.
At last, 2019 is my next visit to P3A circuit. I like to revisit amp projects when I think I have got new knowledge and design secrets that can be used to improve the circuit.
We can make improvement only if we know what to be improved, right? For me CFP designs always sound nice but to be considered as one of the best they usually lack something. They sound unnatural and they lack micro drive (whatever that means). All of them, including the P3A and Circlophone.
I have built more than a thousand amps, so I'm interested only with the best ones. So, when I say that an amp sounds 'bad', I suggest most people to not take it so seriously
First secret that I tried to bring into this circuit is the output transistors. I have researched all transistors sold in US suppliers (thru their parametric search engines) and those sold locally, and I couldn't find anything better than my two pairs. One is Panasonic B1317/D1975 and the other one I forget and I haven't tried yet.
Next secret is the importance of 'stability' to amplifier and how it can be fixed thru compensation techniques. The unnatural sound of CFP is indeed caused by the nature of its instability trait (plus a little from the lack of drive).
To fix the drive issue without changing the drive circuit (to keep it a P3A), I can see that the driver current should be increased. At least the collector resistor should be reduced to below 100 Ohm (original is 220 Ohm). All CFP that I have seen (commercial or not) have this low value collector resistor (especially those with better drive). But after studying and fiddling I can appreciate why Rod chooses 220 here. I don't want to reduce it either. Stability is paramount in CFP to make the amp sound natural. The issue with drive, I will solve with other approach, even if I have to accept that it is not the best amp at that quality parameter.
I choose 30V for supply because I want to be able to use devices with 50V Vceo. The thing that I disagree with Rod and Sakis and most people here is that I need high bias current for the amp to sound good. I'm now using 0.47 output resistors, and I tried to keep 60mv across it, so 128mA of bias current!
First setup that is considered as 'worth to keep' has these: bc546 input, b649 vas, c5171 driver, c2632-r bias, B1317/D1975 output. 100n across drivers' bases.
Do we need luck?
Do we need luck to have a good sounding P3A? I'm not comfortable with the answer. Even tho a simple build will sound good, it is not good enough for me to keep it. I hate it that i don't fully understand the recipe. Here is the story...
I started with using the parts mentioned in above post, except with output. I tried C5200 then C2922 then my secret part D1975 where i found the natural sound for the first time with CFP. Beside using the D1975, my other goals were to use C5171 as driver (which i'm using right now) and i wanted to try C3953/A1538 (400MHz) as driver, probably with different output transistor and more modification to the circuit.
My first effort to improve on above setup was to try various transistors in the VAS position. But nothing were better than the previous B649. Actually, the high capacitance of B649 is what make it fit in the circuit if natural sound is wanted. This is in disagreement with others who worship on very low Cob.
After having no more parts to try in the VAS, I wanted to put back the D649. But first I tried the D649 with the highest and the lowest hfe in my collection (165 and 128). And to my surprise i didn't find the reason why i wanted to keep the amp! I'm sure that i have good ears. I'm sure thay it was great before and it is not with the last two transistors. Fingers crossed, i put back the previous mid hfe D649 and... i was so relieved that i was not wrong. This is the best sound i have ever heard through my beloved 2 inches test speakers.
So that's it. If it is not for this specific D649, i was not sure i want to keep the amp. Was it luck? I haven't studied how such a small difference can make or break the amp. A little mystery for me.
I'm reluctant to change anything right now. I'm now building the other channel to beat the above setup. My plan was to use A1537 (700MHz) as VAS transistor. It should be a winner!
Do we need luck to have a good sounding P3A? I'm not comfortable with the answer. Even tho a simple build will sound good, it is not good enough for me to keep it. I hate it that i don't fully understand the recipe. Here is the story...
I started with using the parts mentioned in above post, except with output. I tried C5200 then C2922 then my secret part D1975 where i found the natural sound for the first time with CFP. Beside using the D1975, my other goals were to use C5171 as driver (which i'm using right now) and i wanted to try C3953/A1538 (400MHz) as driver, probably with different output transistor and more modification to the circuit.
My first effort to improve on above setup was to try various transistors in the VAS position. But nothing were better than the previous B649. Actually, the high capacitance of B649 is what make it fit in the circuit if natural sound is wanted. This is in disagreement with others who worship on very low Cob.
After having no more parts to try in the VAS, I wanted to put back the D649. But first I tried the D649 with the highest and the lowest hfe in my collection (165 and 128). And to my surprise i didn't find the reason why i wanted to keep the amp! I'm sure that i have good ears. I'm sure thay it was great before and it is not with the last two transistors. Fingers crossed, i put back the previous mid hfe D649 and... i was so relieved that i was not wrong. This is the best sound i have ever heard through my beloved 2 inches test speakers.
So that's it. If it is not for this specific D649, i was not sure i want to keep the amp. Was it luck? I haven't studied how such a small difference can make or break the amp. A little mystery for me.
I'm reluctant to change anything right now. I'm now building the other channel to beat the above setup. My plan was to use A1537 (700MHz) as VAS transistor. It should be a winner!
Yes, indeed. I have mine pulled from a new (working) NAD amplifier
The other transistor i was talking about before, were, are also in a working Sony amplifier. That's why i'm not too curious to see how it perform in P3A, because the Panasonic D1975 already performs as expected
Last edited:
Planning On Using A1537 As VAS in P3A
So we have heard people worship on high fT low Cob for a VAS transistor? A1538/C3953 (400MHz) video transistor is the driver in my best EF amplifier. Such transistors, especially in a VAS, can give you details you wont ever hear from ordinary transistors.
During my search for better than D649 in the VAS of P3A, I tried this A1538 (400MHz) and A1537 (700MHz). I don't remember which one gave me the most details, but A1538 was too far from sounding natural, while A1537 was only slightly unnatural. Untrained ears i believe will think that this transistor is more natural than the D649 due to the inner details it produced, which was very beautiful and would have been my preference if only it sounded more natural. It really has potential and i thought to solve the issue i would need to insert some resistors into the circuit, which unfortunately not provided in the PCB. I'm not sure of the best way to incorporated extra resistors to the already crowded PCB but I'm sure this will be in another league than previous setup!!
So we have heard people worship on high fT low Cob for a VAS transistor? A1538/C3953 (400MHz) video transistor is the driver in my best EF amplifier. Such transistors, especially in a VAS, can give you details you wont ever hear from ordinary transistors.
During my search for better than D649 in the VAS of P3A, I tried this A1538 (400MHz) and A1537 (700MHz). I don't remember which one gave me the most details, but A1538 was too far from sounding natural, while A1537 was only slightly unnatural. Untrained ears i believe will think that this transistor is more natural than the D649 due to the inner details it produced, which was very beautiful and would have been my preference if only it sounded more natural. It really has potential and i thought to solve the issue i would need to insert some resistors into the circuit, which unfortunately not provided in the PCB. I'm not sure of the best way to incorporated extra resistors to the already crowded PCB but I'm sure this will be in another league than previous setup!!
I think that the better correlation between sound quality and VAS transistor properties is Cob, though Ft is usually related to this too. Transistors typically used by hobbyists and supplied in kits are common types selected for Vce max, power rating and low cost but not much else. For some builders who like sound effects though, even crazy choices can sound good, as they seek to enhance the splash of cymbals or details like audience, player and instrument noises which they find entertaining.
As DIYs we have choices, provided the amplifier stability margins allow for the use of anything from obsolete video driver transistors (which offer the lowest Cob and best linearity) to audio output stage drivers and general purpose or switching transistors which may have relatively high Cob and often poor linearity in this role. D667/B647 for example, are drivers with a Cob about 20pF so you would expect different results to the best video drivers. Bear in mind that Cob is a voltage dependent capacitance (Varicap) which is in parallel here, with Cdom, the miller compensation cap. Think about what swapping VAS transistor types and the effect of wobbulating, extra capacitance does to the compensation and signal
And how it relate to the other gain devices too as VAS is not the only gain stage defining the sound. Low Cob is good when the circuit and other gain devices can 'support' it.
Fortunately tho, as long as we are looking for the 'music' in the sound, we will never get lost. Music doesn't always need a cymbal splash, and a cymbal splash doesn't always musical, but a musical cymbal splash is cool
HF can be entertaining, but along with wide bandwidth and the fact that transistor gain is falling at HF, we will found issues that we may realize or worse not. Hard to prove anything here tho.
What will be the difference if the circuit is designed to 'hide' it.
Think about what swapping VAS transistor types and the effect of wobbulating, extra capacitance does to the compensation and signal
It can create severe effect which is called oscillation, or much less severe effect in the form of unnatural sound. But like always, it depends on how good we can hear, otherwise everything sound the same and there is no point of designing the best sounding amp as it has been done long time ago.
Why not use a EF-CFP output configuration?
I read the discussion about the difference between a EF and CFP output stage. Why not combine the best of those two worlds?
More than 10 years ago I published the Ultimate Suplifier. This is a class A bridged power amplifier with no global feedback. It has a EF-CFP output stage. Later I found that the Audiolab/Cambridge 8000 serie has the same output configuration.
Hereby the schematics. Of course the values of the resistors etc. have to be adjusted.
I read the discussion about the difference between a EF and CFP output stage. Why not combine the best of those two worlds?
More than 10 years ago I published the Ultimate Suplifier. This is a class A bridged power amplifier with no global feedback. It has a EF-CFP output stage. Later I found that the Audiolab/Cambridge 8000 serie has the same output configuration.
Hereby the schematics. Of course the values of the resistors etc. have to be adjusted.
Unfortunately that it is not a P3A anymore. But i think that it seems the right way to do it, i.e driving the CFP with more 'powerful' VAS. Tho honestly i still don't fully understand why a CFP can sound musically the way it does.
Also check how it is done in Naim Nait XS which is based on a chip for driving CFP (LM319?). But i think that the simple topology used by P3A is still capable to perform sufficiently when built carefully. Idem with LTP.
Last edited:
So the keyword is linearity. My best EF amp is based on this concept of linearity from input to output. But it doesn't sound like a CFP or P3A. But what i like about it (the EF one) is the sensation of watching live music in much larger scale than the size of the speaker. But i think this P3A can impress more people That's why i plan to use it at work to show off.
That's why i plan to use it at work to show off.