Calvin,
I have assumed that the schematic of post #12 was correct compared to post #6, since it does not have the darlington's on the output. However, the post #12 schematic does not have the C11 capacitor. Is this actually on the kit board, or is it your an add on?
Also, I am curious about the R35 & L1 network. Do you feel these are needed and does the simulation tell you that they are?
I have assumed that the schematic of post #12 was correct compared to post #6, since it does not have the darlington's on the output. However, the post #12 schematic does not have the C11 capacitor. Is this actually on the kit board, or is it your an add on?
Also, I am curious about the R35 & L1 network. Do you feel these are needed and does the simulation tell you that they are?
Hi,
I´ve quickly written up some notes on the LJ12-2.
I´ve used a few modules over the Years from the same source and there have been minor differences between different badges.
@Ian
The difference between two-pole and 3-pole Miller is the additional cap in series to the resistor. If You ´stretch´ the 2-pole compensation towards higher OL-bandwidth the phase angle will drop to levels close to -180°.
See the dip in phase response around 50kHz in file "... OL-gain 3-pole-Miller-comp". Without C18 the dip would approach -180°.
Anyway its just sims so far. I don´t know so far if the higher OL-gain results in any practical advantage, either sonically, or in measurement and stability.
Btw. I reached an age where it becomes increasingly difficult to remember the meaning of abbreviations.
TMC means Twist My Cap right?
IIC I think has a similarity to I2C or I2S
OIC is that Olympic International Committee?
jauu
Calvin
I´ve quickly written up some notes on the LJ12-2.
I´ve used a few modules over the Years from the same source and there have been minor differences between different badges.
@Ian
The difference between two-pole and 3-pole Miller is the additional cap in series to the resistor. If You ´stretch´ the 2-pole compensation towards higher OL-bandwidth the phase angle will drop to levels close to -180°.
See the dip in phase response around 50kHz in file "... OL-gain 3-pole-Miller-comp". Without C18 the dip would approach -180°.
Anyway its just sims so far. I don´t know so far if the higher OL-gain results in any practical advantage, either sonically, or in measurement and stability.
Btw. I reached an age where it becomes increasingly difficult to remember the meaning of abbreviations.
TMC means Twist My Cap right?
IIC I think has a similarity to I2C or I2S
OIC is that Olympic International Committee?
jauu
Calvin
Similar.
In my opinion, any resistance in series a capacitance. Can make the fixed frequency signal through.
The principle is like this.
Less filter amplifier, or compensator. Change can reduce signal loss.
The result is reduced distortion.
Hi,
I´ve quickly written up some notes on the LJ12-2.
I´ve used a few modules over the Years from the same source and there have been minor differences between different badges.
@Ian
The difference between two-pole and 3-pole Miller is the additional cap in series to the resistor. If You ´stretch´ the 2-pole compensation towards higher OL-bandwidth the phase angle will drop to levels close to -180°.
See the dip in phase response around 50kHz in file "... OL-gain 3-pole-Miller-comp". Without C18 the dip would approach -180°.
Anyway its just sims so far. I don´t know so far if the higher OL-gain results in any practical advantage, either sonically, or in measurement and stability.
Btw. I reached an age where it becomes increasingly difficult to remember the meaning of abbreviations.
TMC means Twist My Cap right?
IIC I think has a similarity to I2C or I2S
OIC is that Olympic International Committee?
jauu
Calvin
B: yes. This is basically correct circuit.
In the design of the back I have to cancel the C14. But I don't use the L1, R35.
From my point of view, analog amplifier in access to the inductive load. Such as speakers.
Don't need to use L1. L1 can produce resistance. May be greater than 0.05 ohms.
The effect of it only in the use of capacitive load.
I think, even can cancel R13, the R20.
Try every means to reduce the output resistance. The purpose is to make the best amplifier. Not the most conservative amplifier.
The difference between it. Sound box line equivalent to 100 USD, with 1000 USD of sound box line.
Last edited:
Heh...heh.Yes, I get I get it, thanks Calvin. I'm just amazed to see it used in audio. My own age worries me too, as I try to follow the buzz around the engineers' threads of Bob Cordell, Douglas Self and their cohorts in recent years which has mainly been about advanced compensation forms, as introduced and presented by Edmond Stuart quite some time ago. In terms of incredibly long threads, Ihan's about Bob Cordell's book gravitate to advanced compensation and in epic proportions. Just how long you run one thread is what seems to be the goal now, because accessing the useful content will take forever, if the info. is really in that thread after all.
As you ask, TMC is Transitional Miller Compensation, covered in Bob Cordell's publications, thread contributions, Cordell Audio website references, Book and Edmond Stuart's numerous contributions, since it has been Stuart's that established it as a credible and effective scheme with few vices. several of our best designers now use TMC routinely.
It is beyond my full comprehension and math skills, but according to several members, the net result of TMC is surprisingly similar to TPC (yep, that's Two Pole Compensation)!!
OIC is a term agreed on by Douglas Self to describe his own crystallization of Peter Baxandall's earlier proposals for Output Inclusive Compensation. This turned up in D. Self's first contribution in Volume 1 of the bookzine and Baxandall's papers are published at The Douglas Self Site and in a special tribute volume by Jan Didden. See www.linearaudio.net - home Like it suggests, the feedback encloses the output node and VAS in a dual loop.
IIC Is another term coined by D. Self, I understand, and means Input Inclusive Compensation. It essentially is also a dual feedback loop that this time, encloses the input stage as the feedback summing point with the Voltage amplifier and output nodes providing feedback.
Each method seems to have advantages and Douglas Self covers these 2 in his latest, 6th edition of "Audio Power Amplifier Design Handbook"
Dave Zan, in a guest editorial in Vol.6 of the bookzine, neatly defines them all with the aid of diagrams.
It seems we can never have enough acronyms, especially those that don't translate well between different languages. Sorry about this but lucky for me, I had nothing to do with it
Last edited:
Hi,
it took the computer a whole day to simulate, but here are two curves for THD over frequency.
One with the Miller-compensation C11=100pF and without R30/C14.
The second with the 3-pole-Miller-compensation.
As expected (because it´s so typical for amps with this compensation) the Miller compensated amp shows an increase in THD at a breakpoint of ~1kHz.
Also as expected from the higher and more extended OL-gain, the THD for the 3-pole-Miller stays low above 1kHz and starts rising just before 10kHz when OL-gain starts to drop.
Also attached are Output impedance plotsfor Miller and 3-pole-Miller-compensation, each without the output LR-Network R35/L1 and with it.
L1 a 2.2µH, 100mOhm inductance.
As one can see influences the LR network just at frequencies >10kHz and it reduces output imedance phase shift, thereby increasing stability against capacitive loading.
Within the audio range the Zout is mainly the DCR of the inductance. Without it the output impedance is close to 3mOhm, as close to a perfect voltage source as one could wish for.
In praxis 3mOhm or 100mOhm don´t matter and I´d prefer the additional safety headroom gained by the use of the R35/L1.
jauu
Calvin
it took the computer a whole day to simulate, but here are two curves for THD over frequency.
One with the Miller-compensation C11=100pF and without R30/C14.
The second with the 3-pole-Miller-compensation.
As expected (because it´s so typical for amps with this compensation) the Miller compensated amp shows an increase in THD at a breakpoint of ~1kHz.
Also as expected from the higher and more extended OL-gain, the THD for the 3-pole-Miller stays low above 1kHz and starts rising just before 10kHz when OL-gain starts to drop.
Also attached are Output impedance plotsfor Miller and 3-pole-Miller-compensation, each without the output LR-Network R35/L1 and with it.
L1 a 2.2µH, 100mOhm inductance.
As one can see influences the LR network just at frequencies >10kHz and it reduces output imedance phase shift, thereby increasing stability against capacitive loading.
Within the audio range the Zout is mainly the DCR of the inductance. Without it the output impedance is close to 3mOhm, as close to a perfect voltage source as one could wish for.
In praxis 3mOhm or 100mOhm don´t matter and I´d prefer the additional safety headroom gained by the use of the R35/L1.
jauu
Calvin
Last edited:
I have Doug Self's APADH (4th ed.). Pages 183-5 claim that two pole compensation is clearly beneficial with EF output stages but with CFP (Sziklai) not. With CFP (like L12) all you need is classic Miller comp. For three pole compensation, I don't know. I never had a chance to see or hear any audio amp with it.
Hi,
First thing to be noted is, that the diagrams You refer to show, that the CFP with simple Miller-compensation shows very similar amplitude and frequency response of the THD-curve as the other output stages with 2-pole-compensation.
Self says, that the same 2-pole compensation applied to the CFP resulted in a margnal improvement.
The 2-pole compensation after D.Self didn´t change the first cap, meaning that the UGLF didn´t change either.
To expect further improvement I follow therefore, that You have to raise the UGLF without sacrificing too much on phase-reserve.
This is exactly what I have done in the sims.
If You compare the file ".... OL-gain wo R30_C14" -which shows the Miller compensation with C11=110pF- and file " ... OL-gain 3-pole-Miller-comp" that shows the 3-pole-Miller-compensation, You may notice that the UGLF raises more than twice from 418kHz (-98°) to 890kHz (-114°).
At least in the sim this resulted in a visible reduction of THD above 1kHz.
Still more potential lies in a Darlington-VAS-stage thst further raises OL-gain by at least 10dB.
Therefore the emitter resistor R28 of Q12 should be increased to 1k and a second 2N5551 be connected as Darlington to Q12.
jauu
Calvin
First thing to be noted is, that the diagrams You refer to show, that the CFP with simple Miller-compensation shows very similar amplitude and frequency response of the THD-curve as the other output stages with 2-pole-compensation.
Self says, that the same 2-pole compensation applied to the CFP resulted in a margnal improvement.
The 2-pole compensation after D.Self didn´t change the first cap, meaning that the UGLF didn´t change either.
To expect further improvement I follow therefore, that You have to raise the UGLF without sacrificing too much on phase-reserve.
This is exactly what I have done in the sims.
If You compare the file ".... OL-gain wo R30_C14" -which shows the Miller compensation with C11=110pF- and file " ... OL-gain 3-pole-Miller-comp" that shows the 3-pole-Miller-compensation, You may notice that the UGLF raises more than twice from 418kHz (-98°) to 890kHz (-114°).
At least in the sim this resulted in a visible reduction of THD above 1kHz.
Still more potential lies in a Darlington-VAS-stage thst further raises OL-gain by at least 10dB.
Therefore the emitter resistor R28 of Q12 should be increased to 1k and a second 2N5551 be connected as Darlington to Q12.
jauu
Calvin
I do of the experimental results. There are many. Is not consistent with the result of douglas self
So you can find. I L12-2 is not equal to amplifier design.
It has many different places. But their results are not the same.
This is what I found in the test. I don't know whether to reduce the performance. But L12-2 is a very stable and safe.
This aspect it is higher than the book introduction, zero defect amplifier.
It is very easy. Only need to keep on the oscilloscope to find the result is ok. Even we don't need to know why.
Hi,
yes, the primary inductance and the transformed capacitance are the main components of the impedance.
The capacitive part may be as high as 10uF.
Depending on the Audio trannies quality the amp may see a close to ideal capacitance, hence a large phase shift, that may reach values beyond -80°.
The power cube amp measurements only load the amp under test to +-60° and a lot of the testands already fail at easier loads.
The less good the transformers are the lower the phase shift, the easier is life for the amplifier.
So, despite the marketing blabla of using only expensive first class audio trannies, older MLs and Audiostatics used cheapest trannies of lower quality than even simple power transformers to allow for less good amplifiers to be mated with.
As I wrote earlier, a Accuphase A50 simply sucked and some other renowned brands were not better either.
Amps not relying on plain global feedback from the output typically performed better.
So, I didn't expect much from the L12s, but rather expected as they say at the Crystal fireworks: "Stand well clear and enjoy the Show"
I certainly did not expect the L12 to perform so utmost stable and remain sonically so well behaved.
jauu
Calvin
yes, the primary inductance and the transformed capacitance are the main components of the impedance.
The capacitive part may be as high as 10uF.
Depending on the Audio trannies quality the amp may see a close to ideal capacitance, hence a large phase shift, that may reach values beyond -80°.
The power cube amp measurements only load the amp under test to +-60° and a lot of the testands already fail at easier loads.
The less good the transformers are the lower the phase shift, the easier is life for the amplifier.
So, despite the marketing blabla of using only expensive first class audio trannies, older MLs and Audiostatics used cheapest trannies of lower quality than even simple power transformers to allow for less good amplifiers to be mated with.
As I wrote earlier, a Accuphase A50 simply sucked and some other renowned brands were not better either.
Amps not relying on plain global feedback from the output typically performed better.
So, I didn't expect much from the L12s, but rather expected as they say at the Crystal fireworks: "Stand well clear and enjoy the Show"
I certainly did not expect the L12 to perform so utmost stable and remain sonically so well behaved.
jauu
Calvin
Last edited:
hello guys,
I've been following this thread with the utmost interest. I don't remember (but I might be mistaken, sorry in advance) if the input impedance was notified as I was wanting to try a LDR type pre amp with those modules. " TheAttenuator works best with amplifiers that have input impedance greater than about 50k"
All the best.
I've been following this thread with the utmost interest. I don't remember (but I might be mistaken, sorry in advance) if the input impedance was notified as I was wanting to try a LDR type pre amp with those modules. " TheAttenuator works best with amplifiers that have input impedance greater than about 50k"
All the best.
Thanks a lot for this fast and efficient answer.
This remind me that my understanding of things are rather limited. But as I'm not stupid and interested, is there any article of book to advise that will help me understand better the trick you provided? maybe some douglas self audio power amplifier design?
Best regards
This remind me that my understanding of things are rather limited. But as I'm not stupid and interested, is there any article of book to advise that will help me understand better the trick you provided? maybe some douglas self audio power amplifier design?
Best regards
by the way, any thought about tiroidal transformers?
Does the encapsulated ones (http://www.farnell.com/datasheets/1520830.pdf) worth the extra expense compared to regular ones (http://www.farnell.com/datasheets/1661931.pdf)
Any brand to favor? Any recommandation
Would a R-core type instead of tiroidal ones be preferable? Impact the sound quality?
best regards.
Does the encapsulated ones (http://www.farnell.com/datasheets/1520830.pdf) worth the extra expense compared to regular ones (http://www.farnell.com/datasheets/1661931.pdf)
Any brand to favor? Any recommandation
Would a R-core type instead of tiroidal ones be preferable? Impact the sound quality?
best regards.