Favorite High Power Output Transistor

[anatech]

Hi Lyra,
The higher collector voltage makes them more attractive for new design. This is not a factor in your case. The collector currents and power dissipation is the same in both so ....
At least they are not incompatible, you won't hurt anything. If your outputs were blown, I'd use the 2SA1302 or MJL1302 pair. Only because I stock them. There is no compeling reason to change them otherwise.
-Chris

 

[Robin]

The 2SC3281/1302 are not manufactured anymore by Toshiba and most of the devices on the market that are marked with these part numbers are dodgy pirates with sub standard silicon. Toshiba now market their equivalents and On Semi market their MJL alternatives which are identical. Just be careful if you buy parts these days that claim to be Toshiba.

 

[rajeev luthra]

I am planning to make a class AB amp ,
Transformer 66v-o-66v ie DC will be +/- 93v
Power Output approx 450w 8E, 800w 4E , and 1500w 2E
The amp will rarely be used at 2E but I want to over kill ,

Will 8 pairs of 2SA1943/2SC5200 be ok ,
Drivers MJE15034/35 or 2SA1302/2SC3281
Filter caps 30,000MFD per rail

here the Toshiba drivers 2SA1302/2SC3281 are costing 50% more than MJE15034/35 which ones should I choose.

please advise and give your recomendations

 

[Eva]

2SC3281/2SA1302 are also 15A 200V 20Mhz output transistors, they are not intended as drivers

Anyway, Toshiba discontinued them and replaced them by 2SC5200/2SA1943 [improved new versions] so if your amplifier circuit specifies C3281/A1302 as drivers, then you should also use C5200/A1943 instead

Please don't buy C3281/A1302 anymore, they are allways fakes. Currently it's obviouls that there are no remaining stocks of these devices. All the stocks existing when Toshiba stopped production [It was in 1999 I think] have been already sold and used

I bought some pairs of C3281/A1302 from different sources two years ago ang they were all fakes. Actually I got three different types of fakes with different cases, different markings and different die sizes!!! [Not to talk about different transistor parameters]

Currently there are probably a dozen or more counterfeit 'manufacturers' and each one is producing C3281/A1302 fakes in a particular flavor

 

[K-amps]

I don't know the exact model but a Mark Levinson 4 series monoblock (rated at 300 8E, but actually does 500 8E, 9004E and 1.5kW 2E) uses 8 pairs of a Toshiba class TO-3PL devices.

It has 40mF per rail or 80,000uF in total with a 2372 VA toroid.

Your design seems similar to that in terms of power, I'd say as a rough estimate, you are on target.

 

[OliverD]

In turn the Accuphase M-8000, rated at 250W/8Ohm, uses 22 pairs of 150W output devices.

It depends on design philosophy, of course. 8 pairs are probably safe enough, although I used 12 pairs for my 300W/8Ohm power amps.

 

[lumanauw]

Here I experiece the same thing. Drivers like MJE15030/31 has the same price as C5200 pairs.
What is so special about these drivers?

 

[K-amps]

In greater than 4 pairs of outputs, I have seen manufacturers use power devices as drivers. Especially the C5200/A1943's because of low beta droop.

The 15030/31's are ok, but the 15034/35 are much better when mated to the MJL21193/4/5/6.

My experience.

Yes the 5200/1943 are not designed to be drivers but they perform well.

 

[Upupa Epops]

To K - amps : Optimalision of driver depend on typical working current. Modern output devices have flat beta curve to the low current, but by low curren fall ft. I think, that from this case is better use as driver recomended drivers, 'cos they have optimalised ft by lower current.

 

[MikeB]

Yes, take a look into the datasheet from onsemi, these are very
excellent devices, at 100mA they still have ft of ~50mhz (typ).
Also they are very linear up to 1A. (both pnp & npn)
I once built a small poweramp using these as outputdevices.
They are really good ! Onsemi itself recommends them for amps up to 25watts.

BUT: Only use original onsemi devices. I had some from another noname
manufacturer, 2 seconds after powering up these exploded.

I was thinking of paralelling many of them, but this seemed crazy to me.
Here in germany they cost about €0,75. So for a not too hipower
amp they would be a real alternative...

Mike

 

[sajti]

Yes, take a look into the datasheet from onsemi, these are very
excellent devices, at 100mA they still have ft of ~50mhz (typ).
Also they are very linear up to 1A. (both pnp & npn)
I once built a small poweramp using these as outputdevices.
They are really good ! Onsemi itself recommends them for amps up to 25watts.

BUT: Only use original onsemi devices. I had some from another noname
manufacturer, 2 seconds after powering up these exploded.

I was thinking of paralelling many of them, but this seemed crazy to me.
Here in germany they cost about €0,75. So for a not too hipower
amp they would be a real alternative...

Mike

The MJE15032/33 performs very well as output stage. I used 2 pairs/channel in 30W/8ohm amplifier. BD139/140 drive them perfectly, and they give nice sound, even with 4ohm load...
My only warning, that they need 100mA bias at least.

sajti

 

[MikeB]

hi sajti !

Why do they need that much quiscentcurrent ? 100mA per device
or 50mA per device (*2) ?

Did you get crossoverdistortion ?
Maybe they are not optimized for on/off times, as drivers are normally
used ClassA ?

Mike

 

[K-amps]

To K - amps : Optimalision of driver depend on typical working current. Modern output devices have flat beta curve to the low current, but by low curren fall ft. I think, that from this case is better use as driver recomended drivers, 'cos they have optimalised ft by lower current.

Optimization is a relative term isnt it? 😉 My solution or recommendation was provided with 8 or more OP devices running 500-1kW in such, currents demanded are high. For a 25-100 watt amplifier, I'd use the recommended drivers.

As I said, it's all relative. 😎

 

[sajti]

hi sajti !

Why do they need that much quiscentcurrent ? 100mA per device
or 50mA per device (*2) ?

Did you get crossoverdistortion ?
Maybe they are not optimized for on/off times, as drivers are normally
used ClassA ?

Mike

The drivers are run with 30mA class A (single emitter resistor, and 330nF paralell), there was predrivers (BC546B/556B), with 6mA class A. I prefer triple darlington configuration.
The crossover distortion cancelled with about 10mA bias. But I found that the sound is bit sharp. I started to increase the bias, and sound was smoother at about 100mA/device.
I don't understand it, so I measured the B of the MJEs. It starts to fall rapidly under 100mA. It's true for 15030/31, and 32/33. The 34/35 is flat down to 10mA (but I can't get them in my country). This bias looks too much, but for 30W amplifier it means just 3 - 3.5W dissipation for each device...

sajti

 

[sss]

is it ok to use for the output transistors to be faster then the drivers?

 

[MikeB]

to sajti:

Yes,yes, me too ! I love the sound from tripledarlington, it's like it
handle all currents with ease. The ClassA-predriver (i also use
the bc546b/556b) does a great job in reducing load to the vas.
According to sims it can lower thd about factor 100...
I don't get the 34/35 here either... (maybe time to order some samples ?)


to sss:

It shouldn't be a problem, but seems strange to me... Which outputdevice
is faster than drivers ?


Mike

 

[sss]

Yes,yes, me too ! I love the sound from tripledarlington

those circuits tent to oscilate

It shouldn't be a problem, but seems strange to me... Which outputdevice
is faster than drivers ?

if u use mje15035/34 as outputs u can find slower drivers 🙂

 

[sajti]

I think we use the triple darlingtons for different purposes🙂 In my amplifiers I don't use lot of feedback. The maximum is about 14dB. To keep the low distortion, I apply local feedbacks. But local feedback doesn't give low output impedance. Triple darlington does it! To keep the overall feedback low I have to plan the open loop gain load independent. For this purpose I use fixed resistors at the base of the predrivers, to the ground. The value of the resistors are much lower than the input impedance of the triple darlington stage. With this solution I get constant gain without load, or with 2ohms. The change can be less than 10% in practice.

sajti

 

[sajti]

those circuits tent to oscilate

if u use mje15035/34 as outputs u can find slower drivers 🙂

Triple darlington is very stable, if You apply some basestopper resistors. Generally 100-470ohms for predrivers, 10-47ohms for the drivers, and 1-4.7ohms for the output devices. Higher values result more stable stages, but somewhat reduce the output power.

For lower speed drivers, the most common, when MJE340 drive the 2SC5200...

sajti

 

[MikeB]

those circuits tent to oscilate

if u use mje15035/34 as outputs u can find slower drivers 🙂

hmm, i have some opposite experience, i was able to reduce oscillation
with higher bandwidth when using tripledarlington...


Okay, with these outputs... but i would use something like mje243/253 as driver.
these are faster than the mje1503x
Of course you can drive the mje1503x with mje1503x 😉

to sajti:

I like the classA-predriver because it puts very little load to the vas,
and keeps most of the reactive loads away from it...
My circuits tend to use small currents (~2ma) in the vas.

Mike