[PSP]

For the benefit of the AKSA community, I thought I should describe the outcome of an experiment that didn't work out so well.  My purpose in writing this is to save someone from repeating my mistake.

For a long time, I have been interested in biamping my system.  In fact it was the modular nature of the AKSA amps and power supply that first piqued my interest (and then I read some reviews and first impressions... then I was *really* interested!).  

My initial vision was REALLY grand... I wanted to put four 55w AKSA modules and power supplies, the TLP (with source selector and attenuator), and possibly an active XO all in one 2U 17x14 enclosure.  To do this (and this is an important detail for later developments) I needed to lay things out so that I could squeeze four 55w AKSA modules on one 300mm heatsink (actually, I cut the heatsink in half and planned to mount two AKSA modules on each 150mm heatsink half--the heatsink halves mounted on the sides of the enclosure toward the rear).

My design certainly achieved my primary goal (short signal paths) but it failed on noise... in fact, with only two 55w AKSA modules in place (the other two were unbuilt at the time) I had hum in the TLP due to EMI from the AKSA torroids.  After a long struggle with shielding and playing with orientation, I put the TLP in its own enclosure... the hum went away, and the music has been sweet indeed now for many months.

Last month I built the other two 55w AKSA amp modules and mounted them as planned above, two amp modules on each half heatsink.  My rationale (the reason I thought I could get away with this) was that I would still be playing my music at the same SPLs... the only difference would be that I would be using two amps per channel to achieve that SPL... and thus (I reasoned) the overall heat dissipation would not be changed much, if at all.  

So, I mounted the amps, and biased them.  Because I now had two amps on each 150mm heatsink half, I biased both left channel amps simultaneously (so whatever thermal feedback Hugh designed into the amp/heatsink system would be intact)... so I played music through both left channel amps, adjusted P1 on both amps for 55mv bias, then played both channels loud and checked the bias... and repeated this for the right channel.

I played music at very low levels overnight to burn in the new amp and noticed in the morning that the heatsinks were much warmer than I have ever seen (in the past, my heatsinks have remained cool, just barely warm to the touch, no matter how loud or long I play).  

Putting a thermocouple in the middle of each heatsink and letting it equilibrate for at least 30 minutes, I got the following temperatures:

ambient air temp = 21-22C
play low level (70db)overnight, heatsink temp = 35C
play at loud level (80-88db) 30 minutes, heatsink temp = 38C
I measured similar temps on both (150mm long) heatsinks, by the way.

So, it looks like I made a bad assumption... that the total heat dissipation would be relatively unchanged if I used 4 amps to achieve the same SPL vs. 2 amps.  My next move will be to put the second set of amps into its own enclosure (probably run it at 25w and make a tweeter amp out of it, since Hugh says that at 25w "it will sound even better"... he's got to be kidding, but I'm eager to give it a listen!).  

good luck and sweet music to you all,
Peter


forgot something... how did it sound (my passively biamped system)?  Better, but it was more subtle than I might have expected.  First, my speakers (Energy exl26) are 93db efficient, so they don't need a lot of power and I don't listen very loud anyway.  Still, there was a clear sense of more power in reserve and my speaks could perhaps play a bit louder before they begin to shriek (can you guess what I'm planning to upgrade soon, in addion to adding a GK-1?).  Clarity, definition, layering, and bass punch *may* have improved, but again, a subtle effect.  I'm still listening to the 4 amp biamped configuration... perhpaps when I go back to one amp/channel in the near future (while I build the 25w AKSA tweeter amp) the benefits of biamping will be more evident when they are gone.  We'll see.

[mb]

It takes some effort to fully digest your post, so excuse me if I've not got all the points right. I'm also bi-amping, and love the results. I expect would love it even better if my LF amp were also AKSA55n. It's a modest Rotel, but with my mods, it does a creditable job.

With all the modules you've put in, it's not a surprise about the heat you're observing.  You've got more than double the baseline (bias, overhead, etc) heat generating the in a standard amp, all in one chassis. I would have thought that two chassis, each close to the speaker, would be the way to go.

With my 55n driving 200Hz-top, and the Rotel looking after 10-200Hz, with overlap of course, the result is noticeable "ease of drive". Integration was less of a problem than expected, and a new digital EQ with measurement s/w is now focused on managing some room modes (eg. a BIG 56Hz resonance). Results will be posted sometime. Looks promising.

[Larry]

ambient air temp = 21-22C
play low level (70db)overnight, heatsink temp = 35C
play at loud level (80-88db) 30 minutes, heatsink temp = 38C
I measured similar temps on both (150mm long) heatsinks, by the way.

A few things you can check.

Did you connect the two star earths of two PSUs together?
Did you have any signal input's polarity reversed?
How did you connect your LF to the system, through signal line level or amp speaker high level output? It there any possibility the polarity of them has been reserved?

[AKSA]

Hi Peter,

Interesting figures.  Let's assume a 16C rise above ambient.  The half heatsink has a thermal impedance of 0.75C/watt, so if it rises 16C the dissipation is around 12 watts.

From each module that is just 6 watts.  In truth, even without a music signal, and running from 36V rails, this is a quiescent of around 90mA;  or more correctly, a quiescent of 58mA (this gives 4.2 watts of dissipation per module at idle) and about two watts of signal dissipation.  That's tiny, in fact.  There's clearly no oscillation at all;  this is minimal dissipation.

You merely need a little more heatsink, that's all!

Cheers,

Hugh

[Larry]

I had a personal experience with the significant temperature raise of my ASKA100. (It was still playing music as normal when it's hot.) I was panic and turned it off in a hurry. At the end I found I accidentally reversed the polarity of the subwoofer connection. After I rectified the connection error, the temperature returned to normal. So the external conditions, in particular presenting unexpected potentials attached, could affect the quiescent currents of an amplifier considerably.

I had another personal experience with seperate PSUs for each channels when it was connected to a common preamp that apparently provided a ground loop to the amp, which caused one channel running at higher temperatue than the other. After I connected the two star earths of the PSUs of the both channels, the temperature returned to normal and balanced. So when two seperate PSUs are connected through one common signal ground, it is possible that some quiescent currents are also affected, not necessarily channel vs channel but possibly PSU vs PSU.

[AKSA]

Larry,

I may be wrong here, but the quiescent current, and thus the quiescent dissipation, is set solely by the bias generator, and the six output stage devices on the 100W AKSA;  the two drivers, and the two pairs of complementary outputs.

The polarity of any attached load, as long as it is floating like a speaker driver, should have no effect.

I would suggest that the additional heat might have been for some other reason than the polarity of the connected speaker.  It's very odd....

Cheers,

Hugh

[Larry]

The polarity of any attached load, as long as it is floating like a speaker driver, should have no effect.

I would suggest that the additional heat might have been for some other reason than the polarity of the connected speaker.  It's very odd....

I said the polarity of the subwoofer's connection, which is an active subwoofer connected through amp's speaker outputs. If it's a passive speaker, I agree with you.

When it's acitve, it's different. Imagine that if the postive ouput of one amp finds an easy way(wishfully not directly), into the ground of another amp through it's negetive input, what could happen? I did not check how the subwoofer's amp is designed, but after rectified the polarity of the connection, the problem was gone.

[PSP]

Thank you all for your comments... I fully acknowledge that my design pushed the envelope a bit too far... that was the point of the post... to document my mistake so that some other poor sod would not be doomed to repeat it.

I agree that I need more heatskink and I've ordered parts to make the changes.  Nor have I given up on biamping... I guess I was  expecting a major improvement right out of the gate; instead the improvement was more subtle but real nontheless... "impact of biamping", like many things, will likely be highly system specific.  Like mb noted, the biamped system sounded more muscular, showed more "ease of drive" for sure.  Maybe my expectations were unrealistic... when I did the Nirvana upgrade on my TLP the improvement was huge but not so with my initial biamping experiments.  I will continue to work on things and post my results as I go.  If I reach biamping Nirvana that will be good; if not I'll have an outstanding amp to drive the rears.  

Peter