Hi Woj,
I have looked at this schematic a number of times and concentrated my efforts on the output end. I never saw your R13 or it's significance.
The LTP is sometimes done as a single ended input and in this topology the feedback is (almost) always taken to the emitter of the input transistor.
It appears that ONsemi have really thought about this, since they are combining AC and DC feedback.
Would someone like to model this and vary the value of R13 to see it's effectiveness as a feedback controller?
ACD,
C5 as an alternative to Miller comp cap is probably about right. There is a lot more current (implying lower impedance) available here to charge the cap. It is normal to see it a few times larger in value than the Miller comp cap. It is often paralleled with a series RC to trim the HF response of the global circuit, but ONsemi have omitted this option. Look at JLH 80W FET amp or Crimson quasi BJT.
I have looked at this schematic a number of times and concentrated my efforts on the output end. I never saw your R13 or it's significance.
The LTP is sometimes done as a single ended input and in this topology the feedback is (almost) always taken to the emitter of the input transistor.
It appears that ONsemi have really thought about this, since they are combining AC and DC feedback.
Would someone like to model this and vary the value of R13 to see it's effectiveness as a feedback controller?
ACD,
C5 as an alternative to Miller comp cap is probably about right. There is a lot more current (implying lower impedance) available here to charge the cap. It is normal to see it a few times larger in value than the Miller comp cap. It is often paralleled with a series RC to trim the HF response of the global circuit, but ONsemi have omitted this option. Look at JLH 80W FET amp or Crimson quasi BJT.
I think, that R13 makes current source with R8. I've noticed, that
R13/R8 = R28/R5
R8 with R13 makes near ideal current source for powering LTP with Q1 and Q2.
I don't think that C7 and C8 are too big. I even think, that C3 and C4 are too small ... but I am sure, that designers of this schematic had done proper calculations about that
.
Shuldn't C3 be connected between colector and base Q4? And what about C5?
And I still wanna links to other schematics with R8 R13 like that.
R13/R8 = R28/R5
R8 with R13 makes near ideal current source for powering LTP with Q1 and Q2.
I don't think that C7 and C8 are too big. I even think, that C3 and C4 are too small ... but I am sure, that designers of this schematic had done proper calculations about that
.Shuldn't C3 be connected between colector and base Q4? And what about C5?
And I still wanna links to other schematics with R8 R13 like that.
connecting C3 as you describe invokes Miller comp cap. Many designers claim this is bad for sound quality. D.Self only designs this way to reduce distortion and I believe this is one of the main reasons his amps measure technically very well but reputedly sound bad.wojtek5001 said:I think, that R13 makes current source with R8. I've noticed, that
R13/R8 = R28/R5
R8 with R13 makes near ideal current source for powering LTP with Q1 and Q2.
I don't think that C7 and C8 are too big. I even think, that C3 and C4 are too small ... but I am sure, that designers of this schematic had done proper calculations about that
Shuldn't C3 be connected between colector and base Q4? And what about C5?
And I still wanna links to other schematics with R8 R13 like that.
C5 provides the alternative feedback as mentioned earlier. It is feedback over two stages instead of just one and avoids loading the LTP collector excessively.
This pseudo constant current source seems to me a curiosity than anything else. Any divergence in the divsion factor of of R8/(R13+R8) and R5/(R28+R5) would modulate the intended constant current.
Miller capacitor sounding bad is an idea (I would say a legend) propagated by John Linsley-Hood. Most op-amps use Miller compensation, not all are said to be that bad.
According to Self, you cannot even avoid a bit of Miller compensation because of the Cbc and surrounding parasitic capacitor. Other ways to compensante feedback amplifiers have their own pitfalls, this has been the subject of many articles and many posts on this forum.
I think the main reason why Self amp have a bad sounding reputation is due to the lack of their own footprint when compared with more flattering schemes.
Miller capacitor sounding bad is an idea (I would say a legend) propagated by John Linsley-Hood. Most op-amps use Miller compensation, not all are said to be that bad.
According to Self, you cannot even avoid a bit of Miller compensation because of the Cbc and surrounding parasitic capacitor. Other ways to compensante feedback amplifiers have their own pitfalls, this has been the subject of many articles and many posts on this forum.
I think the main reason why Self amp have a bad sounding reputation is due to the lack of their own footprint when compared with more flattering schemes.
by forr:
Exactly. But lack of R13 also will cause "intended" constant current to be modulated by input voltage/R8. Smaller R13 is like a negative feedback, larger like a positive, but not strictly. My real problem about that is a higher frequencies and what will happen then. What, if for example output voltage will be out of phase with input voltage? There may be risk of positive feedback, and then oscilations. Well, precise calculations are needed about that .
Exactly. But lack of R13 also will cause "intended" constant current to be modulated by input voltage/R8. Smaller R13 is like a negative feedback, larger like a positive, but not strictly. My real problem about that is a higher frequencies and what will happen then. What, if for example output voltage will be out of phase with input voltage? There may be risk of positive feedback, and then oscilations. Well, precise calculations are needed about that
.
R13 provides feedback, yes, but only positive.
The idea of the scheme is based on maintaining DC voltages between base of Q1 (= almost equal to voltage at base of Q2 at low frequencies) and connexion R6, R7, R8, R13 constant.
I think we should say that emitters of Q1-Q2 and resistors R6 -R7 are submitted to the bootstrap action provided by R8-R13.
Just for fun, I wonder if instead of using R13 we could change the value of the feedback network to make it of lower impedance and add a high value bootsrap capacitor between the base of Q2 and the common point of R6, R7 and R8.
Wojtek5001
--- But lack of R13 also will cause "intended" constant current to be modulated by input voltage/R8.---
I agree. A very often seen scheme, which should not be used, at least in non-inverting configurations.
---Well, precise calculations are needed about that.---
Standard constant current sources are much more dependable than this circuit whose simplicity is only apparent, once analysed with frequency in mind.
The idea of the scheme is based on maintaining DC voltages between base of Q1 (= almost equal to voltage at base of Q2 at low frequencies) and connexion R6, R7, R8, R13 constant.
I think we should say that emitters of Q1-Q2 and resistors R6 -R7 are submitted to the bootstrap action provided by R8-R13.
Just for fun, I wonder if instead of using R13 we could change the value of the feedback network to make it of lower impedance and add a high value bootsrap capacitor between the base of Q2 and the common point of R6, R7 and R8.
Wojtek5001
--- But lack of R13 also will cause "intended" constant current to be modulated by input voltage/R8.---
I agree. A very often seen scheme, which should not be used, at least in non-inverting configurations.
---Well, precise calculations are needed about that.---
Standard constant current sources are much more dependable than this circuit whose simplicity is only apparent, once analysed with frequency in mind.
So has anyone built the amp circuit from On-semi's schematic.
I saved a copy of the jpg and I want to build it and see how it works.
I imagine for further tweaking, you could parallel resistors with the diodes to fine-tune your bias mA to what you would like. I think that the thermal sensing of the diodes would still work.
Also, instead of running a resistor to -15V like in the schematic, I Think a CCS or a different resistor to the -ve rail would work fine.
I wonder if that -15V rail is regulated or not and how it affects CMRR of the amp?
I saved a copy of the jpg and I want to build it and see how it works.
I imagine for further tweaking, you could parallel resistors with the diodes to fine-tune your bias mA to what you would like. I think that the thermal sensing of the diodes would still work.
Also, instead of running a resistor to -15V like in the schematic, I Think a CCS or a different resistor to the -ve rail would work fine.
I wonder if that -15V rail is regulated or not and how it affects CMRR of the amp?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.