spreading bias over more MOSFETS in aleph 5

after taking a break for the summer, :) I'm back to working seriously on my aleph 5. I'm making two monoblocks, each with 750VA transformers and 35v rails (thats why I'm calling it an aleph 5) My problem is with the heatsinks. I have a lot of small heatsinks that can safely dissipate about 10W each. I cannot buy bigger heatsinks, and I've looked into adding a copper backing to make a big heatsink, and it is not possible. so... I can keep the output mosfets at a safe temperature if I spread the bias out over 7 output pairs, instead of the original 3, thus assigning one heatsink to each mosfet. My question is how will this affect performance? I know that you can do this to some extent, as seen in the aleph 4. But I don't know if it matters that I have a lower rail voltage, or if I'm above the upper limit of mosfets, or if the alignment of the planets is wrong for this type of thing... btw, this is assuming 2A total bias.. I could do with 5 or 6 output devices if I lower the bias.

Thanks for your help!

Evan
 
Evan,
Spreading the bias out over more devices will work, but the resulting circuit will probably have somewhat higher distortion. Not enough to blow you out of the water. Your problem will be that you'll have to recalculate the rear end of the amp. The extra source resistors will throw the circuit out of balance. This isn't the end of the world, either. I believe Nelson gives the 'recipe' in his patent.

Grey
 
Grey, thanks for your reply. Disappointed by the reply, I looked over my numbers again (thats my kind of engineering - if you don't like the answer, then ignore it :) ) and realized that I made a big mistake... I was assuming 50 to 55 deg C for junction temperature... I don't know where I came up with that number now, but its very prominent on my notes :) I think I confused some of the items that I read on the postings of heatsink temp vs junction temp. If I assume 100 deg C for junction temp, then I should be ok.

But I'm still curious... where does the extra distortion come from? I'm asking because its done in the aleph 4, with presumably similar (to aleph 5) distortion.


Evan
 

cp642

Member
2001-08-17 6:00 pm
Evaas,
In the Aleph 5, 3 MOSFETS are required the share the 2A bias current, which is ~.6A pr device ( the optimum figure? ). When you inrease the devices, you have to increase the bias proportionally. Less bias across the devices you'll end up with more distortion

...the aleph 4 runs with 6 devices at a bias current of 3A in case you didn't notice

[Edited by cp642 on 12-09-2001 at 02:18 AM]
 
In the Aleph 5, each Hexfet is biased with about 650mA, not 400mA.
But 400mA is the drain current of each Hexfet in the Aleph 60 design.
In my opimion the correlation between distorsion and single device biasing is not so simple, and the optimum figure is quite wide, as regards the sound quality, at least from a certain point.

[Edited by tortello on 12-09-2001 at 01:58 AM]
 
Yes, I also think that the main parameter is the junction temperature, firmly related to the power dissipation and the efficiency of the heat sink (and in second rate to the thermal resistance Hexfet-heat sink).
It seems that the different dissipation figures between the PL Alephs (as GR said 22-25W) and the Volksamp versions (15-17W per Hexfet) is due to the less dissipation capability of the Volks heat sinks.
 

Nelson Pass

The one and only
Paid Member
2001-03-29 12:38 am
As usual, Grey nails it down. At 25 watts or
so, the transistor "never" fails. At 50 watts,
we see some failures.

As a rule, the higher bias figures per device
are desirable, and you could optimally look for
1 amp bias per device with supplies at +-25 to
30 volts or so. At the same time, paralleling
devices at somewhat lower bias can give you greater
transconductance, and this was chosen in designs
like the Aleph 60 as it gives more bottom end control.

When you parallel devices in an Aleph but want to
vary the bias per device, you look to the value
of the Source resistances on the current source
(the bank on the positive half), and also (referring
to the Aleph 60 schematic) resistor R19.

After you get the DC figure you are looking for, you
want to adjust the AC gain of the current source so
that the current source provides about 50% of the
output AC current.

The easiest way to do this is to build the circuit
without R21 and operate it at 10 watts or so into a
load while measuring the AC voltage across R46-51
which are the Source resistors on the negative half
of the amp. Put in a value for R21 which halves
the AC voltage across R46-51, and you'll know that
the current source is doing half the work.

Another issue that comes up with more devices in
parallel is that the capacitance of the circuit
goes up, and with it the nonlinearity of this
capacitance, which at high frequencies starts
showing up in the distortion curve. Somewhere
around 12 devices in parallel, you have to start
modifying the circuit to deal with this, depending
of course on the devices. With IRF250, this would
be around 6 devices.