I use BC Components from Farnell, and Nichicon from Mouser. Nichicon is better, especially the Super Through, even it's bit expensive.
I built rectifier using HFA15TB60s. They sound much better than the common integrated bridges.
Sajti
Attachments
SA2015: using vertical mosfets IXTH80N20L and IXTH48P20P
SA2015 reloaded: using vertical mosfets from IXYS:
IXTH80N20L and IXTH48P20P
Dave Zan has provided some preliminary spice models. Here are first simulation results. Many thanks Dave for the models and the idea to use those power horses!
In the simulation the Rgate is as low as 22R - hopefully we have no oscillations here...
BR, Toni
SA2015 reloaded: using vertical mosfets from IXYS:
IXTH80N20L and IXTH48P20P
Dave Zan has provided some preliminary spice models. Here are first simulation results. Many thanks Dave for the models and the idea to use those power horses!
In the simulation the Rgate is as low as 22R - hopefully we have no oscillations here...
BR, Toni
Attachments
Thanks.
Those Super Through sure are expensive at 30 Euro a piece.
Perhaps the Nichcon Gold Tune will do too? They are about half the price.
For the SA2015 70W amplifier you would need only one pair of those mosfets and it would be save running 1.5R load.
No matching required and smaller pcb size possible.
If I get those parts running I will design a small SMD pcb using one of those power horses. Good for building active speaker systems...
BR, Toni
Last edited:
Thx!
Holy smoke is very uncommon in my lab - hopefully it will continue to be rarely seen ...
BR, Toni
SOA of the IXTH80N20L holds up to 80Vds
A 1pair output stage kept fairly cool with low bias could achieve a maximum power upto 250W.
If you go for a higher bias as needed by FETs, then a reasonable maximum power would be around 150W from a 1pair output stage biased to 200mA to 250mA
A 40-0-40Vac transformer should take that 1pair to around 130W to 150W into 8ohms.
A 1pair output stage kept fairly cool with low bias could achieve a maximum power upto 250W.
If you go for a higher bias as needed by FETs, then a reasonable maximum power would be around 150W from a 1pair output stage biased to 200mA to 250mA
A 40-0-40Vac transformer should take that 1pair to around 130W to 150W into 8ohms.
Each of these brands are well reputed. Chose a capacitor of enough capacity, enough voltage rating, enough ripple current capability and low ESR. I don't believe in different »sounds« of capacitors of the same parameters.
On using Schottkys, you will observe some less dissipation.
Best regards!
Your simulation looks broadly in line with mine, at 100 W @ 20 kHz THD is only a few PPM, pretty nice number
In real life will be dominated by board layout etc. but should be exceptionally low.
Very pleased to be able to help with the models, I will test then more carefully in the next two days if you can wait a little.
Use a few pairs and build a hi-fidelity arc welder
That is exactly what I had in mind
Best wishes
David
SA2015: using vertical mosfets IXTH80N20L and IXTH48P20P
SOA simulations
There is always something to learn new.
During design of the dual slope VI limiter for the new power mosfets I added 2 virtual "loudspeakers" parallel at outputs to simulate overload behaviour when too many speakers are attached. The VI limiter does it's work - but the speakers too. The back-EMF leads to short SOA violation pulses (see peaks around 101V). Not only this may kill your mid and treble speakers but may kill your OPS stage too...
A "normal" short circuit is no problem - the VI limiter does the work perfectly here.
SOA simulations
There is always something to learn new.
During design of the dual slope VI limiter for the new power mosfets I added 2 virtual "loudspeakers" parallel at outputs to simulate overload behaviour when too many speakers are attached. The VI limiter does it's work - but the speakers too. The back-EMF leads to short SOA violation pulses (see peaks around 101V). Not only this may kill your mid and treble speakers but may kill your OPS stage too...
A "normal" short circuit is no problem - the VI limiter does the work perfectly here.
Attachments
