Keep the follower, change to fets. This will be better. Next, for advanced students, figure out how to direct couple, reduce parts and reduce distortion.
Take a look at http://www.borbelyaudio.com/index15.htm. I think he even has an example of one of John Curls followers. Very simple, low parts count. I'm going to build some of these and have a listen. I want to use some of these as followers for active crossovers. I tried a constant current source version and the output impedance was about 40 ohms and the JFET was biased at about 10 mA.
BDP
BDP
Well, if you can accept a slightly higher output Z (like an extra hundred ohms or so), you can do that with two fets and two resistors- but they have to be well-matched fets with tight thermal coupling. And well-matched resistors, but that's easier.
Re: Re: Discrete and opamp chips and all that...
Boy! I seem to have started something here! 🙂
Per-Anders, will you please try to give me some inputs about my diamond buffer vs ultra-linear VAS circuit question? I'm going round and round in circles on that one, ever since I saw those (very interesting) threads on diamond buffers. Even any vague pointers, suspicions, hints, will do.
Tarun
Boy! I seem to have started something here! 🙂
Totally true. In fact, if we use the word "devices", to mean individual R, L, C, diodes and transistors, most ICs will make the device count go through the roof, I think.peranders said:
Per-Anders, will you please try to give me some inputs about my diamond buffer vs ultra-linear VAS circuit question? I'm going round and round in circles on that one, ever since I saw those (very interesting) threads on diamond buffers. Even any vague pointers, suspicions, hints, will do.
Tarun
One very good reason is: It's more fun with discrete electronics! (not neccessarily better).tcpip said:
Tarun, sit down take a deep breath. Then start to explore the alternatives. The first suggestion of a simple emitter follower isn't bad, remember that.
I suggest also that you fool around with LTSpice which is a freeware. Since a couple of us use this software (including myself!) we can discuss things both at simulation level and in real life circuits.
AKSA said:
Tarun,
How would you rate this advice against the other advices saying -use a diamond buffer, that's great, or -use such and such a component, that's the best, or -simulate it, that will open your world? Remember, do your own thinking.
Jan Didden
BDP said:
You added a period "." to end your sentence and it glitched your link but yes, John Curl is mentioned in Borbely's article here:
http://www.borbelyaudio.com/ae699bor.pdf (bottom of the fourth page, numbered 19).
This is in regards to using complementary JFETs (2sk170/2sj74) for a source follower.
JF
My follower is too complex for this application. It is best to use example C in fig 15 on p 18. 10 ohm resistors will work better than 22ohms. Trust me.
Can you enlighten me, please? I may have seem it somewhere but can't remember really.john curl said:
Personally, Jan, I think Hugh's advice is very sane.janneman said:
I'd only add one rider to this listen-and-you-shall-know approach. I believe our brain can fool us into believing that we're hearing something which we aren't. Therefore, if I have to draw conclusions by listening, I'll probably only trust blind tests. This is the only reason that I'm yet to buy caps costing fifty dollars, even after such glowing reports in the forums. Having said that, I have no problem trying Hugh's suggestions (i.e. actually build and listen), provided both alternatives are inexpensive.
Regarding the "simpler is better" point of view, I guess all of us know that there is such a thing as "too simple." I work with software, Unix OS, and networks, and I live with simplicity-versus-power tradeoffs every day. In electronic circuits, I guess the same choices need to be made. My two-transistor buffers may be good enough for some applications, but too simple for very demanding systems. For similar reasons, the 16-bit resolution of CDs may be adequate and 24-bit digital audio may be more of undetectable improvement and hype. (Sampling rate is a different story altogether. 44KHz???? What were they thinking?)
Behind all this is something I firmly believe in, which I first read in the Great Linkwitz' site. He says audio reproduction is not like wine or tea tasting; it's not totally subjective. Since it aims merely at reproduction of an original, we can (and should) have objective measures of assessing the "goodness" of an audio reproduction system by comparing its reproduced output with the original. Only if two systems behave equally closely on objective scales should subjective preferences come in. Therefore, I would not bother with some "nice-sounding" super-duper platinum-plated amp if it didn't have very low measured higher-order harmonic distortion, irrespective of how nice it sounded. This is the primary reason I prefer solid-state amps to most valve amps... most valve amps have audible levels of distortion. Maybe, with time, I'll learn to place more value on parameters other than closeness-to-original, but for the time being, this is how I feel.
It's actually a little strange to see people opt for high-distortion valve amps, and then eschew active preamps because they want "clean, untainted" sound. 🙂 If that's what they like, it's okay with me, but I can't understand the rationale. Similarly, it's strange to see people fight over that last digit of distortion in opamps, and then, without demur, use speakers which generate 1% or more distortion.
To come back to the topic of my simple buffer... I will try listening to this circuit, and compare it with an NE5532-based buffer, and actually see what I can hear, within the limits of my ears, brain, and music system. Since the components are inexpensive, it should be fun trying. 🙂 I know my friend Angshu tried replacing TL072 with OPA2134 in a lovely little phono preamp he built, and heard no difference. I guess people like Angshu and me are simply the audiophile-equivalent of tone-deaf (can I invent a new term? "component-deaf"??), but it's always interesting to experiment. 🙂
And then, if my PCB-making skills prove adequate, I'll probably make a PCB of Randy Slone's super-duper input+VAS circuit, and listen to that too. 🙂
I keep remembering SL's Website, where he takes all this trouble to design and build the Phoenix and Orion, and then says you can drive them with an array of LM3886 poweramps. I wonder whether he can hear the difference between OPA627 and AD797? 😀
Bottom line: "whatever floats your boat." 🙂
Hope my shooting my keyboards off made sense to at least some of you.
Tarun
Totally agree! 🙂 That's the reason I started fiddling with this two-transistor thingie in the first place... somehow seemed that there must be easier ways to create a simple unity-gain buffer than throwing a whole opamp at it. Masochism?? Perhaps! 😀peranders said:
Hey, I'm calm enough, and sitting down too. 😀 I really didn't realise, when I first started this thread, that some of you would consider this topic to be something which needs deep breaths... 🙂
I will certainly try this. Do any of you know whether there's anything on Linux?
Tarun
Tarun,
I agree with almost everything of your previous post (isn't it boring when people keep on agreeing with you??).
There often is this standing on two legs: it sounds good, but we would even feel better if it also measured good.
I think that most people, some more than others, need some kind of external confirmation that they are right if they find something that feel or sounds good. The measurements can provide that external confirmation. Another factor known to provide this external confirmation is the price you pay. It's so expensive, it MUST be good.
Let me tell you, when I started digging into what really motivates people, it scared the wits out of me. Including my own motivation.
But that's for an entirely different forum.
Jan Didden
I agree with almost everything of your previous post (isn't it boring when people keep on agreeing with you??).
There often is this standing on two legs: it sounds good, but we would even feel better if it also measured good.
I think that most people, some more than others, need some kind of external confirmation that they are right if they find something that feel or sounds good. The measurements can provide that external confirmation. Another factor known to provide this external confirmation is the price you pay. It's so expensive, it MUST be good.
Let me tell you, when I started digging into what really motivates people, it scared the wits out of me. Including my own motivation.
But that's for an entirely different forum.
Jan Didden
tcpip said:
Serch the net for "Linux" and "Spice" and you will find a lot. The
official Spice from Berkeley runs on Unix AFAIK, so I suspect the
Linux versions are based on this original core.
BTW, maybe I shouldn't give you more options to choose from, but
it seems to me that so far the discussion has only been about
either using an op amp as buffer or using a discrete buffer of
one type or another. There is also the common solution of using
an op amp plus a discrete buffer, with the buffer within the feedback
loop. That is even more complex, of course, so it counts as a variant
on the op amp solution.
On spicy Unix and opamp'd buffers
I'll add some more... Randy Slone says that his input+VAS circuits are essentially opamps, with two inputs and one output. So his circuits, which I was referring to as examples of discrete buffers earlier, are actually opamps. 🙂 In fact, I've been tempted to build a preamp and tone controls using his "opamp" circuits. Maybe I will, someday. 18 transistors for a mono tone control circuit... that will be something. 😀
And then Walt Jung's buffer is actually a sort of output stage grafted on to an opamp, to get specially low Zout. And there are a lot of power amp designs which have an opamp acting as input+VAS, with discrete power transistors as OPS, the whole thing tied together with GNFB. I guess that would be a beefed-up version of your buffer?
I remember a lovely piece about computer architecture somewhere, where the author argued that the divide between hardware and software is totally artificial and a matter of convenience and commercial issues. You can convert an entire business application into microcoded hardware, or implement a microprocessor using a few logic gates and lots of software. When I see this discrete-and-opamp thing, I remember that article. Finally, the only thing that matters is the resulting electrical circuit behaviour, right?
Tarun
Thanks for the pointer, but I was hoping for an easy way out, asking. 🙂 I know Spice, like a lot of American university developments, was first done on Unix, usually resulting in good, portable C code. Since I'm a product of an academic environment where we used Unix heavily, and I continue to use Unix at my workplace, I love this class of software. However, often such stuff is weak on interactive GUI, so I was wondering whether the GUI schematic capture tools of the Unix versions would be as powerful as the newer Windows-based stuff.Christer said:
Join the Confusion-For-Fun-and-Profit Club. 😀
I'll add some more... Randy Slone says that his input+VAS circuits are essentially opamps, with two inputs and one output. So his circuits, which I was referring to as examples of discrete buffers earlier, are actually opamps. 🙂 In fact, I've been tempted to build a preamp and tone controls using his "opamp" circuits. Maybe I will, someday. 18 transistors for a mono tone control circuit... that will be something. 😀
And then Walt Jung's buffer is actually a sort of output stage grafted on to an opamp, to get specially low Zout. And there are a lot of power amp designs which have an opamp acting as input+VAS, with discrete power transistors as OPS, the whole thing tied together with GNFB. I guess that would be a beefed-up version of your buffer?
I remember a lovely piece about computer architecture somewhere, where the author argued that the divide between hardware and software is totally artificial and a matter of convenience and commercial issues. You can convert an entire business application into microcoded hardware, or implement a microprocessor using a few logic gates and lots of software. When I see this discrete-and-opamp thing, I remember that article. Finally, the only thing that matters is the resulting electrical circuit behaviour, right?
Tarun
tcpip said:
LTSpice runs under WINE. See the following thread in the sci.electronics.cad newsgroup: http://www.google.com/groups?hl=en&...b4nl2f%24i7k%40dispatch.concentric.net&rnum=1
No, I could really get used to a bit more of this. Please keep it up. 😀janneman said:
And continuing an earlier thread of "simpler is better, provided it's not too simple", one constant sanity-check I have is the Nak ZX-9 that I have at home. After I was seduced by its unbelievable sound into buying a used ZX-9 a couple of years ago, I decided to read more about it, and about Naks in general. As you know, a good Nak can extract unheard-of performance from the cassette medium. The more I read about Naks, the more I realised how much complexity has been added by its designers just to extract that last 10% performance. Since then, I always remember that that "fantastic sounding" system may be a product of huge complexity, not minimalistic simplicity at all. Sometimes I suspect that some of the people (not all, just some) who believe in "simpler is better-sounding" are in the same class as people who believe in a mythical Olden Days when Life was Better (conveniently forgetting how many died of typhoid, cholera, and pneumonia). A balance between simplicity and superior performance is difficult, and ruthless listening tests, by listeners who did not build the systems, are probably the only way out.
This is affected by the academic backgrounds of the builders, I suspect. Those with traditional engineering qualifications will probably always want to measure performance using some standard measurement techniques, and those with more of an arts or music background will probably trust their ears more. At least that's what I find among the few audio-loving friends I have.
Beautifully put. I totally agree.
You forgot brand hype! I have four friends with Bose Acoustimass systems, three with Acoustimass 5s and one with Acoustimass 10 (the 5.1 channel version). You know what's worse? I know more about the strengths of these Bose speakers than they do, even though I'd never buy one. Such is the power of branding... it can make you spend money without even teaching you what's good about itself.
🙂 I think we can swap stories on this one. 🙂
Thanks for everything,
Tarun
Comments invited elsewhere... please???
Dear all,
Can I please shamelessly ask for your comments on another thread I've posted, please? No one has commented on it all since it was created, and any help would be greatly useful. The thread is about my proposed remote volume control using motor pots. I've posted it here.
Thanks for any comments, however trivial.
Tarun
Dear all,
Can I please shamelessly ask for your comments on another thread I've posted, please? No one has commented on it all since it was created, and any help would be greatly useful. The thread is about my proposed remote volume control using motor pots. I've posted it here.
Thanks for any comments, however trivial.
Tarun
Another Buffer
A plagiarized version of the LH0063C 'ultra damn fast buffer'.
http://cache.national.com/ds/LH/LH0033.pdf.
A plagiarized version of the LH0063C 'ultra damn fast buffer'.
http://cache.national.com/ds/LH/LH0033.pdf.
Attachments
Tarun,
In all human endeavour, there is a politically correct solution. We are herd creatures, so we are powerfully driven to follow this solution. We join a team, we strive to 'fit in'. This is fine and dandy if you are in a team.
However, this is DIY, for yourself, and not for others necessarily. You will try to create something with your imprimatur, which sounds good to you, and suits your musical choices.
Who was it who said that progress depends on unreasonable men? He was probably trying to say you should listen to your own ears, your own ideas, and your own preferences, while keeping an open mind for the ideas of others. You are fortunate that people like Nelson, John C., Jan Didden and PMA are here; they will give you wonderful choices, but it's a fact that only dead fish go with the flow.......
Cheers,
Hugh
In all human endeavour, there is a politically correct solution. We are herd creatures, so we are powerfully driven to follow this solution. We join a team, we strive to 'fit in'. This is fine and dandy if you are in a team.
However, this is DIY, for yourself, and not for others necessarily. You will try to create something with your imprimatur, which sounds good to you, and suits your musical choices.
Who was it who said that progress depends on unreasonable men? He was probably trying to say you should listen to your own ears, your own ideas, and your own preferences, while keeping an open mind for the ideas of others. You are fortunate that people like Nelson, John C., Jan Didden and PMA are here; they will give you wonderful choices, but it's a fact that only dead fish go with the flow.......
Cheers,
Hugh
tcpip:
Re: running LTSpice under linux, have a look at this: http://www.brorson.com/gEDA/SPICE/intro.html
I rarely run linux these days so I haven't tried it out myself yet, but this guy recommends using gschem as a schematic capture front-end for LTSpice.
Enjoy!
Re: running LTSpice under linux, have a look at this: http://www.brorson.com/gEDA/SPICE/intro.html
I rarely run linux these days so I haven't tried it out myself yet, but this guy recommends using gschem as a schematic capture front-end for LTSpice.
Enjoy!
- Status
- Not open for further replies.