Pushing OnSemi transistors to the limit

[EWorkshop1708]

How hard can you REALLY push a single pair of OnSemi 4281/4302?

They are 230W spec, can you drive over 200W speaker at 4 or 2 ohms with a single pair for a short time, and have them live? How high could you run the rails? I was thinking of trying a pair on +/- 30 - 40 V at 2 ohms.

Like if you wanted to use them in a car amp, and use as few pairs as possible, what can you get away with?

[Jan Dupont]

Have you tried looking at the SOA curve in the datasheet ?

[wg_ski]

Originally posted by EWorkshop1708
How hard can you REALLY push a single pair of OnSemi 4281/4302?

They are 230W spec, can you drive over 200W speaker at 4 or 2 ohms with a single pair for a short time, and have them live? How high could you run the rails? I was thinking of trying a pair on +/- 30 - 40 V at 2 ohms.

Like if you wanted to use them in a car amp, and use as few pairs as possible, what can you get away with?

I had a homebrew PA amp miswired for *over a year* before the distortion at 2 ohms got annoying enough to look into it. It was a darlington triple, with multiple MJ15024 outputs with the same driver, and a high-beta Jap predriver. +/- 70 volt rails. On one channel the output bank was shorted on the PCB base to emitter output rail. It was running off the driver! Not only that, the current limiting was effectively disabled. Lightly loaded, it sounded fine. Heavily loaded there was noticeable distortion, and I always wondered why the bias pot didn't like the same setting on both channels. A single 15024 was taking 2 ohm loads at +/-70v rails! It has since been fixed, but that does give some insight into how rugged these devices are. How long it would have taken it I don't know, but s/b is usually a quick death if it happens and by all rights should have happened the first time I hooked 4 speakers to each channel.

[EWorkshop1708]

Wow! That's impressive. The 4281 seems to be somewhat similar to MJ15024 in power handling.

I'm not looking for HT or PA reliability, just reliable enough to last without too much abuse if kept cool. I'm using a large, thick piece of aluminum for heat sink. Also, the least paralleling I do, the more transistors can be used for other amp channels.

I sure won't be running 70 Volt rails in a car amp, but if a single pair at lower rails will push 2 ohms, then that's great news to save space and still have a decent amp. I've seen smaller transistors used with 100-120W RMS stages before, so I figure that the BIG TO264 transistors can probably handle it.

If I do parallel.....
Also, if I closely match transistors by voltage drop and Hfe, is it necessary to use emitter resistors when paralleling?

[wg_ski]

Don't sweat it at low voltages. With 40V rails you're really only limited by average junction temperature. Provide plenty of cooling and don't skimp on the driver unless you want to run out of HFE and get distortion. A sinlge unit would hold - they push way beyond rated SOA for brief instances on most store-bought amps these days.

I wouldn't parallel without emitter resistors. You only need a tenth of an ohm if you bother to match them at all and if they're physically mounted within an inch of each other on the same heat sink.

[AndrewT]

using one pair of MJL4281, 200W into 2ohms 60degree phase angle is only just inside the 100mS SOAR at 25degC.
This was modeled at +-37.1Vdc, +-75mF of smoothing, Re=0r1

1r0 resistive just does not work at all.

Increasing Tc to 32degC brings the SOAR down to coincide with both the max current limit and max power limit for 100mS.

200W into 4ohms 60 degree phase angle is outside the 100mS Tc=25degC SOAR for the secondary breakdown voltages (60Vce @ ~7.3Apk) using +-48.9Vdc

[wg_ski]

Originally posted by AndrewT
using one pair of MJL4281, 200W into 2ohms 60degree phase angle is only just inside the 100mS SOAR at 25degC.
This was modeled at +-37.1Vdc, +-75mF of smoothing, Re=0r1

If you analyze a known good amp, like the PLX3402, the SOA of it's 4 output devices per bank are exceeded by a *huge* amount. Peak power dissipation of 612 watts (per device, half way up the upper rail, 2 ohm resistive load). You shouldn't be able to operate it at all at 2 ohms at 20 Hz, but people do and it don't blow up until you start running it into thermal limit a few times.

Peak power dissipations of 200 watts (resistive) with a bit more reactive are no big deal for a 200W device. Those old Marantz boat anchors used to run 4 ohms off 40V rails (and some use to push lower) using old-school 100 watt Jap devices. And they held just fine.

[Kevin_Murray]

I have no problem with advice offered so far, I thought I'd add some before you get far in the design phase. Running low impedance loads is a throw-back to the old days of car audio when only 12V was available and switching power supplies where for the rich. I don't understand why modern setups continue to push impedances lower, except of course in competition where amp ratings are fixed and it's a way to "cheat".

If you are building your own amp maybe you could design for higher voltage rails and use higher impedance speakers? A low current/higher voltage design is typically more efficient and easier on the output stage. Distortion is also improved at lower output currents. Of course this is all useless info if your low impedance system is already installed and paid for. High impedance like used in home audio is the way I'm going with my current build. I'm using 6-8ohm speakers and my amp will have no trouble with them. In my opinion even 4 ohm speakers are dinosaurs.

[wg_ski]

Originally posted by Kevin_Murray
If you are building your own amp maybe you could design for higher voltage rails and use higher impedance speakers? A low current/higher voltage design is typically more efficient and easier on the output stage.

Yep. You can go to a quad 40V rail and go class H . Then two pairs will do it at 4 ohms, you don't need to bridge, and it runs a LOT cooler.

I've been toying with the idea of cloning a Behringer EP2500 circuit for car audio. Rewind the existing trafo for a 12V primary, run it at 400 Hz old school inverter style and crank up 650 watts per channel with low heat.

[Kevin_Murray]

Originally posted by wg_ski


Yep. You can go to a quad 40V rail and go class H . Then two pairs will do it at 4 ohms, you don't need to bridge, and it runs a LOT cooler.

I've been toying with the idea of cloning a Behringer EP2500 circuit for car audio. Rewind the existing trafo for a 12V primary, run it at 400 Hz old school inverter style and crank up 650 watts per channel with low heat.

Now that's a project!