If you've never designed a class D amplifier before, start at a discrete 25W circuit and see if you can sort out all control / EMI / reliability related design issues. Next, move up a station to 100W. This will bring up a fresh set of problems. The next step can be 1000W, and, if your even mildly critical about switching behaviour, this will show you every corner of MOSFET physics you can dream of.
Class T Option
You may consider building a Class T amplifier. It also has a switching output topology but the creation of the switching signal follows a digital route. Tripath, www.tripath.com, make the necessary processing chips all you have to do is add the output FETs. Again it is a very power efficient design which results in low weight but high power. There was a 300W per channel, bridge-able to 600W, described in Elektor Magazine starting June 2004 based on Tripath's TA3020. If you would like the articles I have them in soft format, I can email them to you.
You may consider building a Class T amplifier. It also has a switching output topology but the creation of the switching signal follows a digital route. Tripath, www.tripath.com, make the necessary processing chips all you have to do is add the output FETs. Again it is a very power efficient design which results in low weight but high power. There was a 300W per channel, bridge-able to 600W, described in Elektor Magazine starting June 2004 based on Tripath's TA3020. If you would like the articles I have them in soft format, I can email them to you.
Im also thinking of class d amp (1kW and up) for frequencies up to 500hz-1khz for some time now, i would like it to be as simple as it could be, quality is not an issue because i would use it for PA..
With which pwm modulators could i achieve this?
Would it need nfb (i dont care for the distortion so much as for loudness, i think i could get away with distortions up to 1-3%.. what do you guys think?) ?
If anyone has some ideas (or schematics..) which would help me start this project i would be very grateful
regards,
skaara
With which pwm modulators could i achieve this?
Would it need nfb (i dont care for the distortion so much as for loudness, i think i could get away with distortions up to 1-3%.. what do you guys think?) ?
If anyone has some ideas (or schematics..) which would help me start this project i would be very grateful
regards,
skaara
Skaara: If you only need up to 1 kHz, you can get away with quite a lot of dead time, making the output stage simple, and the efficiency high.
If you are talking 8 Ohm 1kW you need a rail voltage of around 135 V DC. This means you are looking at 400V MOSFET's.
The peak current is 16 Ampere @ 8 Ohms, 32 Ampere at 4 Ohms etc. You can get very powerful 400 - 500 V MOSFET's in a small package. One of the best one in TO220 package is STP20NM50 from ST. A pair of these could be made to produce 1kW in 8 Ohms, if cooled properly. I have run them up to 400 kHz with 150 V rails with no problems.
If you want to cover 2kW in 4 Ohms, you have to look at IXYS or APT. They are in a whole other ballgame of power!
If you are talking 8 Ohm 1kW you need a rail voltage of around 135 V DC. This means you are looking at 400V MOSFET's.
The peak current is 16 Ampere @ 8 Ohms, 32 Ampere at 4 Ohms etc. You can get very powerful 400 - 500 V MOSFET's in a small package. One of the best one in TO220 package is STP20NM50 from ST. A pair of these could be made to produce 1kW in 8 Ohms, if cooled properly. I have run them up to 400 kHz with 150 V rails with no problems.
If you want to cover 2kW in 4 Ohms, you have to look at IXYS or APT. They are in a whole other ballgame of power!
AMP2 from 41hz.com would be close?
Voltage DC*1 VA rating*2 Speaker impedance Output Comment
+/-45V 750 4 ohm 2x300W Standard solution
+/-45V 400 8 ohm 2x150W Standard solution
+/-45V 750 8 ohm bridged 1x600W Standard solution
+/-60V 750 8 ohm 2x300W Standard solution*3
+/-60V 1500 4 ohm 2x600W Requires component upgrading*4
+/-60V 1500 8 ohm bridged 1x1200W Requires component upgrading*4
Voltage DC*1 VA rating*2 Speaker impedance Output Comment
+/-45V 750 4 ohm 2x300W Standard solution
+/-45V 400 8 ohm 2x150W Standard solution
+/-45V 750 8 ohm bridged 1x600W Standard solution
+/-60V 750 8 ohm 2x300W Standard solution*3
+/-60V 1500 4 ohm 2x600W Requires component upgrading*4
+/-60V 1500 8 ohm bridged 1x1200W Requires component upgrading*4
Did I post my 4KW full-range, audiophile class-D amp schematic yet, with zero-dead time and 99.99% efficiency? No? Well, that's because I would have a hard time designing it in the first place, as Bruno correctly pointed out.
What I am wondering about though is what's with these people that seem to think designing and building a class-D amp is like ordering a LM3886 and soldering a few parts to its leads to make it work?
Or am I missing the point here?
Sander Sassen
http://www.hardwareanalysis.com
What I am wondering about though is what's with these people that seem to think designing and building a class-D amp is like ordering a LM3886 and soldering a few parts to its leads to make it work?
Or am I missing the point here?
Sander Sassen
http://www.hardwareanalysis.com
Why stop at 4 kW?
Go for 10 kW. Why not? Gets easier the higher the voltage/power requirements. Especially with a plethora of MOSFETs and caps to chose from at the higher voltages. And if you don't care about >10% distortion, the efficiency will go way up. EMI? Who cares....their ears will be bleeding by then anyway.
Yep, lot easier than building a chip amp.
Jocko
Go for 10 kW. Why not? Gets easier the higher the voltage/power requirements. Especially with a plethora of MOSFETs and caps to chose from at the higher voltages. And if you don't care about >10% distortion, the efficiency will go way up. EMI? Who cares....their ears will be bleeding by then anyway.
Yep, lot easier than building a chip amp.
Jocko
OT.
Lars Clausen,
Can somebody from your company read email
lc@lcaudio.dk
thank you and sorry about off-topic.
Lars Clausen,
Can somebody from your company read email
lc@lcaudio.dk
thank you and sorry about off-topic.
I dont know why some people make fun of attempting high power low frequency class d.. I built quite a few smps but never built class d for real use and now i`d try it because it works on same principle. If i understand that properly it wouldnt need that high switching frequency for amplifying signals up to 500hz (or 1 khz)..
Would 10khz switching frequency be enough?
And is negative feedback necessary?
I tried once building class d with sg2524 for fun, it was very simple and it worked but i wonder if this would work for that power too?
Would 10khz switching frequency be enough?
And is negative feedback necessary?
I tried once building class d with sg2524 for fun, it was very simple and it worked but i wonder if this would work for that power too?
Skaara: If you use 10 kHz, you will probably have a pack of howling dogs around your house ( = poor S/N ratio )
I would recommend using higher switching frequency, otherwise you have to have a big choke on the output, to filter it. ( = high output impedance).
I would say 250 - 400 kHz is a good natural place.
I would recommend using higher switching frequency, otherwise you have to have a big choke on the output, to filter it. ( = high output impedance).
I would say 250 - 400 kHz is a good natural place.
The first reason is the behavior of music signal: it is mostly low level, and there is sometimes a big "bang". For big bang You must use high rail voltage, even for a tiny woofer like this: http://www.hszk.bme.hu/~sp215/elektro/Sony_hangfal.jpg , wich eats 80 V peak voltage (8 ohm nominal impedance per speaker), without any problem! The effective current is only a fraction of voltage/nominal impedance, because of the resonance near the dominant freq in music, but you still have to ensure the full current capability.
This is why a 1 kW amp is not useless. And why classD? Practically a classD amp's efficiency is not only better a little, as 90% vs. 65% tells, but a huge! First of all, the woofer's impedance is very reactive, so a lot inefficient power dissipating on a ClassAB amp. The music signal's mentioned high dinamics makes ClassAB's efficiency even much lower, since ClassD's efficiency drops only a few. In practice: a 15", 400 W Eminence woofer in a sealed box can be overload with a ClassD amp, powered from a DC/DC converter (with 2*15000uF/35V puffer capacitor), and a 12V*16A current limited power supply, and current was pulsating around 8 A. In ClassAB, I couldn't drive fully this speaker with the same supply.
10 kHz is toooo low, 50 kHz is minimally recommended. Feedback too.
This is why a 1 kW amp is not useless. And why classD? Practically a classD amp's efficiency is not only better a little, as 90% vs. 65% tells, but a huge! First of all, the woofer's impedance is very reactive, so a lot inefficient power dissipating on a ClassAB amp. The music signal's mentioned high dinamics makes ClassAB's efficiency even much lower, since ClassD's efficiency drops only a few. In practice: a 15", 400 W Eminence woofer in a sealed box can be overload with a ClassD amp, powered from a DC/DC converter (with 2*15000uF/35V puffer capacitor), and a 12V*16A current limited power supply, and current was pulsating around 8 A. In ClassAB, I couldn't drive fully this speaker with the same supply.
10 kHz is toooo low, 50 kHz is minimally recommended. Feedback too.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.