at 53V im using 100ma per transistor for class AB so how much do you recommend?
at 22V im planning to use at 1Amp but will the bias will be stable using lme49810?
at 22V im planning to use at 1Amp but will the bias will be stable using lme49810?
A very high end name brand amplifier uses 10 pair 5200/1943 at ~200mA each (60V rails), but that is a fully discrete circuit. Class A 250x8, operates to 1 ohm at 1kW output (lower impedance breaks Class A though, so technically it is Class AB amp).
I've seen some old Japanese amplifiers operate the same transistors at ~15mA each at roughly the same voltage, so I guess the answer is that it depends on what you want to achieve and the circuit topology.
The last 4702 implementation I did sounded best at about 25mA for single pair output. Higher bias simply made the sound muddier. The bias was stable over eight hours of continuous operation and no, the chip was not connected to the main heatsink. jackinnj has done some research on this chip family with the summary being that they work best when disconnected from the output sink.
OTOH the Vbe multiplier needs to be on the main sink. There is always some lag between the bias adjustment and the actual readings, because of the 'thermal delay' as the multiplier catches up with the main output. Class A amps, for example, need a lot of warm-up time to get the bias fully stable.
There's a big thread on bias and its effect on sound, where one conclusion was that bias voltage between the transistor bases was more important than the current, at least for Class AB operation with minimum distortion. Another conclusion was that low bias of this sort measured better but subjective preference was for higher bias (I think NP brought that one up).
I've seen some old Japanese amplifiers operate the same transistors at ~15mA each at roughly the same voltage, so I guess the answer is that it depends on what you want to achieve and the circuit topology.
The last 4702 implementation I did sounded best at about 25mA for single pair output. Higher bias simply made the sound muddier. The bias was stable over eight hours of continuous operation and no, the chip was not connected to the main heatsink. jackinnj has done some research on this chip family with the summary being that they work best when disconnected from the output sink.
OTOH the Vbe multiplier needs to be on the main sink. There is always some lag between the bias adjustment and the actual readings, because of the 'thermal delay' as the multiplier catches up with the main output. Class A amps, for example, need a lot of warm-up time to get the bias fully stable.
There's a big thread on bias and its effect on sound, where one conclusion was that bias voltage between the transistor bases was more important than the current, at least for Class AB operation with minimum distortion. Another conclusion was that low bias of this sort measured better but subjective preference was for higher bias (I think NP brought that one up).
Last edited:
agreed regarding the sound getting muddier with higher bias in class AB especially with 2sc5200 1943 pair. I was thinking to bias higher with the transistors just to achieve 1A of total bias at 22V.
yes the bias took longer time to settle infact the bias of my current amplifier is taking about 20mins to sound good and smoother. There is about 8cm from the output transistor from the vbe multiplier.
yes i was inclined to use 5 transistors with 200ma bias to achieve 1A per rail and 2A peak at 40V but never tried as I am worried that it might blow off.
once i was running at 53v and biased at 300ma and the bias crawled up and ended up in thermal runaway and finally my transistor blew up. I couldnt even observe that was happening as I was in deeply enjoying the sound of higher bias and suddenly puff....!!
believe me I was using massive heatinks even then it blew away.
so what Im thinking is to use 5 in parallel and bias at 200ma that should do the job i believe?
whats your opinion?
yes the bias took longer time to settle infact the bias of my current amplifier is taking about 20mins to sound good and smoother. There is about 8cm from the output transistor from the vbe multiplier.
yes i was inclined to use 5 transistors with 200ma bias to achieve 1A per rail and 2A peak at 40V but never tried as I am worried that it might blow off.
once i was running at 53v and biased at 300ma and the bias crawled up and ended up in thermal runaway and finally my transistor blew up. I couldnt even observe that was happening as I was in deeply enjoying the sound of higher bias and suddenly puff....!!
believe me I was using massive heatinks even then it blew away.
so what Im thinking is to use 5 in parallel and bias at 200ma that should do the job i believe?
whats your opinion?
The way I see it, you should only bias as much as the transistor would take without any heatsink, which for TO247 devices would be a couple watts.
Thermal lag is real. When you set bias the device warms up, and warms the heatsink. The thermal compensator, of whatever kind, then heats and pulls down the bias. The delay is directly related to the thermal resistance of the heatsink, the current temperature of ambient and sink, and of the bias levels at the time. This is why transistors with inbuilt thermal compensation diodes are so great, practically no lag.
3 to 4 watts is the maximum I would trust these devices when taking potshots in the dark. You can work out the current from the rail voltage.
Thermal lag is real. When you set bias the device warms up, and warms the heatsink. The thermal compensator, of whatever kind, then heats and pulls down the bias. The delay is directly related to the thermal resistance of the heatsink, the current temperature of ambient and sink, and of the bias levels at the time. This is why transistors with inbuilt thermal compensation diodes are so great, practically no lag.
3 to 4 watts is the maximum I would trust these devices when taking potshots in the dark. You can work out the current from the rail voltage.
The required amount of bias depends on the output stage topology and the value of the RE resistors used. That is if you are after a supposedly 'optimally' biased class AB amplifier.
Exactly! It does not depend on the transistor type or supply voltage. See Douglas Selfs recommendation for the optimal bias voltage across Re for the different output stage topologies.
Jan
Jan
- Status
- Not open for further replies.