A75 Question

[N-Channel]

A question for the master Audio God himself:

Nelson,

OK, this might be sacriledge, but I am wondering, can I run the A75 in high-bias AB mode? I have two A75 boards that will push/pull 4 IRFP140/9140 pairs (each channel) from two +/- 37V and +/-47V half-bridge current-mode switching power supplies.

The plan is this: One A75 board/heatsink-module (2 channels - total 8 IRFP140s and 8 IRFP9140s) will be run, biased Hi-AB, in a bridge mode, powered by one off-line SMPS, while the other board-module is run the same way, from the other Off-line SMPS. Both modules and SMPSs in the same 19" wide chassis. The only common section to the will be the PFC'ed front end. Inputs will be switchable between balanced and unbalanced mode.

I chose the A75 for this project for several reasons: ease of switching between balanced and unbalanced mode, true complimentary differential input sections, and elegant simplicity of the topology. If this works, I plan to do a 2-channel version for the car, again switchable between balanced and unblalnced inputs.

I'm shooting for a peak power capacity of 500W from each switcher, as I would like the A75s (Now AB75s) to truly double-down into a 4-ohm load, 150W, Hi-AB. The MOSFETs will be mounted on a piece of Thermalloy heatsink with a 0.7C/W thermal resistance (I don't remember the stock number, but I will find it).

I will post pics later when I get to my camera. Each amp module has both the board and the output MSOFETS mounted on the heatsink, for simplicity of construction, ease of assembly and maintenence.

Thank You Much!


Steve

[Nelson Pass]

Class AB for that circuit works just fine. Nice thing about the
A75 is that the output stage was built much larger than needed,
just because I had those many holes in the heat sinks.

[BobEllis]

I can vouch for that. I rebiased my A75s (yes, 4 channels) when I couldn't take the heat last summer. Afraid of thermal runaway, I mounted the Vgs multiplier on top of one of my output devices for thermal compensation.

They seem overcompensated now, it takes a couple of minutes for the bias to come up enough that the crossover distortion becomes inaudible. It used to take about 30 seconds, the time it takes the VAS CCS to come up. So much for instant on solid state.

The good news is I cannot hear a difference between full and reduced bias. Maybe I need to upgrade my speakers, but my 10W peak in class A takes care of 98% of my listening.

[Nelson Pass]

I have a scope on my system, and its a loud day when
I hit 8 volts peak.

pass - plans on keeping his hearing for a little while longer.

[BobEllis]

Master,

Your Kleinhorns are probably at least 6dB more sensitive than my MTM's so we are in the same ballpark as far as output level. Just the occasional bomb blast watching movies busts that threshold.

Bob
/also enjoys hearing the music and wore earplugs and hearing protectors while working the flight deck

[lumanauw]

Hi, Mr. Pass,

A75 is that the output stage was built much larger than needed,

I've got a question. What is the relation/equation of bias current needed in VAS (minimal bias) and number of output devices? Is it different (minimal) bias needed for 1 pair of output and 10 pair of output mosfets?

I try to use formula I=VxCxF, where I=bias current, C=internal mosfet capacitance, F=max frequency desired.
But which one is C value (from datasheet)? With total C value of mosfet datasheets, the bias comes very big, while in A75 you only use about 20mA for 10 pair of mosfet outputs.

[lumanauw]

It's about bias in VAS, not in the output stage

[N-Channel]

I will post them tomorrow. I have to figure how to get them below 100K. It is 1am here and I am tired. I will comment on them further tomorrow.

[N-Channel]

I will post them tomorrow. I have to figure how to get them below 100K. It is 1am here and I am tired. I will comment on them further tomorrow.

[Nelson Pass]

Originally posted by lumanauw
I've got a question. What is the relation/equation of bias current needed in VAS (minimal bias) and number of output devices? Is it different (minimal) bias needed for 1 pair of output and 10 pair of output mosfets?

I try to use formula I=VxCxF, where I=bias current, C=internal mosfet capacitance, F=max frequency desired.
But which one is C value (from datasheet)? With total C value of mosfet datasheets, the bias comes very big, while in A75 you only use about 20mA for 10 pair of mosfet outputs.

I don't usually bother to calculate this. I start with a nominal
value for the VAS bias and see what I get for speed and
distortion, and then I adjust it accordingly. In the A75, this is
quite easy because the circuit allows easy adjustment of the
VAS bias, and also it's a push-pull circuit that can deliver more
current to the output stage than the bias figure. Of course you
can make your adjustments based on the sound, if you prefer.