AndrewT said:
Well, the problem with taking the mosfet out is that much of the performance goes away (lower loop gain) and no stability is gained. In fact this was the first thing that I tried when I saw myself forced to up the current through the power bjt (in the darlington). I said this earlier, I've already tested the idea.
IMHO, the salas regulator has less of an oscillation problem because it has an acceptable phase margin in reality. The cetoole reg I think has some severe phase lag.
Meanwhile back on the farm.........
Hi Iko,
Is loop gain all that important and is there a danger that excessive amounts and the corresponding feedback could cause overshoot problems........there must be a point lowering the output impedence gains no sonic improvement and might actually be detremental.
Maybe listening tests should be in order.............with all due respect to the shunt gods (notice that gods is spelt in lower case as not to offend the Almighty).
Regards,
Jam
P.S. How about replacing the darlington with a triple?
Hi Iko,
Is loop gain all that important and is there a danger that excessive amounts and the corresponding feedback could cause overshoot problems........there must be a point lowering the output impedence gains no sonic improvement and might actually be detremental.
Maybe listening tests should be in order.............with all due respect to the shunt gods (notice that gods is spelt in lower case as not to offend the Almighty).
Regards,
Jam
P.S. How about replacing the darlington with a triple?
Re: Meanwhile back on the farm.........
My short answer would be this. First, let's keep in mind that this version is an exploration for me, in part it is because it helps me learn new concepts and in part is because I simply like trying out new things. I cannot stress enough, salas v1 is a really good regulator which has been tested by many people and they liked it sonically. That is the version to build, IMHO.
Now, let's get back to the discussion about loop gain, etc. The salas v1 and all the other versions I've been exploring do not have explicit feedback. They also do not use a differential amp. Normally you would look at the output, compare it with a reference voltage, amplify the difference between them, and apply a correction. The all-FET Borbely regulator does that, and you can easily identify the diff amp, feedback path, etc. In both cases, whether there is a diff amp or not, the error output ripple is amplified X times. That's the loop gain. If the load draws a large current all of a sudden, the output voltage will drop. If it drops significantly, then you need to be able to apply a correction large enough to bring the voltage to the steady state dictated by the reference. If the loop gain is not large enough, the regulator will not be able to do that. In a phono stage you will most likely not see that. Some (Andrew?) point out that in a Class A amp you will not see that. But there will be cases where the load will draw current at different rates. Then the input voltage to the amp will vary (by how much, depends) and if large enough, it may be sonically perceptible.
I cannot think of a case where a low output impedance would be detrimental.
But I don't want to convince any of you that in anyway it is worthwhile to get lower output impedance and larger bandwidth. Personally I don't have another outlet for the things I'm investigating than this thread.
jam said:
Hi Iko,
Is loop gain all that important and is there a danger that excessive amounts and the corresponding feedback could cause overshoot problems........there must be a point lowering the output impedence gains no sonic improvement and might actually be detremental.
Maybe listening tests should be in order.............with all due respect to the shunt gods (notice that gods is spelt in lower case as not to offend the Almighty).
Regards,
Jam
P.S. How about replacing the darlington with a triple?
My short answer would be this. First, let's keep in mind that this version is an exploration for me, in part it is because it helps me learn new concepts and in part is because I simply like trying out new things. I cannot stress enough, salas v1 is a really good regulator which has been tested by many people and they liked it sonically. That is the version to build, IMHO.
Now, let's get back to the discussion about loop gain, etc. The salas v1 and all the other versions I've been exploring do not have explicit feedback. They also do not use a differential amp. Normally you would look at the output, compare it with a reference voltage, amplify the difference between them, and apply a correction. The all-FET Borbely regulator does that, and you can easily identify the diff amp, feedback path, etc. In both cases, whether there is a diff amp or not, the error output ripple is amplified X times. That's the loop gain. If the load draws a large current all of a sudden, the output voltage will drop. If it drops significantly, then you need to be able to apply a correction large enough to bring the voltage to the steady state dictated by the reference. If the loop gain is not large enough, the regulator will not be able to do that. In a phono stage you will most likely not see that. Some (Andrew?) point out that in a Class A amp you will not see that. But there will be cases where the load will draw current at different rates. Then the input voltage to the amp will vary (by how much, depends) and if large enough, it may be sonically perceptible.
I cannot think of a case where a low output impedance would be detrimental.
But I don't want to convince any of you that in anyway it is worthwhile to get lower output impedance and larger bandwidth. Personally I don't have another outlet for the things I'm investigating than this thread.
Iko,
I ment to say local feedback. I do not mean to take away anything from the wonderful work you are doing but rather determine some sort of cutoff point where there might be no sonic gains with increased loop gain and bandwidth.
The problem with CFP's is a real one, both Charles Hansen and Mr.John Curl will attest to that, great care needs to be used when designing with them.
I have to say that a lot of us in the forum , regard the work both you and Salas put into this invaluable and are looking forward for you next investagitive effort.............maybe gain stages using the shunt.
Regards,
Jam
I ment to say local feedback. I do not mean to take away anything from the wonderful work you are doing but rather determine some sort of cutoff point where there might be no sonic gains with increased loop gain and bandwidth.
The problem with CFP's is a real one, both Charles Hansen and Mr.John Curl will attest to that, great care needs to be used when designing with them.
I have to say that a lot of us in the forum , regard the work both you and Salas put into this invaluable and are looking forward for you next investagitive effort.............maybe gain stages using the shunt.
Regards,
Jam
I thought John & Charles preferred not to use CFP's because they prefer the sound of amps that do not use loop feedback.
I do not doubt that they both could easily implement CFP's if they had the slightest inclination to do so - it's not that difficult and these guys are not lacking in the ability department.
If we follow Charles's philosophy we would not be designing regulators that use any loop feedback whatsoever so it seems a bit odd to single out one FB element as bad in a design that uses FB anyway.
As you may have guessed I have used them and liked the results.
I found that mounting the control transistor on the actual legs of the power transistor as close to the body as possible was quite a good way to avoid any oscillation.
I would not want people to be put off before they have even tried CFP's because of scare stories . . . . . because CFP's do offer some clear advantages and some people say that they actually prefer their "sound".
I do not doubt that they both could easily implement CFP's if they had the slightest inclination to do so - it's not that difficult and these guys are not lacking in the ability department.
If we follow Charles's philosophy we would not be designing regulators that use any loop feedback whatsoever so it seems a bit odd to single out one FB element as bad in a design that uses FB anyway.
As you may have guessed I have used them and liked the results.
I found that mounting the control transistor on the actual legs of the power transistor as close to the body as possible was quite a good way to avoid any oscillation.
I would not want people to be put off before they have even tried CFP's because of scare stories . . . . . because CFP's do offer some clear advantages and some people say that they actually prefer their "sound".
Hi Salas,
Newbie question here..
Tried V1 p2p using all component values in the schematic except a 24V zener (1N4749A). Power from a 24V trasformer with full bridge rectification, I measured 28V at the output when connected to a B1.
Is there any changes that I need to make (component, component values) to get 24V output?
Regards
Tham
Newbie question here..
Tried V1 p2p using all component values in the schematic except a 24V zener (1N4749A). Power from a 24V trasformer with full bridge rectification, I measured 28V at the output when connected to a B1.
Is there any changes that I need to make (component, component values) to get 24V output?
Regards
Tham
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.