Certainly not much originallity in putting those generic modules in a cute case, but seems many are doing just that.Charles Hansen said:
How can you have no feedback? Even individual active devices have some internal feedback (and that goes for tubes as well)
roender said:
Why do you have caps across the diodes? They just slow down the parasitic ringing, and do not damp it. You should use a proper snubber.
PMA said:
No, I was speaking of preamplifiers. I could do it for power amplifiers also, but only if running in pure class A. For practical applications, this means an upper limit of 25 watts or so. I haven't done it yet, but someday will. It would be very interesting when used with a high-sensitivity speaker.
Bob Cordell said:
Debater? Perhaps. Or perhaps it's just that I feel that Charles was ill-used in the thread about his letter to Stereophile. Know this: He, John, and Nelson do not stand alone. I will defend them. Many talk; those guys do. When they talk, I listen. I've learned more from those three than from all the pointless nattering of the fifty most spec/simulation-obsessed fools put together.
I will freely grant that others were guilty of poor behavior, but you were the one who took it upon himself to write a letter to Stereophile with the express intent of publicly taking Charles to task. That's why I'm handing it back to you in particular, although others are just as deserving of being raked over the coals.
Shall I take your attempt to throw the burden of proof back on me as an oblique confession that you cannot back up your statements regarding passive components in your letter to Stereophile? If so, we will have dispensed with roughly half of your letter and given that the first few paragraphs were an expression of your emotions (I'm shocked! Shocked, I say!), they are pretty much exempt from demands for empirical proof. At which point we will be down to perhaps ten or twenty percent of your letter most, if not all, of which has been dealt with by Charles in pointing out that the chip stuff was in his cheaper product and sorely out of date at any rate.
Should that be the case, the actual content of your letter to Stereophile will have been reduced to tatters, excepting, as I say, the first paragraph or two regarding your feelings, which you are certainly entitled to. However, I think most people would agree that your letter would read very differently indeed had you simply stopped after expressing your shock and dismay.
Speaking only for myself, I generally use the Vishay/Dale RN60 1% metal film resistors for smaller applications. They were--at least until recently--fairly cheap in 100 qty boxes (.16 ea.), which was actually less than I was paying locally for NTE 2% resistors ($1.20 for a package of six, i.e. .20 ea.) until my favorite parts house went out of business. The Dales are reputed to sound pretty good, but I will not put my reputation (such as it may be [stop laughing, dammit!]) on the line until I've actually done comparisons.
I have used the Caddock power resistors in some applications and the Panasonic "blue" metal film 3W resistors in others when higher Pd is required. The Caddocks are more expensive, but they will take more than 3W and I'm working towards tightening percentages, just to remove the whole Monte Carlo thing from my mind. I'd like to get it down to the point where the actives are the only real source of variability in my circuits (yes I know, they already are, even with 10% resistors)--and those I can either match by hand or compensate for with pots instead of fixed resistors. For what it's worth, the Panasonic resistors are much tighter than their 5% tolerance would lead you to believe. I've measured quite a few and can't remember ever finding one over 2%; many are within 1%.
Sad to say, I haven't done a serious comparison of passive components in quite some time, partly because of the cost and partly because of the time required to do a good job. Time, unfortunately, is my most rare and precious commodity. With that in mind, I can only say that I have no opinion if asked about comparisons between Vishay/Caddock/Roderstein (sp?)/Mills/et. al. I am slowly working towards doing a comparison between my "standard" Dale resistors and the Caddock MK132, but there's no telling when I'll actually get it done.
If you know of a post where I have made a firm statement about preferring one brand of resistor over another for sonic reasons, I'd appreciate it if you would point it out. I try to stay with "I have no opinion" until I've heard something myself or gotten word from someone whose ears I trust (very few people on that list), in which case I try to remember to specify that it's not personal knowledge, but does come from a source that I trust.
How about it? Ready to yield on that Stereophile letter or do you have data on those passives? And, yes, I really am quite seriously interested in the relative quality of chip passives vs. discrete components. The Vishay/Dale RN60 data is readily available, as is the Caddock data for the MP915 and MP930. I believe there's a link from Digikey to a datasheet on the Panasonic "blue" power resistors.
Oh, and regarding caps, I generally use polystyrene or polypropylene. Mica on occasion. Electrolytic when I must (no, I don't use anything super fancy like Black Gate caps--I try to design the circuit so that if I must use an electrolytic, the voltage is as close to DC as I can manage). I think representative data is available to anyone who cares to look for it. I do sometimes use the MIT RTX polystyrene caps--that's a holdover from the last time I did an actual listening comparison of caps. It's possible that some of the more recent caps may be better, but until I have a chance to compare again, I'll stick with the MIT. Expensive, but damned good.
Grey
andy_c said:
Actually an emitter follower with a capacitive load is inherently unstable. It turns out to be non-trivial to stabilize. (Why do you think so many power amps have output inductors and Zobels? But it would be really silly to put an output inductor on a power supply!)
I've used the super fast Sanyo parts (Ft = 400 MHz) as predrivers in an emitter follower triple. But when I tried to use them as emitter follower in a modified capacitance multiplier I could not get it stable. I tried for about a week and gave up and used some 40 MHz parts instead.
Adding global feedback to such a circuit doesn't make it any easier to stabilize...
Charles Hansen said:
Well, it depends on the transistor and the load. A lot of low-power devices have enough base spreading resistance that you may not even need a base stopper (BC550=130 to 170 Ohms for example).
Yow, I didn't realize they made such fast medium-power transistors! Sounds like a lot of trouble.
That's what I was trying to say. The reg with global feedback will be more difficult to stabilize. There's also the tendency for some people to "upgrade" their regulators by using low ESR caps on the output - a bad idea, especially for circuits with global feedback.
Bob Cordell said:
A few ideas, but nothing substantial. I gave up trying to explain things years ago. It's kind of like teaching a pig to sing. It just wastes your time and annoys the pig. And trying to figure out exactly why the high-feedback regulators sounds worse is a similar waste of time.
In the time that it would take to figure it out (indeed, if that is even possible), I could dozens of other listening tests and learn even more about making amplifiers sound good.
andy_c said:
Base stoppers are of very little use in stabilizing an emitter follower in my experience. Without feedback, Zo = 26 / Iload (in mA). But we must also include the driving impedance divided by the beta of the transistor. I'm not familiar with the BC550, but many transistors have a beta of around 100. So with a base spreading resistance that high, Zo would be over 1 ohm no matter how much load current it was supplying.
What's worse is that I've had almost no luck in stabilizing emitter followers with base stopper resistors. Even going to values so high that they degrade the performance of the circuit doesn't seem to make them stable under all conditions.
Charles Hansen said:
Depends a lot on the application - transistor type, load, circuit topology, etc. A zobel can do a nice job for VHF oscillation into inductive, but not capacitive loads. Different solutions for different situations. I have heard you mention your "neutralization" technique. I've heard of that for narrowband RF amps, but not audio. Congrats on a really nice solution there - though I admit I don't know what the circuit looks like.
Sounds like this could be a problem for a circuit without feedback.
Bob Cordell said:
It's simply not possible. The numbers don't mean squat. See below for more info.
GRollins said:
Grey, I can save you a fair amount of time. By far the best sounding resistor for the money was the Roederstein. In quantity, they cost around $0.01 each. They sound better than the Dales or any other resistor less that $5 each, with one exception. By the way, they used to be made at the Draloric factory. Vishay closed down the Draloric factory recently after they bought Beyschlag. The parts made in the Beyschlag factory sound slightly different, and to my ears not quite as good. However they are still excellent and the best value.
The only reasonably priced resistor that sounds better are the PRP PR9372 audio grade resistors. They cost around $0.20 depending on the quantity. You can get them from Michael Percy and PartsConnexion. Here's the story on PRP resistors:
Over 10 years ago Jeff Rowland recommended them to me. I was interested because they are completely non-magnetic, while the Roedersteins have steel end caps. (I still have no idea how they can get those resistors to sound that good with steel end caps.) I ordered some samples, and while they did some things quite well, overall the Roedersteins were better.
When I told the PRP people this, they were intrigued and asked if I would do some listening tests for them. I agreed and they sent me ten sets of resistors that were identifiable by only a single color coding band on each resistor. Each had some difference in the construction details (or maybe not -- they could have put in some identical "ringers" to see if they could fool me), but they did not tell me anything about what the differences were (even to this day).
I spent about ten days listening to the resistors and taking notes. I then summarized my findings and sent them to the folks at PRP. (Only one of the variations sounded better overall than the Roedersteins. Several of the variations were so similar that the differences were insignificant.)
I can only speculate, as I was not privy to the internal dialogs at PRP. I would assume that they thought I was crazy when I told them I would do listening tests. But they must have found some sort of internal consistency between my report regarding the sound and the actual differences in construction. In other words, if several resistors sounded very similar and the only difference was the grade of ceramic core, then they would surmise that the grade of ceramic core was not an important factor. Or if there were several resistors that varied in the composition of the metal film and I reported large differences in these, they must have concluded that the composition of the metal film is very important. (I have no idea what the true differences were or which ones are important -- I am just making up examples to illustrate the point.)
At any rate, they put enough credibility in my findings to then produce their special "audio grade" resistor. And it sounds pretty darned good.
There are two "take home" lessons to be learned here:
a) Grey, if you want to save a lot of time and money, just use either the Roederstein or PRP resistors. Anything that might sound better is going to be silly expensive.
b) Bob, it really doesn't matter which factors affect the sound of those metal film resistors or what the exact mechanism is at work in affecting the sound. The only thing that matters is knowing which ones sound good and which ones don't. And I can guarantee you that there are no specs to be found on any datasheet in the world that will tell you that. It's kind of like gravity. It has many practical applications, but nobody has the slightest idea how it actually works.
PMA said:
Agreed, although it sounds noticeably less good than our discrete circuit from five years ago, which in turn is significantly less good sounding than our latest discrete circuit.
KSTR said:
That is the basic "trick", although there are some strange behaviors that can crop up. You just have to play around with it and make sure it is really doing what you want it to do. After all, it was not designed to be used with a load at the Tz pin. Barrie Gilbert was fairly horrified that I was doing so. (He didn't believe me when I told him it sounded better this way.)
I don't use the "super-symmetric" trick because it is another form of feedback. Besides, Nelson has a patent on it and if you are interested in using it you would have to license it from him. (I think TI announced something to that effect over a year ago, but have not seen any actual products released yet.)
SY said:
Ah, if only the world were so simple...
But I guess if it were that easy, then the Japanese majors would be building $199 receivers that sounded every bit as good as the best high-end gear.
Edit: Wait a minute! The $199 receivers *measure* better than most high-end gear -- so why is it that they don't *sound* better?
Oh, that's where we came in on this thread, isn't it?
Charles Hansen said:
At the risk of sounding like a parrot, woo-hoo & thanks! After reading the adjectives used in the complete slagging these resistors took on another board - colourless, lacking life and air (metrics usually associated with added 'spice') - I had to try them and was delighted with the results. Better than Mills to my ear/application.
Keep arguing guys.
- Status
- Not open for further replies.