The whole idea behind P3A is to use Sziklai. Don't change the most important aspect of the design. I'd use old transistors because the design is suited for them, especially if you have many and can match them. The only aspect where old transistors are inferior to new ones is power dissipation. But if you stick with nominal 35V rails everything will be OK. It is my experience that the safest way with regards to stability with P3A is to use slower output transistors like TIP36C/TIP36C or higher powered MJL 21193/21194. If you use superior modern transistors proper testing on the scope is essential.
Hello Guys.
did anybody tried to use a BC550 instead of a BC546 on a P3A?
BC550 is the low noise version of BC546.
if yes, should I notice any difference and it worth to replace it on my two P3A I already have.
Thanks
https://www.fairchildsemi.com/datasheets/BC/BC546.pdf
did anybody tried to use a BC550 instead of a BC546 on a P3A?
BC550 is the low noise version of BC546.
if yes, should I notice any difference and it worth to replace it on my two P3A I already have.
Thanks
https://www.fairchildsemi.com/datasheets/BC/BC546.pdf
The datasheet numbers refer to a specific Vce of 5V and very low bias current which means that the improvement in noise may not be as great as you imagine - perhaps only up to a couple of dB. There is also the matter of the basic design's modest PSRR, which will overshadow transistor noise.
Understanding what datasheets really say, is like reading between the lines. You need to distinguish between numbers that refer to parts under optimum operating conditions that highlight the best possible performance, from the numbers more likely in your application.
The P3a input transistors already operate at a Vce near the + supply rail voltage and since many beginners plan to squeeze the max. possible 8 Ohm power by using transformers with 2 x 30V windings, the safe limit for BC550 operation is then easily exceeded with an unregulated power power supply and normal mains power fluctuations. With the recommended 2 x 25V transformer, I don't see any problem though.
Last edited:
Difference in distortion is actually negligible for these particular ones. Noise figure diffetence at this level is also unnoticeable for the power amp with voltage gain of around 30db.
It becomes important for the microphone preamp, with the gain of, say, 60db. Total gain of the whole (preamp + power amp) system becomes 90db (1000 times higher than the power amp's alone) - that's where the input stage noise level is vitally important.
For using those in P3A kind of circuit, what matters - is SOA, as Ian already mentioned, so my preference is BC546.
Cheers,
Valery
It becomes important for the microphone preamp, with the gain of, say, 60db. Total gain of the whole (preamp + power amp) system becomes 90db (1000 times higher than the power amp's alone) - that's where the input stage noise level is vitally important.
For using those in P3A kind of circuit, what matters - is SOA, as Ian already mentioned, so my preference is BC546.
Cheers,
Valery
In principle the IPS transistors have only half the total voltage between collector and emitters, and given that the bases are DC grounded this suppress about all risks, that being said i must admit that personaly i m not at ease with such conditions, that s why i could only recommend transistors like the 120V 2SA992/C1845, 2SA970/C2240 or eventually 2SA872A/C1775A, the latter pair is about the best IPS dedicated transistors but is limited to 300mW.
somebody is doing some thinking here but as usual is based on calculation and not real life conditions One should look Japanese amplifiers of the 70's 80's and understand a few things more about cooking at higher temperatures ....spectrum analyzers also on a given circuit /psu/structure will also have an opinion to add ....
Kindest regards
Sakis
Kindest regards
Sakis
Accuracy of schematic shown on the ESP web site
The schematic shown on Rod's web site appears to be different from the photograph of the populated PCB. The first obvious differences are the resistors on the output transistor collectors. The schematic gives a value of 0r33 ohms. The PCB picture has 0r22 resistors shown. This may have occurred to optimise the performance with different output devices. I also believe there have been a number of changes to the value of other resistors. The PCB is Rev. C. So it is likely the schematic refers to to earlier version. Do anyone know more about this?
I think the only definite way to identify the changes is to order a PCB from Rod. This will be my first DIY amplifier since I built an ETI480 in 1977. Should be fun and it just so happens I need a new amp for the study.
Regards ....
The schematic shown on Rod's web site appears to be different from the photograph of the populated PCB. The first obvious differences are the resistors on the output transistor collectors. The schematic gives a value of 0r33 ohms. The PCB picture has 0r22 resistors shown. This may have occurred to optimise the performance with different output devices. I also believe there have been a number of changes to the value of other resistors. The PCB is Rev. C. So it is likely the schematic refers to to earlier version. Do anyone know more about this?
I think the only definite way to identify the changes is to order a PCB from Rod. This will be my first DIY amplifier since I built an ETI480 in 1977. Should be fun and it just so happens I need a new amp for the study.
Regards ....
ESP makes and sells the PCBs for his published projects.
He states that all required BoM and set up data will be supplied with the PCB and gives access to restricted area of his site for his customers.
I would expect ESP to protect his IP by deliberately conceaaling the correct component values. He does much for free, why try to take away his source of income?
He states that all required BoM and set up data will be supplied with the PCB and gives access to restricted area of his site for his customers.
I would expect ESP to protect his IP by deliberately conceaaling the correct component values. He does much for free, why try to take away his source of income?
No complaint intended or implied
Fair enough! I have already ordered a number boards with the intention of doing side-by-side comparisons of different modifications. No slighting or diminishing of Rod is intended.
Just interested to hear of others experience with different circuit versions.
Fair enough! I have already ordered a number boards with the intention of doing side-by-side comparisons of different modifications. No slighting or diminishing of Rod is intended.
Just interested to hear of others experience with different circuit versions.
Hello Guys.
I did some tests on one of my P3As, So I tried IMD tests using 6.5kz and 7khz and
with a high biasing like 0.5 amps the performance on this test was much better.
at 50ma - the IMD was 0.045.
at 500ma - 0.016.
So my question this improvement is due to more bias on the output transistor or more bias on the drivers?
did anybody managed to increase the drivers bias without increasing the output bias?
Thanks
ps. I did the same test on a LM3886 ( IMD 0.065 ). they use the same test conditions. using the same psu and all other parameter.
I did some tests on one of my P3As, So I tried IMD tests using 6.5kz and 7khz and
with a high biasing like 0.5 amps the performance on this test was much better.
at 50ma - the IMD was 0.045.
at 500ma - 0.016.
So my question this improvement is due to more bias on the output transistor or more bias on the drivers?
did anybody managed to increase the drivers bias without increasing the output bias?
Thanks
ps. I did the same test on a LM3886 ( IMD 0.065 ). they use the same test conditions. using the same psu and all other parameter.
Last edited:
Member
Joined 2009
Paid Member
Just reduce R12 and R13, the collector resistors on the drivers.
You can see how I increased the current in this thread (a bit complicated though): http://www.diyaudio.com/forums/solid-state/245619-tgm8-amplifier-based-rod-elliot-p3a.html
In the original P3A circuit these resistors have unusually high value (low drivers bias) and one of the suggested tweaks is to increase bias by reducing the value of these resistors. Rod's idea was to make P3A as reliable as possible, therefore he chose low bias in order to reduce any thermal strain on the components.