Digital amplifiers

[Dr.EM]

I've been reading about digital amplifiers, and have a small idea of how they work now. I was thinking, arn't they ideal for subwoofers? The dynamic range isn't so vital for subwoofers, and your going to be filtering the high frequency signals either way. Also, it allows for a higher power in a smaller space without the need for such excessive heatsinks.

Is it actually possible to design a digital amp as a project, without buying the highly specialised ready made chips? They are all surface mount anyway by the looks of it.

 

[Nigel Goodwin]

I've been reading about digital amplifiers, and have a small idea of how they work now. I was thinking, arn't they ideal for subwoofers? The dynamic range isn't so vital for subwoofers, and your going to be filtering the high frequency signals either way. Also, it allows for a higher power in a smaller space without the need for such excessive heatsinks.

Is it actually possible to design a digital amp as a project, without buying the highly specialised ready made chips? They are all surface mount anyway by the looks of it.

It's quite possible, have a look at https://www.diyaudio.com/forums/forumdisplay.php?forumid=49 for various discussions about them.

You should be aware though, that the advantage of smaller heatsinks may well be outweighed by the higher cost of the semiconductors used, which need to be far faster than normal audio ones!.

What sort of power are you looking for?, I'm VERY :lol: :lol: HAPPY with my Behringer mixer amplifier, 400W per channel stereo for only £216.99 from https://www.bluearan.co.uk/. It's class D and switchmode PSU, so it's very small and light (for 800W).

 

[Dr.EM]

Cool, that looks nice , very compact for that sort of power yeah.

I probably won't actually build a proper one, but might experiment a bit. Are these MOSFETs suitable?

**broken link removed**

IRF530 looks good, not too expensive. I'd need 4 wouldn't I, for one channel.

I'm not too sure about the filter. I have a passive crossover, that will cut below 1khz, that would work wouldn't it, for a subwoofer? I would add an electronic one on the input if I was doing it propely of course, because 1khz is a bit high.

 

[Nigel Goodwin]

Cool, that looks nice , very compact for that sort of power yeah.

I probably won't actually build a proper one, but might experiment a bit. Are these MOSFETs suitable?

**broken link removed**

IRF530 looks good, not too expensive. I'd need 4 wouldn't I, for one channel.

I suggest you have a look at the forum I posted above, and see what they are doing.

I'm not too sure about the filter. I have a passive crossover, that will cut below 1khz, that would work wouldn't it, for a subwoofer? I would add an electronic one on the input if I was doing it propely of course, because 1khz is a bit high.

You need a lowpass active filter on the input, there are plenty of designs on the net.

 

[audioguru]

Hi Dr. EM,
A class-D digital amplifier is extremely difficult to make from scratch. It operates by switching its output transistors fully on and off in series with the supply at a very high frequency, then modulating the duty-cycle with the audio. If N-channel Mosfets are used then the upper ones must have a voltage booster for a high enough voltage for their gates.
Mosfets have a very high gate capacitance that needs to be charged and discharged by a fairly high current for them to switch quickly. If they don't switch quickly then they melt. Additional circuitry is needed to prevent the upper and lower transistors from conducting simultaneously.
A fairly complicated scheme must be used to prevent overmodulation or the output is continuously the full DC supply voltage.

TI has a tiny little surface-mount IC that provides 240 Whats output. It has two accurately matched 120W bridged amps that can be operated in parallel.

 

[drrogla]

Nice powermixer Nigel

 

[Nigel Goodwin]

Nice powermixer Nigel

I'm very pleased with it :lol:

 

[Styx]

Hi Dr. EM,
A class-D digital amplifier is extremely difficult to make from scratch. It operates by switching its output transistors fully on and off in series with the supply at a very high frequency, then modulating the duty-cycle with the audio. If N-channel Mosfets are used then the upper ones must have a voltage booster for a high enough voltage for their gates.
Mosfets have a very high gate capacitance that needs to be charged and discharged by a fairly high current for them to switch quickly. If they don't switch quickly then they melt. Additional circuitry is needed to prevent the upper and lower transistors from conducting simultaneously.
A fairly complicated scheme must be used to prevent overmodulation or the output is continuously the full DC supply voltage.

TI has a tiny little surface-mount IC that provides 240 Whats output. It has two accurately matched 120W bridged amps that can be operated in parallel.

and all this is possible.
All a CLassD amplifier is a 4-quadrant H-bridge inverter + external filter
I built a 1kW 4-quadrant inverter for my final year project and univeristy so it is not that hard, the switching freq was 10khz which was ideal for what we were using it for

Now we have a 150kw drive at work with a switchnig freq of 12khz. For a 4-quadrant inverter to be used as an audio amplifier that switching freq needs to be increased to 100kHz at least.

In increasing the switching freq the output power comes down drastically.
THe use of isolated DCC to provide floating supplies to the upper gatedrive ensure addiquate local gate-drive potential as well as enough power to blat the gate with

All the interlocks to be done in software.
so, this is all very feasable from home if you have enough time and money

 

[Dr.EM]

Yeah, now you mention about how an increase in frequency will reduce the power, but for use in a subwoofer amplifier where the highest audio frequency is likely top be only 150hz. Then mabye 30khz switching would be quite enough, using a 12db, 100hz filter on the output and a 24db variable electronic on the input.

A 150Kw amplifier, now that would be crazy :lol: :twisted: