Jocko,
I have to agree with your observations with most CFB circuits I have built. They sound clean, fast, have low distortion but for want of a better word the seem to lack a sense of 'swing' when the music gets going. Increasing the bias does not seem to help much. The same circuit with no global feedback seems to be much better.
If possible, how about posting your non-global feedback design?
The design below seems to have good reviews and was posted on the forum sometime back.
Regards,
Jam
I have to agree with your observations with most CFB circuits I have built. They sound clean, fast, have low distortion but for want of a better word the seem to lack a sense of 'swing' when the music gets going. Increasing the bias does not seem to help much. The same circuit with no global feedback seems to be much better.
If possible, how about posting your non-global feedback design?
The design below seems to have good reviews and was posted on the forum sometime back.
Regards,
Jam
Attachments
Jocko, what was your opinion about CFB? Your client seems to be very "sugar pill" sensitive.
I think that CFB seems to be qutie alright as long as the amp works as intended. The korean guy Hyun Joon Rho nearly wets his pants in thrillment (is this a real word?) when he built my Alexander type of amp.
http://www4.head-fi.org/forums/showthread.php?s=&threadid=35074
I myself have so far no objections against CFB.
I have fooled around more with SPICE and notice that you can gain a little if the current mirror has a small current gain. 3-5 is good for the performance.
I appreciate that this thread has started a mature information exchange compared to an another one...you know which I'm talking about.
I think that CFB seems to be qutie alright as long as the amp works as intended. The korean guy Hyun Joon Rho nearly wets his pants in thrillment (is this a real word?) when he built my Alexander type of amp.
http://www4.head-fi.org/forums/showthread.php?s=&threadid=35074
I myself have so far no objections against CFB.
I have fooled around more with SPICE and notice that you can gain a little if the current mirror has a small current gain. 3-5 is good for the performance.
I appreciate that this thread has started a mature information exchange compared to an another one...you know which I'm talking about.
peranders said:
Just curious - what type of improvements are you referring to (bandwidth, offset etc.)?
Also, you mentioned about the Wilson vs. conventional current source not making a difference. The LTSpice audioamp.asc in Examples\Educational shows how to measure the loop gain for stability analysis. I use this a lot. I'm guessing you'd see higher low-frequency loop gain with the Wilson than without. I'd also venture a guess that the distortion would be somewhat lower as well.
If you're interested, I posted in another thread about how to do a somewhat primitive but reasonably effective harmonic distortion analysis using some information I got from the LTSpice Yahoo user's group about the FFT. It's at:
http://www.diyaudio.com/forums/showthread.php?postid=196461#post196461
The post that follows contains the text of the Yahoo discussion.
I'm a little bit skeptical about SPICE. A plain diff stage has a bandwidth to 10 MHz with 40 dB gain. If I add a cascode the stage won't get any faster. This isn't just true in real life. It seems that the simualtions models are too ideal. What about the Miller effect?
I would say the real life diff stage goes to 1-4 MHz without the cascode and to approx. 10 MHz WITH.
It seems that the results are too good.
I would say the real life diff stage goes to 1-4 MHz without the cascode and to approx. 10 MHz WITH.
It seems that the results are too good.
which transistor models are you using?
what about you source, is the source impedance zero?
I can get both results...
40db gain and Bandwith of 10MHz with a perfect source, but also a lot lower bandwidthwith a source impedance of 1k. try insert a resistor in series with you source.
what about you source, is the source impedance zero?
I can get both results...
40db gain and Bandwith of 10MHz with a perfect source, but also a lot lower bandwidthwith a source impedance of 1k. try insert a resistor in series with you source.
I used BC547. Ok, I did use zero impedance of the signal source. I'll guess you can do many stupid things disconnected from the real world.
Jam's findings "sound" about right. Very frustrating that it did so much right. Even more frustrating that nothing I did affected the sound; unlike VFB, where one can do any number of things to "tweak" the results.
Hey P-A.......the "clients" were the dealers.....and there were lots. Enough of them that I had to fix it, and fix it fast, or I would lose them.
In the case of the power amp, we (my buddy and I) each made several different ones. We both started out with a version of the Alexander amp. He liked it.....I wasn't so sure. We then went on to build several different CFB amps made with discrete components. They were all close in design, but each reflected our individual thoughts and the evolution of the project as each saw it.
(Basically, we built a bunch of amps........compared notes, picked each other's brain, and stole ideas.)
After we decided on what all the bugs were, and how to solve them, he put his version into production. Mine went in the junk bin. I had already been down that road once before. His stayed in production for a few years. I don't believe it was very succeessful.
Jocko
Hey P-A.......the "clients" were the dealers.....and there were lots. Enough of them that I had to fix it, and fix it fast, or I would lose them.
In the case of the power amp, we (my buddy and I) each made several different ones. We both started out with a version of the Alexander amp. He liked it.....I wasn't so sure. We then went on to build several different CFB amps made with discrete components. They were all close in design, but each reflected our individual thoughts and the evolution of the project as each saw it.
(Basically, we built a bunch of amps........compared notes, picked each other's brain, and stole ideas.)
After we decided on what all the bugs were, and how to solve them, he put his version into production. Mine went in the junk bin. I had already been down that road once before. His stayed in production for a few years. I don't believe it was very succeessful.
Jocko
You mean sonics?????? I guess it was ok. Clean, fast, the timbre of the instruments sounded right. Aside from the sound, I felt that the amp had too many little problems from a design standpoint. My involvement was just to help work out the bugs. And they "bugged" me.
Like I said......too many problems to build it as opposed to what we were making then.
Go build one.....and you tell us. I assume that you have the AD apps note for the Alexander version. (You won't need to model it first!)
(Yes, this thread is more interesting. However, as I have been ill the last week or so......needling you-know-who in between temperature spikes was a fun diversion from how lousy I felt. Still not better, yet.)
Jocko
Like I said......too many problems to build it as opposed to what we were making then.
Go build one.....and you tell us. I assume that you have the AD apps note for the Alexander version. (You won't need to model it first!)
(Yes, this thread is more interesting. However, as I have been ill the last week or so......needling you-know-who in between temperature spikes was a fun diversion from how lousy I felt. Still not better, yet.)
Jocko
I have help a friend to build an Alexander amp with very good results, I would say. Electrically "pure" performance and soundwise OK but I hadn't the opportunity to really test it.
Jocko, could you tell me your problems with it, more exactly? Was it an Alexander amp-like amp or the basic symmetrical topology like my picture in the first post?
I will build a CFB amp but now I'm sort of gathering opinions but still I'm a little bit confused because noone seems to really point out the weakness' of the amp. Quite in contrast to some people which make headphone amps with CFB and nearly wet their pants in delightment.
Maybe I'll ask Mike
? ..... 
Jocko, could you tell me your problems with it, more exactly? Was it an Alexander amp-like amp or the basic symmetrical topology like my picture in the first post?
I will build a CFB amp but now I'm sort of gathering opinions but still I'm a little bit confused because noone seems to really point out the weakness' of the amp. Quite in contrast to some people which make headphone amps with CFB and nearly wet their pants in delightment.
Maybe I'll ask Mike
? .....
Don't do that!!!!
Are you crazy man!!
Problems:
(Apply to the Alexander type and the one that you posted.)
It needed an inductor on the output. Loads over about 10 nF made it go nuts.
Inverting version is more stable.
Discrete versions needed expensive parts like MAT-02s, -04s.
Not a problem, but.....it could not be "tweaked". Nothing....even bias seemed to make any diffenence.
BTW.....check your e-mail.
Jocko
Are you crazy man!!
Problems:
(Apply to the Alexander type and the one that you posted.)
It needed an inductor on the output. Loads over about 10 nF made it go nuts.
Inverting version is more stable.
Discrete versions needed expensive parts like MAT-02s, -04s.
Not a problem, but.....it could not be "tweaked". Nothing....even bias seemed to make any diffenence.
BTW.....check your e-mail.
Jocko
not to correct you or anything like that.
what GPBW have you set it to?
You see i have one at home working with no output inductor and no expensive matched parts. actually they are just thrown into to the board.
on the other hand it have a DC servo.
I have just designed a new one.
it will be working next week. we have etched some dobbeltsided board today.
To start with we will make 3 stereo sets (six channels in all) ...
So we will see how it works out.
By the way i have settled for a GPBW of ~ 7 - 8MHz.
what GPBW have you set it to?
You see i have one at home working with no output inductor and no expensive matched parts. actually they are just thrown into to the board.
on the other hand it have a DC servo.
I have just designed a new one.
it will be working next week. we have etched some dobbeltsided board today.
To start with we will make 3 stereo sets (six channels in all) ...
So we will see how it works out.
By the way i have settled for a GPBW of ~ 7 - 8MHz.
I have a gut feeling that matched parts aren't essential (although it doesn't hurt).
I plan to have some DC-servo
but I haven't desided yet where I will inject currents.
Sonny, exciting. Keep me posted. I have an open mind and I'm interested in all info and opinions I can get.
Jocko, about your email: Thanks for your offer.
I plan to have some DC-servo
but I haven't desided yet where I will inject currents.Sonny, exciting. Keep me posted. I have an open mind and I'm interested in all info and opinions I can get.
Jocko, about your email: Thanks for your offer.
You think that I can remember that far back? The Alexander did not require matched parts, just the discrete one. Should have been more clear on that point.
I did not have problems with capacitive loads. But my buddy had dealers that sold cables that were very capacitive. Made by a company that he used to be buddies with.
I thought that you were getting married, and didn't have time for this stuff anymore??????
Jocko
I did not have problems with capacitive loads. But my buddy had dealers that sold cables that were very capacitive. Made by a company that he used to be buddies with.
I thought that you were getting married, and didn't have time for this stuff anymore??????
Jocko
she understand my needs for this hobby. :O)
I am married now.... 14. june
There was a big bunch of preparation before the wedding, which took all my sparetime.....
So i start out slow...
i am a bit ashamed, the is some mails from this boards members i haven't answered.... SORRY!
I am married now.... 14. june
There was a big bunch of preparation before the wedding, which took all my sparetime.....
So i start out slow...
i am a bit ashamed, the is some mails from this boards members i haven't answered.... SORRY!
Translation:
"The honeymoon is over. I ain't getting any. Time to screw with audio again."
Hey, pal......welcome back to reality. Snicker, snicker.
Jocko
"The honeymoon is over. I ain't getting any. Time to screw with audio again."
Hey, pal......welcome back to reality. Snicker, snicker.
Jocko
😀 I am getting, but there is time to screwing again.
We was one week at one of our beaches near the city "løkken", great weather, warm water.
We was one week at one of our beaches near the city "løkken", great weather, warm water.
Sonny, watch out, there is a double, well more than double, meaning to the verb to screw...
Jocko, I wonder if the quick fix you needed in order keep your customers would have been adding a line level single ended buffer to generate more second harmonic. Maybe the problem only was that your design was too good.
As I said, according to some of those few magazines which I believe to have a certain amount of thruth in their subjective reviews, seem to like CFB, so it does not seem that CFB per se is flawed.
Regards,
Eric
Jocko, I wonder if the quick fix you needed in order keep your customers would have been adding a line level single ended buffer to generate more second harmonic. Maybe the problem only was that your design was too good.
As I said, according to some of those few magazines which I believe to have a certain amount of thruth in their subjective reviews, seem to like CFB, so it does not seem that CFB per se is flawed.
Regards,
Eric
- Status
- Not open for further replies.