Hi friends.. according to Rod Elliott, he fixes Bias current at 75 mA measuring a Voltage drop av about 50mV between the collectors of the output transistors. I fixed at 60mA, I think I can increase Bias current a little more.
It will be hard to hear it for most people, unless at the extremes (too low or too high bias).
Set the bias low, listen to HF sound such as cymbals and increase the bias until you can hear a better/natural cymbal, then you will find out that the bias is higher than recommended. Fatigue is actually very related with HF, so when HF is problematic, expect a higher level of fatigue. This is the problem with low bias amplifiers.
When you increase the bias in a CFP, HF becomes better, the sound becomes smoother and less disjointed. At the same time when bias is higher, the crossover distortion becomes more severe (in EF crossover distortion is a 'dead band' where both sides do not conduct, in CFP crossover distortion is when both sides conduct at the same time). Here you lost 'definition', soundstage and the vocal can become smoothed out to the point where pronounciation is not clear.
If you think the low bias version sounds okay then stick with it. P3A is designed for such low bias (which you can see from 220R driver collector load, unlike other CFP designs which have R below 100R). It is the point where P3A is very 'impressive' sounding.
If your ears are sensitive to fatigue (for example, there are those who cannot accept anything less than class-A amplifiers) then the only way to increase bias without sacrificing the sound too much is to increase driver current by lowering the 220R.
When you change the bias (or the driver current), stability may change also, and this affect the sound. In my P3A, more bias more stable. But when you reduce the 220R I think there will be more challenge for stability.
Good explanation Johnego..then you play with the 220R resistors to vary the Bias current..I am gonna take your explanations in account .
I've been experimenting with quiescent current settings, after leaving my P3A at 47mV/70.5mA. I tried 20mV/30mA.
To quote RodE: "When you are satisfied that all is well, set the bias current. Connect a multimeter between the collectors of Q7 and Q8, at the power transistor end of R13 and R14 - you are measuring the voltage drop across the two 0.33 ohm resistors. The desired quiescent current is between 25 and 100mA, so the voltage you measure across the resistors should be set to between 16.5mV and 66mV. The setting is not overly critical, but at lower currents, there is less dissipation in the output transistors. Current is approximately 1.5mA / mV, so 65mV will be 98mA."
And: "Testing has shown that the quiescent current can be as low as 20mA without evidence of visible or audible crossover distortion (13mV across the two 0.33 ohm resistors)."
This is for the rev b, for rev c he mentions a low setting of 16mV/24mA.
There is no single, optimal recommendation in Rod's testing procedures. I find lowering it to 30mA provides improvements in overall ease and openness of sound to my ears. Of course, there's also less steady state output device dissipation, which should increase lifespan and save money for those of us who like to leave our amps on.
Cam
To quote RodE: "When you are satisfied that all is well, set the bias current. Connect a multimeter between the collectors of Q7 and Q8, at the power transistor end of R13 and R14 - you are measuring the voltage drop across the two 0.33 ohm resistors. The desired quiescent current is between 25 and 100mA, so the voltage you measure across the resistors should be set to between 16.5mV and 66mV. The setting is not overly critical, but at lower currents, there is less dissipation in the output transistors. Current is approximately 1.5mA / mV, so 65mV will be 98mA."
And: "Testing has shown that the quiescent current can be as low as 20mA without evidence of visible or audible crossover distortion (13mV across the two 0.33 ohm resistors)."
This is for the rev b, for rev c he mentions a low setting of 16mV/24mA.
There is no single, optimal recommendation in Rod's testing procedures. I find lowering it to 30mA provides improvements in overall ease and openness of sound to my ears. Of course, there's also less steady state output device dissipation, which should increase lifespan and save money for those of us who like to leave our amps on.
Cam
This would depend on the build. Whether you use exactly the same parts or whether your transistors have similar specs. IF every other parameters are on the right track, best quiscent is the lower one because this represent the better crossover behavior. My current setup can achieve the 'sweetspot region' with 13.3mv (steady at 14mv) at my usual listening level (signal on, because i monitor the bias in real time).
Quiscent current itself means nothing. It is not the quiscent itself that makes or breaks the sound 😉
Because when you turn on the amp, there will be time until the 'steady' current is achieved, you can actually turn on the amp and listen to the music while the quiscent current slowly increasing to its steady state. When the music starts to sound 'proper' check the current. Mine it is 13.3mv. From cold start it takes at least half an hour to reach the steady 14mv.
In part since my standard-build P3A sounds noticeably better at 30mA than 71mA, I'll try the advice to go even lower and listen as bias is slowly increased. Lowest point imentioned in Rod's documentation is 13mV for rev b, 16.5mV for rev c, so I may as well start at 13mV for my rev b.
Johnego, how low was your beginning voltage, and is your amp a P3A, or? And how do you determine the point when bias is TOO low?
Johnego, how low was your beginning voltage, and is your amp a P3A, or? And how do you determine the point when bias is TOO low?
I built Rod's circuit (as it is) many years ago. This year I have built the original cct for a few hours or days but my supply is 31v, where the original cct should've worked better and differently with higher voltage.
I think with the original cct, because it doesn't conform to my standards, it have many compromises such that the preferred bias will depend on your taste. I'm used to class-A smoothness so if I were you I would pick the highest recommended bias which is 70mA. But i think most people will prefer lower value as the sound will be more impressive. Don't think it too hard though, as your bias preference with CFP amps might change with different SPL.
My CFP (0.33 output resistor version) started around 3mv and stay below 10mv for 3 minutes. I think class-B will be more noticeably hard on the ears on violins so that might be what you want to listen to. I don't just use ears to determine the minimum bias but I measure the cct (but it is a different cct than P3A).
You don't need to fuss about waiting for things to warm up before bias settles. Bias is controlled by the driver transistors which warm up promptly so all you need do is strap the Vbe multiplier to one of the drivers, as you see in Rod's pic of the latest P3a revision.
Previously, you'll see DIY versions of P3a where a small heatsink is used for both drivers and also the Vbe multiplier transistor. See Sakis' long comparison thread for his version at the OP.
Previously, you'll see DIY versions of P3a where a small heatsink is used for both drivers and also the Vbe multiplier transistor. See Sakis' long comparison thread for his version at the OP.
Yes, understood. I have done both ways. It is not the issue of thermal coupling alone. The mounting and thermal coupling is not yet my priority. I think I have got the pinnacle sound with 120R of driver's resistor, which is hotter than the original 220R, and my next improvement will be with smaller resistor (higher driver bias) which will require bigger heatsink (the 3 transistors are less than 1cm from each other, on common heatsink).
I like jazz and classical, simply put, also folkish music. Nowadays there's more difficulty in naming and tightly categorizing musical styles. Much of the new jazz I listen to is very atmospheric, with solid bass lines and interesting soundstaging. The classical is contemporary, with uncoventional composition and sonic surprise.
I settled for now on a bias of 17mV, and that seems the lowest that still provides a more vivid and dynamic sound with very good staging at low to moderate volume settings. I feel something is lost at extremes of 13mV and 66mv. Again, this is with a P3A built very closely to basic Elliott component and circuit specs.
Cam
I settled for now on a bias of 17mV, and that seems the lowest that still provides a more vivid and dynamic sound with very good staging at low to moderate volume settings. I feel something is lost at extremes of 13mV and 66mv. Again, this is with a P3A built very closely to basic Elliott component and circuit specs.
Cam
Yes, for your music preference it is better to go near the minimum. Actually, I think minimum bias is the better option. With different implementation, optimum bias will be different from one to another. But there is always a MIN bias and a MAX bias. This range in my amp is very narrow (unlike in original P3A). MIN bias is rather easy to detect as it is like a quasi-saturation curve, i.e. there is an abrupt change from too low a bias to its MIN bias.
At minimum bias everything you like (vivid tubey sound or in-your-face sound or bass dynamic/PRAT) is maximal, except 'distortion', as if you're listening to class-B crossover distortion or a high distortion full-range speakers. I like to watch TV/movie and music at the same time. With this MIN setup I feel disturbed with the music when I need to focus on watching the movie. So I see the need to fix this by increasing current.
At random high bias (say 100mA), the amplifier's strengths (tubey sound, bass dynamic/PRAT) are completely missing but the sound is smoother, pushed to the back like a normal flat multi-way speakers. The problem is that this tend to be fatiguing/boring in the long run. From this setup the bias should be reduced to its MAX bias value where you start to hear the vocal magic and bass dynamic again, even tho not as much as what you can hear at the MIN value. And I personally opt for this MAX bias point.
I can't agree that the sound deteriorates as much as losing its strengths as bias is increased above a nominal 17mV with the P3a. I find like Rod and other folk here, that there is a compromise region around 70mV that sounds the best and is less fatiguing than at low bias, where higher harmonics sometimes irritate, depending on the programme material.
What supply voltage, power transistors and drivers are you currently using, johnego?
What supply voltage, power transistors and drivers are you currently using, johnego?
I said there is MIN bias and there is MAX bias. Above the MAX bias it sounds bad. Not above MIN bias. The MIN/MAX values are really dependent upon the circuit. Mine it is from 3.2mA to 7.7mA (surprisingly low!). P3A MAX point is I think 70mA (I always prefer the MAX value). Crossover distortion (cross-conduction, gm-doubling) is higher with higher bias. Really depends on the circuit.
I have several versions, currently the best sounding one is version_12B. I'm trying to milk the sound quality little by little. Power supply is currently 30V. Later if the amp is the better one than the other topology then it will be redesigned for 36V supply.
Output transistors will be B317/D1975 (already tried) and A1672/C4387 (later, for 3-way active speaker). Driver will be A1930/C5171 and will try later A1538/C3953.
Thanks for the detail. You have a variety of transistor types there, with some likely copies from China too, I think. Assuming they are genuine OEM type though, at least some would have to be removed from old equipment. Are they alleged NOS (i.e. old stock)?
My Ali-Express and Ebay purchases of some of those semis were falsely advertised as genuine original parts. Now I avoid their semis wherever possible. They were actually shiny, new style transistors that were allegedly like, 30 years old??) and there were also new transistors with those part numbers that looked the same but were are simply offered as copies.
It is not clear to me with Alibaba, Ebay and resellers of parts from such sources, just what specification parts we would be getting so I stay with authorized distributors because long term, it's cheaper to use original, specified parts so I only have to do the job once and I'm in no doubt about the results.
Sometimes, fakes and even bad copies work fine in our circuits too but when we share results with others, we may be comparing notes about very different semis. I ask because there are many threads where the DIY or kit builder has no idea about the semis he really has, nor does he know why his clone may seem so different to the original or other builds discussed on the forum.
I now use the more robust NJW0281/0302 with the recommended nominal +/-35V supply and 60-70mA bias. It runs quite warm but sounds as good as it gets in my experience too 🙂
My Ali-Express and Ebay purchases of some of those semis were falsely advertised as genuine original parts. Now I avoid their semis wherever possible. They were actually shiny, new style transistors that were allegedly like, 30 years old??) and there were also new transistors with those part numbers that looked the same but were are simply offered as copies.
It is not clear to me with Alibaba, Ebay and resellers of parts from such sources, just what specification parts we would be getting so I stay with authorized distributors because long term, it's cheaper to use original, specified parts so I only have to do the job once and I'm in no doubt about the results.
Sometimes, fakes and even bad copies work fine in our circuits too but when we share results with others, we may be comparing notes about very different semis. I ask because there are many threads where the DIY or kit builder has no idea about the semis he really has, nor does he know why his clone may seem so different to the original or other builds discussed on the forum.
I now use the more robust NJW0281/0302 with the recommended nominal +/-35V supply and 60-70mA bias. It runs quite warm but sounds as good as it gets in my experience too 🙂
Yes, Ian, they are old stock. I owned them before ebay was invented. And in my city I can purchase bad copies easily so I had the chance to compare. The use of cheap Chinese components had been adopted since around 5 years ago, not only by individual but government-owned companies. I have purchased lots of these copies (often with many times the normal price), all is terrible.
Member
Joined 2009
Paid Member
You have to be quite careful these days, sometimes these fakes are sold from other countries such as the US in an attempt to give the buyer a false sense of provenance. If you can avoid using NOS parts and be happy with the great choice of super performing parts available from trusted sources of current manufacturers (I favour Digikey in my neck of the woods) - assuming you are willing to pay the cost. Or you have to do some dumpster diving and find some old equipment built back when parts were good.
Yes, I'm also aware of that. From diyaudio site I also noticed that many have been satisfied with copies of K170 and J74. I have purchased a few for my stock, never used them and I'm curious if it is anywhere near the original. If only by measuring the basic parameters, many of these copies are even better than the original. Not until you 'listen' to it.