ostripper said:
Those are cool..
An externally hosted image should be here but it was not working when we last tested it.
I will keep them in mind for my small amps (BTF50) or auxiliary supplies. What I want is something very "findable" (scrap heap , mouser , digikey , chinese) , and replaceable.
OS
Mouser
576-0251010.MXL
Hi Piersma,
In your schematics, C10 is in a wrong place,
it shunts any AC negative feedback from the TL431 output to ground.
AndrewT
---the input cascode can be referenced to signal ground or to input pair emitter/source tails. I don't know which, if any, is better.---
Or even different reference for each of them, to their own emitter :
http://www.kevinaylward.co.uk/ee/micampdesign/ComplementaryCurrentSourcedCascodeMicAmp.GIF
In your schematics, C10 is in a wrong place,
it shunts any AC negative feedback from the TL431 output to ground.
AndrewT
---the input cascode can be referenced to signal ground or to input pair emitter/source tails. I don't know which, if any, is better.---
Or even different reference for each of them, to their own emitter :
http://www.kevinaylward.co.uk/ee/micampdesign/ComplementaryCurrentSourcedCascodeMicAmp.GIF
I MUST put my 2 cents in on this. I started with the original sym (mpsa 92/42 VAS) I, too liked its sound. In my scaling (upgrade path), I added the wilson CM to the VAS and even entered the folded cascode (roenders) level of design.
ALL had a similar sound , the folded design , I must admit , was the best ..... but not by much. So , all these improvements add many more parts , but with diminishing returns. In the real world , the simple LED CCS gives inaudible crosstalk , even at insane volumes. I have tested this amp with dual 1500VA trafos vs. running both on just one PS... NO difference that you can hear.
You can measure or see a slight difference on a DMM or CRO , but where it counts ,in the listening room , nothing.
A much bigger return on the sound quality is realized with manipulation of the open loop gain , feedback , component choice , and compensation. The advanced techniques are academically simulating , might refine the sound slightly , but are they worth the added complexity?
In the case of adding the cascode to the LTP , this should make the circuit more "bulletproof" electrically while giving us many more options... input wise. ( a good return at a very low cost) Even on the DIY level , an amp one can customize will make for a better building and learning experience.
OS
@Andrew
Yes, I meant to signal GND. As always, there are alternatives already mentioned here. But the first cascode OS proposed here was not entirely correct.
Cascodes are not always the solution for making an amplifier with low distortion figures as the LTP is (usually) not the main contributor of distortion in it. Sometimes they are needed only to reduce Vce to acceptable value for devices used in LTP.
Yes, I meant to signal GND. As always, there are alternatives already mentioned here. But the first cascode OS proposed here was not entirely correct.
Cascodes are not always the solution for making an amplifier with low distortion figures as the LTP is (usually) not the main contributor of distortion in it. Sometimes they are needed only to reduce Vce to acceptable value for devices used in LTP.
You are right on , Bluesky. My last simulation turned into big pulse generator with a tail referenced cascode. A local feedback loop no doubt. Ground it is.
Yes, it is THAT stable . With what's in the first picture , it will run 2 paralleled fisher 15" 3-ways full volume ,all day with NO zobel , through 25' lamp cords. when I built it , I knew with my huge 77V
supply, an unstable amp would be VERY bad.
I have run C6(Cdom) down to a 33pF , but the sound was best at 56 - 68 pF.
Hi ostripper,
Reference the cascode to the common emitter of the diff pair. The entire idea is to lock the C-E (or D-S) voltages. Why allow a common mode signal to upset the apple cart? A common mode signal could be even signal peaks at high volumes where you are comparing input to feedback.
For more information, look at many Japanese amplifiers (this is one place they do it properly), or even Doug Self's book. My lowly bench experiments years ago showed this to be better at higher signal levels. A cascode arrangement seems to sound cleaner to my ear. This could simply be that the input pair are running cooler and isolated that little bit from the rails.
Please, don't reference to ground for those. Sim some other successful designs to see if there is something about your simulator. Unless you are messing with the tail current big time, there should be not effect that would cause it to be unstable. Try increasing your tail current to compensate for the reference current for the cascode voltage set supply?
All you have to do is remove the dropping resistors and connect your high quality regulated supply at that point. No board changes other than that. Those of you who wish to try this can use an external supply (regulated) just to try this out.
-Chris
Reference the cascode to the common emitter of the diff pair. The entire idea is to lock the C-E (or D-S) voltages. Why allow a common mode signal to upset the apple cart? A common mode signal could be even signal peaks at high volumes where you are comparing input to feedback.
For more information, look at many Japanese amplifiers (this is one place they do it properly), or even Doug Self's book. My lowly bench experiments years ago showed this to be better at higher signal levels. A cascode arrangement seems to sound cleaner to my ear. This could simply be that the input pair are running cooler and isolated that little bit from the rails.
Please, don't reference to ground for those. Sim some other successful designs to see if there is something about your simulator. Unless you are messing with the tail current big time, there should be not effect that would cause it to be unstable. Try increasing your tail current to compensate for the reference current for the cascode voltage set supply?
Want to really improve the performance of an amplifier? Run the voltage amp section off it's own regulated power supply or supplies. This has the added benefit of isolating the gain stages from outside AC supply noise more effectively. It also eliminates any effects caused by output currents through the supplies back. Now it's really hard to see any interactions with a scope or otherwise.
All you have to do is remove the dropping resistors and connect your high quality regulated supply at that point. No board changes other than that. Those of you who wish to try this can use an external supply (regulated) just to try this out.
-Chris
Measure the damping factor, no load in output then loaded with 2R or 4R.ostripper said:
Just to show an unrealistic output and load, (2R and 120v p-p)and how this loads the VAS at 4ma, the (attached) file shows (top plot)
current through driver basestoppers and (bottom plot) total Ic of the Mje15032/33 , not even close to their SOA , even at 100 C.
I would not want to be the speaker or touch the wire at this level..
OS
http://www.diyaudio.com/forums/showthread.php?s=&threadid=22107
Hi Raphael.luc,
---Measure the damping factor, no load in output then loaded with 2R or 4R.---
I used to often do that kind of measurements.
The inverse way is easier for calculation :
Measure with load first.
For 4 Ohm, set the output to 4 V with the volume control, so the current delivered is 1 A.
Then disconnect the load and measure the new slightly higher voltage.
THe difference indicates the internal resistance seen at the output.
The funny thing is when you get a lower voltage without load !
---Measure the damping factor, no load in output then loaded with 2R or 4R.---
I used to often do that kind of measurements.
The inverse way is easier for calculation :
Measure with load first.
For 4 Ohm, set the output to 4 V with the volume control, so the current delivered is 1 A.
Then disconnect the load and measure the new slightly higher voltage.
THe difference indicates the internal resistance seen at the output.
The funny thing is when you get a lower voltage without load !
Suggestion for C7
Ostripper -
The earlier discussion on C7 resulted in what might be its premature removal from your design. As pointed out earlier, C8 gets the benefit of the Miller Effect because there's voltage swing at the collector of Q5, while C7 does not show the same effect because there was very little voltage swing at the collector of Q4. Have you tried connecting C7 between the base of Q4 and the collector of Q6? That's where the voltage swing is.
regards,
Mike
Ostripper -
The earlier discussion on C7 resulted in what might be its premature removal from your design. As pointed out earlier, C8 gets the benefit of the Miller Effect because there's voltage swing at the collector of Q5, while C7 does not show the same effect because there was very little voltage swing at the collector of Q4. Have you tried connecting C7 between the base of Q4 and the collector of Q6? That's where the voltage swing is.
regards,
Mike
I was talking to measure in simulator, With what you measured, the real amplifier, multimeter portable?forr said:
My multimeter of accuracy to 1% for voltage, it is true RMS.
I get the value the simulation is for exemple 24,30 V with load, 24,29 V no load. I believe that with the multimeter to measure is not correct.
Here is the Ic across C7 with Q4 base to Q6 collector (attached) I- C7. there is a little more "swing" there , but just 1.4v, since it is cascoded. The Ic on C8 ... I(C8) is 100 times greater. I don't think
C7 is needed with the cascode. With the original otala or hitachi design ,(resistor instead of cascode -Q6) C7 would have miller characteristics.
OS
Attachments
Do you mean like the pioneer M90 ?? (attached) It seems so simple how they did it with a simple voltage divider from rail to tail , but it simulates perfectly and even shaves a hair off the THD.
That is already planned for , A large pad on the voltage amp end of the PCB can either accommodate the resistor or a totally separate supply.
OS
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.