Im working on a high power amplifier that can offer peak power reaching up to around 500 watts, but im still not sure which kinda drive im gonna use.
I've simulated both BJT drive and MOSFET drive, both offer the same performance as the same power.
The output stage consist of five MJW21195 and five MJE21196, The bjt drive consist of a MJL4281/MJL4302 driven by a MJE340/MJE350.
The mosfet drive would be a IRF540/IRF9540 or a IRFP240/IRFP9240.
Which drive to choose ? the performance difference is a whopping 0.01% between the two, bjt version does 0.02% at 550Wpeak at 20kHz in 4 ohms, mosfet drive version does 0.03% at same power, frequency and load.
Main purpose of this amp would be bass, so low THD at higher freqs is not mandatory.
Schematics:
http://i.imgur.com/EjIDd.png
http://i.imgur.com/Z6ae6.png
I've simulated both BJT drive and MOSFET drive, both offer the same performance as the same power.
The output stage consist of five MJW21195 and five MJE21196, The bjt drive consist of a MJL4281/MJL4302 driven by a MJE340/MJE350.
The mosfet drive would be a IRF540/IRF9540 or a IRFP240/IRFP9240.
Which drive to choose ? the performance difference is a whopping 0.01% between the two, bjt version does 0.02% at 550Wpeak at 20kHz in 4 ohms, mosfet drive version does 0.03% at same power, frequency and load.
Main purpose of this amp would be bass, so low THD at higher freqs is not mandatory.
Schematics:
http://i.imgur.com/EjIDd.png
http://i.imgur.com/Z6ae6.png
I would take BJT here.
There is no compare of how many Bipolar is around.
They are easily found and good price.
You just get more to select in BJT and they are also more 'complementar pairs' than MOSFETS (Hexfet).
There is no compare of how many Bipolar is around.
They are easily found and good price.
You just get more to select in BJT and they are also more 'complementar pairs' than MOSFETS (Hexfet).
Oki.
Though triple EF may not be needed, it was just my simulation that dident like VAS driving the 4281/4302 directly.
The MJE340/350 drivers are likely capable of driving the outputs on their own anyways without any help.
Though triple EF may not be needed, it was just my simulation that dident like VAS driving the 4281/4302 directly.
The MJE340/350 drivers are likely capable of driving the outputs on their own anyways without any help.
My opinion,
- 3EF is difficult to stabilise (for me as a rookie)
- 2EF will load the VAS and IPS to much
- FET's as drivers solves in my opinion the above disadvantages. This when the the VAS current is not to low. To have Ciss as low as possible is not a bad though I think when choosing good FET's for drivers.
In my current design, when I switched drivers from BJT to FET's, the high order harmonics in the FFT decreased a lot, THD1 went from 0.00x to 0.000x in LTSPICE.
I didn't use HEXFET's but small lateral's, see attachment.
PS: BOB216 and BOB79 are improved models for 2SK216 and 2SJ79.
Gr
- 3EF is difficult to stabilise (for me as a rookie)
- 2EF will load the VAS and IPS to much
- FET's as drivers solves in my opinion the above disadvantages. This when the the VAS current is not to low. To have Ciss as low as possible is not a bad though I think when choosing good FET's for drivers.
In my current design, when I switched drivers from BJT to FET's, the high order harmonics in the FFT decreased a lot, THD1 went from 0.00x to 0.000x in LTSPICE.
I didn't use HEXFET's but small lateral's, see attachment.
PS: BOB216 and BOB79 are improved models for 2SK216 and 2SJ79.
Gr
The problem with MOSFETs is the voltage required to turn them on.
The bias voltage is likely to be around 7 volts so that is 3.5 colts from each rail that wont be used.
I suppose you could compensate by having a slightly higher rail.
The bias voltage is likely to be around 7 volts so that is 3.5 colts from each rail that wont be used.
I suppose you could compensate by having a slightly higher rail.
Thanks, I've just downloaded it. I was wondering about the improved SPICE models that you mentioned, but I see you included them as well.
You are right when the OPS are MosFET's, those are voltage driven.
When BJT's are used in the OPS, the voltage swing at the output is not effected while these are current driven.
Not?
MOSFETs drivers will Limit at VGS ~4V.
If VAS is using same supply rail as output, this means a loss in output voltage, too.
BJT drivers have a drop of 0.7 Volt only.
If VAS is using same supply rail as output, this means a loss in output voltage, too.
BJT drivers have a drop of 0.7 Volt only.
Member
Joined 2009
Paid Member
have you considered a CFP driver ? yes it can be unstable, but when I used it I simply added a small capacitor between base-collector of the lowest power driver device. The CFP provides a low impedance drive to the output.
the driver stage I refer to is here:http://www.diyaudio.com/forums/solid-state/111756-rmi-fc100-single-stage-audio-power-amplifier.html
my implementation is here: http://www.diyaudio.com/forums/solid-state/167369-designing-tgm3-output-triples.html
(go to the last post for final schematic)
the driver stage I refer to is here:http://www.diyaudio.com/forums/solid-state/111756-rmi-fc100-single-stage-audio-power-amplifier.html
my implementation is here: http://www.diyaudio.com/forums/solid-state/167369-designing-tgm3-output-triples.html
(go to the last post for final schematic)
I agree. Vertical fet VgsTh will require extra headroom if used in common drain in order to fully drive the output BJTs. The supposed high current gain of the mosfets may be mitigated by the input capacitance (particularly the nasty non-linear Cgd) in loading the VAS and pre-drivers more at higher frequency. This may be part of the reason for the extra 20KHz distortion with the fets. In addition IRF540 & IRF9540 are not even remotely similar in Gm and thus Vgs vs Id transfer. Just to point out, the mosfets need gate-source Zeners to protect the fragile oxide layer. Using Lateral fets would cut VgsTh down to ~1V.
then again ...dont know if this a generic schematic and will be advanced in the near feature but i dont think that you can design a such a high powered amplifier without protection ( s) inside the amplifier ...Vi limmiters / input limmtier/ some way to guard the VAS and so on ...
Since schematics like the above we have seen plenty ( or more than plenty ) i think engineers should focus on sophisticated methods of protection since all the rest we all ready seen before ...
Yet again since my theory background is bad i dont ask my job to be done by other people i wish i had the resources to do that but i dont .
thankfull regards
sakis
Since schematics like the above we have seen plenty ( or more than plenty ) i think engineers should focus on sophisticated methods of protection since all the rest we all ready seen before ...
Yet again since my theory background is bad i dont ask my job to be done by other people i wish i had the resources to do that but i dont .
thankfull regards
sakis
I absolutely HATE HATE HATE VI limiters, they introduce distortion when activated as pointed out by rod elliott. And they really just dumps the drive current into the speaker output rather than into the power devices bases = drivers pretty much become outputs and can potentially blow up, taking the rest of the amp with them.
The VI limiters only shift the failure point to the drivers so they dont protect the amp from overdrive in any way and are therefore somewhat useless, they may provide a degree of protection against loads below 4 ohms, but not for a dead short.
When an amp blows, it'll blow no matter if you have VI limiters or not. However a DC protect is mandatory on higher power amplifiers.
Yes i know, i should assume that others may build what i publish and therefore include these vi limiters.
The VI limiters only shift the failure point to the drivers so they dont protect the amp from overdrive in any way and are therefore somewhat useless, they may provide a degree of protection against loads below 4 ohms, but not for a dead short.
When an amp blows, it'll blow no matter if you have VI limiters or not. However a DC protect is mandatory on higher power amplifiers.
Yes i know, i should assume that others may build what i publish and therefore include these vi limiters.
Bensen, you could call this the amp that keeps on giving. Even with a 0.5 ohm load it is still going strong. It clips at just over +30V but all that is needed is to back off the input amplitude. The collector currents of the output devices are 12A each, well below the maximum rating of 17A in the datasheet. There seems to be no need for a VI limiter.
Take another look at the SOA curve. At 67V rail voltage, each transistor can only take up to 2A safely, that is 400 watts into 4 ohms, which is above what i can get with the transformer beeing 500VA, i'd be lucky to reach 300 watts.
Allready said that my theory is bad..... but my dear Tekko with a fancy signature i have to question yours now ... None of the VI limiters i ve seen works after the drivers ...all of them work BEFORE the drivers ...their job is to limmit drive .
now from the expirience of repairing an average of 200 amps of many types per year i have never seen blown drivers or items in the VI limmiter area caused by any failure
at 95% of blown amps the failure is around the output area and its caused mostly by thermal stress
Hate Hate Hate Vi limiters ? good !!! i actually do to .... so come up with something less distortion and more inovation !!!!!or focus on designing sophisticated methods of protection than spending time on simulating an existing circuit .
seen a cheap Gemini amplifier that protection is based on the curent flow of vas ... i can post a schematic but dont understand exactly how is working ...
there is only 2 ways to go in a high power amplifier ( high power = anything above 120W ) either you go with VI limmiter or other similar protection and then limmit also the input .... or overdesign everything to be on te safe side
winners are the designers that found proper balance between safety and quality ....
kind regards
sakis
Last edited:
A sneakpeek
My heatsink have arrived so work on this big amp can begin.
Preliminary layout of the outputs and drivers:
http://i.imgur.com/uP4z7.jpg
My heatsink have arrived so work on this big amp can begin.
Preliminary layout of the outputs and drivers:
http://i.imgur.com/uP4z7.jpg
Tekko,
Do I see it good, have you choosen for BJT's as drivers?
What was your main raison for this?
One reason why I'm choosing for FET's as drivers is that I never tried it, I wan't to see and hear what difference this makes.
Ingenieus,
Thanks for the feedback. I'm goiing to leave 1 output pair out of it. A good frequency compensation for this amp is a hard job, if someone has suggestion, I'll glad to hear it.
Do I see it good, have you choosen for BJT's as drivers?
What was your main raison for this?
One reason why I'm choosing for FET's as drivers is that I never tried it, I wan't to see and hear what difference this makes.
Ingenieus,
Thanks for the feedback. I'm goiing to leave 1 output pair out of it. A good frequency compensation for this amp is a hard job, if someone has suggestion, I'll glad to hear it.
- Status
- Not open for further replies.