I am planning to build Class A Push-Pull, something between ZEN 5 and CSX, with Sony VFETs, with no feedback, 4 Ohm output impedance. In this design, the supply ripple is going to be amplified by a factor of 3, so it's very important to use a PSU with as clean voltage as possible.
It's going to draw a constant current of approximately 1.5A per rail (neg or positive).
Guys, how to improve this design to reduce the ripple? Another MOSFET in series? Smoothing voltage on the Mosfets gate doesn't help much. Any proven ideas?
It's going to draw a constant current of approximately 1.5A per rail (neg or positive).
Guys, how to improve this design to reduce the ripple? Another MOSFET in series? Smoothing voltage on the Mosfets gate doesn't help much. Any proven ideas?
Last edited:
😱
If that’s the case, did you consider linear regulated power supply which would provide only several uV ripple and noise?
Hello,
Or you could do a search on the incredible amount of cap multiplier posts and information already on this site, and others.
Or you could do a search on the incredible amount of cap multiplier posts and information already on this site, and others.
Just add a synchronous bridge to the Salas' L-Adapter type circuit - this is what I use with the F6 and spent bigger on 'better quality' caps, no choke
You don't actually need anywhere near those oversize capacitors or the series choke for a 1.5A load or the voltage drop/power consumption of the IRFP 240/9240 in this type of circuit - suggest you have a look at 'starkey's' suggestion above - you don't need to reinvent the circuits.
You don't actually need anywhere near those oversize capacitors or the series choke for a 1.5A load or the voltage drop/power consumption of the IRFP 240/9240 in this type of circuit - suggest you have a look at 'starkey's' suggestion above - you don't need to reinvent the circuits.
This not a real cap mult, it is a regulator without an output feedback, but they have the same fundamental limitation (~60dB PSRR): the finite lambda of the MOS.
Whilst it is possible to compensate for that and remain open-loop (I have shown somewhere how to proceed) it is simpler and more effective to go closed loop by tying R11 to the output.
For the other side, things are going to be somewhat more complicated
Zen5 in of itself is a fet cap multiplier with modulated ground refference and set gain...you might consider put two of them in series , one having no modulation or some anti-phase feedback by placing the audio amplifier in the loop around the cap multiplier...
Thank you guys for your inputs, appreciate your help. I am going to check that out during the weekend. Cheers!
It might be "modern" to use an isolated DC-to-DC converter module like this one (12VDC in, 15VDC out), to create a low current, boosted supply that's 15 volts above the raw and ripply DC input. Aha! You've now defeated the number one big problem with power MOSFET based capacitance multipliers: huge Vin-to-Vout voltage drop because of huge MOSFET threshold voltage. Especially the Pchannel MOSFET in the negative supply capacitance multiplier. There's no such thing as a "logic level gate, low threshold PMOS". Delightfully, this solution only costs you $2.50 per capacitance multiplier.
Put your ripply (Vin + 15V) boosted supply through a multistage CRC filter and use that ultra clean boosted voltage, to derive your cap-mul MOSFET's gate voltage. Presto: only 1.0 volts dropped from input to output, because the Nchannel MOSFET's gate voltage is actually higher than its drain voltage. At 1.6 amps of load current, that's only 1.6 watts dissipation, which you could accommodate with a modest PCB-mounted heatsink. No need to bolt the MOSFET to the chassis.
Or maybe you'll want to use the ultra clean boosted voltage, to supply an error amplifier and a Vref generator. Now you can build yourself an actual voltage regulator with feedback from the regulated output, and get > 70dB attenuation of 100 Hz ripple. With only 1 volt drop from input to output.
_
Put your ripply (Vin + 15V) boosted supply through a multistage CRC filter and use that ultra clean boosted voltage, to derive your cap-mul MOSFET's gate voltage. Presto: only 1.0 volts dropped from input to output, because the Nchannel MOSFET's gate voltage is actually higher than its drain voltage. At 1.6 amps of load current, that's only 1.6 watts dissipation, which you could accommodate with a modest PCB-mounted heatsink. No need to bolt the MOSFET to the chassis.
Or maybe you'll want to use the ultra clean boosted voltage, to supply an error amplifier and a Vref generator. Now you can build yourself an actual voltage regulator with feedback from the regulated output, and get > 70dB attenuation of 100 Hz ripple. With only 1 volt drop from input to output.
_
Last edited:
If he goes this way then this is pretty good for contant current class A circuits: Baxandall converter strange output