I would say the same . One day when I get equipment sensitive enough I will look further .
I think unfortunately the sound quality of circuits often have a faith verses truth logic . I heard recently that faith is philosophically a discipline above truth . Truth being that which is held to be true mostly . If so the statement is correct as better truth is sometimes available from those who had faith . Newton's proof of light spectrum was an excellent example . It is correct to ask if pretty colours seem when light passes through glass might be aberration . Newton found a simple if difficult to execute way to prove otherwise .
People often had wrong ideas about capacitor distortion . This set of tests saved me a fortune and recently help squeeze a deign into a small space . Anyone expressing opinion as fact before these tests would be equally wrong regardless of faith or fact . The real truth is the faith guys were mostly right . Biasing electrolytics turned out to be best avoided . Before it was assumed 1/3 to 1/2 maximum voltage would be ideal . < 0.4V turns out to be ideal .Non polar electrolytics almost as good as Audiophile types . With a little caution ( money ) what harm does two resistors make ? At worse it must be the same ?
http://www.waynekirkwood.com/Images/pdf/Cyril_Bateman/Bateman_Notes_Cap_Sound_1.pdf
I think unfortunately the sound quality of circuits often have a faith verses truth logic . I heard recently that faith is philosophically a discipline above truth . Truth being that which is held to be true mostly . If so the statement is correct as better truth is sometimes available from those who had faith . Newton's proof of light spectrum was an excellent example . It is correct to ask if pretty colours seem when light passes through glass might be aberration . Newton found a simple if difficult to execute way to prove otherwise .
People often had wrong ideas about capacitor distortion . This set of tests saved me a fortune and recently help squeeze a deign into a small space . Anyone expressing opinion as fact before these tests would be equally wrong regardless of faith or fact . The real truth is the faith guys were mostly right . Biasing electrolytics turned out to be best avoided . Before it was assumed 1/3 to 1/2 maximum voltage would be ideal . < 0.4V turns out to be ideal .Non polar electrolytics almost as good as Audiophile types . With a little caution ( money ) what harm does two resistors make ? At worse it must be the same ?
http://www.waynekirkwood.com/Images/pdf/Cyril_Bateman/Bateman_Notes_Cap_Sound_1.pdf
I have been giving some thought to resistor sound . I think anyone making commercial amplifiers would be a fool to consider it nonsense . Capacitor sound was considered important or nonsense depending on the perception of the person . Out of an attempt to understand it trends were seen that echoed perceptions . Amongst these test results were very cheap and excellent devices which I note still have no acceptance in the tweaky component world . Ceramic was said to be awful and most designers who use them do so secretly . Tantalum is a device from any point of view I would not use . Panasonic FC being a totally acceptable substitute of greater reliability .
One thing I dislike is a manufacturer who has a components list like an old war veteran with medals . Soviet general perhaps ? An amplifier I designed now is full of such stuff . The person who did that counts it as perhaps 90% their design now . I just think it is very sad . The idea is to find the cheapest way to do an excellent job . I was told the current mirror I use is the most inspired one ever . Took me 2 minutes to find it in an op amp and I claim no credit for it . When bored I read internal op amp circuits .
DF you say CC resitors have poor linearity . It would be interesting to debate that . The usual mechanisms for that do not exist , as far as I know inductance and capacitance are low . It must by how they send the electrons along , diode effects ? . Personally I do not think they sound better for having faults . They seem to make a better compromise . Foil resistors I like . I can not afford to use them .
Formula 1 could be based on a 90% identical car . I would allow aerodynamic parts and cylinder heads / injection to be bespoke . Cylinder block and gearbox generic . It would allow even a small team to stand a chance . Whilst of the optimist school I would firmly side with DF96 if having to take sides . I hate nonsense even if I talk it myself . My nonsense is about saving money . I do not like the F1 overtones of hi fi , jewelry . I suspect India's Mars mission cost less than a middle rank F1 team for a years racing ?
One thing I dislike is a manufacturer who has a components list like an old war veteran with medals . Soviet general perhaps ? An amplifier I designed now is full of such stuff . The person who did that counts it as perhaps 90% their design now . I just think it is very sad . The idea is to find the cheapest way to do an excellent job . I was told the current mirror I use is the most inspired one ever . Took me 2 minutes to find it in an op amp and I claim no credit for it . When bored I read internal op amp circuits .
DF you say CC resitors have poor linearity . It would be interesting to debate that . The usual mechanisms for that do not exist , as far as I know inductance and capacitance are low . It must by how they send the electrons along , diode effects ? . Personally I do not think they sound better for having faults . They seem to make a better compromise . Foil resistors I like . I can not afford to use them .
Formula 1 could be based on a 90% identical car . I would allow aerodynamic parts and cylinder heads / injection to be bespoke . Cylinder block and gearbox generic . It would allow even a small team to stand a chance . Whilst of the optimist school I would firmly side with DF96 if having to take sides . I hate nonsense even if I talk it myself . My nonsense is about saving money . I do not like the F1 overtones of hi fi , jewelry . I suspect India's Mars mission cost less than a middle rank F1 team for a years racing ?
Inductance and capacitance are not "the usual mechanisms" for nonlinearity; both are generally linear!! CC are nonlinear because of their resistance. Note that there is no a priori reason why a resistor should be linear. Ohm's Law is not a law of nature, but a description of what (approximately) happens with most conductors. Given good conductors and mechanically stiff construction, there are good reasons why air-based inductors and capacitors are linear. Introducing materials adds nonlinearity. A resistor necessarily has materials so you need to choose materials which are ohmic and then ensure ohmic connections at the ends. There is no debate about CC nonlinearity - it is measurable.nigel pearson said:
We had teacher who would only allow plurals of the fundamental units as others were derived . I think Joules and Coulombs were OK . It was 40 years ago . The foil resistors fit your picture . A nice homogenous material which by diligence has the low inductance it champions.
Peter Platt the teacher also told me a very valuable thing . " You audio engineers always think of things being in the signal path . Kirchhoff says everything to a greater or lesser extent is " . Bless him and this was 40 years ago and he an RAF instructor . We gave him a breakdown which I am sure comes as no surprise . He resumed when well and even came down the pub with us after that . He found leaving the RAF too big a jump . I remember he called my ideas " red herring time " . He thought it was a diversion activity . One day he embarked on proving something about capacitors due to my red herring . Stopping in mid sentence he said " this is way above degree level " . He was angry for being duped ( not my intention ) . The RAF guys said continue and insisted it was easier to understand at the advanced level . The thing is they were time served engineers wanting a piece of paper . The deal was from then that red herrings were discussed if all wanted to do it . I think 70% were allowed . My greatest affection goes to the RAF guys who kept us in line , they were fascinated by electronics as much as anyone could be . I missed college and I missed the girls at college when I left .
Peter Platt the teacher also told me a very valuable thing . " You audio engineers always think of things being in the signal path . Kirchhoff says everything to a greater or lesser extent is " . Bless him and this was 40 years ago and he an RAF instructor . We gave him a breakdown which I am sure comes as no surprise . He resumed when well and even came down the pub with us after that . He found leaving the RAF too big a jump . I remember he called my ideas " red herring time " . He thought it was a diversion activity . One day he embarked on proving something about capacitors due to my red herring . Stopping in mid sentence he said " this is way above degree level " . He was angry for being duped ( not my intention ) . The RAF guys said continue and insisted it was easier to understand at the advanced level . The thing is they were time served engineers wanting a piece of paper . The deal was from then that red herrings were discussed if all wanted to do it . I think 70% were allowed . My greatest affection goes to the RAF guys who kept us in line , they were fascinated by electronics as much as anyone could be . I missed college and I missed the girls at college when I left .
I have no idea what plurals have got to do with it. "Ohm's" in "Ohm's Law" is a possessive, not a plural. Homogeneity and ohmic behaviour are largely unrelated.nigel pearson said:
'Signal path' nonsense is still being routinely trotted out by people who don't understand circuits. It is one of those concepts which can only be safely used by people who know enough to manage without it.
I am interested to hear that red herrings have been your stock in trade for years. You certainly seem experienced in serving them!
Red herrings are like holidays . Once in a while really worthwhile . Mostly interesting . Sometimes a mistake . I am far more conservative in my work . Audio is special and I allow a bit of speculation . It is a bit like art . We can buy paint , does it mean we can paint . We could say why bother when a photograph is always better . The rest is personal and when making a compromise a matter of taste . I like draftsmanship . I see the specification as exactly that . I think audio despite all efforts to make it photographic is still at best art . The box speaker almost guarantees that . The speakers I use have no box . Distortion similar to a good 1947 amplifier . People who visit say they dominate the room . To me they are they absolute minimum requirement . Any discussion about hi fi is invalid with most speakers .The new KEF LS 50 almost good enough if space is a problem . People who own Spendor BC1 might think they are in this club . Much as like them they do not sound like real music and the measurements say why . I have used the BC1 drive units outside the box and found they can work very well ( understatement ) . The box in question although one of the best still colours the sound drastically . In fact much like a strong version of a carbon composition resistor .
I love that song . These debates remind me of college . Often I knew I was out on a limb with something . The only way to learn was by risking a daft idea . Often being told off was the only way someone would share the true reason . How many times I was told to beleive something because it was in the syllabus . The higher level explanation often far easier to grasp . As some chap from the 1920's said in Hi Fi world . When he learnt electronics it was physics . As time went on it was fiddle factors to get things to work . I think by fiddle factors he meant no one was prepared to continue his education . It caused him to leave the industry . His letter to Mr Keywood was about having a voice again when SE valve amps .
At college I asked a simple question ( I was 17 ) to Peter who could teach above degree level if asked ( and he could ) . The question was about optimum capacitor size for a power supply . I had suggested 1000 uF . Peter did all the calculations and got to about 22 uF . He was greatly troubled by this as he knew as well as I did that 4700 uF would be easily within the possible options . Out of despair he called in Mr Mills who was passing . " You forget Perter there is no such thing as a perfect capacitor luckily for us " was his answer ( more to it than that , good enough for a 17 year old ) .
I was forced to do electrical engineering in addition to electronics . Mr Mills being from the electrical side . From then on I took more notice . We visited a power station as part of that and can say it was the high spot . The tour taken by wives of the workers ( Didcot , just closed alas ) . One interesting part was standing in one of the chimneys ( air intake ) . The lady said some people said being there made them feel happy . 5.7 Hz she said ( 40 years ago so as memory has it ) . One felt nothing obvious and heard even less . Like resistors in parallel effect I did feel happy . Who knows if I imagined it . Equally such frequencies are said to make people feel ghosts are around . Increasingly the mechanical reason taken to be the true one rather than the spirit world . High pressure gas pipes and wind-farms being typical causes .
At college I asked a simple question ( I was 17 ) to Peter who could teach above degree level if asked ( and he could ) . The question was about optimum capacitor size for a power supply . I had suggested 1000 uF . Peter did all the calculations and got to about 22 uF . He was greatly troubled by this as he knew as well as I did that 4700 uF would be easily within the possible options . Out of despair he called in Mr Mills who was passing . " You forget Perter there is no such thing as a perfect capacitor luckily for us " was his answer ( more to it than that , good enough for a 17 year old ) .
I was forced to do electrical engineering in addition to electronics . Mr Mills being from the electrical side . From then on I took more notice . We visited a power station as part of that and can say it was the high spot . The tour taken by wives of the workers ( Didcot , just closed alas ) . One interesting part was standing in one of the chimneys ( air intake ) . The lady said some people said being there made them feel happy . 5.7 Hz she said ( 40 years ago so as memory has it ) . One felt nothing obvious and heard even less . Like resistors in parallel effect I did feel happy . Who knows if I imagined it . Equally such frequencies are said to make people feel ghosts are around . Increasingly the mechanical reason taken to be the true one rather than the spirit world . High pressure gas pipes and wind-farms being typical causes .
Nige, I have a request.
I have attached two schematics below.
Number 1 shows the „Virtual Battery“ PSU as I first saw it in a schematic from Technics way back, say 12 or 13 years ago. So far, it has served me well when executed as shown. I generally use it for stabilizing power amp input and VAS stages, plus the predriver. This means its current requirements seldom exceed 50 mA worst case.
Your discussion on noise distribution got me thinking.
What if, as in 1, instead of using a single ttransistor, I used a driver and an output transistor, an in 2? Theoretically, this should lower my output impedance, which is a good thing, but I have not the faintest idea how this may impact noise and its distribution.
So, when you have the time, could you please compare these two both between themselves, and your existing database, and let me know how they stand? No hurry, when you can.
BTW, I was somewhat dismayed to discover that number 2 circuit was used by Dan d’Agostino of Krell in his early 90ies models. On the other hand, this was also good news, as it seems to indicate that at least I’m moving in the right direction – made mine just like he did, and he ain’t just anybody. As far as I am aware, no copyrights are being breached.
I have attached two schematics below.
Number 1 shows the „Virtual Battery“ PSU as I first saw it in a schematic from Technics way back, say 12 or 13 years ago. So far, it has served me well when executed as shown. I generally use it for stabilizing power amp input and VAS stages, plus the predriver. This means its current requirements seldom exceed 50 mA worst case.
Your discussion on noise distribution got me thinking.
What if, as in 1, instead of using a single ttransistor, I used a driver and an output transistor, an in 2? Theoretically, this should lower my output impedance, which is a good thing, but I have not the faintest idea how this may impact noise and its distribution.
So, when you have the time, could you please compare these two both between themselves, and your existing database, and let me know how they stand? No hurry, when you can.
BTW, I was somewhat dismayed to discover that number 2 circuit was used by Dan d’Agostino of Krell in his early 90ies models. On the other hand, this was also good news, as it seems to indicate that at least I’m moving in the right direction – made mine just like he did, and he ain’t just anybody. As far as I am aware, no copyrights are being breached.
Looks like a good pre-amp regulator ....
It is indeed, Wayne.
I always thought it had a better S/N ratio of all the samples I tried before, except for one shunt regulator I used and still use, however, that's a very complex, bulky and expensive regulator. I sometimes wonder if the difference in performance is worth the difference in price and trouble.
Actually, just this kind of reasoning, right or wrong, is what made me ask Nigel for a noise readout and comment.
And, from personal experience, coupling drawing 1 with AD's OP37 (Old School) or 829 (Newer School) will produce some uncanny quiet stages with excellent resolution, given what those op amps need in terms of current and the regulator's output capability.
That regulator will handle anything up to about 1.5 Amps, so anything below 100 mA seems like a joke to it. Almost doesn't even register.
That regulator will handle anything up to about 1.5 Amps, so anything below 100 mA seems like a joke to it. Almost doesn't even register.
Virtually identical to my recent design . I tend to use the Panasonic FC series caps that are cheap and need fewer . I probably would use a 10 uF in place of 220 uF across the zener or 1 uF polyester . If retaining the 220 uF I would use a 100 nF polyester in addition although I doubt it would show on test as 220 uF even of the lowest grade will filter 20 kHz + noise . If you find as I do the noise will be lower than a commercial regulator for very little extra soldering .
Now a question to all . We assume low impedance is required for pre- amp regulators . Make it too low and sometimes a hum loop is created . Naim for all of it's massive transformers had 27 R breaking the loop . In one pre amp I made I used two times 24R followed by 6800 uF to power a bunch of op amps . That fed by simple 7812's ( two in series like batteries and twin rectifiers ) . The idea being breaking the loop and converting to a simple RC filter . Whilst I understand why low impedance helps I don't imagine it is a prime requirement ? In fact I fed one pre amp from dual constant current sources and found it sounded very good . There were zener diodes set at the op amp limit to catch it as it started up . These were on LED's to indicate faults . The op amps themselves present a near constant load so using a current source will also set a voltage . Admittedly it wondered by about two volts as it settled . I choose +/- 12 volts as a good compromise .
I have often wondered if shunt regulators have an advantage .They certainly seem to have more difficulties if another thread at DIY Audio which exceeded 4000 entries is any guide . The zener shunt with transistor current amplifier seems to offer the best of both worlds with minimal complexity and little chance of instability . If a small amount of output voltage variation is allowable then it begs the question why do it any other way ? As I said before this Ugly Duckling actually has lower noise than commercial regulators ( > 10 dB if LM317 ) . There might be a small advantage using a Darlington transistor along with a small noise penalty . High gain , high Ft seems best . Here are two that set good benchmarks .
ZTX792A - DIODES INC. - TRANSISTOR, PNP, E-LINE | Farnell UK
ZTX853 - DIODES INC. - TRANSISTOR, NPN, E-LINE | Farnell UK
The similar capacitance multiplier is also a great idea . When a zener has a capacitor across it does both jobs . It also reduces the noise the zener makes ( prime function ) . Whilst it is anyone's guess as to at what frequency a regulator of this sort stops working effectively , the transistor Ft can be > 50 MHz and the zener plus capacitor must be much higher . I suspect when the transistor gain is nearing 10 it is still doing useful work . From what I know shunt regulators give up long before then ( 5 MHz ? ) . I suspect most op amps like it quiet into the upper MHz ? With PCB space often cramped these options are still possible .
BTW . Building 2 x NPN regulators connected in series is a possible option . 2 x bridge rectifiers as the only complication . The gain + ft is usually better with NPN . I recently connected 2 x LM 317 this way without problems . Possibly better than LM317/337 combination .
ZTX792A - DIODES INC. - TRANSISTOR, PNP, E-LINE | Farnell UK
ZTX853 - DIODES INC. - TRANSISTOR, NPN, E-LINE | Farnell UK
The similar capacitance multiplier is also a great idea . When a zener has a capacitor across it does both jobs . It also reduces the noise the zener makes ( prime function ) . Whilst it is anyone's guess as to at what frequency a regulator of this sort stops working effectively , the transistor Ft can be > 50 MHz and the zener plus capacitor must be much higher . I suspect when the transistor gain is nearing 10 it is still doing useful work . From what I know shunt regulators give up long before then ( 5 MHz ? ) . I suspect most op amps like it quiet into the upper MHz ? With PCB space often cramped these options are still possible .
BTW . Building 2 x NPN regulators connected in series is a possible option . 2 x bridge rectifiers as the only complication . The gain + ft is usually better with NPN . I recently connected 2 x LM 317 this way without problems . Possibly better than LM317/337 combination .
Here are some ideas to think about- One, use LM317/LM337 as pre-regulators/ short circuit protection. Reducing the ripple and noise going into a regulator can only help. Another, close tolerances for voltages are usually not necessary in these analog circuits. Stability against input and load changes can be important however. two regulators are always better than one shared regulator. Make sure your grounding doen't push the load currents back into the regulators in effect. If the load current puts a voltage drop across the ground between the system reference point and the power supply your not getting the full benefit of the regulation. This is where shunt regulators can help. The reference and the regulated node are clearly defined. Often less obvious is twhat happens with series regulators and the sensing points. All active regulators are effectively inductive at their outputs. No active device has constant gain to infinite frequency or anything near. Don't expect it to provide a low output Z at 1 MHz or above. To some degree its also true of caps. That is why some regulators oscillate with really low ESR caps.
Here is a simple circuit that can get the noise down to less than 1 nV/rtHz. http://www.diyaudio.com/forums/digi...b-interface-audio-widget-103.html#post2884358
Here is a simple circuit that can get the noise down to less than 1 nV/rtHz. http://www.diyaudio.com/forums/digi...b-interface-audio-widget-103.html#post2884358
This is the „short form“ shunt power supply which is in regular use since 2009, when my friend and associate Oli (Oliver) and I dreamt it up.
It is the end result of much experimenting with almost all parts, and especially the active devices, where quite a number of transistors, standard BJT, Darlingtons and MOSFETs, were experimented with. The version illustrated is the final form, as we both felt it provided the best overall results.
It is used in my headphone amp/preamp versions with BJTs and FETs/MOSFETs. While we claim the right to be honorably mentioned, feel free to use it at will for your own non-commercial use as you see fit.
We do not claim that this is the best there is – rather, we believe it is a versatile circuit primarily intended for preamps and headphone amps, which will strike a good balance between the very best there is and what can be made for reasonable money. For those who want the best bar none, I suggest you look up the Shunty – it is far nmore complex and expensive, but it was given thumbs up by both Nelson Pass and Erno Borbely, who both simply said what amounts to that’s it, don’t fiddle with it any more, just make it.
As for this regulator, one has to use headphones powered to the near maximum to hear any noise as such. Please note that for C3 and C6, rather than 2,200 uF, I use 4,700 uF/40V caps from Fisher & Tausche, although using Panasonic caps will produce no discernible difference.
Regarding resistors, I routinely use German made Beyschlag 1% metal film resistors, each rated for 0.6W, so I have no idea which can be used with smaller power dissipation rates, that’s up to you to work it out. As noted, R1 and R14, which define the current output, have to be worked out case by case. In my case, I runk the headphone amps in class AA, or „enriched“ class A, a fancy way of saying that if my maximum current requirements are say 40 mA, the circuits are bised for twice that, not a hard thing to do with headphones as they are relatively very efficient, and 1 mW usually yields well over 96 dB SPL, even at 30 Ohm impedances, or less.
Nevertheless, please do not forget to work out the actual dissipation of the two trannies, it may not be as small as you might think at first glance. Also include the two LM regulators, as they too will be dissipating not inconsequential power levels, again depending on how much current you want.
And by all means, do compare it with the usual power supplies. It won’t take long to make up your mind.
It is the end result of much experimenting with almost all parts, and especially the active devices, where quite a number of transistors, standard BJT, Darlingtons and MOSFETs, were experimented with. The version illustrated is the final form, as we both felt it provided the best overall results.
It is used in my headphone amp/preamp versions with BJTs and FETs/MOSFETs. While we claim the right to be honorably mentioned, feel free to use it at will for your own non-commercial use as you see fit.
We do not claim that this is the best there is – rather, we believe it is a versatile circuit primarily intended for preamps and headphone amps, which will strike a good balance between the very best there is and what can be made for reasonable money. For those who want the best bar none, I suggest you look up the Shunty – it is far nmore complex and expensive, but it was given thumbs up by both Nelson Pass and Erno Borbely, who both simply said what amounts to that’s it, don’t fiddle with it any more, just make it.
As for this regulator, one has to use headphones powered to the near maximum to hear any noise as such. Please note that for C3 and C6, rather than 2,200 uF, I use 4,700 uF/40V caps from Fisher & Tausche, although using Panasonic caps will produce no discernible difference.
Regarding resistors, I routinely use German made Beyschlag 1% metal film resistors, each rated for 0.6W, so I have no idea which can be used with smaller power dissipation rates, that’s up to you to work it out. As noted, R1 and R14, which define the current output, have to be worked out case by case. In my case, I runk the headphone amps in class AA, or „enriched“ class A, a fancy way of saying that if my maximum current requirements are say 40 mA, the circuits are bised for twice that, not a hard thing to do with headphones as they are relatively very efficient, and 1 mW usually yields well over 96 dB SPL, even at 30 Ohm impedances, or less.
Nevertheless, please do not forget to work out the actual dissipation of the two trannies, it may not be as small as you might think at first glance. Also include the two LM regulators, as they too will be dissipating not inconsequential power levels, again depending on how much current you want.
And by all means, do compare it with the usual power supplies. It won’t take long to make up your mind.
- Status
- Not open for further replies.