OK ok, thats only for 20 second before 16A slow-blow fuse gives up. 
setup
kludge power stage
highly refined modulator stage
Crown-BCA topology with full-bridge operated from +300V rail. 4.5++ kW to 5 ohms for short perioids, 2.2kW continuous. BIG 230/230v isolation transformer required for powering this up.
Still somewhat high losses on output stage, 2.2kW out with 85% effiency. Or something, havent calibrated my current shunts to measure efficiency any better than 5%
setup
kludge power stage
highly refined modulator stage
Crown-BCA topology with full-bridge operated from +300V rail. 4.5++ kW to 5 ohms for short perioids, 2.2kW continuous. BIG 230/230v isolation transformer required for powering this up.
Still somewhat high losses on output stage, 2.2kW out with 85% effiency. Or something, havent calibrated my current shunts to measure efficiency any better than 5%
ezis666 said:Where is schematic???
Dont have such a thing yet.
well, here's output stage anyways. bit lazily drawn.
http://www.ee.oulu.fi/~masaj/class-d/output%20stage.JPG
9Amp mosfet drivers from TI. these have proven to be very robust even without clumsy shottky diodes at driver chip output that are usually necessary to keep driver chips alive.oldroborg said:Looks awesome, exactly what i'm looking for, what drivers do u use for the gate drive transformers? What is the switching frequency? How many mosfets have u turned to smoke
cheers,
Rob
100khz swithing frequency as this is subwoo-amp. Gate drive DC-restoration circuitry limits full power -3db to around 1khz.
zero smoked mosfets, these are robust welding-machine models
IVX said:very impressive, however driver pluged by too long wires..
too long wires between mosfets and driver or driver and modulator board? Yes, wires between mosfets and drivers are longish and their inductance is bit too much, but IMHO transformer leakage inductance is dominating and difficult to reduce.
I measured 260nH leakage(total) inductance in gate drive circuitry wich is already limiting factor in reducing switching times.
Modulator stage is something ugly, pre-prototype but seems to work considerably well.
mzzj said:
recalculation from heat sinks thermal mass gives effiency of 93-95% at 2.2kW and i am much more satisfied with that than 85%
1% current shunts and 1% voltage measurements give worst-case error of 4% wich is 8% max error in effiency. Now i calculated losses from aluminium blocks heat rise, errors are much smaller this way.
(voltage measurements maybe were even worser than 1% as RF from power stage was making interference to meters.)
mzzj said:
9Amp mosfet drivers from TI. these have proven to be very robust even without clumsy shottky diodes at driver chip output that are usually necessary to keep driver chips alive.
100khz swithing frequency as this is subwoo-amp. Gate drive DC-restoration circuitry limits full power -3db to around 1khz.
zero smoked mosfets, these are robust welding-machine models
too long wires between mosfets and driver or driver and modulator board? Yes, wires between mosfets and drivers are longish and their inductance is bit too much, but IMHO transformer leakage inductance is dominating and difficult to reduce.
I measured 260nH leakage(total) inductance in gate drive circuitry wich is already limiting factor in reducing switching times.
Modulator stage is something ugly, pre-prototype but seems to work considerably well.
Awesome, -3db at 1Khz is half an order of magnitude better than i'm aiming for
I'm looking to build sub-amp(s) for <80hz work. For some reason i'd subconsciously discounted using transformer drive (winding transformers frightens me ) but now after seeing yours in the flesh i'm back on track, checked out the mosfets too, look interesting! Gonna see if i can use some torriods out of PC smps's to save winding them myself.....cheers
Rob
PS any chance of a schematic of your modulator too
Cumming soon, maybe even this evening. Its christmas-time for high-power fanaticsoldroborg said:
PS any chance of a schematic of your modulator too
Gate transformers solve lots of problems, ones I am using provide more than 4kV insulation for example... use kynar or teflon wire-wrap if you want good insulation.
For mosfets I may recommend SPW47N60C3 or similar rated coolmos, no need to parallei them anymore. Reason that I am using STWxxxx mosfets is that I happen to have bunch of them in my drawers.
I have also 55Amp IRF´s warp speed IGBT´s but I havent tried them(yet).Edit: there is no use for toroids in pc-power supplies, iron powder is not suitable for GTDs and they are too small for output filter chokes. I was planning to use 4pcs Amidon T-200A toroid of -26 material in final version, with around 30-50 turns of wire i believe. Or gapped EE55 cores as output toroids are pure pain in 4ss to wind.
ezis666 said:Where is schematic???
conceptual schematic:
http://www.ee.oulu.fi/~masaj/class-d/modulatorska.jpg
idle current adjustment/dead time adjustement R15 is different from my own prototype but should? work
Hi mzzj!
First of all, congratulations for the amp!
I'm collected a few questions about the design:
1. Had you got any problem during testing with natural voices?
Some of the instruments (like piano) have non-zero d.c. component in their frequency spectrum. Thus, the driving transformers should to saturate.
2. Are you applying repose duty factor controller, or is it constant?
3. Where could you buy SiC schottky diodes?
Keep it up!
Gyula
First of all, congratulations for the amp!
I'm collected a few questions about the design:
1. Had you got any problem during testing with natural voices?
Some of the instruments (like piano) have non-zero d.c. component in their frequency spectrum. Thus, the driving transformers should to saturate.
2. Are you applying repose duty factor controller, or is it constant?
3. Where could you buy SiC schottky diodes?
Keep it up!
Gyula
1. DC-blocking caps on GDT primary prevent saturation. In fast transients, ie 1khz square wave there is probably momentary saturation but operation with 1k square is already prohibited because of dc-clamp on secondary. Too fast transients cause ugly gate voltage waveforms resulting partial turn-on and exessive dissipation. This limits usability to subwoofer only, i have 400hz low-pass filter on input to keep it sane.Gyula said:Hi mzzj!
First of all, congratulations for the amp!
I'm collected a few questions about the design:
1. Had you got any problem during testing with natural voices?
Some of the instruments (like piano) have non-zero d.c. component in their frequency spectrum. Thus, the driving transformers should to saturate.
2. Are you applying repose duty factor controller, or is it constant?
3. Where could you buy SiC schottky diodes?
Keep it up!
Gyula
2. repose? duty factor is not constant and it makes dc-restoration clamp on GDT secondary necessary.
3. digikey, farnel, mouser etc.
hyperfast PFC-diodes could be 10 times cheaper alternative tough.
DC sound, now I have heard it all (pun intended)
Sound cannot have a DC component. Anyways, it the PWM signal which is presened to driving transformers and so is switching at your carrier frequency. A PWM signal will also always have duty cucle < 50% for a half dridge configuration.
Gyula said:
Sound cannot have a DC component. Anyways, it the PWM signal which is presened to driving transformers and so is switching at your carrier frequency. A PWM signal will also always have duty cucle < 50% for a half dridge configuration.
"DC sound, now I have heard it all (pun intended)"
No, you thought it. That isn't what I wrote.
If you calculate Fourier-series from a null-symmetric limited signal containing not equal fields on the positive ordinata and the negative ordinata, the series' zero frequency coefficient will be non-zero.
Beam it?
I just would like to ask from mzzj if he has ever listened to the amp with music. Or there was only just a hiss-test.
No, you thought it. That isn't what I wrote.
If you calculate Fourier-series from a null-symmetric limited signal containing not equal fields on the positive ordinata and the negative ordinata, the series' zero frequency coefficient will be non-zero.
Beam it?
I just would like to ask from mzzj if he has ever listened to the amp with music. Or there was only just a hiss-test.
Gyula said:
nope, havent listened anything more than 1k sine on 2" pc speaker. BtW, for some reason it broke.
Casing and emi filtering comes next when i have enough interest again as now it blocks AM and FM radio on same room completely. talking about serious EMI.
I cant drag kilowatt range subwoofers to university where i am playing with that toy.
Modulators guts:
http://www.ee.oulu.fi/~masaj/class-d/P1040036.jpg
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.