Re: cool idle temp.
With +/- 28V rail voltages, the amplifier will dissipate less power for the same current (20-35% less) which may seem relatively cool when idle.
Also, we can see from the ESP article that the 5W power resistors are now 0R22 rather than 0R33 which the rest of the notes, schematic and replies here ignore. Be certain you are using the actual resistor values fitted for calculating current.
With +/- 28V rail voltages, the amplifier will dissipate less power for the same current (20-35% less) which may seem relatively cool when idle.
Also, we can see from the ESP article that the 5W power resistors are now 0R22 rather than 0R33 which the rest of the notes, schematic and replies here ignore. Be certain you are using the actual resistor values fitted for calculating current.
Can you provide a link to this article? The ESP project page I'm familiar with shows these resistors as 0R33.
Hi brian92fs. The project page is actually what I referred to. Sorry to confuse by calling it an article. The pic. of the new PCB and part of the text indicate 0R22 but the schematic shows 0R33 which isn't the only variation in the latest revision C. To get the full details, you have to buy the PCBs to get them in the documentation. That's how ESP stays in business.
People here that follow the Figure 1 schematic will get an older version of the design which still works very well as the replies here tell us but you'll need to spend money to get the latest revision to the design, PCB and parts. This becomes clear when you sign up and read the ESP forum pages.
People here that follow the Figure 1 schematic will get an older version of the design which still works very well as the replies here tell us but you'll need to spend money to get the latest revision to the design, PCB and parts. This becomes clear when you sign up and read the ESP forum pages.
Hmm. So it's only in Rod's pic. that 0R22 resistors are fitted. I don't have the rev.C PCB or BOM to compare and I stand corrected, though DIYs still need to use the actual values in calculations, which is the point I was making. I'd be interested to see what he uses in his own active loudspeakers which he has said are P3a.
Indeed, a similar comment has been in the text for some time. I assumed it was a sensible precaution for those who are new to DIY and could have problems, especially with some of the power supplies, heatsinks, fans and enclosures (or lack of) that I've seen cobbled together to complete the amplifier.
Bias stability is supposed to be not an issue, whatever the R value. From DC operating points pov, no difference whatever R you choose (as long as you adjust the DC OP with the pot). But the sound will be different. I prefer higher R (0.47), tho it means less output transconductance. With high R, the drive can be won back from increasing LTP current (with its accompanying issues) but the preferred distortion spectrum is hard to dial in.
I'd read up on bias stability for both CFP and EF output stages. Neither are completely stable with varying temperature unless some criteria can be met. When you use 75mA bias current, this often is an issue.
The simplified conclusions and procedures in this paper may help gain some understanding: http://www.guitarscience.net/papers/biasdsgn.pdf
The simplified conclusions and procedures in this paper may help gain some understanding: http://www.guitarscience.net/papers/biasdsgn.pdf
Bias stability is always an issue and it has a lot to do with the thermal design of the project in total PCb also will play some role as you can read bellow .
Beyond choice of resistor and may be a few other electronics based considerations how the machine is made plays a quite big role .
For a CFP that senses bias on the driver (s) even more ...Idea is that output may warm as much as it likes as long there something monitoring the bias in the driver area so this area is to be kept away from the outputs to avoid overcompensation of the vbe multiplier
So in a way that means that the enclosure should be able to dissipate heat bulit in the outputs but also from the enclosure to ensure if possible a steady temperature plus ambient on the board area.
That is a few comments regarding thermals obviously there is a few other things related to electronics ...
75ma no matter how much power supply will produce some heat One good idea will be to adjust bias if possible ( or measure it ) with the amplifier at a normal operating temperature or pretty warmed up after 15 minutes of playing at 80% ( that will also include cover on the top and so on. So expect it to be a complicated procedure and complication level is Audiolab or Stasis PA7 )
Some of the P3A i made had a hatch down under on the belly and trimmers located under the board specifically for the above reasons .
Kind regards
Sakis
Beyond choice of resistor and may be a few other electronics based considerations how the machine is made plays a quite big role .
For a CFP that senses bias on the driver (s) even more ...Idea is that output may warm as much as it likes as long there something monitoring the bias in the driver area so this area is to be kept away from the outputs to avoid overcompensation of the vbe multiplier
So in a way that means that the enclosure should be able to dissipate heat bulit in the outputs but also from the enclosure to ensure if possible a steady temperature plus ambient on the board area.
That is a few comments regarding thermals obviously there is a few other things related to electronics ...
75ma no matter how much power supply will produce some heat One good idea will be to adjust bias if possible ( or measure it ) with the amplifier at a normal operating temperature or pretty warmed up after 15 minutes of playing at 80% ( that will also include cover on the top and so on. So expect it to be a complicated procedure and complication level is Audiolab or Stasis PA7 )
Some of the P3A i made had a hatch down under on the belly and trimmers located under the board specifically for the above reasons .
Kind regards
Sakis
Last edited:
Many ways to solve it, so the bias stability is not really an issue. We should choose output R not based on the one that will give the best/easiest bias stability. Otherwise, just build with latfet EF, or the Blameless.
Based on initial prediction, i guessed that the optimal R was slightly above 0.33. But I started with 0.47 because it is the most stable option. I gradually improved the design by ears. When i tried 0.33, I didn't have the same patient to improve it, mainly because I don't have components that i needed, so it couldn't better my 0.47 one.
The hardest or most tricky thing in upgrade process is to find the correct bias value, because it can make or break the design. Everytime there is any component change, the bias should always be rechecked again by ears. Not enough just defining the standard value for base-to-base voltage of the drivers.
Last edited:
johnego, you seem confused because you say stability is not a problem, but then you say you started with 0.47 because it's the most stable option. Also, to this point we are only concerned with bias stability. Let's not mix it up with satisfying our personal and subjective sound quality requirements.
Obviously, if you are using lower than 75 mA bias current, the temperature will be reduced and the drift slower - maybe stable enough for home use but maybe not in hot weather, so we need to know what degree of stability is necessary and how to ensure it. That's the first priority.
Obviously, if you are using lower than 75 mA bias current, the temperature will be reduced and the drift slower - maybe stable enough for home use but maybe not in hot weather, so we need to know what degree of stability is necessary and how to ensure it. That's the first priority.
Of course I understand the relationship between output R and stability. Hence, when I said bias stability shouldn't be an issue, it means that bias stability is not a problem that cannot be solved. Many people mentioned latfet is the best option because of its temp coefficient. For me it is something strange to mention when everyone knows how to work with vfet or bjt. It will be silly if I choose 0.47 simply because it is the most stable. I choose the most stable options as a starting point. Another example is the Miller cap, I always start with 100pF (this fits with BD139/140), at the end it will be the lowest possible. Some topology is harder because the compensation value must be precise (higher or lower is not stable).
Take a look at P3B. Same circuit, class-A. For me it is not a first priority, may be last one.
Class A current control needs are different to those of AB and can be quite simple. All that is required is sufficient bias to ensure the transistors are at or above the Q point at which they remain in the linear region of their characteristic curve at all times. Temperatures are constantly high and small changes or drift of the bias should make little or no difference to performance. Some designers go to a lot of effort to reduce the high dissipation by various tricks with tracking bias but this is another topic.
Class AB, where a lesser and in this case arbitrary amount of bias is applied, should be controlled as best we can to avoid changes in sound quality with power supply voltage and temperature because the Q point will be changing with it and so too, the harmonic spread of distortion products. So generally, the bias current controller needs to be well matched to the crossover characteristic of the output stage and also quite stable.
Many old style class AB amplifiers tend to be simple and sensitive to temperature variations so audiophiles can be fussy about their warm-up period. Some go to the extreme of leaving their equipment on 24/7 to ensure it sounds right whenever they use it. This is another reason why we need to get class AB biasing under tight control and avoid the risky practice.
Class AB, where a lesser and in this case arbitrary amount of bias is applied, should be controlled as best we can to avoid changes in sound quality with power supply voltage and temperature because the Q point will be changing with it and so too, the harmonic spread of distortion products. So generally, the bias current controller needs to be well matched to the crossover characteristic of the output stage and also quite stable.
Many old style class AB amplifiers tend to be simple and sensitive to temperature variations so audiophiles can be fussy about their warm-up period. Some go to the extreme of leaving their equipment on 24/7 to ensure it sounds right whenever they use it. This is another reason why we need to get class AB biasing under tight control and avoid the risky practice.
Last edited:
Yes, and i have a few up my sleeve, so i'm not worry at all. I have only a few months experience with CFP but it is not as difficult as i thought it was. Packaged darlington was very hard. High output R was mandatory. But like anything else, if we can get access to other people previous works then we know how to set expectation.
Hi, is it possible to use 2SC2837-2SA1186 pair as output transistors for p3a?
fT is fairly high at 70mhz...
Edit: sorry. Someone asked this question long ago at the different thread. I just found it.
Transistor choice for P3A
fT is fairly high at 70mhz...
Edit: sorry. Someone asked this question long ago at the different thread. I just found it.
Transistor choice for P3A
Last edited:
Capacitors
Hello, I'm assembling a p3a, Alexmm PCB, I have a question about capacitors: what would be more suitable for c + and c- is ceramic or polyester?
And for C2, filter, a polyester or common ceramic ?
Those are the options for now, quality polypropylene are difficult to obtain.
Hello, I'm assembling a p3a, Alexmm PCB, I have a question about capacitors: what would be more suitable for c + and c- is ceramic or polyester?
And for C2, filter, a polyester or common ceramic ?
Those are the options for now, quality polypropylene are difficult to obtain.
Attachments
Polyester caps are only just ok in quality audio circuits - seldom specified now. The function of those 100n caps in parallel with "C+" and "C-" is to augment the filter performance of those electrolytic bypass caps at high frequencies, where electrolytics can be less effective at filtering noise. So it figures that unless the film caps have much better high frequency performance, they probably don't help at all. In fact, they are seldom specified nowadays and you don't see them on schematics here at DIYaudio.
In most cases, just fitting modern, good quality electrolytic caps alone means the 100n film caps are redundant. There would be little point to fitting the film caps unless they were high quality types such as polypropylene film, not cheap polyester or ceramic caps since they there to improve performance, not just come along for the ride. Otherwise, they are not necessary.
The 100p capacitor is an RF filter cap. so it needs to be effective at frequencies much higher than audio, so cheap ceramics and polyester caps are out. Most audio amplifiers used polystyrene fim caps in this role but it should also be OK to use genuine NPO or COG grade ceramic caps there. I doubt 100pF value polyester or polypropylene film types time will even be be available anyway.
As always, beware of fakes and low grade components unless you use reliable, commercial parts distributors for your components.
In most cases, just fitting modern, good quality electrolytic caps alone means the 100n film caps are redundant. There would be little point to fitting the film caps unless they were high quality types such as polypropylene film, not cheap polyester or ceramic caps since they there to improve performance, not just come along for the ride. Otherwise, they are not necessary.
The 100p capacitor is an RF filter cap. so it needs to be effective at frequencies much higher than audio, so cheap ceramics and polyester caps are out. Most audio amplifiers used polystyrene fim caps in this role but it should also be OK to use genuine NPO or COG grade ceramic caps there. I doubt 100pF value polyester or polypropylene film types time will even be be available anyway.
As always, beware of fakes and low grade components unless you use reliable, commercial parts distributors for your components.