Hey
yes I was working on it, its all done now, but can't say how you parallel outputs, in any case there is schematic for it, so you now how its wired, and see how paralleling output would be possible
I know what you mean, with DIY, but are you sure you could make good pcb for it?Coz any class D, if not any, at least my could be made to work on 1R, with multiple output fets..
yes I was working on it, its all done now, but can't say how you parallel outputs, in any case there is schematic for it, so you now how its wired, and see how paralleling output would be possible
I know what you mean, with DIY, but are you sure you could make good pcb for it?Coz any class D, if not any, at least my could be made to work on 1R, with multiple output fets..
judtoff said:
Hi Judtoff,
Would that be 3 sets of 4 fets? You bring up an interesting point about Rds(on) increasing with temperature. I wonder, however, how fast Rds(on) increases with temperature, and how that increase relates to the average Rds(on) for a given batch of fets.
I feel you are correct in suggesting temperature increase will naturally share current amongst the devices. However, one problem with this I see is that the device which had the lowest Rds(on) at room temperature, will run the hottest under load. Consequently, the temperatures will be different across the devices, and the whole system will be thermally limited by the device with the lowest Rds(on) at room temperature.
Perhaps your fets were carefully matched?
Hey guys,
Would anyone know of a method of scaling power in a modular class D design? So, taking a 100 watt mono design, is it possible to run 10 such blocks together to reach 1 kw?
I would imagine this scaling would take place in terms of power supply size and number of output mosfets. But that is a guess!
Would anyone know of a method of scaling power in a modular class D design? So, taking a 100 watt mono design, is it possible to run 10 such blocks together to reach 1 kw?
I would imagine this scaling would take place in terms of power supply size and number of output mosfets. But that is a guess!
In addition to redesigning the entire front end and driver circuits. Ensuring that your waveforms are just as clean when you double the # of outputs and half the load Z, for instance, may not be trivial. It's really a new design.
Would the entire front end have to be redesigned, or would it just be the gate driving section?
Another question: what would stop one from using an industrial mosfet array for switching duties? Those destined for industrial motor control seem to have current capability to spare. Would it be gate capacitance and slew-rate that make such devices unsuited for amplifier output stages? (edit: for the purposes of technical debate let's disregard the outlandish price!)
As an example, this industrial half-bridge package:
http://www.microsemi.com/datasheets/APTM10AM02FG-Rev1.pdf
As a novice I find the thought intriguing. But I'm sure it is merely a matter of time before the enthusiasm of my ignorance is replaced with the factual why-nots of reason.
You might be able to salvage the triangle wave generator and comparator. The gate driver circuits would need overhaul, and propagation delays will be longer requiring re-compensation.
Those big motor controllers are way too slow. You'd do as well or better with a bank of cheap TO-247's. I've played around with building class D circuits (nothing serious) with old slow IRFP hexfets and they work - *at low clock frequency*. Try cranking it up for performance and efficiency goes to pot. The 2nd to last figure on page 6 doesn't exactly inspire confidence as to the suitability of those modules for this application.
Those big motor controllers are way too slow. You'd do as well or better with a bank of cheap TO-247's. I've played around with building class D circuits (nothing serious) with old slow IRFP hexfets and they work - *at low clock frequency*. Try cranking it up for performance and efficiency goes to pot. The 2nd to last figure on page 6 doesn't exactly inspire confidence as to the suitability of those modules for this application.
Hi there wg_ski,
That's interesting that you've tried class D stuff with big industrial fet's. By 2nd to last figure, do you mean 'operating f vs drain current'?
No, I was using IRFP2907's. High current lowish voltage. They're really good rail switches for class H where the switching just isn't that fast, and not too bad for DC-DC converters of 60 Hz inverters with 36V batteries. But for class D they're not very clean. Dirty or slow waveforms = heat. Like I said, proof of concept at low frequency. The big industrial FETs will be even worse.
'operating frequecny vs. drain current' says it all - it just hits a wall. The curve is probably where it hits max junction temperature.
There are two limits in this kind of specifications:
1: junction temperature,
2: where sum of delays exceeds 10 % (or any pre-defined fraction) of the period.
The curve represents the lower one.
Thinking about these dinosaur FETs is pointless, since there are cheap, small, powerful and fast ones too. The weakest link is the PCB.
hi freeman are you still there ? i used the below circuit to power a 2ohm load
very succesfully , i used it loud for long hours . the big challenge is using more than one circuit in one casing . it just fails
it is ucd class d amp the build details and comments pics are in the below link . all the best
ucd 25 watts to 1200 watts using 2 mosfets
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.