VBE multiplier

[fglabach]

wq ski

The original luxman was fine into 4 ohms with 6 sanken output transistors. But when I ran it into 2.67 ohms with the Kinergetics woofer tower (six 16 ohm drivers in parallel) it got too hot tripping the heat protection sensor. That is why I converted it into a monblock. See discussion above.

If I eliminate one pair of output transistors from each heat sink I would have 4 on each heatsink and could bias them much higher.

But then I would be concerned with beta drop from eight 17 amp Sanken transistors.

The amp made for the speaker used about twelve Sanken 17 amp transistors but this may have been overkill. It never sounded as good as the Luxman, which has great bass, body and bass imaging. No hardness in this amp.

Am I better off reducing the number of transistors, underbiasing or increasing the value of the emitter resistors?


I am going to try the great idea of moving a led current reference in the VAS to the heatsink this weekend but I don't think this will solve all of the problems.

 

[AndrewT]

Hi,
look at the power into 8r0 and 4r0.
The amplifier was struggling to drive 4r0. The output fell by ~-2dBV into the 4ohm load. Driving an 8ohm reactive loadismore stressfullthan driving a 4r0 load.

It can never be recommneded to drive a reactive impedance ~2.6ohm.
Removing transistors will make the situation even worse.

I would estimate that this amplifier is only suitable for 8ohm speakers now that you have removed some of the devices..

 

[fglabach]

I finally got this thing to work.

I tried all of your ideas but this is what did it.

I used artic silver conductive grease between the sensor and the top of the output transtor, created a sytrofoam insultor to prevent any cooling of the sensor.

The amp is now thermally stable.

I added the Self current compensation resistor to the design but I found it extremely difficult to figure out the value so I smply used a variac as suggested by Self. Sloan siad it was based on R prime but Self said it depended on the multiplication factor.

The bias circuit had high quiscent current and high mlutiplication factor. ANd the current plataued with a 10 ohm resistor, which seems very low, but worked.

The two transistor vbe multiplier actually overcompenstated too much.

I tried Kronos idea but it made very little difference. I was considering using a chain of diodes to get more negative tempco but I think I am better off without it. I think the answer for me was to keep the sensor from cooling by enclosing it in insulation as suggested by Self.

Thanks for all of your help. ''

 

[EchoWars]

I finally got this thing to work.

I tried all of your ideas but this is what did it.

I used artic silver conductive grease between the sensor and the top of the output transtor, created a sytrofoam insultor to prevent any cooling of the sensor.

The amp is now thermally stable.

I added the Self current compensation resistor to the design but I found it extremely difficult to figure out the value so I smply used a variac as suggested by Self. Sloan siad it was based on R prime but Self said it depended on the multiplication factor.

The bias circuit had high quiscent current and high mlutiplication factor. ANd the current plataued with a 10 ohm resistor, which seems very low, but worked.

The two transistor vbe multiplier actually overcompenstated too much.

I tried Kronos idea but it made very little difference. I was considering using a chain of diodes to get more negative tempco but I think I am better off without it. I think the answer for me was to keep the sensor from cooling by enclosing it in insulation as suggested by Self.

Thanks for all of your help. ''

Glad you got it.

I'm not too suprised about the LED temco not working too well. LED's Vf doesn't change much with temperature.