Hi!
I completely agree that CD is somewhat cave technology when we have GB+ storage media today, but this is a "Blowtorch preamplifier" thread.
I suggest to open a new thread to discuss "CD ripping" there and let John & other to continue all the preamplifier fun stuff
I would really like to proceed where ended, "we" have H topology "done" and I am very curious what come next... there is a low-distorted current signal and it has to be transformed back to a voltage... cascoded mirrors perhaps?
Cheers,
Matej
I completely agree that CD is somewhat cave technology when we have GB+ storage media today, but this is a "Blowtorch preamplifier" thread.
I suggest to open a new thread to discuss "CD ripping" there and let John & other to continue all the preamplifier fun stuff
I would really like to proceed where ended, "we" have H topology "done" and I am very curious what come next... there is a low-distorted current signal and it has to be transformed back to a voltage... cascoded mirrors perhaps?
Cheers,
Matej
Off topic discussion on format merits and digital ripping moved to digital sources here:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=121740
http://www.diyaudio.com/forums/showthread.php?s=&threadid=121740
Jan, your question on circuit breaker protection should be addressed. Yes, for many amps, quality dual DC circuit breakers will work in the DC supply. The original JC-3, the Gale 250W/ch, the Symmetry 100W/ch, and the VMPS 250W/ch (with ring emitters) worked well with the correct type. However, I had failures with VFet and Mosfet output stages, even well within their nominal design limits. Something certainly was different.
PMA, I do have your model running on Nelson's PC. Now, I have still to relearn the commands, since it is so different from the old Microcap program that I am used to. However, it looks good and I hope to cut and try simple circuit topologies. I am also looking hard for the analytical comparision of bipolar and fet differential circuits published in one of the IEEE publications that I subscribed to for many years. It is either in 'Solid State Circuits' or 'Circuits and Systems', or 'Instrumentation and Measurement'. It has been several years so I have lost track of it. It would be interesting to see if the complementary differential can be made to bypass some of the compromises found in the study.
Dear all,
When I read the discussions I took from them that the bandwidth was too low for the amount of NFB.
Meaning, if I wanted 40dB of NFB I could start with, say 250khz BW, and add 100x BW for the 40db of desired NFB meaning 25MHz BW would be appropriate. Not practical, but it's how I interpreted things.
I wonder how that would work.
Is the issue NFB or BW?
myhrrhleine said:
To me they are separate and distinct. Negative feedback is the traditional way to define the performance of a circuit. Set the ratio of two resistors and you're done...what could be simpler? Unfortunately, in the There's No Such Thing As A Free Lunch sense, for all the good things that negative feedback provides, it also brings unwelcome traits as well, such as reduced ambience and distortion products that many believe represent "detail." All it takes is familiarity with the sound of real music to know that's not true.
To me the price to be paid for the benefits just isn't worth it and I've been steadily decreasing the feedback I use over time.
Bandwidth is, or can be if you are careful in your design, a separate issue. I think most people today would agree that you need more than 20kHz bandwidth in the finished product. I don't know if anyone has done a formal survey of the bandwidth of all the preamps and amplifiers on the market, but if I had to guess, I'd say that most products tend to cluster around 100kHz or so. For most people, it's a decent compromise between wide bandwidth, stability (in terms of NFB), and staying out of RF trouble.
But...
I used to play with tubes. I forsook the tube world and came over to solid state, gravitating to FETs because bipolar base current just doesn't feel right to me (in tubes, if you're drawing grid current, you're courting disaster). I very soon discovered that I was having trouble getting solid state circuits to sound as open as my main tube amps (ca. 10dB NFB, ca. 130kHz bandwidth). A little more experimentation led me to conclude that--for reasons unknown--FET circuits require a minimum of 1.5 to 2 times the bandwidth to sound as relaxed and open as tube circuits. Charles Hansen recently posted a similar observation; he feels that bipolar circuits require more bandwidth to sound as open as FETs. I have not spent enough time with bipolar circuits to have an opinion, but if you put his and my observations together, you end up with a series of increasing bandwidths: tubes, FETs, bipolars, to get more-or-less equivalent sound.
To me, NFB and feedback are separate and lead to separate results, sonically. The trick is to get others to view things that way. Why? Because so many people design solid state circuits to have, say, 100Hz or 1kHz open loop bandwidth, then open up the bandwidth using negative feedback. To people accustomed to thinking this way, the two are inseparable.
Fortunately, they are very much separable. I recently started a thread on an amplifier I called the GR-25, which has 0dB feedback and 500kHz bandwidth. Just for fun, I threw in a 0dB feedback, wide bandwidth line stage, as well. If anyone feels that they just can't live without feedback, I left a couple of signposts pointing the way. The bandwidth would be easy enough to trim if you wanted to do so. I drew the schematic (and for that matter, made up the circuit boards) so I'd be able to put in some caps if I had RF or something. I don't, so I'm not using them, but in theory you could put caps in any number of places in the circuit and try bandwidth-limiting the circuit down to lower figures.
There's at least one potential fly in the ointment. I will pose it as a thought question: Is bandwidth limited by capacitors the same as bandwidth limited by the active devices themselves?
Grey
myhrrhleine said:
Hmmm. I thought I hit that hard enough...perhaps not.
There are a thousand posts on this site about the negative aspects of high rates of negative feedback. On one hand, it increases bandwidth, reduces THD, and lowers noise. So far, so good. Some people believe that it's a Free Lunch sort of thing--that you get all that for nothing. Ain't so. If you use a lot of feedback, your karma gets balanced by a forward upper midrange/lower treble and a loss of the lower level detail stuff like hall ambience and the tiny details that tend to get overlooked by people who don't listen carefully.
(Given that about a half-million posts were just surgically removed from this thread, people who read this in the future will have no idea what lousy source devices are being used by many here who claim to care about sound quality. That quite a few people insist there's no difference to be heard in a lot of these circuits is unsurprising, given MP3 files and the like used as listening material. It may be acceptable for parties and background listening, but it won't cut the mustard when it's time to do serious evaluation. If all you ever do is non-critical listening, then take your downloaded files and be happy. Do so with my blessing. Just don't come flaming others who use higher resolution source formats, afterwards. It won't wash.)
The wide bandwidth thing is a little harder to define. Some call it air, some call it ambience, some call it "relaxed" or "open." If you close your eyes and listen to your bedroom, then close your eyes the next time you're in a meeting room, you should be able to hear the difference in cubic volume. If you then go to a decent concert hall when it's empty and there's no music, you'll be able to hear that you're in an even larger space. As an analogy that still falls short, though. There are other aspects to the sound of a wide bandwidth circuit that don't quite equate to the size of a room. Part of it is the quality (not quantity) of the high frequency reproduction. Part of it is image. Part of it is...who knows what? For all the marvelous advances in measuring things, no one can measure even such a simple, basic, and obvious thing as image. The best you'll get from the measurements-are-everything crowd is a bunch of ill-tempered bluster about how NFB reduces all distortions so obviously it must make image better, too. The problem is, it doesn't. Or at least is doesn't when you design a decent enough amp for comparison. How the devil these people claim to know how their amps sound without feedback is something I can't fathom. So...what?...you're listening to an amp with 100Hz bandwidth without NFB, then applying feedback and saying that 100kHz sounds better than 100Hz? Now there's a shocker!
Given that feedback and bandwidth can be done independently--and have different sonic signatures--it's not an impossible comparison. Just difficult. Because you're no longer comparing apples to apples. The best you can do is listen to as many circuits as possible and eventually you can sort of triangulate the sound.
Or you could take a walk on the wild side and build a tube circuit and turn the feedback up and down with a knob. The sonic cost/benefit scenarios are the same in a tube circuit as they are in solid state. The thing tubes have is stability. A lot of solid state pieces will go ape on you if you try such a stunt. Oscillations out the wazoo. It's not hard to build a tube circuit that will let you find out for yourself what feedback does to the sound. Nobody says you have to live with it afterwards. Just treat it as an experiment, then shelve the circuit when you've heard for yourself the tradeoffs involved in the use of feedback.
I can sit here and describe this stuff all night long. It doesn't matter to me whether you simply take my (and John's, and Charles's, and...) word for it or whether you go listen for yourself. An open loop amplifier and an amplifier with 50dB of NFB--both with 100kHz bandwidth--don't sound the same. An open loop amplifier and an amplifier with 50dB of NFB--both with the same THD--don't sound the same. Go listen.
Just not with a lowest common denominator source.
Bletch!
Grey
I appreciate and respect your input.GRollins said:
My basic question was whether sufficient OL bandwidth could compensate for NFB. Is the undesirable sound from a limited bandwidth.
Why is global NFB different from emitter degeneration?
My only guess is BW/phase.
It seemed wise to ask.
Mr. Curl has tested somewhat.
To my ears, no open loop bandwidth does not compensate for the things NFB does wrong. Now, John's view may be different from mine because he seems to feel that wider bandwidth does to some degree mitigate the effects of feedback. Or at least he used to, judging from some of the TIM-related things he's said.
Is degeneration feedback?
There have been numerous threads here at DIY where people argued the semantics of this for pages and pages...it gets tedious. I used to feel that it was feedback but am persuaded that it isn't because of two things:
1) There's no back to feed to. It's right there in real time. No time delay, no nothing. It's all forward (without trying to imply that it's some kind of forward error correction, either).
2) Others have made this point, but the version that stuck in my mind came from Charles Hansen. In any active device, there is impedance in the cathode/Source/emitter that is intrinsic to the device. You can't get rid of it. It's glued in there by the basic laws of the universe. Do you count that as feedback? Most people don't, no matter how hidebound they may be about other things. So what's the moral difference between the impedance that's native to the device and a little pure resistance that you add on the outside? Well, none, really. If you had some way that you could reach up inside a device and twist a knob to adjust that impedance depending on the needs of a particular circuit, you wouldn't hesitate to do so, would you? And some of the time, you'd be turning it up. So we turn it up on the outside. Same difference.
Other feel differently, of course, but if this turns into a big semantic argument just because you asked a loaded question, I'm going to put a hex on your cat.
Grey
Is degeneration feedback?
There have been numerous threads here at DIY where people argued the semantics of this for pages and pages...it gets tedious. I used to feel that it was feedback but am persuaded that it isn't because of two things:
1) There's no back to feed to. It's right there in real time. No time delay, no nothing. It's all forward (without trying to imply that it's some kind of forward error correction, either).
2) Others have made this point, but the version that stuck in my mind came from Charles Hansen. In any active device, there is impedance in the cathode/Source/emitter that is intrinsic to the device. You can't get rid of it. It's glued in there by the basic laws of the universe. Do you count that as feedback? Most people don't, no matter how hidebound they may be about other things. So what's the moral difference between the impedance that's native to the device and a little pure resistance that you add on the outside? Well, none, really. If you had some way that you could reach up inside a device and twist a knob to adjust that impedance depending on the needs of a particular circuit, you wouldn't hesitate to do so, would you? And some of the time, you'd be turning it up. So we turn it up on the outside. Same difference.
Other feel differently, of course, but if this turns into a big semantic argument just because you asked a loaded question, I'm going to put a hex on your cat.
Grey
- Status
- Not open for further replies.