I have tried that on simulator with no success. Don't know if actual implementation will be different, but if there is no candidate values from simulation, then what value to implement?
Does it always conform? Do you also measure the ESR/ESL of the generic cap you use? Easier (and accurate of course) for me to plug and play because I have many parts.
And different brand has different performance.
Best preamp for P3A
This is slightly off-topic but I guess still in the spirit of getting the best out of P3A.
Rarely we seen here people simulating pre+amp (while actually there is usually matching issues between pre and amp). So here is my little experimentation with B1+P3A in simulator. B1 is chosen because it is popular, easy to build and seems very suitable with P3A (a buffer shouldn't change much the character of an amp).
I started with very simple questions: what potentiometer value to be chosen, where to put the potentiometer, before or after the B1.
What I'm looking for is the minimum distortion at around 5W output (I think it is the most power used in domestic listening). This is associated with 0.3V input into the P3A (slightly less than 5W actually).
The B1 is "specified" with 25K pot at the input. I tried 50K, 25K, 20K, 10K (I believe a CDP has no impedance matching issue). The result is: the lower the potentiometer the better.
Now I put the potentiometer after B1 (I believe CDP performance has small effect). The result is better than putting the potentiometer in front of B1. And the best result is for 20K potentiometer.
This is slightly off-topic but I guess still in the spirit of getting the best out of P3A.
Rarely we seen here people simulating pre+amp (while actually there is usually matching issues between pre and amp). So here is my little experimentation with B1+P3A in simulator. B1 is chosen because it is popular, easy to build and seems very suitable with P3A (a buffer shouldn't change much the character of an amp).
I started with very simple questions: what potentiometer value to be chosen, where to put the potentiometer, before or after the B1.
What I'm looking for is the minimum distortion at around 5W output (I think it is the most power used in domestic listening). This is associated with 0.3V input into the P3A (slightly less than 5W actually).
The B1 is "specified" with 25K pot at the input. I tried 50K, 25K, 20K, 10K (I believe a CDP has no impedance matching issue). The result is: the lower the potentiometer the better.
Now I put the potentiometer after B1 (I believe CDP performance has small effect). The result is better than putting the potentiometer in front of B1. And the best result is for 20K potentiometer.
Have you tried bipolar electrolytics for C1 & C3? Less distortion with zero bias than a normal electrolytic. 40v/63v rated caps seem to be best....
Ref: http://www.waynekirkwood.com/Images/pdf/Cyril_Bateman/Bateman_Notes_Cap_Sound_5.pdf
Ref: http://www.waynekirkwood.com/Images/pdf/Cyril_Bateman/Bateman_Notes_Cap_Sound_5.pdf
Last edited:
well ... The B1 has less dynamics in this level than any preamplifier with a bit of gain Also to my ears the B1 tends to round a bit the ends of the graph
A discrete preamp with proper semis and proper psu if made properly can beat the B1 in how real and how clear the sound can be and add the missing dynamics in low levels .
One thing here that is important and remains a complain among the people that listen to P3A is that the amplifier is way too real ..IE ..signal is getting amplified without anything changing on it so in a good production it will rock while in a bad production it will reveal all the negatives from it and throw them in to your ears ...With a B1 as a pre things are better at this point since the B1 tends to round the ends as i said and in a way ""cover mistakes in this area""
Generally for this amount of power i would go with a preamp with some little gain use a 50K pot at the input of the pre ...and skip a DC block capacitor either in the amp or the preamp
This type of setup i listen for 4 years now and in a range of 400-500 amplifiers per year that i repair and listen next too , i have found very very few that can beat the combo ...
A discrete preamp with proper semis and proper psu if made properly can beat the B1 in how real and how clear the sound can be and add the missing dynamics in low levels .
One thing here that is important and remains a complain among the people that listen to P3A is that the amplifier is way too real ..IE ..signal is getting amplified without anything changing on it so in a good production it will rock while in a bad production it will reveal all the negatives from it and throw them in to your ears ...With a B1 as a pre things are better at this point since the B1 tends to round the ends as i said and in a way ""cover mistakes in this area""
Generally for this amount of power i would go with a preamp with some little gain use a 50K pot at the input of the pre ...and skip a DC block capacitor either in the amp or the preamp
This type of setup i listen for 4 years now and in a range of 400-500 amplifiers per year that i repair and listen next too , i have found very very few that can beat the combo ...
Perhaps, distortion will decrease but this is measurement, sonically you may prefer the bd as many prefer the sound of distortion. Look at tubes, some produce 2 3 percent distortion and some people love it that way. As these parts cost max around a dollar I dont think its expensive at all to just try and listen.
The Transistors that I used for my P3A:
Q1,Q2,Q3=2N5551- input stage.current source
Q4=2SB649- VAS stage
Q5=2SC2073 -npn driver
Q6=2SA940-pnp driver
Q7=2SA1941-pnp output
Q8=2SC5198-npn output
Q9=BD139-Vbe Multiplier
For me I'm satisfied with the combination of these trannies.The tweeters sounded so good, Even the bass response is perfect for me. The sound of cymbals so crisp that's why I used these drivers 2SC2073 and 2SA940. Having a high frequency drivers the sound of its tweeters gives a thin sound. I prefer using a low ft 4mhz driver to have a balance sound thus, giving a clear transparent sound that I like. I tried 2SB649 and 2SD669 for the drivers but I prefer the lower ft driver. It is perfect for these output 2SC5198 and 2SA1941. Thanks.
Q1,Q2,Q3=2N5551- input stage.current source
Q4=2SB649- VAS stage
Q5=2SC2073 -npn driver
Q6=2SA940-pnp driver
Q7=2SA1941-pnp output
Q8=2SC5198-npn output
Q9=BD139-Vbe Multiplier
For me I'm satisfied with the combination of these trannies.The tweeters sounded so good, Even the bass response is perfect for me. The sound of cymbals so crisp that's why I used these drivers 2SC2073 and 2SA940. Having a high frequency drivers the sound of its tweeters gives a thin sound. I prefer using a low ft 4mhz driver to have a balance sound thus, giving a clear transparent sound that I like. I tried 2SB649 and 2SD669 for the drivers but I prefer the lower ft driver. It is perfect for these output 2SC5198 and 2SA1941. Thanks.
Last edited:
Member
Joined 2009
Paid Member
I have always thought that the CFP is a poor choice for Class AB output due to the awful switch-off behaviour of the CFP pair. I'm not alone in this regard. Yet this amplifier has a good reputation. Mr. Self found that the distortion of CFP is lower than double EF at high signals but was worse at very low signals. I'd be interested to know whether the CFP is in reality worse, better or equal to double EF ?
CFP amplifiers of this range will play far better than EFP the bad reputation is next to them mostly because people do not know how to make them properly and most of the times look at a CFP schematic but always think inside of them EFP wise ... Since EFP is more common ..That alone is a very bad practice and most of the time and for most people with just a quick look it is impossible to understand the principal of operation for a CFP amplifier .
Consumer wise only an amount of 2-5% of consumer amplifiers was made like that .
As about bigger amplifiers CFP( with many outputs ) made i think that this cannot be done properly with today's available technology /schematics
Kind regards
Sakis
i built more than a dozen of Daniel Meyers' Tigersaurus amps in the 80's and 90's...they indeed have punchy bass even with just 10,000fd/100v filtering for a stereo amp.....but the super leach amp beats it overall....
so maybe i can agree that for small amps like Rod's P3A, CFP can do....
CFP crossover distortion is indeed less than EF but the crossover region is narrow, which results in low levels of high order distortion extending well beyond audible limits. EF crossover distortion is typically of greater magnitude but lower order because the crossover region is broad, even if the discontinuity is worse.
Subjectively, we prefer low order distortion but the P3A design is way overbiased at 70 mA or more. The result is different to what you may be guessing and by using a sound card or spectrum analyser with low, < 500Hz tones, it is quite interesting to compare the design as specified to when using "optimal", 13 mA bias. That is where the sound quality becomes identifiably CFP.
Randy Slone discussed and illustrated this well in his High Power Amplifier Construction Manual. Others discuss it too but I think his is the simplest and easiest to understand presentation. In a nutshell, don't judge the amplifier simply by it's generic design. SQ is not defined simply by topology or power supply etc. For example, we often percieve bass strength to be greater, not because it is, but there are more upper harmonics above bass frequencies - around 100-200 HZ that give it greater presence and definition. It may be quite a synthetic effect but we tend to like it and you do hear guys ranting about it as a positve in reviews with some music types.
Subjectively, we prefer low order distortion but the P3A design is way overbiased at 70 mA or more. The result is different to what you may be guessing and by using a sound card or spectrum analyser with low, < 500Hz tones, it is quite interesting to compare the design as specified to when using "optimal", 13 mA bias. That is where the sound quality becomes identifiably CFP.
Randy Slone discussed and illustrated this well in his High Power Amplifier Construction Manual. Others discuss it too but I think his is the simplest and easiest to understand presentation. In a nutshell, don't judge the amplifier simply by it's generic design. SQ is not defined simply by topology or power supply etc. For example, we often percieve bass strength to be greater, not because it is, but there are more upper harmonics above bass frequencies - around 100-200 HZ that give it greater presence and definition. It may be quite a synthetic effect but we tend to like it and you do hear guys ranting about it as a positve in reviews with some music types.
Last edited:
Member
Joined 2009
Paid Member
Hi Gareth
I couldn't say what rode's criterion for the P3's 70mA nominal bias was. High bias certainly was the norm for older commercial CFP designs like Hitachi and Sonys' which people still speak well of for SQ on these usually hot running amps. we are told that the P3 genesis was many years ago too.
According to Self though, the "optimal" CFP bias for a single pair, R(e) = 0R33 is: Iq 8.5 mA and Vq 5.6 mV
"optimal" EF2 bias for a similar pair, Re = 0R33 is: Iq 74 mA and Vq 48 mV
The latter may be more familiar figures but there are fewer junctions to bias with CFP in any case, so Vbias is only half the EF value to start with. All the same, many people here treat bias as arbitrary but an experienced majority seem think 100mA is about right for an EF pair.
13mA is cited by Randy Slone and is about "optimal" for Re=0R22 so to be consistent, I quoted his nominated bias current for the CFP output stage.
The point I was trying to make though, was that if the bias is so far from the textbook level, there's going to be little correlation with the model performance and the SQ won't be what we might expect. It's certainly true of the P3 and similar designs anyway.
I couldn't say what rode's criterion for the P3's 70mA nominal bias was. High bias certainly was the norm for older commercial CFP designs like Hitachi and Sonys' which people still speak well of for SQ on these usually hot running amps. we are told that the P3 genesis was many years ago too.
According to Self though, the "optimal" CFP bias for a single pair, R(e) = 0R33 is: Iq 8.5 mA and Vq 5.6 mV
"optimal" EF2 bias for a similar pair, Re = 0R33 is: Iq 74 mA and Vq 48 mV
The latter may be more familiar figures but there are fewer junctions to bias with CFP in any case, so Vbias is only half the EF value to start with. All the same, many people here treat bias as arbitrary but an experienced majority seem think 100mA is about right for an EF pair.
13mA is cited by Randy Slone and is about "optimal" for Re=0R22 so to be consistent, I quoted his nominated bias current for the CFP output stage.
The point I was trying to make though, was that if the bias is so far from the textbook level, there's going to be little correlation with the model performance and the SQ won't be what we might expect. It's certainly true of the P3 and similar designs anyway.
Last edited:
For P3A 70mA bias is certainly overbias condition where gm doubling causes distortion to rise, not to fall. Could it be that shifting bias level up (relative to optimal) serves as some kind of subtle tone control, a kind of aural exciter? May be that is the reason why most people prefer overbias to optimal bias?