Hey Telstar,
Nope. Only Class A, where the cap multiplier seems more appropriate, as it saves on capactitance. Class A hogs current, so even with giant caps ripple is still a problem. Eventually I found that CLC worked the best; for Shaan's circuit (which I ran at 3A 50V) I used 22,000uF/15mH/22,000uF.
Hugh
Nope. Only Class A, where the cap multiplier seems more appropriate, as it saves on capactitance. Class A hogs current, so even with giant caps ripple is still a problem. Eventually I found that CLC worked the best; for Shaan's circuit (which I ran at 3A 50V) I used 22,000uF/15mH/22,000uF.
Hugh
Thank you.
I guessed that... those damn inductors!
Thanks for your answer, it saved me lot of time in experimenting myself.
Any suggestion in caps brand/model for Shaan circuit (so we're back on topic
?Would performance (Distortion?) improve if the mostfet was changes for a lataral audio mosfet? Or does it not matter when the switch transistor is in the sorcefollower coupling, especially when it is run in class A.
I consider using them in a class A push pull design. Are switch type of mosfets suitable there?
Aksa, what mosfets did you use in you tube/mosfet design?
I consider using them in a class A push pull design. Are switch type of mosfets suitable there?
Aksa, what mosfets did you use in you tube/mosfet design?
Caps: I used Fischer Tausch (sp?), but large Philips computer grade are just fine......
Inductor: Custom wound locally, DCR 0.25R (about 60mm on a side, quite small)
Mosfets: IRFP140 and IRFP9140.
Laterals can be used, BUT, the hexfets have about five times more transconductance, so there is less variation in Vgs during operation, leading to less H2. Additionally I have found the hexfets extremely thermally rugged, with all four unchanged since construction about 1995. In my circuit, no fan is used, heatsinking is large, and dissipation in each device is 38W continuous.
Hugh
Inductor: Custom wound locally, DCR 0.25R (about 60mm on a side, quite small)
Mosfets: IRFP140 and IRFP9140.
Laterals can be used, BUT, the hexfets have about five times more transconductance, so there is less variation in Vgs during operation, leading to less H2. Additionally I have found the hexfets extremely thermally rugged, with all four unchanged since construction about 1995. In my circuit, no fan is used, heatsinking is large, and dissipation in each device is 38W continuous.
Hugh
Thanks Aksa. What makes these Mosfets (IRFP140/IRFP9140) most Qualified. What about IRFP240/IRFP9240? Farechild or ixys devices might also be interesting?
I dont remember whitch one, N or P channel, but one of them from IR should not be so good, is that something you have heard about?
Hexfets are not always easy to find Downunder; the IRF devices are easier to get. NP uses them widely too, so I'm not sure they are that bad! What you want is high transconductance, at least 5S (since there is no feedback in this circuit), low Rds, and not too high Cgd (not over 800pF). The 240/9240 series are identical to the 140/9140 except rated to 200V Vds max, but this does mean their Cgd is higher. Certainly the higher voltage rating is wasted on an amp with a single 40V rail!
Hope this helps,
Hugh
Hope this helps,
Hugh
About gyrators...
Apparently, based on research made by Teddy Pardo et al. and posted on Pink Fish Media, the C's quality on the gyrator is extremely important on sound and measurements, with its current leakage being the most important (or that is what I understood) parameter. His circuits use Tantalum ( parallel low C SMD ceramics). I am using his PowerReg as I recommended before, but I confess I never used an electrolytic based gyrator, to compare...
I found some mil-spec tantalum that are said to work until 50VDC and those are what I will use on the power regulator for Pavel's Buffer.
Apparently, based on research made by Teddy Pardo et al. and posted on Pink Fish Media, the C's quality on the gyrator is extremely important on sound and measurements, with its current leakage being the most important (or that is what I understood) parameter. His circuits use Tantalum ( parallel low C SMD ceramics). I am using his PowerReg as I recommended before, but I confess I never used an electrolytic based gyrator, to compare...
I found some mil-spec tantalum that are said to work until 50VDC and those are what I will use on the power regulator for Pavel's Buffer.
Attachments
Lumba-
Got it. You rock, and stoppers too.
Give me a few decades to live with-out stoppers. I'll come back to you with prayers of help when the FETs start oscillating. Until then please be at peace and let others be.
Sorry I didn't tell you what the improvement is, neither will to any foollish abusive time wasters.
Seriously dude, what's wrong with you?
Others-
Placing a 100ohm stopper made it oscillate. So I'm using it this way, in case anyone wondered.
Thanks.
Got it. You rock, and stoppers too.
Give me a few decades to live with-out stoppers. I'll come back to you with prayers of help when the FETs start oscillating. Until then please be at peace and let others be.
Sorry I didn't tell you what the improvement is, neither will to any foollish abusive time wasters.
Seriously dude, what's wrong with you?
Others-
Placing a 100ohm stopper made it oscillate. So I'm using it this way, in case anyone wondered.
Thanks.
Last edited:
Shaan,
there is a small positive fb loop from the supply via the 4K7 to the gate of the mosfet CCS. With positive going signal, supply sags a little, this pulse is passed through the 4K7 to the gate, and output rises just a little. The effect is small, but non-linear. It could be eliminated by splitting the 4K7 into two 2K2 in series, with a 47uF bootstrap cap from the mid-point to the ground. This ensures that the supply for the lower gate is referenced to ground, and will it will reduce distortion.
I do suggest gate stoppers too, it may be OK in the present form, but gate stoppers take care of the hexfet tendency to oscillate by slowing charge transfer to and from the gate junction, which is a cap, in fact. You can't be too careful here, particularly as the amp must drive a highly reactive speaker.
Nice circuit! Have you had a good listen? The harmonic profile of these very simple circuits is the hallmark of NP's designs, and very nice to listen to.
Cheers,
Hugh
there is a small positive fb loop from the supply via the 4K7 to the gate of the mosfet CCS. With positive going signal, supply sags a little, this pulse is passed through the 4K7 to the gate, and output rises just a little. The effect is small, but non-linear. It could be eliminated by splitting the 4K7 into two 2K2 in series, with a 47uF bootstrap cap from the mid-point to the ground. This ensures that the supply for the lower gate is referenced to ground, and will it will reduce distortion.
I do suggest gate stoppers too, it may be OK in the present form, but gate stoppers take care of the hexfet tendency to oscillate by slowing charge transfer to and from the gate junction, which is a cap, in fact. You can't be too careful here, particularly as the amp must drive a highly reactive speaker.
Nice circuit! Have you had a good listen? The harmonic profile of these very simple circuits is the hallmark of NP's designs, and very nice to listen to.
Cheers,
Hugh
I don't think so. It would be better if quiescent current of the whole output stage goes down when the amp is switched off, instead of abrupt discharge of the coupling capacitor through the CCS and a speaker..
Because such a pole can cause oscillations, while without it the CCS may be stable. According to what Shaan writes it is stable.
WB,
I have built this circuit's CCS using both Shaan's transistor drive and using LEDs stacked as a voltage reference. In both cases it is stable with or without the gate stoppers using the IRF n type mosfet.
He should build it, and since he has, he is in a position to ignore 'good' advice from the stalls. I particularly enjoyed his seminal advice for Lumba, appropriate I thought....
I have built this circuit's CCS using both Shaan's transistor drive and using LEDs stacked as a voltage reference. In both cases it is stable with or without the gate stoppers using the IRF n type mosfet.
He should build it, and since he has, he is in a position to ignore 'good' advice from the stalls. I particularly enjoyed his seminal advice for Lumba, appropriate I thought....
AFAIK, the spikes are worsened by the gate current at a transient, which produces a voltage drop across the resistor, increasing Vgd, which increases Is, or vice versa (whether or not this is negligible will depend on frequency). (may have gotten my GDS mixed up, I don't use MOSFETs much).
Without a gate stopper, current spikes will be less harsh because Ig transients don't affect Vgd.
EDIT: oh, didn't see the 6 posts after that one... I guess I'm a little late.
- keantoken
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.