Hugh,
...........you mean we have to listen to this stuff?
Just kidding! We should have some listening testresults soon. I have high hopes for the patented current mirror so does homemodder,it seems.. This has been aq very useful investigation into current mirrors and I am finding that what is being used in the audio indusry is usually far from optimal.
Regards,
Jam
...........you mean we have to listen to this stuff?
Just kidding! We should have some listening testresults soon. I have high hopes for the patented current mirror so does homemodder,it seems.. This has been aq very useful investigation into current mirrors and I am finding that what is being used in the audio indusry is usually far from optimal.
Regards,
Jam
There are some high tech solutions that results good into calculations or simulators
Some of them (not all them of course... and i am happy to see this) do not sounds fine into real life amplifiers.
You see this just building and comparing...so...there's no "closed doors" to anyone of us.... everyone of us can check that stuff.
If sophistication, special inputs, CCS, mirrors, sinks, diodes and all those things was really good we were all listening to chip amplifiers..because they use all those things without any kind of economy or restritions...so....
regards,
Carlos
Some of them (not all them of course... and i am happy to see this) do not sounds fine into real life amplifiers.
You see this just building and comparing...so...there's no "closed doors" to anyone of us.... everyone of us can check that stuff.
If sophistication, special inputs, CCS, mirrors, sinks, diodes and all those things was really good we were all listening to chip amplifiers..because they use all those things without any kind of economy or restritions...so....
regards,
Carlos
Jam,
I never failed to be impressed by your use of the web, you travel everywhere..... can you use your flybuys on Second Life?
I tried the diode mod on Alex's advice some time back and liked it. I did extensive listening tests, and it improved decay in particular. I've no reason to believe Glen's 3 transistor circuit (which Bob Cordell says has been around for years, nothing is new under the sun) is any worse, and good reason from Ostripper's work to think it's better, and I can see that the use of a transistor rather than a diode reduces current matching errors. The cascode idea you came up with from the net is very clever, and loads down both sides evenly, BUT there are now [Vdegen + mirror junction + lower cascode junction + Vcb + upper cascode junction + Vcb] = 2 volts minimum above the rail, which may be too much for many VAS drive points, so in practice the cascode idea might need to be dropped, reverting to the 3 transistor topology.
For instrumentation, I'd go for the cascoded Glen mirror, for audio, I'd leave off the cascoding transistor and live with the tiny error on the base bias for the center transistor, probably less than 2uA based on a beta 150 device.
Alex, I've been thinking about your argument that Vce for the LTP transistors should be matched within 5mV. On the theory, this comes back to Early effect, since Vbe match of the LTP (and hence accuracy of error extraction) is changes at around 1/10,000th of the Vce change. But the effect is slight; 0.01%, and I can't see it affecting the sound quality. I reckon there's something else there.... and as Roender suggests, it's probably current matching which is more important here.
Cheers,
I never failed to be impressed by your use of the web, you travel everywhere..... can you use your flybuys on Second Life?
I tried the diode mod on Alex's advice some time back and liked it. I did extensive listening tests, and it improved decay in particular. I've no reason to believe Glen's 3 transistor circuit (which Bob Cordell says has been around for years, nothing is new under the sun) is any worse, and good reason from Ostripper's work to think it's better, and I can see that the use of a transistor rather than a diode reduces current matching errors. The cascode idea you came up with from the net is very clever, and loads down both sides evenly, BUT there are now [Vdegen + mirror junction + lower cascode junction + Vcb + upper cascode junction + Vcb] = 2 volts minimum above the rail, which may be too much for many VAS drive points, so in practice the cascode idea might need to be dropped, reverting to the 3 transistor topology.
For instrumentation, I'd go for the cascoded Glen mirror, for audio, I'd leave off the cascoding transistor and live with the tiny error on the base bias for the center transistor, probably less than 2uA based on a beta 150 device.
Alex, I've been thinking about your argument that Vce for the LTP transistors should be matched within 5mV. On the theory, this comes back to Early effect, since Vbe match of the LTP (and hence accuracy of error extraction) is changes at around 1/10,000th of the Vce change. But the effect is slight; 0.01%, and I can't see it affecting the sound quality. I reckon there's something else there.... and as Roender suggests, it's probably current matching which is more important here.
Cheers,
Carlos and Hugh, I use the combination circuit for some time now, sonically it is superior to just a normal mirror, there is a problem using std mirrors and thats why you see all the different mods by different members, Hugh you perhaps use the diode on the mirror , I can see by sims that it indeed has some effect, the same I can say about SandyK vertical diode. Both bring the ltp sides in closer balance and both of you agree that it is beneficial to the sound, as do I and Self does recommends this in his book too. Im just doing it by different means but to get the same effect and offcourse if there are other benefits to be had its even better. In the case of current mirrors their are many other benefits to be had if the effect of bringing the ltp to better balance can be done with improved mirrors. Two 20 cent trannies and little extra complextion is not to bad.
Hugh is right we have to listen to the results and in the next blameless amp I build Im going to listen to that patented circuit as it also seems to work just as well as the combination circuit I use, a big plus here is that it measures better too like the case of the combination circuit. In some cases what we measure and what we hear corrolates and its the case here.
Another example is with current sources, I ve been told by another member that placing a resitor between current source and ltp has no effect. The thing is using seinheisers I can clearly hear that this resitor does something. By simming it I can tell that it not only affects gain but also THD and most importantly in this case the distortion distribution. Its small db levels in measurement but they are there and I can pick them up subjectively using headphones.
Jam, the current sources mainly because I feel theres some space for improvement here too. A big plus here would be increased PSRR. Will try get Dr Bora involved with this as he has some good ideas and showed some interest in the matter. How do you find all these patents??
Hugh is right we have to listen to the results and in the next blameless amp I build Im going to listen to that patented circuit as it also seems to work just as well as the combination circuit I use, a big plus here is that it measures better too like the case of the combination circuit. In some cases what we measure and what we hear corrolates and its the case here.
Another example is with current sources, I ve been told by another member that placing a resitor between current source and ltp has no effect. The thing is using seinheisers I can clearly hear that this resitor does something. By simming it I can tell that it not only affects gain but also THD and most importantly in this case the distortion distribution. Its small db levels in measurement but they are there and I can pick them up subjectively using headphones.
Jam, the current sources mainly because I feel theres some space for improvement here too. A big plus here would be increased PSRR. Will try get Dr Bora involved with this as he has some good ideas and showed some interest in the matter. How do you find all these patents??
I think you sum it up well, AKSA.
You take into account practical reality, as well as eventual more theoretical matters.
In the end, nobody can force anyone to use this or that mirror, if they do not believe in it.
But nobody can deny, that the more you know about one detail
and the more angles, sides you can look at one object, matter from,
the better you may be to able to make your own decision.
This is the beauty with brain-storming where people that have different viewpoints get together around one issue
There are some, a few, expressing that only looking from one way, using one approach
is the politically correct for Diy People.
Now thank my god, these persons are not so many.
Most people can tolerate differences and even see the value in
that we are and do things differently
---------------------
Personally, I have no problems with adding the diode in the mirror
like you suggested, AKSA,
and Mister Stochino uses in this originally Glen Kleinschmidt Amplifier attachment
which I reposted in my Post #61
http://www.diyaudio.com/forums/attachment.php?s=&postid=1657096&stamp=1226567907
I thought first that 1N4148 working at say 1uA-5uA is operating in a very un-linear region.
But looking at Fairchild datasheets I am far from convinced.
The BAX16 diodes used by Stochino in his LTP pairs shows same charateristics at very low forward currents.
So, I think we should not fear, too much, that adding such a diode would suddenly ruin one LTP input stage.
And I recommend all serious Diy Builders to at least give this a try, a listen.
Because, even if my attachment here shows DC currents diagram
I can not think that the very small AC current variations in such a diode
would matter more in a negative sense
than many, many other details in one Power Amplifier.
Lineup
You take into account practical reality, as well as eventual more theoretical matters.
In the end, nobody can force anyone to use this or that mirror, if they do not believe in it.
But nobody can deny, that the more you know about one detail
and the more angles, sides you can look at one object, matter from,
the better you may be to able to make your own decision.
This is the beauty with brain-storming where people that have different viewpoints get together around one issue There are some, a few, expressing that only looking from one way, using one approach
is the politically correct for Diy People.
Now thank my god, these persons are not so many.
Most people can tolerate differences and even see the value in
that we are and do things differently
---------------------
Personally, I have no problems with adding the diode in the mirror
like you suggested, AKSA,
and Mister Stochino uses in this originally Glen Kleinschmidt Amplifier attachment
which I reposted in my Post #61
http://www.diyaudio.com/forums/attachment.php?s=&postid=1657096&stamp=1226567907
I thought first that 1N4148 working at say 1uA-5uA is operating in a very un-linear region.
But looking at Fairchild datasheets I am far from convinced.
The BAX16 diodes used by Stochino in his LTP pairs shows same charateristics at very low forward currents.
So, I think we should not fear, too much, that adding such a diode would suddenly ruin one LTP input stage.
And I recommend all serious Diy Builders to at least give this a try, a listen.
Because, even if my attachment here shows DC currents diagram
I can not think that the very small AC current variations in such a diode
would matter more in a negative sense
than many, many other details in one Power Amplifier.
Lineup
Attachments
homemodder said:
When members of a group eagerly and unquestionably embrace any flowery subjective evaluation, regardless of how implausible, ridiculous or incompetently executed for the express purpose of gathering corroborative evidence for their equally implausible and ridiculous claims and imaginings, that is groupthink.
Hugh I have been thinking about Roenders question and he has a point, the sim results arent bad for the current imbalance either though. I think the voltage versus current imbalance can only be assesed by listenining but the problem here is that some may prefer one and others the other option.
Groupthinking groups tend to:
1 Fail to adequately determine their objectives and alternatives,
2 Fail to adequately assess the risks associated with a decision,
3 Fail to cycle through discarded alternatives to reexamine their worth after a majority of the group discarded the alternative and not seek advice,
4 Select and use only information that supports their position and conclusions, and
5 Does not make contigency plans in case their decision and resulting actions fail.
1 Fail to adequately determine their objectives and alternatives,
2 Fail to adequately assess the risks associated with a decision,
3 Fail to cycle through discarded alternatives to reexamine their worth after a majority of the group discarded the alternative and not seek advice,
4 Select and use only information that supports their position and conclusions, and
5 Does not make contigency plans in case their decision and resulting actions fail.
Re: Re: Re: CM Discussion
I have taken D. Self investigation into distortion in power amplifier
as a good starting point.
And his conclusion was that the equal current balance in the input pair is very important.
Target value for current sharing should be below 1%.
If looking, for example at the old original AKSA 55, we can easily get more imbalance than this.
The balance is also dependent on power supply voltage.
In this sense, and if Douglas Self is right,
then any type of current mirror that can equalize
so that input pair share close to 50%-50% would be considered an improvement.
Now, in reality, I would say there are many, many well regarded Audio Power Amplifiers
where the simplicity of the input stage does not make any equal current sharing.
-------------
So, my conclusion is:
From a low distortion perspective only, 50-50 is desirable.
But using global negative feedback and have imbalance input pair,
will not make distortion so bad, that The Power Amplifier will not be able to perform very good Audio.
roender said:
As for the second question, should we be more concerned about voltage than current imbalance
I have taken D. Self investigation into distortion in power amplifier
as a good starting point.
And his conclusion was that the equal current balance in the input pair is very important.
Target value for current sharing should be below 1%.
If looking, for example at the old original AKSA 55, we can easily get more imbalance than this.
The balance is also dependent on power supply voltage.
In this sense, and if Douglas Self is right,
then any type of current mirror that can equalize
so that input pair share close to 50%-50% would be considered an improvement.
Now, in reality, I would say there are many, many well regarded Audio Power Amplifiers
where the simplicity of the input stage does not make any equal current sharing.
-------------
So, my conclusion is:
From a low distortion perspective only, 50-50 is desirable.
But using global negative feedback and have imbalance input pair,
will not make distortion so bad, that The Power Amplifier will not be able to perform very good Audio.
May I add to the question of 50-50 share of current in input pair
that many op-amps
uses a DC-offset correction, balancing method
that rely on that you upset the current balance in the input pair!
Have a look at the pins of NE5534.
But this is true for many more modern 'hi-fi' opamps, too.
Actually, you create this imbalance, that Douglas Self does not like.
To get zero DC-offset at output.
that many op-amps
uses a DC-offset correction, balancing method
that rely on that you upset the current balance in the input pair!
Have a look at the pins of NE5534.
But this is true for many more modern 'hi-fi' opamps, too.
Actually, you create this imbalance, that Douglas Self does not like.
To get zero DC-offset at output.
homemodder said:
Thanks, but you're still completely wrong. The only criteria necessary:
"Groupthink is a type of thought exhibited by group members who try to minimize conflict and reach consensus without critically testing, analyzing, and evaluating ideas."
And no “critically testing, analyzing, and evaluating ideas” does not include making a current mirror modification that may yield a 0.00001% linearity improvement to a 0.05% amplifier and then reporting a hugely improved “soundstage” and much better imaging, in agreement with some other guys subjective evaluation.
