Since my aussieamps N-channel still is my favorite (DIY) amp...
I just wondered why there are so few(?) quasi designs?
What is good and what is bad/difficult with such designs?
Arne K
I just wondered why there are so few(?) quasi designs?
What is good and what is bad/difficult with such designs?
Arne K
Hi Cobra,
The two main Difficulties Arised when Designing the Quasi amp with N-channel Mosfets are:
Cross-Conduction,
Getting Rail to Rail Swing of output voltage
regards,
K a n w a r
The two main Difficulties Arised when Designing the Quasi amp with N-channel Mosfets are:
Cross-Conduction,
Getting Rail to Rail Swing of output voltage
regards,
K a n w a r
Kanwar, Arne,
I've not done a quasi-comp yet, but they are beginning to really appeal to me.
R2R Swing: No problem on the bottom rail, since if heavily driven an N-channel vertical mosfet will swing to within a volt of its source.
Top rail is different, however. Perhaps the best way to achieve close to the top rail on the positive half cycle would be to use a bootstrap to lift the midpoint of the BS resistors well above rail. This would easily accommodate the five or six volts bias (and source resistor drop) needed to ensure that Drain-Source voltage is down to about a volt. There is much to be said for an amp with 63V rails which can swing 124Vpp across a speaker!
Cross conduction is caused by asymmetric drive delays between top and bottom of output stage. Perhaps the best way to avoid this problem is to drive each block along identical signal path lengths, and preferably from opposite outputs of the LTP input stage rather than inserting an extra phase shifting stage right at the driver level. This is the usual drive topology, but it's costly of rail efficiency since the driver base (and hence its emitter which is one diode from output level) cannot go lower than the full mosfet drive required, which could be as much as 5 or 6 volts. Quasi, a South Australian designer in the High Power Quasi thread, has done this very cleverly and used a CCS to supply the second LTP for the twin VASs which even accommodates temperature compensation.
Arne, I know Anthony Holton and have heard his N-channel amp. I agree with you, it's a fine amp with a sweet, engaging sound.
Cheers,
Hugh
I've not done a quasi-comp yet, but they are beginning to really appeal to me.
R2R Swing: No problem on the bottom rail, since if heavily driven an N-channel vertical mosfet will swing to within a volt of its source.
Top rail is different, however. Perhaps the best way to achieve close to the top rail on the positive half cycle would be to use a bootstrap to lift the midpoint of the BS resistors well above rail. This would easily accommodate the five or six volts bias (and source resistor drop) needed to ensure that Drain-Source voltage is down to about a volt. There is much to be said for an amp with 63V rails which can swing 124Vpp across a speaker!
Cross conduction is caused by asymmetric drive delays between top and bottom of output stage. Perhaps the best way to avoid this problem is to drive each block along identical signal path lengths, and preferably from opposite outputs of the LTP input stage rather than inserting an extra phase shifting stage right at the driver level. This is the usual drive topology, but it's costly of rail efficiency since the driver base (and hence its emitter which is one diode from output level) cannot go lower than the full mosfet drive required, which could be as much as 5 or 6 volts. Quasi, a South Australian designer in the High Power Quasi thread, has done this very cleverly and used a CCS to supply the second LTP for the twin VASs which even accommodates temperature compensation.
Arne, I know Anthony Holton and have heard his N-channel amp. I agree with you, it's a fine amp with a sweet, engaging sound.
Cheers,
Hugh
Quasi...
Thanx for the explanations...
Oki, I understand that there is some special considerations, but (exept for the fact that quasi uses only N-ch / npn output devices) what are the good points?
(Well, I know mine sounds good ;-)
And a bit efficiency-loss is no problem?
Arne K
Thanx for the explanations...
Oki, I understand that there is some special considerations, but (exept for the fact that quasi uses only N-ch / npn output devices) what are the good points?
(Well, I know mine sounds good ;-)
And a bit efficiency-loss is no problem?
Arne K
Hi, Cobra2,
I've build quasicomp with Vmosfet and the most stable one is something like in patent #5,783,970 (Joseph J Pleitz).
This configuration is very clever, very stable. Look at the arrangement of T6, T7 and R4. And T4,T5, like AKSA said, both legs of differential are used here.
I've build quasicomp with Vmosfet and the most stable one is something like in patent #5,783,970 (Joseph J Pleitz).
This configuration is very clever, very stable. Look at the arrangement of T6, T7 and R4. And T4,T5, like AKSA said, both legs of differential are used here.
Quasi...
@ lumanauw ; do you have a link to the patent?
(btw, nice car-amps! )
Let me rephrase my question; what is the advantage of a quasi amp compared to a N + P type?
Arne K
@ lumanauw ; do you have a link to the patent?
(btw, nice car-amps! )
Let me rephrase my question; what is the advantage of a quasi amp compared to a N + P type?
Arne K
Hi Aksa,
Any retraction or am I misinformed again?
MIRACLE would be my comment unless the condition you conveniently did not state was >=100ohm speaker and signal duration not exceeding 100uS.
Any retraction or am I misinformed again?
Hi AndrewT,
Nice try!! I don't give retractions because I'm not writing text books or magazine articles here, but if you:
1. Drive the upper output block with a bootstrap, so that input at the gate can easily exceed the upper rail, and thus bring source within a volt of drain, and
2. Can measure 1V drop drain-source on a Class A lower N-type CCS at clip with constant bias, then it can be done with adequate drive (I have made this measurement, BTW, on a working amp).
I see no reason therefore why it could not be done with a stiff power supply, miracle or otherwise, and I feel this is something worth aiming for. Of course, this would be equally possible with dual bootstraps on a fully complementary, too.
Cheers,
Hugh
Nice try!! I don't give retractions because I'm not writing text books or magazine articles here, but if you:
1. Drive the upper output block with a bootstrap, so that input at the gate can easily exceed the upper rail, and thus bring source within a volt of drain, and
2. Can measure 1V drop drain-source on a Class A lower N-type CCS at clip with constant bias, then it can be done with adequate drive (I have made this measurement, BTW, on a working amp).
I see no reason therefore why it could not be done with a stiff power supply, miracle or otherwise, and I feel this is something worth aiming for. Of course, this would be equally possible with dual bootstraps on a fully complementary, too.
Cheers,
Hugh
Hi Hugh Dean,
There are some excellent ways through which a Quasi NVMOS amp could be made with Great performances....
1..Quasi adopted the Siemens design and Developed into a Great design with thermal compensation + Cascode driver stage...very good indeed.....
2. QSC uses OPAMPs to drive the VFETS directly in PL6.0II amp...
3. Workhorse uses Push-Pull Drivers to drive the VFETs with Floating power supplies and THE output peak voltage swing KISSES the Sweet & Sexy Rails with a margin of just 550mV Max Driving 2 ohm loads..
4. Meyer Sound uses Summed Differentials to drive the VFETs
regards,
K a n wa r
There are some excellent ways through which a Quasi NVMOS amp could be made with Great performances....
1..Quasi adopted the Siemens design and Developed into a Great design with thermal compensation + Cascode driver stage...very good indeed.....
2. QSC uses OPAMPs to drive the VFETS directly in PL6.0II amp...
3. Workhorse uses Push-Pull Drivers to drive the VFETs with Floating power supplies and THE output peak voltage swing KISSES the Sweet & Sexy Rails with a margin of just 550mV Max Driving 2 ohm loads..
4. Meyer Sound uses Summed Differentials to drive the VFETs
regards,
K a n wa r
Kanwar brings passion to Indian Audio -
..." uses Push-Pull Drivers to drive the VFETs with Floating power supplies and THE output peak voltage swing KISSES the Sweet & Sexy Rails with a margin of just 550mV Max Driving 2 ohm loads".
Bagwan Kanwar.
..." uses Push-Pull Drivers to drive the VFETs with Floating power supplies and THE output peak voltage swing KISSES the Sweet & Sexy Rails with a margin of just 550mV Max Driving 2 ohm loads".
Bagwan Kanwar.
I've used the IRF type FETs with a push-pull driver, a high value gate resistor, with a diode bypass for gate charge. Common source with an opamp drive and a common base level shift.
Very nice for 100Khz square waves.
Very nice for 100Khz square waves.
amplifierguru said:Kanwar brings passion to Indian Audio -
..." uses Push-Pull Drivers to drive the VFETs with Floating power supplies and THE output peak voltage swing KISSES the Sweet & Sexy Rails with a margin of just 550mV Max Driving 2 ohm loads".
Bagwan Kanwar.
GreG😀 😀 😀 😀 😉 😉
You know Amplifier Designing is indeed a Passion which engulfs both Art, Skill, Crafting & Science as necessary ingredients.....
🙂 🙂 😀 😀 😉 😉
K a n w a r
Greg, I agree!
Kanwar brings Bollywood exuberance to mosfet design, and knows a thing or two........
Half a volt from each rail into 2R, you say???????? WOW!!!
Kanwar, would you like to tell us why you think N-type quasi-comps are superior to full complementary designs?
Cheers,
Hugh
Kanwar brings Bollywood exuberance to mosfet design, and knows a thing or two........
Half a volt from each rail into 2R, you say???????? WOW!!!
Kanwar, would you like to tell us why you think N-type quasi-comps are superior to full complementary designs?
Cheers,
Hugh
amplifierguru said:Kanwar brings passion to Indian Audio -
..." uses Push-Pull Drivers to drive the VFETs with Floating power supplies and THE output peak voltage swing KISSES the Sweet & Sexy Rails with a margin of just 550mV Max Driving 2 ohm loads".
Bagwan Kanwar.
Boys, Boys.... all this talk is sending my biasing crazy. I gotta sit still for a minute and let my heatsinks cool.
In some respects a FET N-channel is more symmetrical given sufficient drive on the positive rail. If matched correctly you can have FETs with the same characteristics on both rails.
Cheers
AKSA said:Half a volt from each rail into 2R, you say???????? WOW!!!
I guess, given sufficient drive you are limited only by the on resistance of the FET, at the on-set of clipping.
Hi,
these rail to rail achievements have surprised me. I suppose quasi with CFP and adequate (bootstrapped) drive result in VERY good utilisation of available rail voltage.
So now we all know that VFETs can achieve good rail utilisation.
What if we use these VFETs as a bootstrapped quasi CFP output stage in a ClassA amp. Can we approach 50% efficiency at maximum output and achieve quality as well?
these rail to rail achievements have surprised me. I suppose quasi with CFP and adequate (bootstrapped) drive result in VERY good utilisation of available rail voltage.
So now we all know that VFETs can achieve good rail utilisation.
What if we use these VFETs as a bootstrapped quasi CFP output stage in a ClassA amp. Can we approach 50% efficiency at maximum output and achieve quality as well?
- Status
- Not open for further replies.