Spent xmas/newyear playing with BCA architecture amps & have come up with a fairly OK full bridge sim (will post tomorrow, really...) ltspice reckons it can deliver 5Kwpeak sine, 40hz, into 4ohms with dual 100v supply. Getting flukely low THD at 1Khz <0.001%, much more at 40hz though!
I want to start prototyping something soon, so have to order some fets & diodes. These are the only parts i envisage having to long-distance order.
I'm obviously looking for the best price/performance solution, For longevity/pricing is it best to go for an ID of greater than the mean or around peak current?
What are peoples MOSFET recommendations/preferrences for peak current handling of around 30A, mean of 10A, with voltage rating >200v, also need some beefy schottkys with similar ratings.
Ultimately fast switching speeds aren't too important as the switching frequency is <100Khz & shootthru isn't an issue.
cheers
Rob
I want to start prototyping something soon, so have to order some fets & diodes. These are the only parts i envisage having to long-distance order.
I'm obviously looking for the best price/performance solution, For longevity/pricing is it best to go for an ID of greater than the mean or around peak current?
What are peoples MOSFET recommendations/preferrences for peak current handling of around 30A, mean of 10A, with voltage rating >200v, also need some beefy schottkys with similar ratings.
Ultimately fast switching speeds aren't too important as the switching frequency is <100Khz & shootthru isn't an issue.
cheers
Rob
oldroborg said:
Usually you have to select mosfets by rdson, ID rating is greatly misleading and pretty much worhtless. calculate losses from rdson and see if conduction losses are tolerable.
over 200V shottkies...there aint many. Onsemi has some brand new line of 250v shotkies, these should be most cost-effective solution. APT and possibly others have 300v GaAs-shottkys but they are somewhat pricey. Some hyperfast recovery "ordinary" si-diode could be also usefull, but then you have reverse recovery issues.
Cant wait for your shematic, how did you implement circulating current control? Without it you wont get past 5% thd.
(check what Guyla is writing in here)
http://www.diyaudio.com/forums/showthread.php?s=&threadid=69429
This circulating current is also causing pretty high losses on output stage, beware.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=68636
Before idle current adjustement my output stage was taking more than ampere from 300v supply, resulting over 400W idle losses without any load or signal. Circulating current trough inductors was more than 35amps, burnt my first set of inductors pretty smokey.
for mosfets http://www.fairchildsemi.com/ds/FQ/FQA62N25C.pdf
looks pretty nice for high-power applications.
looks pretty nice for high-power applications.
Re: Re: suitable mosfets/schottkys for BCA +-100v
Actually i haven't specifically implemented any controls on anything deliberately, it just runs! The output stage is basically the same as yours save the paralleled fets & the modulator isn't quadrature phase, just 180 phase doubled up, if you see what i mean.
In earlier incarnations of the circuit with different fet drive etc the current thru the coils was being exhausted however now the opposite is occuring & there is a constant current of around 30+ amps 'circulating' at full rating. Idling has a few amps or so i think, would prolly do the smoke thing with the inductors like you had
I wonder if a suitable choice of output coil would allieviate the problem or do i need a big rethink & redesign. I'm not happy with the gate drive either, its very mosfet sensitive & pulling the GDT bjts base drive out of those poor opamps will probably not work too long in reality! I feel like this circuit runs (apparently well, as far as power delivered to the load & THD) more out lucky choice of spice component models rather than a robust design.
I'd like to keep the gate drive as discrete as possible as i dont fancy waiting around for shipping on driver chips i'm likely to destroy often.
nice mosfet, wonder if i can get 20 of them out here without paying import duty 😱 generally small packets of electronics get thru customs without the mandatory 50%+ slapped on them
cheers
Rob
mzzj said:
Usually you have to select mosfets by rdson, ID rating is greatly misleading and pretty much worhtless. calculate losses from rdson and see if conduction losses are tolerable.
over 200V shottkies...there aint many. Onsemi has some brand new line of 250v shotkies, these should be most cost-effective solution. APT and possibly others have 300v GaAs-shottkys but they are somewhat pricey. Some hyperfast recovery "ordinary" si-diode could be also usefull, but then you have reverse recovery issues.
Cant wait for your shematic, how did you implement circulating current control? Without it you wont get past 5% thd.
(check what Guyla is writing in here)
http://www.diyaudio.com/forums/showthread.php?s=&threadid=69429
This circulating current is also causing pretty high losses on output stage, beware.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=68636
Before idle current adjustement my output stage was taking more than ampere from 300v supply, resulting over 400W idle losses without any load or signal. Circulating current trough inductors was more than 35amps, burnt my first set of inductors pretty smokey.
Actually i haven't specifically implemented any controls on anything deliberately, it just runs! The output stage is basically the same as yours save the paralleled fets & the modulator isn't quadrature phase, just 180 phase doubled up, if you see what i mean.
In earlier incarnations of the circuit with different fet drive etc the current thru the coils was being exhausted however now the opposite is occuring & there is a constant current of around 30+ amps 'circulating' at full rating. Idling has a few amps or so i think, would prolly do the smoke thing with the inductors like you had
I wonder if a suitable choice of output coil would allieviate the problem or do i need a big rethink & redesign. I'm not happy with the gate drive either, its very mosfet sensitive & pulling the GDT bjts base drive out of those poor opamps will probably not work too long in reality! I feel like this circuit runs (apparently well, as far as power delivered to the load & THD) more out lucky choice of spice component models rather than a robust design.
I'd like to keep the gate drive as discrete as possible as i dont fancy waiting around for shipping on driver chips i'm likely to destroy often.
mzzj said:
nice mosfet, wonder if i can get 20 of them out here without paying import duty 😱 generally small packets of electronics get thru customs without the mandatory 50%+ slapped on them
cheers
Rob
Re: Re: Re: suitable mosfets/schottkys for BCA +-100v
I'd encourage you to try their samples program. Also before you do try and find a few more you think might work and get a few samples of those too.
They cover the shipping.
Regards,
Chris
oldroborg said:
I'd encourage you to try their samples program. Also before you do try and find a few more you think might work and get a few samples of those too.
They cover the shipping.
Regards,
Chris
Sorry for the "noob" question, but what is this "circulating current" stuff in full bridge Class-D amplifiers? How does this circulating current evolve?
Applies only to BCA-topology.Danko said:
When you drive BCA-topology with more than 100% total pulse widht current starts to circulate trough inductors, switces and then returning back to supply caps. Inductors can easily have 30A dc-component if not designed properly...
To prevent gross non-linearity in output stage circulating current should be at least half of peak current delivered to load at highest power.
Lazy ordering ferrites/mpp-cores. 🙂Danko said:
Until now i have been testing it with 2pcs 50mm OD toroids and 2pcs EE65 cores. I need more cores.
I had 3.5kW PFC-stage also on my mind, maybe I move to that project next. 🙂 (I have plenty of unfinished and paperprojects...)
I'm not understanding the operation of the circuit i posted a few posts back 
I was working on a 'half bridge' BCA as outlined in that Crown BCA paper, just doubled up (so more like a 'normal' push-pull full bridge) the idea being only to apply twice the PD across the load & to allieviate supply pumping by taking the power back thru the other side.
There's no attempt to reduce ripple by driving the other half of the output stage 90degrees out of phase.
At the moment a circulating current of 30+amps (if thats what it really is) builds up in about 10ms after power-up, that just doesnt seem like something i want an amp to do, but I'm fairly clueless as to how to keep the circulating current back low. Need some feedback somewhere, but no idea what needs controling.
I would gladly trade a few % THD for stopping that infernal current from wasting power & producing a china-syndrome output stage. After all the speaker it will be driving will be producing up to 20% 2HD throat distortion @ full rating (a few extra harmonics from the amp will prolly make it sound louder too 😱, this is an amp purely for un-hifi PA use...
Any more knowhow info on this type of design would be greatly appreciated b4 i prototype something.
cheers
Rob
I was working on a 'half bridge' BCA as outlined in that Crown BCA paper, just doubled up (so more like a 'normal' push-pull full bridge) the idea being only to apply twice the PD across the load & to allieviate supply pumping by taking the power back thru the other side.There's no attempt to reduce ripple by driving the other half of the output stage 90degrees out of phase.
At the moment a circulating current of 30+amps (if thats what it really is) builds up in about 10ms after power-up, that just doesnt seem like something i want an amp to do, but I'm fairly clueless as to how to keep the circulating current back low. Need some feedback somewhere, but no idea what needs controling.
I would gladly trade a few % THD for stopping that infernal current from wasting power & producing a china-syndrome output stage. After all the speaker it will be driving will be producing up to 20% 2HD throat distortion @ full rating (a few extra harmonics from the amp will prolly make it sound louder too 😱, this is an amp purely for un-hifi PA use...
Any more knowhow info on this type of design would be greatly appreciated b4 i prototype something.
cheers
Rob
oldroborg said:
Now, finally... 🙂
Try creating variable dc-offset for U2 in your schematic. Quick and dirty as i have in my prototype.
That does the trick! Stops the circulating current in its tracks, i found i only needed a tiny offset to prevent it building up, doesn't add too much distortion either, can still get much less than 0.5%.
Only problem is this creates a significant dc offset (not too suprising really!) So would need some dc blocking caps on the output or maybe (possibly cheaper) an isolation transformer as an output transformer! A blocking capacitor is desirable as it would offer DC protection for the speakers & also allow for a hard wired low frequency rolloff.
I wonder if its possible to alter some other aspect of the GDT signals to prevent current circualtion without adding a dc offset?
cheers
Rob
Only problem is this creates a significant dc offset (not too suprising really!) So would need some dc blocking caps on the output or maybe (possibly cheaper) an isolation transformer as an output transformer! A blocking capacitor is desirable as it would offer DC protection for the speakers & also allow for a hard wired low frequency rolloff.
I wonder if its possible to alter some other aspect of the GDT signals to prevent current circualtion without adding a dc offset?
cheers
Rob
you won't get 5kw into 4R with +/- 100v rails, especially not with the distortion figures you mentioned.
You will need maybe +/- 250v rails for that !
You will need maybe +/- 250v rails for that !
Dunkyboy said:
...Full-bridge....
My class-d BCA with +300v single-rail supply will hit over 4.5kW to 5R load for short perioids and limiting factor here is 230v 16A circuit breakers. With 3-phase 25A supply I expect some more... 😀
3-phase wiring would also lower ripple on caps greatly, less bulk capacity needed.
Hey mzzj what do u reckon to using an isolation transformer as an output transformer? I want 'hard wired' dc speaker protection. Will eliminate any dc offset in the output stage too. Not so bothered about burning off a few hundred watts in the output stage so small dc offset is tolerable (should still get ~90% efficiency full rating 😎 )
Would also make me feel alot more confident about bridging more than one amp too.
For its intended use, sub-woofer cooker 20-80hz 24dB/octave i reckon a few kVA isolation transformer would be interesting, cheaper than a suitable bank of bi-polars too (caps would probably just go off like fire crackers if the output stage failed & prolly feed dc to the speaker anyway.)
By playing around with using 220-110 transformers could probably better impedance match the load & amp, squeeze a few hundred more watts out of the amp!
cheers,
Rob
Would also make me feel alot more confident about bridging more than one amp too.
For its intended use, sub-woofer cooker 20-80hz 24dB/octave i reckon a few kVA isolation transformer would be interesting, cheaper than a suitable bank of bi-polars too (caps would probably just go off like fire crackers if the output stage failed & prolly feed dc to the speaker anyway.)
By playing around with using 220-110 transformers could probably better impedance match the load & amp, squeeze a few hundred more watts out of the amp!
cheers,
Rob
Dunkyboy said:
Keeping the drive roughly 10-90% duty cycle I can get around 2-3Kw average with 5-6Kw peak output into 4 ohms, depends alot on the choice of mosfet & how much clipping distortion can be tolerated. If i allow it to go into clipping it delivers alot more, but like that its not really an audio amplifier!
The mosfets i'll use will be >300v rated so that once tested at a much lower supply i'll increase it to the 220v mains line (so feed it with ~300v.)
cheers,
Rob
Install another trimpot for offset trimming(add some offset to inverted input signal for example) or use some sort of feedback/dc-servo to keep it zero.oldroborg said:
I think circulating current control is easier to do before comparators. Traditional class-d amps use speedup diodes on mosfet gate resistors to prevent shoot-trough but to actually make it adjustable its way more easy to do before comparators.
oldroborg said:
Well, why not if you like. Should be fine for PA as yours. I would put it before amplifier to keep things simple and do single-rail full-bridge or +-dual-rail half-bridge. For strictly under 100hz operation I think traditional class-d should be easy alternative. Maybe even with fast IGBT's if low-cost is preferred. 🙂
- Status
- Not open for further replies.