A question for Swordfishy and AKSA.
How easy is it to get a stable bias with the bootstrapped VAS?
As d.self points out, the current through the VAS stage changes a great deal more with the bootstrapped VAS than with an VAS with an Current source.
I should be honest with you all. I have never build an amp with a Bootstrapped VAS my self before.
How easy is it to get a stable bias with the bootstrapped VAS?
As d.self points out, the current through the VAS stage changes a great deal more with the bootstrapped VAS than with an VAS with an Current source.
I should be honest with you all. I have never build an amp with a Bootstrapped VAS my self before.
A question for Swordfishy and AKSA.
How easy is it to get a stable bias with the bootstrapped VAS?
As d.self points out, the current through the VAS stage changes a great deal more with the bootstrapped VAS than with an VAS with an Current source.
I should be honest with you all. I have never build an amp with a Bootstrapped VAS my self before.
Yes it does bounce around a bit, but not enough to concern me.
I can log the variation for you if you like.
That would be nice. the VAS shouldn't care being an integrator
if you mean the output bias V spreader could vary with bootstrap current then you can just make the Vbe multiplier a CFP for better regulation
if the front end has a higher V regulated/filtered supply run the bootstrap from the more stable supply V
if you mean the output bias V spreader could vary with bootstrap current then you can just make the Vbe multiplier a CFP for better regulation
if the front end has a higher V regulated/filtered supply run the bootstrap from the more stable supply V
Sonny,
The bootstrap does move the current around a bit, though this is minimised with a RC filter to its supply. The range is roughly +/-15% at full power, depending on the RC values chosen.
Bias is absolutely no problem, no difference I've ever noticed.
In practice this tends only to increase H2, H3 and H4, NOT the higher order artefacts.
You can examine it in LTSpice; the distortion profiles are different, and indeed they sound different, with the bootstrap having more 'warmth', whatever that is....
A second consideration is clip. A bootstrap is far more benign, and will in fact punch well above the rail. A CCS can, in certain situations and with certain designs, latch up, and the Early effect of the CCS will change its current very slightly too.
Lastly, the advantage of a bootstrap is in compensation. As frequency rises, the electro becomes inductive and the load seen by the VAS starts to increase, reducing OLG and with it, loop gain. This aids stability.
All in all, much to recommend.... and since only two resistors and a cap, less to go wrong.
I have almost finished a pcb for this circuit. It is 230mm x 69mm, designed to bolt straight onto a 300mm flat backed heatsink. I'm gunning for 1A quiescent with 36V rails. This delivers the first four watts of power in Class A, Class B thereafter. I'm using the lateral output mosfets.
Cheers,
Hugh
The bootstrap does move the current around a bit, though this is minimised with a RC filter to its supply. The range is roughly +/-15% at full power, depending on the RC values chosen.
Bias is absolutely no problem, no difference I've ever noticed.
In practice this tends only to increase H2, H3 and H4, NOT the higher order artefacts.
You can examine it in LTSpice; the distortion profiles are different, and indeed they sound different, with the bootstrap having more 'warmth', whatever that is....
A second consideration is clip. A bootstrap is far more benign, and will in fact punch well above the rail. A CCS can, in certain situations and with certain designs, latch up, and the Early effect of the CCS will change its current very slightly too.
Lastly, the advantage of a bootstrap is in compensation. As frequency rises, the electro becomes inductive and the load seen by the VAS starts to increase, reducing OLG and with it, loop gain. This aids stability.
All in all, much to recommend.... and since only two resistors and a cap, less to go wrong.
I have almost finished a pcb for this circuit. It is 230mm x 69mm, designed to bolt straight onto a 300mm flat backed heatsink. I'm gunning for 1A quiescent with 36V rails. This delivers the first four watts of power in Class A, Class B thereafter. I'm using the lateral output mosfets.
Cheers,
Hugh
Last edited:
A question for Swordfishy and AKSA.
How easy is it to get a stable bias with the bootstrapped VAS?
As d.self points out, the current through the VAS stage changes a great deal more with the bootstrapped VAS than with an VAS with an Current source.
I should be honest with you all. I have never build an amp with a Bootstrapped VAS my self before.
You can take a look at this thread http://www.diyaudio.com/forums/newreply.php?do=newreply&p=2537223 and you can have both, bootstrap and CCS VAS.
dado
Direct Coupled Version
All,
I just tried the direct coupled version shown below.
Initial impressions are that it has the potential to be the best sounding of all the ones I have tried so far.
However the DC offset stability appears to be pretty terrible (as expected).
I need to do a more in depth investigation tomorrow, so stay tuned.
If you want to throw something together just as a proof of concept to see how nice this amplifier has the potential to sound, this is the circuit to use.
Lineup, I think a FET ccs for the input might be a good compromise between complexity and stability.
All,
I just tried the direct coupled version shown below.
Initial impressions are that it has the potential to be the best sounding of all the ones I have tried so far.
However the DC offset stability appears to be pretty terrible (as expected).
I need to do a more in depth investigation tomorrow, so stay tuned.
If you want to throw something together just as a proof of concept to see how nice this amplifier has the potential to sound, this is the circuit to use.
Lineup, I think a FET ccs for the input might be a good compromise between complexity and stability.
Attachments
Last edited:
Great Swordfishy,
So happy you at least tried it - less gain will help, I think X16 will work for me
I'm interested to know the spread of DC offset you measured.
Personally I will persevere with whatever it takes to make this version work well.
Most of my parts are here now - just await jfets
So happy you at least tried it - less gain will help, I think X16 will work for me
I'm interested to know the spread of DC offset you measured.
Personally I will persevere with whatever it takes to make this version work well.
Most of my parts are here now - just await jfets
Last edited:
Great, can't wait to hear your impressions. I really hope you like it. Do you think you'll go for the ccs VAS or bootstrap version? Sonically there's good arguments for both.
Tomorrow I will do a proper dc offset test similar to the last ones I posted. To give you an idea though, tonight I set the offset after initial turn on to 0mV over the course of maybe 5-10 minutes. Then I played music for ten minutes or so and the offset was over half a volt.
However, with this circuit it will require a lot of time to level out, so I will do a full hour test tomorrow and log it every 0.5s like last time.
I am running 20mA in the VAS and 0.5A bias so everything runs pretty warm and it takes some time to get up to temperature.
I really do believe the 9610 sounds smoother at 20mA.
Oh, by the way, this one was also stable without compensation. Nice to know I didn't fluke it last time. Also, there's still no gate stopper on the VAS. As long as leads are kept really short it seems to work ok.
Tomorrow I will do a proper dc offset test similar to the last ones I posted. To give you an idea though, tonight I set the offset after initial turn on to 0mV over the course of maybe 5-10 minutes. Then I played music for ten minutes or so and the offset was over half a volt.
However, with this circuit it will require a lot of time to level out, so I will do a full hour test tomorrow and log it every 0.5s like last time.
I am running 20mA in the VAS and 0.5A bias so everything runs pretty warm and it takes some time to get up to temperature.
I really do believe the 9610 sounds smoother at 20mA.
Oh, by the way, this one was also stable without compensation. Nice to know I didn't fluke it last time. Also, there's still no gate stopper on the VAS. As long as leads are kept really short it seems to work ok.
I will try both BS & CCS VAS
I will also use an o/p fuse as a precaution but this is really just to protect in extraordinary circumstances.
Hugh has already said he thinks 200mV is a tolerable DC offset
but if you look at the maths for 0.5 volts it's still not that bad.
Assuming 4 ohms DC then:
0.5V / 4ohms = 0.125 amps
0.125 amps x 0.5V = 0.0625 Watts.
And if the 0.5 volts is spread evenly around zero we only see 0.25V.
and that is before we even tried any measures to reduce it !
anyway, I await your test results with interest.
mike
I will also use an o/p fuse as a precaution but this is really just to protect in extraordinary circumstances.
Hugh has already said he thinks 200mV is a tolerable DC offset
but if you look at the maths for 0.5 volts it's still not that bad.
Assuming 4 ohms DC then:
0.5V / 4ohms = 0.125 amps
0.125 amps x 0.5V = 0.0625 Watts.
And if the 0.5 volts is spread evenly around zero we only see 0.25V.
and that is before we even tried any measures to reduce it !
anyway, I await your test results with interest.
mike
Many types.
rock, fusion, jazz, classical, world, soul, male & female vocal, folk, acoustic & electronic
what about you ?
Fet VAS
Dear all,
so here We are, then when after quite a while, some kind of BJT was the VAS answer...
Now it's the IRF9610?
It's like nobody has cared much to listen to others humble experience?
Just wish some of You would have the courage to try the the Toshiba 2SJ313 for VAS.
Please forget the the 2SJ79 from Hitachi for this application - it sounds amazingly 'Glassy or quantiziced', if that possibly 'ring's a bell' with anyone...
But In more or less in the the admittedly 'round about' words of Hugh from AKSA: Hearing it is believing it, when You have been through it Yourself?
Yes the ZVP4424 could do an amazing job here, but if You want 20mA to go through it even at +/-25V DC, then some kind of added, possibly rather elaborate, cooling arrangement is called for?
All the best intentions from here
DocO
Dear all,
so here We are, then when after quite a while, some kind of BJT was the VAS answer...
Now it's the IRF9610?
It's like nobody has cared much to listen to others humble experience?
Just wish some of You would have the courage to try the the Toshiba 2SJ313 for VAS.
Please forget the the 2SJ79 from Hitachi for this application - it sounds amazingly 'Glassy or quantiziced', if that possibly 'ring's a bell' with anyone...
But In more or less in the the admittedly 'round about' words of Hugh from AKSA: Hearing it is believing it, when You have been through it Yourself?
Yes the ZVP4424 could do an amazing job here, but if You want 20mA to go through it even at +/-25V DC, then some kind of added, possibly rather elaborate, cooling arrangement is called for?
All the best intentions from here
DocO
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.