I read somewhere that Sanken transistors are specifically designed for audio applications but they're difficult to acquire (Japan only mostly, and some fakes on AliExpress which I won't touch).
Are there any other BJT transistors recommended for the output stage of an audio amplifier?
Are there any other BJT transistors recommended for the output stage of an audio amplifier?
A very wide selection of answers you will get here!
You do not state what power, voltage or design.
Sanken produced a lot of power switching transistors and special semiconductors as well as power low frequency audio types.
For brute force at up to 250 volts at 16Amps MJW21195G and its compliment MJW21194G, MJH11022G NPN Darlington Transistor, 15 A 250 V HFE:100 and its compliment.
Then there's IGBT switching 600volts at 200Amps an ideal class D setup.
Etc etc etc
You do not state what power, voltage or design.
Sanken produced a lot of power switching transistors and special semiconductors as well as power low frequency audio types.
For brute force at up to 250 volts at 16Amps MJW21195G and its compliment MJW21194G, MJH11022G NPN Darlington Transistor, 15 A 250 V HFE:100 and its compliment.
Then there's IGBT switching 600volts at 200Amps an ideal class D setup.
Etc etc etc
Hi,
For class AB or class A audio linear power amplifiers, any BJT with Ft around 30MHz will be as good as the Sanken ones.
The other parameters such as maximum Ic, maximum Vce, SOA and maximum power, will depend on your goals and circuit design you plan to develop.
One BJT pair very popular and easy to find is the Toshiba 2SC5200/2SA1943, if you find them original of course.
For class AB or class A audio linear power amplifiers, any BJT with Ft around 30MHz will be as good as the Sanken ones.
The other parameters such as maximum Ic, maximum Vce, SOA and maximum power, will depend on your goals and circuit design you plan to develop.
One BJT pair very popular and easy to find is the Toshiba 2SC5200/2SA1943, if you find them original of course.
If it were only a matter of getting sufficient stock I would lean toward OnSemi, as they have the most choices with suitable parameters, and the regular distributors get them in regularly. Even Digikey has trouble keeping Sanken parts in stock. But places where Sanken is unobtainium also have a hard time with Toshiba or Onsemi. Unfortunate but true. If you get somebody else’s 2SC5200 it’s a roll of the dice. Fakes are everywhere, new makers seem to come out of the woodwork monthly these days. If there were universal 3rd party pieces that are every bit as good as originals, available in every 2nd 3rd and 4th world country on Earth, without a credit card or resale license, outside standard distribution networks, unimpeded by governments efforts to control imports, they would be screaming it from the mountaintops.
MJL3281/MJL1302
2SC5200/2SA1943
On paper, 2SC5200/2SA1943 is easier to drive comparing to MJL3281/MJL1302, due to lower Cob.
2SC5200/2SA1943
On paper, 2SC5200/2SA1943 is easier to drive comparing to MJL3281/MJL1302, due to lower Cob.
Yes the 3281G/1302G are a classic pair, current gain pretty independent of collector current upto 8A or so, and in production. Multiply sourced I think.
Always check the SOA whatever devices you pick, so secondary breakdown is avoided.
Always check the SOA whatever devices you pick, so secondary breakdown is avoided.
There is a great deal to learn about amplifiers before you worry about the best choice of output transistors. Once you understand the finer points of how to, and how not to use power transistors, then you can review the choices and make your own decisions. There is a great deal to be learned by running spice simulations, but you have to include thermal parameters and simulate overloads and abuse. Semi vendors web sites offer "audio" transistors and distributors publish prices and selector guides. You should have reviewed 100's of existing designs before you make one of your own. Unfortunately, there are many kits and modules for sale that are not reliable solid designs, and the choice of output transistors is a small part it.
The ON 3281/1302 are probably the strongest “fast” audio outputs available. The shrunk versions (0281/0302) are comparable in die size and Cob with current production Jap types. The only question is whether OnSemi is any more available than the Sankens - is it just a cost and stock issue or whether the supply chain is in fact limited by forces beyond your control.
You are right! I simulate a couple of them and if using the recommended power supply and load, in many situations the transistors exceeded the SOA considering a reasonable real temperature (80C case temp for example) and reactive load.
My brother bought two of these kits with 3 pairs of 2SC5200/2SA1943 each for amp with around 350W/4ohm per channel.
I took the schematic and simulated it in Spice.
I had to make some changes to improve currents, bias control etc, and I recommended to add 3 pairs more so as to keep the transistor within the SOA.
In addition, I added a VI protection since there was nothing about it on the PCB.
It is working really fine - but with 6 pairs of 2SC5200/2SA1943 on each channel. Total of 24 output transistors.
Hi bigpilot,
I have a couple of dozen of Sankens available.
If interested send me a PM.
Thanks for all your answers. I've decided (for now) to use OnSemi D44H / D45H for the output stage. Power output is about 50W.
Careful with those. Second breakdown capability isn’t particularly good. Supply voltages above +/-25 are iffy because current/power capability falls like a stone above Vce=15.
I think you need a transistor pair with a more extended SOA (safe operating area).
For 50W @4ohms, you need around +/-26V unregulated power supply and if @8Ohm, you need +/-35V - considering 10% voltage drop and 3V losses in the circuit. This is just rough numbers to show you how to calculate things. You can use your own numbers.
Considering just resistive load, which is the best case, we have:
@4ohm: 26*0.9=23.4V -> 23.4/2 = 11.7V -> 11.7/4 = 2.9A. Look at the SOA using 11.7V and 2.9A - it's very close to the limit.
@8Ohm: 35*0.9 = 31.5 -> 31.5/2 = 15.7 -> 15.7/8 = 2A. Look at the SOA using 15.7V and 2A - it's very close to the limit.
This is just a very basic sanity checking. Best way is to simulate using a reactiive load of at leas 30deg off.
If we consider real reactive load, things get worse, since current will not be in phase in voltage and the SOA will be exceed depending on what volume you listen, how much hot the transistor is, how reactive your speaker is in the frequency range your music is focussed on etc.
I know the SOA considers duty cycles, but I prefer to be below the DC area, specially if entering second breakdown region.
There are too many variables to consider, so best way to go is to use a good SOA margin and have a robust amplifier that can handle any volume, any music and can safely work on the temperature limit you've planned.
If you plan to simulate it, keep posting here and we can discuss the results.
I have had first hand experience with D44/45 blowing up at +/-30V, 8 ohm load. They will not hold up. TIP41/42 do, but performance overall is mediocre at best. For low supply voltage operation, say +/-15 to 18 you cannot make a better choice than the D44. Beta remains high down to Vce=1V, getting the most possible out of low voltage supplies. The MJL3281 can’t make that claim, and the 4281 is even worse about that.
Well a 250V rated transistor isn't made with 15V rails in mind really... The D44/45 are designed for switching as well as general purpose amplification, so they have a low Vsat which probably goes with good low-voltage beta. Optimizing for switching would allow poor SOA as switching doesn't require much SOA.
If you take a look at the typical SOA curves for a power transistor you will see that it doesn't take a very-long pulse for it to approach the transistor's DC characteristics. Since reproducing low frequencies demands the most power output, I look at the transistor's DC performance to determine if I want to use it in an output stage.
I'm very aware of SOA issues because I've blown up my share of outputs due to that. It doesn't take very long to kill a transistor once its SOA is exceeded. It isn't the same as simply exceeding its power dissipation. The transistor(s) will blow before the fuses do, it happens that fast.
I'm very aware of SOA issues because I've blown up my share of outputs due to that. It doesn't take very long to kill a transistor once its SOA is exceeded. It isn't the same as simply exceeding its power dissipation. The transistor(s) will blow before the fuses do, it happens that fast.
Rule of thumb: single pair TO220 => 20W rms (+/-20V); single pair TO-3, TO-264, TO-247 etc. =>50W; Two pair TO-3 etc. => 100W, etc. (50W per pair)
Note that you can do better if you go class-H or class-D.
Of course, some people think constantly rebuilding amps is par for the course. Not me.
Note that you can do better if you go class-H or class-D.
Of course, some people think constantly rebuilding amps is par for the course. Not me.