P3A Comparison table ( long .... )

Member
Joined 2010
Paid Member
2SD669A/B649A are found all over Ebay/Aliexpress and they are passable copies, right down to a faked, old style Hitachi logo. They only come in one Hfe range, unlike the original Hitachi and Renesas parts. You will also find smaller TO92L versions, 2SD667/B647 in abundance and cheap.

Altogether, they share many similarities and uses with BD139/140 and BC639/640 but the voltage rating is much higher and gain linearity is improved. Very useful parts for the money, but not as tough or linear as the originals, IMHO.
 
I disagree. The purpose of this build was to see if slower output transistors will work without oscillations with parts values from the original schematics. Based on my experiences, I would recommend to builders to use such transistors, especially if they don't have scope to verify the results. Fast Japanese transistors are not recommended if the builder is not experienced in modifying the schematics and if does not have scope. TIP35/36 (for 35V supplies) and MJL21193/4 (for 42V supplies) are perfect candidates.


Your thoughts are wrong in both cases TIP 35 is too slow for audio amplifier of today's standards, SOA is poor , or if you like the average Japanese standard where 30MHZ output transistors used from 1970...But what actually forgive you is that this amplifier will play fine even with those or even worst transistors and regardless if the Vas current is 1ma or 30 ma

21193 is What i call an Audio Karma transistor IE direct replacement for almost everything ....Yeap its built like a tank fairly fast but most of us know that 150XX family drivers and 21193 outputs will never sound as smooth and mild as a classic Japanese transistor ...Ideally if i was buliding a P3A for a listener with a classic rock music content 15030 and 21193 will be my first choice .....Other than that 1302-3281 is the king combo !
 
Hi Sakis,

If your preference is 1302-3281 for the outputs (are MJLxxxx On-Semi versions desirable here?), which drivers would give the 'king combo'?
I'd suggest that you don't buy Toshibas since most are likely to be fakes.
That leaves ONsemi, or Sanken, but even these get faked, so buy from a reputable supplier that can guarantee genuine product and can fulfil warranty claims.
 
I'd suggest that you don't buy Toshibas since most are likely to be fakes.
That leaves ONsemi, or Sanken, but even these get faked, so buy from a reputable supplier that can guarantee genuine product and can fulfil warranty claims.
Exactly. I would never consider using non-reputable sources. Future Electronics appear to be ok and have ON-Semi MJLxxxxAG at a reasonable price (£1.64 ea, min qty 5).
 
Hi Sakis,

If your preference is 1302-3281 for the outputs (are MJLxxxx On-Semi versions desirable here?), which drivers would give the 'king combo'?

Hi I have used mjl4281/4302, njw3281/1302 and also njw 0281/0302 in this amp. I didn't face any issue.
3281/1302 is running for almost a year . mostly at normal listening levels but occassionally at loud level.
 
Last edited:
Hi I have used mjl4281/4302, njw3281/1302 and also njw 0281/0302 in this amp. I didn't face any issue.

May I ask which drivers you used with these outputs and how did you find the sound quality? The reason I ask is just to get the best possible sound with the most readily available transistors as some combinations reportedly sound better than others.
 
Last edited:
Hi Sakis,

If your preference is 1302-3281 for the outputs (are MJLxxxx On-Semi versions desirable here?), which drivers would give the 'king combo'?


If common heatsinked and closely matched BD829-830 combo was unbeatable both measuring and listening.
Though i wouldn't push those drivers above 35+35 There is plenty of Japanese drivers available outputs also for a bit lower power 4467-1694 is a very nice choice fast and reliable Plus that it seems that nobody has been messing up with them ....

Ian the 1047 combo is no longer obtainable from a trusted manufacturer There is plenty of them out there but none of them is to trust Not to forget that originally those was made by Sanyo which is no longer active as far as i know ....
 
Forgot that ...
Yes the Toshiba's have been cannibalized by many yet all of us should learn to recognize them there is many ways to do it ...
Point is that at east electronics we are in the repair business so we have been played around a few times but in most cases we manage since we have to order plenty and a lot of them ....

In the Local market you may get 1943-5200 starting from 0.80 euro each to 1.5 then to 2.5and then to 4 euro all of them look alike so yes then the choice is yours
Yet again there is many ways to do it
For example when a shop tags 1943 for 2.5 euro and 5200 for 1.5 that is the number one pointer to tell you that something is wrong
There is other ways also
Kindest
Regards
Sakis
 
Last edited:
May I ask which drivers you used with these outputs and how did you find the sound quality? The reason I ask is just to get the best possible sound with the most readily available transistors as some combinations reportedly sound better than others.

I didn't try different drivers. I was happy with the sq with bd's. Much better in subjective sq over vssa (strictly my personal perception).

Reason for not trying different drivers, is Rod's recommendation on his p3a page. (2nd paragraph)
 
Last edited:
Forgot that ...
Yes the Toshiba's have been cannibalized by many yet all of us should learn to recognize them there is many ways to do it ...
Point is that at east electronics we are in the repair business so we have been played around a few times but in most cases we manage since we have to order plenty and a lot of them ....

In the Local market you may get 1943-5200 starting from 0.80 euro each to 1.5 then to 2.5and then to 4 euro all of them look alike so yes then the choice is yours
Yet again there is many ways to do it
For example when a shop tags 1943 for 2.5 euro and 5200 for 1.5 that is the number one pointer to tell you that something is wrong
There is other ways also
Kindest
Regards
Sakis
You're in the fortunate position of being able to cannibalise for original Toshibas, however I don't have that luxury! That being said, reputably-sourced ON-Semi MJL1302AG/3281AG with BD829/830 drivers should produce reasonable results from what I can gather, unless you suggest otherwise? Thank you.
 
Last edited:
Current flow

I have measured 829 drivers from any possible aspect and listen to them MJL1302 though i have listened them only once and i was very happy with them but the ones i had was made by Motorola far before ON semi took over so i cannot say much about them .
With respect
Sakis
Thanks again, Sakis. I look forward to trying these drivers when I get around to building this amplifier.
Best wishes, Steve
 
For those who didn't read my earlier posts, there was nothing wrong in the sound with 669/649 and 5200/1943 combination. The problem was discovered after putting amp to scope testing. If you build it and just listen to it you don't know if it is OK. Even if you build it with BD drivers and outputs that Rod recommends you don't know if the amp is OK. Even if you buy BDs from ST manufacturer, they have many factories around the world, and all their BD139/140 are not the same. You must use scope to verify your DIY work. Oscillations are supersonic, you don't hear them and when they become apparent it's usually too late.
 
Member
Joined 2010
Paid Member
The ESP website recommends 75 mA current for all reasonable supply voltages, which would include +/- 30V at the lower limit. Assuming you are using the correct 0R33 emitter resistors, the 50mV voltage drop measurement then equates to 75 mA, simply by ohms law.

That bias level is recommended on the basis of best sound quality, for the whole range of voltages but you are free to play with different bias currents from say, a minimum of 13mA to a maximum of 300 mA, bearing in mind that higher current means more heat dissipation and even 75 mA will produce 5W of heat with 30V rails. That's 10W total heating from both channels at idle. Check the heatsink details in the article. 60-80W Power Amplifier
 
Last edited:
Current is NOT the determining factor for optimal ClassAB biasing.

The optimal bias for ClassAB is a VOLTAGE.

Sergiog has it correct
the bias is always 50mV

Optimal ClassAB is to minimise crossover distortion.
The advice being given by Ian F is nonsense
you are free to play with different bias currents from say, a minimum of 13mA to a maximum of 300 mA,