Hi Guys
I think it is incorrect to limit or skew the acceptability of RC compensation of the LTP outputs to just "approved by golden ears". Measurement confirms the positive benefits and respected (not fringe) designers use this method. Cordell clearly demonstrated this approach to be effective as a part of one that separates slew rate from compensation values.
Have fun
Kevin O'Connor
I think it is incorrect to limit or skew the acceptability of RC compensation of the LTP outputs to just "approved by golden ears". Measurement confirms the positive benefits and respected (not fringe) designers use this method. Cordell clearly demonstrated this approach to be effective as a part of one that separates slew rate from compensation values.
Have fun
Kevin O'Connor
/OT
Hi Harry,
The primary signal source is the DAC of the sound card. For low level THD measurements I also use bandpass filters. In addition, an external signal source can be used.
Hi Dave,
You're right, we better discuss these matters in a new thread, though I feel reluctant to start it right now, as I'm still in the development phase.
@ Bob: Sorry for hijacking your thread.
Hi JCX,
Very true! Though I don't mind, as I'm retired and have all the time of the world.
BTW, the same applies to building your own amp. 😉
Cheers,
E.
Hi Harry,
The primary signal source is the DAC of the sound card. For low level THD measurements I also use bandpass filters. In addition, an external signal source can be used.
Hi Dave,
You're right, we better discuss these matters in a new thread, though I feel reluctant to start it right now, as I'm still in the development phase.
@ Bob: Sorry for hijacking your thread.
Hi JCX,
Very true! Though I don't mind, as I'm retired and have all the time of the world.
BTW, the same applies to building your own amp. 😉
Cheers,
E.
How technical is this book?
Or let's say, if it were a text for a college class what would the pre-reqs for the class be? 🙂
Or let's say, if it were a text for a college class what would the pre-reqs for the class be? 🙂
In my opinion it does a good job of spanning quite a range of understanding and abilities. High-school algebra is probably a requirement to really understand the discussion. Aside from that, a dedicated high school student (especially with an introductory class in basic electronics) would probably find most of it useful and understandable, though some sections would be set aside for future study; and a college graduate electrical engineer would likewise find most of it useful and understandable, though some sections would be quickly skimmed.
Dale
Hi Guys
Bob's book only requires high-school math and schematic reading skills unless you want to venture into the chapters about simulating circuits. Same goes for Self's books and my books without the sim stuff.
Have fun
Kevin O'Connor
Bob's book only requires high-school math and schematic reading skills unless you want to venture into the chapters about simulating circuits. Same goes for Self's books and my books without the sim stuff.
Have fun
Kevin O'Connor
Yes , although with a brute force method that load the input stage abruptly...
Hi Guys
Standard miller compensation seems much more brute-force than the RC across the diff output. The latter comes into play in the 100kHz+ range - often much higher even than this. Miller seems much more like "strangling" the circuitry.
Have fun
Kevin O'Connor
Standard miller compensation seems much more brute-force than the RC across the diff output. The latter comes into play in the 100kHz+ range - often much higher even than this. Miller seems much more like "strangling" the circuitry.
Have fun
Kevin O'Connor
Actually, RC shuntcomp applied correctly has great finesse... But you have to know how to apply it and when. Edmund's TIS is a good example of RC shuntcomp. My old headphone amp is another example, though it is pretty old and not necessarily a good one. Oh yeah and Symasym.
The miller cap applies local feedback around the TIS (VAS) and actually linearizes it. It's actually a very elegant solution.
If you wan more sophistication, then you need to move to MIC, TMC or the like. This will help with speed (MIC) and TMC gets you another 15 dB of loop gain at HF without too much trouble.
>It's actually a very elegant solution.
Not to mention the 'pole splitting' effect, which contributes to stability.
Cheers,
E.
Not to mention the 'pole splitting' effect, which contributes to stability.
Cheers,
E.
What is MIC?
Okay, the VAS's purpose in life is to amplify the error signal subtracted by the differential. It's job is to attenuate the distorted drive current of the output stage. So when you decrease VAS gain, you increase the error signal. Even if it were twice as linear it now has half the gain, so the result is a doubling of distortion assuming the VAS distorts much less than the output stage, which is usually true.
If miller compensation were a good thing, we would be finding ways to make it huge rather than trying to make it as small as possible!
There is no local feedback effect of the VAS to speak of at frequencies that matter. This is because whatever current flows through the Miller cap will immediately be drawn away from the VAS input by the input stage because feedback has to keep a very controlled VAS current.
If there was any local feedback effect of any use then I would expect to find some case where miller compensation at the VAS decreases distortion.
You can't look at that in isolation. The total linearity depends on the ol (non) linearity and the feedback factor.
Decreasing the Vas gain will decrease the ol gain and thus, all other things remaining the same, increases distortion. But if the decrease in Vas gain goes hand in hand with an increase in Vas linearity that linearises the ol gain and may well have as end-effect that the cl is more linear.
jan
Decreasing the Vas gain will decrease the ol gain and thus, all other things remaining the same, increases distortion. But if the decrease in Vas gain goes hand in hand with an increase in Vas linearity that linearises the ol gain and may well have as end-effect that the cl is more linear.
jan
Hi Guys
According to Bob's book, these acronyms have these meanings:
no acronym - input compensation, RC across diff-inputs, not recommended
no acronym - miller compensation, cap around highest-gain stage
MIC - miller input compensation, feedback to the input of the input stage from VAS output, sometimes requires further compensation with the enclosed loop, simpler forms exist as used by Bryston
TMC - transitional miller compensation, like 2-pole but with R tied to output
TPC - two-pole compensation, miller cap split in two with R from junction to AC ground
BTP - bridged 2-pole compensation (adds a third cap)
How is TIS equal to VAS? What does TIS stand for?
Have fun
Kevin O'Connor
According to Bob's book, these acronyms have these meanings:
no acronym - input compensation, RC across diff-inputs, not recommended
no acronym - miller compensation, cap around highest-gain stage
MIC - miller input compensation, feedback to the input of the input stage from VAS output, sometimes requires further compensation with the enclosed loop, simpler forms exist as used by Bryston
TMC - transitional miller compensation, like 2-pole but with R tied to output
TPC - two-pole compensation, miller cap split in two with R from junction to AC ground
BTP - bridged 2-pole compensation (adds a third cap)
How is TIS equal to VAS? What does TIS stand for?
Have fun
Kevin O'Connor
But this is almost never the case. Furthermore the Miller cap has other consequences like loading the VAS, and providing a shortcut across the VAS for output stage distortion.
Lets look at this in hindsight. All the improved compensation schemes eliminate the effect of VAS miller compensation at audio. If local VAS feedback were a good thing, for a a power amp at least, we would not be doing this. We already know how false this idea is, it is strange how the idea of a "local miller feedback linearized VAS" sticks so much. I think the idea is popular because it makes design easier, falsely justifying design flaws that take a lot of work and thought to really solve.
Perhaps in isolation the VAS would be more linear with local feedback. Even so, local miller compensation is the worst way you could apply this feedback. If you think about it, TMC is an example of VAS feedback applied correctly; it legitimately linearizes the VAS and output stage as well. However it still loads the input stage so you benefit from it you will need a very linear input stage.
TMC+MIC
Hi Keane,
That's the reason why I designed the 'super TIS', which combines TMC and Miller input compensation (MIC). 😀
Now you don't need a super linear input stage any longer, as the compensation load has been moved from the IPS output to the IPS input.
Cheers,
E.
Hi Keane,
That's the reason why I designed the 'super TIS', which combines TMC and Miller input compensation (MIC). 😀
Now you don't need a super linear input stage any longer, as the compensation load has been moved from the IPS output to the IPS input.
Cheers,
E.
TIS vs VAS
We mean the same thing. "The purists among us (I'm one of them) regard 'VAS' as a misnomer and prefer to call it a Trans Impedance Stage (TIS).
Cheers,
E.
@Keane: Thx!
We mean the same thing. "The purists among us (I'm one of them) regard 'VAS' as a misnomer and prefer to call it a Trans Impedance Stage (TIS).
Cheers,
E.
@Keane: Thx!
Hi Guys
Considering the popularity of Doug Self's articles and books, VAS seems like a more accepted term today. Cordell adopted it, so how bad can it be? Maybe it falls into the bag with all the other compromises adopted by the use of economic British designers?
Besides, most of the "T" acronyms have to do with compensation, so could be confusing in that regard.
As I've said elsewhere, once I saw Cordell's article in 1981 or so I never used miller compensation again. The other methods are more elegant, to me, and seem less brutal to the circuit while imbuing the good stability and low distortion we seek. They also separate slew rate from the compensation means.
Have fun
Kevin O'Connor
Considering the popularity of Doug Self's articles and books, VAS seems like a more accepted term today. Cordell adopted it, so how bad can it be? Maybe it falls into the bag with all the other compromises adopted by the use of economic British designers?
Besides, most of the "T" acronyms have to do with compensation, so could be confusing in that regard.
As I've said elsewhere, once I saw Cordell's article in 1981 or so I never used miller compensation again. The other methods are more elegant, to me, and seem less brutal to the circuit while imbuing the good stability and low distortion we seek. They also separate slew rate from the compensation means.
Have fun
Kevin O'Connor
