If memory serves, we did discuss this, albeit briefly, a short while ago.
Perhaps I should lay off H/K for a while.
Ah so that was in this thread? Hard to keep up
I have no connection anymore, other than some retained old friends, with harman/kardon, but I'm pleased when you mention them, as there were times, before the suits, the time-to-market worshipers, and the quarterly-results devotees began to dominate, when some good things came out under that brand and related ones.
Now, I'd just LOVE to know how it can be increased, aside from using high bias FETs instead of bipolars in the input stage(s).
Come on John, if you have ideas, get them out, push us in the right direction.
Ah so that was in this thread? Hard to keep up
I have no connection anymore, other than some retained old friends, with harman/kardon, but I'm pleased when you mention them, as there were times, before the suits, the time-to-market worshipers, and the quarterly-results devotees began to dominate, when some good things came out under that brand and related ones.
Brad, you obviously haven't noticed, but if I were to call myself a fan of any one company, it would surely be H/K. I kn ow there is better out there, but I believe H/K did, until about a decade ago, produce some of the best gear for the money it cost. In many ways, a poor man's High End company, although they did have some boo-boos, actually inevitable in a long line of products.
If I were to name any one company which has had profound influence on me, it would surely be H/K, followed by Analog Devices. Nothing else would be accurate or true.
Obviously, a very personal thing. I am a bit out of touch with their current models, but I have studied (literally) their best 1990-2000, and have a rather fat collection of their service schematics.
No, I have noticed, and have been pleased to see it.
I can relate an anecdote I heard from the proverbial horse's mouth about some of the current state of affairs with the Harman brands. The CEO believed that as much as possible manufacturing, and even design, should be moved offshore. He dismantled a whole department in the process. But there is still a group in Northridge and some elsewhere that are doing some design and development. However, they have to get the majority of the products completed and manufactured out of the country.
One product, not yet released, received little in the way of listening tests, and the offshore people said that they had done things their way and preferred their equalization to that of a key person in Northridge, complaining about a lack of bass. Samples were sent to be evaluated here.
It turned out that they had loaded the coefficients backwards. The woofer got the mid-tweet EQ and vice versa.
Now anyone can make a mistake. The more interesting thing was the attitude.
In the long run we will all get better. How long a run that will be I won't hazard a guess.
John,
Just finished "reviewing" that one in paperback (thought it was part of IC OP AMPS IEEE series by R.G Meyer).
But other than my sloppy wording I thought everyone else's comments were right in-line with it.
Thanks
-Antonio
I think that reading Solomon's paper can get us all, in touch with what is important in design for maximizing slew rate in a specific design. I noted confusion between rise time limiting and slew rate limiting, etc. Best that everybody note Solomon's paper that, in 1974, was a REVELATION to me, when I first read it, even though I recently had taken courses from R.G.Meyer, and D.O.Peterson who apparently taught both Solomon and Meyer on nonlinear analysis of linear circuits regarding the same subject matter.
It is his PRACTICAL and INTUITIVE approach to the subject that was the breakthrough, and has made this single paper so often recommended.
It is his PRACTICAL and INTUITIVE approach to the subject that was the breakthrough, and has made this single paper so often recommended.
Has everybody reviewed the tutorial? I am sorry that I still don't know how to add a an external link, but it is easy to Google and get your OWN pristine copy of this almost 40 year old paper, made when IC op amp design was young, and clarity was important. I do recommend it to you, because many here are strong on opinions, and a little weak on practical theory. Trust me, I was in the dark about 5 years, for the most part, before I read this tutorial. There are several other topics, equally interesting, that were included in the tutorial, including: thermal feedback (trust me, its real), and 'tail poles' to name two, relatively obscure topics. If you want to understand why Bob Cordell and I use resistors in SERIES with our active current sources, think 'tail pole' reduction.
Thanks John
I always thought those were just to reduce voltage or power across the current source.
Thanks
-Antonio
Found it in two versions, As the abstract and as a National reprint. Ha! This is about the first time John has sent us to the books and it did not cost me anything but work. "Tail pole reduction" That is going to need some thinking and reading. The DH-120 used it. My $1 book on FETs helped fill in a few missing pieces.
So much to learn. At least my speaker building has stepped up a couple of notches. I thought my new crossover and switch to Seas tweets in my guest room system had hurt the imaging, then an ad came on where a plane flew in from about 10 feet outside the room. Imaging is still there, but much cleaner and much smoother.
Is 'tail pole reduction' just a fancy name for reducing the effect of stray capacitance at a high impedance node, such as a CCS, by adding a series resistor? If so, I thought that was common knowledge. Common knowledge can be hidden by giving it a new name, then the gurus can appear superior to ordinary mortals. All very Gnostic!
I don't do SS these days, but 10-15 years ago when I designed a power amp I remember thinking about stray/parasitic capacitance in the LTP tail CCS. Having estimated its likely effect, I can't remember whether I decided to include a resistor or not. I probably did have a resistor, because I tend to err on the side of caution and a resistor there would seem to do no harm provided it leaves sufficient voltage across the CCS so that device parasitic capacitances don't grow too much.
The same issue can potentially arise whenever a CCS is used so 'tail pole reduction' is an unfortunate name. 'CCS parasitic capacitance isolation' has the merit of being more descriptive and more generic.
The same issue can potentially arise whenever a CCS is used so 'tail pole reduction' is an unfortunate name. 'CCS parasitic capacitance isolation' has the merit of being more descriptive and more generic.
Let me explain to everyone WHY I might sometimes use a series resistor and sometimes I might not. I have presumed that Bob Cordell did the same analysis, when he included it in his phono design, in his latest endeavor.
For me, a single differential pair of devices should be fed by a GOOD current source. However, sometimes a PRACTICAL current source might be a simple 2 pin current source made for the purpose and have a rated output capacitance of 2 pf or less. However they can be noisy, so we might make our own current source, instead, with a degenerated low noise jfet, and this effective output capacitance could be MUCH MORE, so what to do. Well, adding a series resistor in series with the output will reduce the effective capacitance, at least at the frequencies where it counts.
Actually, with a limited power supply range, adding this resistor can actually compromise the actual output impedance of the current source by limiting the voltage across the current source jfet, itself. So it is a relatively sophisticated decision what value to use, because a HIGHER value of series R can actually REDUCE the working impedance of the current source, as well as adding a potential noise contribution.
For me, a single differential pair of devices should be fed by a GOOD current source. However, sometimes a PRACTICAL current source might be a simple 2 pin current source made for the purpose and have a rated output capacitance of 2 pf or less. However they can be noisy, so we might make our own current source, instead, with a degenerated low noise jfet, and this effective output capacitance could be MUCH MORE, so what to do. Well, adding a series resistor in series with the output will reduce the effective capacitance, at least at the frequencies where it counts.
Actually, with a limited power supply range, adding this resistor can actually compromise the actual output impedance of the current source by limiting the voltage across the current source jfet, itself. So it is a relatively sophisticated decision what value to use, because a HIGHER value of series R can actually REDUCE the working impedance of the current source, as well as adding a potential noise contribution.
John,
Thanks for the insight.
Also sharing the power dissipation (or voltage) between the CCS and resistor may allow one to choose different components in the CCS.
I wouldn't expect there to be a need for such an isolating resistor with discrete designs of bias currents above a mA (wouldnt it take near a hundred pico-farads to matter).
BTW, personally these are the kinds of exchanges I really like, a few tips for the millions of details.
Is there really a noise penalty?
-Antonio
Thanks for the insight.
Also sharing the power dissipation (or voltage) between the CCS and resistor may allow one to choose different components in the CCS.
I wouldn't expect there to be a need for such an isolating resistor with discrete designs of bias currents above a mA (wouldnt it take near a hundred pico-farads to matter).
BTW, personally these are the kinds of exchanges I really like, a few tips for the millions of details.
Is there really a noise penalty?
-Antonio
"Tail pole" sounds somehow more potentially salacious
This will please some As far as JFET I sources go, if you can spend enough volts very good results can be gotten up to rather high frequencies with a bootstrapped cascode, which I believe Jung has determined was first shown by Csanky. They operate best when the lower device gm is a good deal more than the reciprocal of the lower source resistor value, as then the upper gate's current mostly flows in the lower source. Watch WJ's website for the details, and possibly some feedback from Csanky himself, who is still with us.
- Status
- Not open for further replies.