Hi OS,I added the EF/TMC to the TIS core. I am literally "blown away".
I have simulated and passed on most of DIY audio's designs.
THIS ONE is truly exceptional !
PS - I'll show "halcro" performance later.
Edmond , you are the "man". ("thinking outside the box")
OS
Isn't that too much praise? I'm just, like many others, "ordinary skilled in the art" of electronics. Perhaps more importantly, I'm not indoctrinated by that "blameless" concept.
Cheers, E.
@Keantoken,
I've now mocked up the output of my VAS which shows how the Wilson mirrors and Hawksford cascode are tied together. I think these two wired up together like this form a very strong combo.
In this picture, I tried to replace Q4 and Q7 each by a Baxandall super-pair. And in this configuration, these didn't offer any improvement THD wise.
I've also tried to include current flows, though I've omitted factoring in base currents. I hope you can make sense of it. And perhaps you could tell whether a BSP should be advantageous here or that it indeed doesn't matter? All currents are summed back together when they go back into the mirrors. The cascode reference is actually a split-rail, taking a fixed current from the mirror outputs.
If I did something wrong, or drew wrong conclusions, just tell.. I can handle that
Attachments
Hi OS,
Isn't that too much praise? I'm just, like many others, "ordinary skilled in the art" of electronics. Perhaps more importantly, I'm not indoctrinated by that "blameless" concept.
Cheers, E.
Well said.
S
Using a super-VAS only make sense if the VAS was already the main source of distortion. If not, then changing the VAS will hardly change anything. It could be that the VAS load is the main source of distortion, not the VAS itself, so it's contribution is swamped. When I have a question like this I have to go through several steps of verifying where exactly the distortion comes from, so treat my response as my best guess, I guess.
As far as "Blameless", really it's best to be using the same dictionary as everyone else if you want to make sense to anyone but yourself. If you don't, or you invent your own definitions then you just make it impossible for people to know what you really mean. That doesn't mean you can't keep in mind the "real" definition, it's just an act of courtesy so that the interaction works out better for everyone.
People who read the books written about your work will use the terminology contained in them, so writing a really good book is the best way to pin down an effective terminology. But it's in the future, and communication takes place now. Polite and courteous people can always work out the communication bugs, it's really not a big deal.
As far as "Blameless", really it's best to be using the same dictionary as everyone else if you want to make sense to anyone but yourself. If you don't, or you invent your own definitions then you just make it impossible for people to know what you really mean. That doesn't mean you can't keep in mind the "real" definition, it's just an act of courtesy so that the interaction works out better for everyone.
People who read the books written about your work will use the terminology contained in them, so writing a really good book is the best way to pin down an effective terminology. But it's in the future, and communication takes place now. Polite and courteous people can always work out the communication bugs, it's really not a big deal.
I tested the VAS with an ideal OPS (infinite input Z) to exclude any loading contributions. Would you agree though that the ccb of Q4 and Q7 are recirculated as it would be with a non-mirrored VAS using a superpair?
I'm sorry, most of this is over my head; I simply repeated what people (of whom Bob C was one, in Doug's thread) explained to me what "Blameless" was about.
When you run the VAS with a perfect output stage you eliminate a source of VAS distortion. The OPS input voltage is nonlinear, and this nonlinear signal leaks through the VAS. This is the source of distortion that the Baxandall cascode rejects better than the Hawksford. So it makes sense to me that the Hawksford would perform better because of the way you tested it, which isn't realistic.
I have to include a little detail though; there was a bit of loading, which were 2 very small value caps (~10pf) from VAS output to both the supply rails to stabilize it for testing. These probably swamped any possible differences. I have to figure out a better way to test
P.S. in the complete schematic the VAS is loaded by Bob's HEC. The VAS input current runs at 21mA quiesent, which results in about 18mA for Ivasc in the picture, leaving 3mA for Ihc, the cascode ref rail.
P.P.S. I've dubbed the setup I've shown the "Wilson Enhanced Hawksford Cascode", or, WEHC. Oh the acronyms *grin*
Last edited:
I'm sorry, I have trouble understanding how this mechanism would work, could you perhaps illustrate how does 'this nonlinear signal' leaks through the TIS exactly?
Provided the LTP is properly linearized and the TIS isn't loaded with a too low impedance one would expect the TIS to generate the bulk of the distortion.
This IPS is a good example how low THD can be with ideal OPS. It uses "super-VAS" and have Loop Gain of 150dB at 10Hz and Open Lopp Gai close to 180dB. Is quite hard to tame the super-VAS, without cascade base stoppers and caps across the bias LEDs it oscillates. It is very sensitive to a output load, and to find good OPS for it almost impossible.
It shows almost no distortion at max output level. I simulated it with the standard EF triple and distortion goes up dramatically, and this is not OPS induced distortion only but the VAS distortion provoked with not ideal VAS load.
THD20k at 56Vpp at the output of the IPS is 0.000004% and goes up with the OPS to the 0.000341% and at max output to the 0.001983%, still good result but far from the super VAS potential.
Damir
Attachments
This is about super-VAS suggested by keantoken. If you want to see more about CCS follow this link http://www.diyaudio.com/forums/solid-state/260308-gainwire-ngnfb-classb-poweramp-6.html#post4028879. I switched to that kind of CCS after Sergio showed it.
My post refers directly to your post1193
I am asking whether replacing a diode (D3) with a red LED would improve performance.
It is a cascoded CCS. It is almost identical to Roender's cascoded CCS.
Last edited:
You don't want you accept my explanation and you don't want to read that tread I provided the link. It is Current mirror on the top and you can't change a diode for a LED, it is not current mirror any more. It is not cascaded CCS
Replace the diode with a BJT connected as diode and that could be more clear then.
Last edited:
D1 R4 Q18 is a CCS.
D3 Q17 adds a cascode to the CCS.
It is a cascoded CCS.
D3 defines the Vce across Q18.
Using a red LED, or two diodes, increases the Vce across Q18
This may make no difference to the performance of the CCS, or it may improve the performance of the CCS.
I don't know which.
I am suggesting you look into the change and after informing yourself you decide to leave as is, or change it to a red LED.
D3 Q17 adds a cascode to the CCS.
It is a cascoded CCS.
D3 defines the Vce across Q18.
Using a red LED, or two diodes, increases the Vce across Q18
This may make no difference to the performance of the CCS, or it may improve the performance of the CCS.
I don't know which.
I am suggesting you look into the change and after informing yourself you decide to leave as is, or change it to a red LED.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.