Bizarre (over)heating problem in big amp

[Stuart Easson]

Hi All, I'm hoping to pick a few brains here...

A close friend has a a weird problem with a pair of Adcom 565 monoblock amps driving accustat speakers, thru 4 ft of kimber 8tc. He's not using the subs on the accustats. Something about this combination is 'not quite right', but he loves the amps and speakers and wants to keep them both...

In a nutshell: Just one 'pole' of the amps output transistors runs at high temp, ie 50-55c, whether idling or blasting. The amps are pretty huge, (300w/8, 450w/4, 650w/2) they don't seem to have any issue with the extra heat, play at any level without apparent stress. The reason this is a problem is that over the course of a couple of years this killed a 565, through (I assume) thermal cycling of the components...having replaced the failed amp, he doesnt want to repeat the performance and kill another.

I'm assuming (you know what they say) if the amps were oscillating the sinks would be equal in temp, but the left hand heatsink, attached to paralleled toshiba transistors that are one pole of the output are barely warm, while the right sink is almost too hot to touch, on both channels.

If anyone has any clues, I'd love to hear 'em. I will be going over there with a multimeter and handheld scope on weds, so I will be checking basic things then...

Thanks for your help

Stuart

[cunningham]

Is this amp direct coupled to the speaker?? If it is, DC, positive or negative, whichever side is heating up.

Occilations could have a DC componant, in other words, only one side is occilating, but you would see this as a DC offset at the output.

[Stuart Easson]

I assume the speaker is DC coupled, I have the service manual with schematics at home, I'll check. I also assumed it would have a circuit to detect DC and trigger the other protection circuits, isolating the speakers...too many assumptions perhaps...

Stuart

[Stuart Easson]

Hi,

In response to the earlier question I checked the schematic, the output transistors are each coupled to the load through a 0.33ohm emitter resistor. There is no series output network, though there are a capacitor (0.047u) and resistor (6.8ohms) in series across the output. Would this make the amp less stable driving a highly reactive load?

The transistors that are heating are the negative output pole, on the right hand side of the amp, looking from the front.

Stuart

[Richard C]

What's the voltage across those 0.33 emitter resistors?

[Stuart Easson]

Sorry, won't know until weds...

[Duo]

The capacitor and resistor from output to ground on the amp are there to help against oscillation.

[Jan Dupont]

It surely seems like you have a DC-offset problem....
Compare the Emitter resistor Voltages on the left and right output, and let us know

[Stuart Easson]

Ok, I visited with the offending equipment, oscilloscope and multimeter in hand. After measuring, voltages, output waveforms and finally current, I have determined the speaker is making the amp sink 0.75 amps. Not good. Quite unusual as far as I know. The amp is studly enough to sink the current and keeps the output offset to a few tens of millivolts. Given the 80+v rails, the heating effect is not a surprise.

To the limits of my handheld scope there are no unexpected waveforms on the output.

Has anyone ever heard of a speaker doing this before?

As a control we tested the amps with a couple of other speakers, some Polks and Klipschs and had no problems at all...

Stuart

[cunningham]

I have designed and built amp circuits before, and I do remember one problem that the reactive componants of the speaker caused RF occilations and made the amp output around 15W at 150KHz!

If there is no current flow when the speaker is not connected, the speaker must be causing something. Try connecting the output accross a non-reactive resistor of rated output impeadence. If this doesn't change the result, then you may have a biasing problem.