SOA and whats realistic

[Ryan_Mc]

I am making a 5.1 Ch. amp and was going to use the Leach low-TIM design. I don't need anywhere near that kind of power but I have two 40V transformers already and the design is well covered and has great reviews. This amp 'as-is' is rated to handle a 4ohm load.

As I said I don't need that kind of power as the satelites are most likely going to be FR125S's. But at anyrate I thought I would just limit myself to 8ohm speaker loads only (whatever they may be) so I could cut down on the output transistors or maybe use a more convenient package like a TO-264 since I am only going to use an 8ohm load.

The xfmrs are 40V so after rectifying guestimate a max Rail Voltage of ~57V but designing in some safety for nasty reactive loads (as I could do some experimenting with this amp using different speakers) the transistors (Vce) could see the full R-R voltage of 114V, and into 8ohms that is 14.25A. Drawing that load line and comparing it to a single MJ15003 SOA and it isn't even close, doubling the MJ15003 doesn't even quite get the SOA above the load line. Tripling the output transistors barely gets it above but there isn't a lot of leeway to fit a protection circuit in there.

Those numbers seem grossly unrealistic to me so I know I must be missing something. Can someone set me straight please! ...and tell me where I went wrong?

Thanks,
Ryan

 

[Ryan_Mc]

Just giving it some more thought...

The nastiest load would be purely inductive or capacitive and put the voltage vs. current out by 90°. So the potential for the transistor to see double the rail voltage is still there but the max current would only be 7.125A (using the values above). This load line seems a lot more realistic to me, am I on the right track?

I'm guessing designing for a 90° phase difference may be on the conservative side too.

 

[wg_ski]

Design for 60 degrees, assuming the transistors are good for about 400 watts on a transient basis at voltages 50-60V and under. At 57V rails, it takes one pair per 8 ohm load. This is why when you buy an amp in the store it has one pair and says not to use 4 ohm speakers. With 4 ohms, one pair will take it for a while if you're careful with it, but two will take most abuse. For 2 ohms, double it again.

 

[Steve Dunlap]

The output transistor will only have the 144V across it when the other transistor in the pair is conducting to the clipping point. At any thing over a volt or so of output the nonconducting transistor will be turned off so no current will flow. The transistor that is conducting only has 57V across it at the zero crossing point or when there is no signal. When it is passing the most current it has the lowest voltage across it. You will find this helps a lot with staying in the SOA.

Steve

 

[wg_ski]

The transistor that is conducting only has 57V across it at the zero crossing point or when there is no signal. When it is passing the most current it has the lowest voltage across it. You will find this helps a lot with staying in the SOA.

Steve

For resistive load. For fully reactive, you get peak current at the zero crossing (57V). That's not realistic either. For 60 degree load you get half of peak current at the zero crossing. That's a good point to design to. Example: For 57V supply you'll get about 50V peak output. At 4 ohms, that's 12.5A. At 60 degrees, worst case on the transistor is hald that current (6.25A) at 57V. That's 356 watts. One transistor will handle that, but not at elevated temp. Two will, so using two would be good fo 4 ohm 60 degrees.

 

[Steve Dunlap]

Of course it is good to design for some inductive load. You must remember that in a typical system (for audio) you will have the emitter resistor and resistance of the voice coil. This should be at least 2 ohms even for an 4 ohm speaker. Once again, this really helps with the SOA. The obvious exception to this (were still talking audio) is the transformer input of an electrostatic speaker.
In this case the resistance may well be below 2 ohms and can be below 1 ohm. This is the resistance only, not the impedance at audio frequencies.

What this means for a DIY builder is:
8 ohm speakers will almost always have a 4 ohm minimum in series with any inductive load. For 4 ohm speakers assume 2 ohms.

There are many commercial amps with on one pair of output transistors for this supply voltage. Two pair will be insurance against the occasional poorly designed speaker.

Steve

 

[CBS240]

It's that elevated temperature that seems to get to BJT's a bit more. A dusty heatsink coupled with the wrong transient across a bad load and you run into that secondary breakdown dilemma. If you design for a specific speaker, you may be able to float on 1 pair. But it is better to design for worst case in worse case conditions, this way a small extra investment will result in greater life expectancy.