TIP35/36C maximum power at one pair

[elektryk]

Hi.
Can one pair of this transistors survive 80W/8R and 120W/4R at 47V supply?
I'm building AB class amp, potentially TIP35C is good for this aplication by SOA, but I don't want to make impredictable amp.


I can use 2SC5200/2SA1943 if TIP35/36C are too weak, but in my Country Toshiba transistors are often fake.

 

[JonSnell Electronic]

The absolute maximum power dissipation of TIP35 is 125W, that is assuming there is sufficient heat dissipation. So good for 70% of total power I would say.

 

[jaycee]

I'd say just one pair of transistors on this voltage is pushing it. Remember a typical speaker is an inductive load. Even with 1943/5200's i'd say use two pairs.

 

[elektryk]

Such calculations are too general, I have counted it like under:
Output voltage for 120W/4R-->22V
Current -->5.5A
Vce=47V-22V=25V
Dissipated power --> 137.5W
For one transistor from pair -->68.75W

Did I do it right?
Probably there will be no secondary breakdown, but I don't know whether this is counted well.

btw Sorry for my English...

EDIT:

For example Technics SU-V85a use one pair 2SC3280/2SA1301 at 57V with 150W/4R, 100W/8R.

 

[jaycee]

You are only calculating resistive loading. Speakers are inductive. This means current and voltage will be out of phase, the worst is a 90 degree phase shift.

Rod Elliot has an excellent article on this: Phase Angle Vs. Transistor Dissipation

 

[jaycee]

Don't judge it by commercial amps - these often use undersized output stages, together with an under-rated power supply that will collapse if the load is too severe.. this protects the output transistors, but also sounds like sh*t.

 

[wg_ski]

Peak dissipation with resistive load = (23.5^2)/4 = 138W (half voltage on load, half on transistor). For 60 degree reactive load, double it. That's 276W @ Vce=47V. That's outside the SOA for even a pair, but you can get away with a pair in parallel because the supply will drop enough under load to keep it under 250. The 276 is (47^2)/8 for 4 ohm load (half the peak current at full VCE).

In reality it's much more complicated than this.You may (probably will) run into a worse load, and there is temperature derating to consider. On the plus side you're not operating at DC conditions, and for practical purposes the 25C DC condition is a wash with pulsed conditions derated for about 60 or 70C. This is pretty much on par with what they assume for store-bought amplifiers.

 

[elektryk]

That's right, but people listen music power and dissipated power will be smaller than with continuous sine wave. In non continuous load this rules applicable too?

I believe that 400VA transformer will be good for this (home) amp, but I'm only worry about transistors, 'cause I have already PCB design, and want to make this device cheap.

 

[wg_ski]

'cause I have already PCB design

If you've got a PCB that's only got room for one output pair and you're committed to it, use the MJL21193/21194 pair. Those are about as strong as you're going to get. Anything else I would use two pair. It doesn't have anything to do with continuous vs. non continuous load - if the transistors are insufficient you can blow them without ever getting them hot.

400VA trafo is plenty. Unlike transistors, a trafo can handle about a 10X overload for a few minutes and it won't hurt anything.

 

[jaycee]

With 400VA of power supply, all you'd need is to have a party, turn it up a bit, and you'd quite easily risk blowing transistors.

You might consider TO-3 package parts if you only want to use one pair - these typically have better SOA due to the better thermal profile of the casing. Alternatively, use MJL21193/4 as recommended. You might be able to get them directly from On Semiconductor as a sample part.

 

[davidsrsb]

Looking at the TIP35C SOA, it can only take about 1A at the 47V supply, so very low power levels at 45+ phase are problematic. The MJL21193 SOA allows 5A at 47V, so far better suited to driving non-resistive loads. You still need SOA protection, but will get useful power levels into real loudspeakers

 

[Conrad Hoffman]

A pair or ever a triple is better. IMO, unless you're the only one using the amp, ever, make sure to include some SOA protection. Paraphrasing Sixteen Tons, "a lotta amps didn't, a lotta amps died."

 

[elektryk]

Thanks, I will remeber about restrictions of one pair and will use other PCB or make weaker amp.

 

[AndrewT]

......................................You might consider TO-3 package parts if you only want to use one pair - these typically have better SOA.............................Alternatively, use MJL21193/4 ................

If you've got a PCB that's only got room for one output pair and you're committed to it, use the MJL21193/21194 pair. .................................

Looking at the TIP35C SOA, it can only take about 1A at the 47V supply, so very low power levels at 45+ phase are problematic. The MJL21193 SOA allows 5A at 47V,......................

A pair or ever a triple is better.........................

The message is LOOK at the SOA of the devices. De-rate them for temperature, then predict what load they can drive.
5pair of 1A@47Vce are needed to match the SOA of a single pair of 5A@47Vce device. If that single device has Tjmax=200°C (mj21193/4), then it will perform even better= more reliable.

de-rated SOAR rules !

 

[davidsrsb]

The lesson from this is that the TIP35 is not very useful at 47V supply rails. At 30V it would be far better

 

[AndrewT]

you could use tip35b @ +-34Vdc and then the DC current capability jumps up to 3A @ 34Vce. (double the SOA cf. 50Vce)
100ms SOA is not shown, but expect it to be only a tiny bit better than the DC plot shown.

 

[davidsrsb]

Typical 4~6 Ohm standmounts hit minimum impedance at almost 0 phase (which is an easy load) but +/- 36 degrees at about 8 Ohms magnitude

 

[elektryk]

How can I calculate current at the worst situation with 4Ohm home speakers at defined voltage?

 

[AndrewT]

use one of the SOAR spreadsheet calculators to model the output stage.
Bensen did one for Mosfets, Jan Didden has one.
I modified Bensen's for both FETs and BJTs.

Or use the simple formula

Maximum output power equals the total output device dissipation capability divided by Factor.

For FETs the Factor ~4
For BJTs the Factor is ~5 to ~6

2 125W devices = 250W of device dissipation.
Maximum output power ~ 250/5 to 250/6 = ~50W to 40W
2Pr gets to 100W to 80W
3pr gets to 150W to 120W. This last requires too high a supply voltage for 8ohms, so only possible into 4ohms speaker.

 

[wg_ski]

How can I calculate current at the worst situation with 4Ohm home speakers at defined voltage?

The simple answer is, for an amp powered by +/-Vcc, Pdiss rating needs to be = Vcc^2/8 at Vce = Vcc. This keeps it inside SOA for 2 ohms resistive, 4 ohms at 60 degrees, and 8 ohms fully reactive. That will handle most speakers. Use the DC curve for audio signals at normal temperature rise. You can go through pulsed response and calculate a transient thermal resistance, and then de-rate for temp but for practical purposes you'll end up back on the 25C DC curve.