sajti said:
Hopefully a few volts left... But you're right, not enough to really solve the problem... Maybe using mosfet for t7 ? Or a voltage divider instead of r21 ?
Mike
Edit: I like the voltage divider, change r21 to maybe 1k, and add a resistor of 10k (2w ?) from base of t7 to gnd. This reduces the swing of the cascode ~10%.
Voltage divider between the lower + rail, and the output should be good, but it must be low impedance...
Or constant 8-10V zener over the output
sajti
Or constant 8-10V zener over the output
sajti
Remember this one folks. This cascode arrangement equalises the voltage across the drivers T5 & T8 to a few equal volts. I would have thought this would improve the linearity of this stage. This also allows for smaller driver transistors too (faster & quieter).
Anyway I do like the idea of the 1k & 10k as suggested by MikeB and I will try that. The 10k will not get too hot with normal music.
Cheers
Q
Anyway I do like the idea of the 1k & 10k as suggested by MikeB and I will try that. The 10k will not get too hot with normal music.
Cheers
Q
I have two remarks:
1, You need more voltage for the drivers, with this cascode MOSFETs, to get rail-to rail output.
2, You can make floating cascode with the MOSFETs too. I saw it in some Mark Levinson amplifier, where the zeners are connected to the emitters of the T8, T5. This means some more current, and You need regulated current source for the zeners, to avoid bias changes
Sajti
1, You need more voltage for the drivers, with this cascode MOSFETs, to get rail-to rail output.
2, You can make floating cascode with the MOSFETs too. I saw it in some Mark Levinson amplifier, where the zeners are connected to the emitters of the T8, T5. This means some more current, and You need regulated current source for the zeners, to avoid bias changes
Sajti
With the extra 10v on the positive rail (100v) this version does achieve rail to rail on the bench, but I found the amp a bit noisy. This is probably because of the reference zener being fed by the 100v positive rail. I going to try a regulator in this rail reduce the noise.
Cheers
Q
Cheers
Q
Hi,
any comment on my proposed irf640 substitution? post278.
Are there any characteristics of the 640 that do not suit the quasi output stage?
any comment on my proposed irf640 substitution? post278.
Are there any characteristics of the 640 that do not suit the quasi output stage?
Comparison IRF640 & IRFP450
Hi Andrew,
No reason at all why you could not use 8 IRF640's per rail rather than 5 IRFP450's. You would end up with 20% less Ciss and slightly higher power handling.
As far as Rds on goes I doubt it would make much difference. Lowest Rds is reached at saturation and this does not happen with the amp.
The IRF630's would be harder to mount with my PCB layout (I use the TO-247 FETs as board spacers). No reason why the board coudn't be changed though.
I have about 100 IRF840's at my disposal and I was thinking about using these too, but the mounting ease of the TO-247's swung me.
Cheers
Q
Hi Andrew,
No reason at all why you could not use 8 IRF640's per rail rather than 5 IRFP450's. You would end up with 20% less Ciss and slightly higher power handling.
As far as Rds on goes I doubt it would make much difference. Lowest Rds is reached at saturation and this does not happen with the amp.
The IRF630's would be harder to mount with my PCB layout (I use the TO-247 FETs as board spacers). No reason why the board coudn't be changed though.
I have about 100 IRF840's at my disposal and I was thinking about using these too, but the mounting ease of the TO-247's swung me.
Cheers
Q
Ok I'm gonna try this schematic;
Just a few slight changes (as suggested by friendly DIYers) that I will try next. I'm going to use 330R and 3K3 on the base of the cascode transistor as the average total dissipation will be well under 1 watt and that is only at full power test conditions.
I'll update over the next few days.
Cheers
Q
Just a few slight changes (as suggested by friendly DIYers) that I will try next. I'm going to use 330R and 3K3 on the base of the cascode transistor as the average total dissipation will be well under 1 watt and that is only at full power test conditions.
I'll update over the next few days.
Cheers
Q
Hi quasi, i just saw it, you connected R12 to signal ground, that's a big 
This way in combination with r2 you create Dc-offset and inject powersupply artefacts into signal gnd. Use normal gnd instead, maybe split r12 into 2 and place elyt to neg rail in the middle. (improving PSRR)
The same for r21, this way you have a positive feedback.
And R14. Use seperated signal gnd (via 10ohms) only for feedback and input.
Maybe you better skip r2 ?
Mike
This way in combination with r2 you create Dc-offset and inject powersupply artefacts into signal gnd. Use normal gnd instead, maybe split r12 into 2 and place elyt to neg rail in the middle. (improving PSRR)
The same for r21, this way you have a positive feedback.
And R14. Use seperated signal gnd (via 10ohms) only for feedback and input.
Maybe you better skip r2 ?
Mike
Hi MikeB,
I spent a quite a bit off time agonising where to return R21 so I'm going to try both.
This board layout allows options to run R21 to quite or noisy ground and for R2 to be replaced with a solid link. This allows for real differences to be determined.
A safer adjustment for the second CCS has been implemented also.
Cheers
Quasi
I spent a quite a bit off time agonising where to return R21 so I'm going to try both.
This board layout allows options to run R21 to quite or noisy ground and for R2 to be replaced with a solid link. This allows for real differences to be determined.
A safer adjustment for the second CCS has been implemented also.
Cheers
Quasi
Hi quasi,
just one another fine tuning:
Remove R12, and R14, and use one single resistor from the base of T5 to the anode of D4. 51-56k looks good. The two separated resistor can inject unwanted noise, and feedback to the signal ground.
I prefer to keep the ground as clean as possible.
Sajti
just one another fine tuning:
Remove R12, and R14, and use one single resistor from the base of T5 to the anode of D4. 51-56k looks good. The two separated resistor can inject unwanted noise, and feedback to the signal ground.
I prefer to keep the ground as clean as possible.
Sajti
Hi,
2sk1058 is an expensive audio grade lateral that completely defeats the cheap Vfet philosophy of these Quasi amplifiers.
It will also lack the high Gfs of the vertical FETs.
2sk1058 is an expensive audio grade lateral that completely defeats the cheap Vfet philosophy of these Quasi amplifiers.
It will also lack the high Gfs of the vertical FETs.
The gain, ratio Vgs change to Ids change. Also called "Forward Transconductance", or GM for BJTs.
Well I think it's finished;
I tested it today with these rails
+/- 75v DC main rails (at idle)
+ 88v DC auxillary rail
The amp delivered.
- symetrical clipping
- output swing to within 2 volts of available rail
- about 490 watts into 4 ohms
- about 266 watts into 8 ohms
So it seems the module is finished. All that remains is to put two in a box...any takers?
Cheers
Q
I tested it today with these rails
+/- 75v DC main rails (at idle)
+ 88v DC auxillary rail
The amp delivered.
- symetrical clipping
- output swing to within 2 volts of available rail
- about 490 watts into 4 ohms
- about 266 watts into 8 ohms
So it seems the module is finished. All that remains is to put two in a box...any takers?
Cheers
Q
Re: Well I think it's finished;
Hey, 😉
Me! Me! Me!
But unfortunately not before the winter time (of my upper hemisphere 😉), as I'm away from home currently...
Thank You for all Your time and effort on those (at least two) very remarkable projects!
Cheers,
Sebastian.
PS: Do You have the possibility to take THD and S/N measurements of Your prototype?
Hey, 😉
Me! Me! Me!
But unfortunately not before the winter time (of my upper hemisphere 😉), as I'm away from home currently...
Thank You for all Your time and effort on those (at least two) very remarkable projects!
Cheers,
Sebastian.
PS: Do You have the possibility to take THD and S/N measurements of Your prototype?
Perhaps 5 in a BOX....
I've got Quasi on the brain....or so it seems. Sounds like an old Willy Nelson lyric? Your other design is so amazing I can't think the reason for this design? But having built the other, I would love to build this too?
What do you have for the final schematic and how/why would you stack it up against the other FANTASTIC QUASI design? Are there any sonic advantages or disadvantages to use this circuit VS the other Quasi? I ask as I do not know. Great work!
Cheers,
Shawn.
I've got Quasi on the brain....or so it seems. Sounds like an old Willy Nelson lyric? Your other design is so amazing I can't think the reason for this design? But having built the other, I would love to build this too?
What do you have for the final schematic and how/why would you stack it up against the other FANTASTIC QUASI design? Are there any sonic advantages or disadvantages to use this circuit VS the other Quasi? I ask as I do not know. Great work!
Cheers,
Shawn.
THD and noise should really only be measured once the amplifier (amplifiers) is in a case and wired neatly. The other amp was slighty noisy on the bench and very quiet in the case. I expect good results and I will confirm this when I build the completed stereo unit.
5 in a box ??? why am I not surprised Shawn. How does it stack up? Well the other one is going to be hard to beat in terms of sonics. This one is more efficient and for the same rails will deliver more power. When I build a stereo unit I will be able to compare directly. Unless you beat me to it.
Cheers
Q
5 in a box ??? why am I not surprised Shawn. How does it stack up? Well the other one is going to be hard to beat in terms of sonics. This one is more efficient and for the same rails will deliver more power. When I build a stereo unit I will be able to compare directly. Unless you beat me to it.
Cheers
Q
- Status
- Not open for further replies.