Tutorials actually, next maybe global feedback vs local networks. The math would lose most folks. But the principle is not too hard to grasp.
If you have a gain stage that follows Out=A*vin + .1*A*vin*vin and you apply feedback the reduction in distortion is well known. The same for Out=A*vin + .1*A*vin*vin*vin.
But when you use Out=A*vin + .1*A*vin*vin * (A*vin + .1*A*vin*vin*vin), the * sign really changes the feedback math.
Of course global feedback works much better when the equation is more like Out=20*A*vin + 2*A*vin*vin *(.05A*vin + .005*A*vin*vin*vin)
So when you compound gain stages of different gain properties but with both contributing to the overall gain global feedback will produce some extra artifacts, better to use local feedback on each stage, but when one stage dominates the gain structure global may be indistinguishable from local.
Note the assumption the total gain open loop and closed loop is the same for these comparisons.
So let us see who is actually reading this thread!
Well Simon, I am reading it, but I look at feedback in the same way, but in much more detail. We have talked about global feedback here, over the years, but usually we get nowhere. This is because of the FM modulation that is ignored, and even denied by many. It took us many decades, but first it was TIM or essentially slew rate limiting, that was shown to be very important. Of course, when you increased slew rate, all else being equal, you usually increased the open loop bandwidth, open loop linearity, and reduced the total amount of feedback. This all helped. Then, about 30 years ago, FIM or PIM became a topic of interest, but it is still put down by many here.
This weekend, I spoke to my associate who has a new method of measuring FIM. I mentioned him about a year ago, and everyone and his brother jumped on me to reveal his info, even before he gave it to me. Because he is pursuing a patent, he got very upset with me for mentioning it at all, but he tells me that he is moving forward by giving a series of tutorials on this subject at a major university. He, himself, is an engineering graduate of UCB, and has worked at AMPEX, Dolby, Sony, and many other telcom co's in Silicon Valley, and worked for me, when he was a student at UCB in the late '70's. I hope for more in future, but that is all there is, at the moment.
This weekend, I spoke to my associate who has a new method of measuring FIM. I mentioned him about a year ago, and everyone and his brother jumped on me to reveal his info, even before he gave it to me. Because he is pursuing a patent, he got very upset with me for mentioning it at all, but he tells me that he is moving forward by giving a series of tutorials on this subject at a major university. He, himself, is an engineering graduate of UCB, and has worked at AMPEX, Dolby, Sony, and many other telcom co's in Silicon Valley, and worked for me, when he was a student at UCB in the late '70's. I hope for more in future, but that is all there is, at the moment.
John,
I was just talking about the distortion caused by nonlinearity in the transfer function and how you compound the errors when there are two about equal in gain but different transfer functions cascaded.
The issues such as slew rate limiting, stored charge, and others can cause great harm if not addressed, but they are different issues and need to be analyzed singly before they are grouped into an audio "Mystery!"
Demian
I have started running you resistors, The loose small square ones were not impressive. The wire-wound was and your Caddock is better than mine! I missed that one of the small round single was 332K.. can't measure that high. Of course if anyone has an amplifier that can produce 150 volts RMS with distortion below say -120 I would be interested in trying!
Also did you get my email?
One could, but should not, more power handling should be less 3rd harmonic distortion. The nicely performing CAL-R's were only rated at .08 Watts so they can be accurately compared to most of the others.
Edit: I did find the answer to my question. So the fundamental is again at 15.81V. But you still didn't describe the measurement procedure. I give up.
Let's see. The first plot shows the fundamental at around -92dBr and the noise floor at about -176dBr, or about -84dB down from 15.81V. So in volts the noise floor would be at 15.81/10^(84-20) = 0.0009975V, or about 997.5 uV. Sounds kind of high for an AP unit. Have I made a mistake?
Let's see. The first plot shows the fundamental at around -92dBr and the noise floor at about -176dBr, or about -84dB down from 15.81V. So in volts the noise floor would be at 15.81/10^(84-20) = 0.0009975V, or about 997.5 uV. Sounds kind of high for an AP unit. Have I made a mistake?
Last edited:
You have to be kidding?
4 opamps and EIGHT elco's in the signal path - nein danke!
Regards, Allen
Allen, the preamp is not the point.
JC was looking for ways to implement variable cartridge loading, and I thought he would appreciate to see one more way to do it.
Graham apparently already worked on this 30 or 40 years ago.
jc
Last edited:
No -176 from 15.81 volts is 10 exp (-15.81/20)*15.81 or 1.58e-09*15.81 or 2.5nV
As to the test method it has already been covered.
As far as cartridge loading is concerned, Mark Levinson and I first worked on the problem in 1974. I even contacted Ortofon and EMT as to their recommendation at the time. They didn't have a clue, and we were in the dark. However, the JC-80 preamp phono had both continuous and switchable phono loading, and the Vendetta Research had variable loading from 10 to 200 ohms, and 47K, built in, with added gold pins, so that ANY resistor value could be used.
While Graham's preamp did indeed address loading, it was not more sophisticated than what we were working with 30 years ago, and put into production.
While Graham's preamp did indeed address loading, it was not more sophisticated than what we were working with 30 years ago, and put into production.
- Status
- Not open for further replies.