Another thing to consider is driver junction temp. If I run a sim at 1/3 power (20W at 8R load), the drivers dissipate around 120mW.
I start the sim with the drivers and output devices at 25C and set the bias at 50mA.
If the drivers do not have heatsinks, their junction temp rises to around 50C and bias rises to 188mA
If I use a 7.5 °C/W heatsink (like Aavid 531202B02500G), their junction temp rises to only 30C and bias rises to 79mA.
Now, running a sine wave 1/3 power is pretty tough and likely harder than anyone would run it it real life (if you want to keep your hearing), but it does illustrate that keeping the drivers cool helps with bias stability
I start the sim with the drivers and output devices at 25C and set the bias at 50mA.
If the drivers do not have heatsinks, their junction temp rises to around 50C and bias rises to 188mA
If I use a 7.5 °C/W heatsink (like Aavid 531202B02500G), their junction temp rises to only 30C and bias rises to 79mA.
Now, running a sine wave 1/3 power is pretty tough and likely harder than anyone would run it it real life (if you want to keep your hearing), but it does illustrate that keeping the drivers cool helps with bias stability
ZTX also has some nice tiny through hole transistors for that place. Anyway i understand your point but because i have better option available i didn't think about 1W types. If you can remember, once very popular DIGI-125 also used BC556 for it's VAS. Btw what is the best approximate value of so called "speed up" capacitor across the bias tr Q9 ? I think it depends on used active components & layout.
I don't think the value of the cap is critical. Anything in the 100n to 220n is commonly used. However, some design will put this in the 10u to 47u range. From what I've read on this, there's doesn't seem to be a lot of consensus on how to size this.
Maybe some far more knowledgeable than me can chime in.
Maybe some far more knowledgeable than me can chime in.
I don't want to build (not even possible..) every amplifier in this galaxy, so i picked couple of design on various topology. P3A is one of them. I have read extensively about it. But as you know it's very hard for amature enthusiasts to possess Exotic components like low noise Ksa992, so for LTP I have no choice other than BC546 (which is fine I guess, although i've 2n5551 but pinout is different. I think BC546 has the edge here because of it's lower VCE). Fortunately I have some excellent transistors for VAS stage. Rest is as Rod recommends, BD for drivers & MJL or TIP for outputs. Because of the nature of CFP I bought medical grade multi-turn Bourns trimmer from RS otherwise normal single turn fat blue trimmers are my go to choice but I thank you for the suggestion on Driver temperature. I'll keep that in my mind.
Last edited:
What do you mean by exotic KSA992? In the US, they are readily available from Mouser and DigiKey for $0.37 each. Buy 10 and they're $0.27 each. This is why I suggested them - easy to source and cheap. BC546 for example is $0.33 at DigiKey. The BD140 that is recommended for this position is $0.72 - nearly twice the price.
By exotic i meant their characteristics & the only good source of ksa992 i have is RS COMP but minimum order quantity is high (& when i buy such transistor i buy in pair). I thought when i have hundreds of genuine Fairchild BC546/56 why buy another type? I also have Fairchild KSA1381 (only pnp part) which you mentioned earlier as well as Sanyo 2SA2911 for voltage amplifying stage. For drivers BD139/40 from ST. So i have everything required for a 1st class P3A except time & intention. But as of now I'm happy with my LM3886 set up
Best Regards
I tried on simulation ksa1381 for VAS with very good result.
https://www.diyaudio.com/community/...rossover-p3a-of-rod-elliot-as-example.402148/
https://www.diyaudio.com/community/...rossover-p3a-of-rod-elliot-as-example.402148/
It doesn't matter, what is matter is the spec. ST is an European giant, has good reputation for their products.
MJE150XX series is like a tank! You don't need such transistors for P3A, also their collector to base capacitance probably higher than the to-126 BD139/40.
When you use progressive power supply caps, you usually use RC filters. Like 1R or 2R for the drivers and 10R or 20R for the Ips/Vas. You have the caps but no resistors