Hi, i mean just use two green LEDs to ground in place of R2. Each contributes about 1.9v, so the output sits at 1.9+1.9+1.25v = 5.00 near as dammit.
Each green led represents about 10-20ohm impedance (depending on how much current you pour through via R1) so the result is really rather 'low noise' - note LEDS are forward-biased PN junctions, not a breakdown mechanism like zener/avalanche diodes.
Each green led represents about 10-20ohm impedance (depending on how much current you pour through via R1) so the result is really rather 'low noise' - note LEDS are forward-biased PN junctions, not a breakdown mechanism like zener/avalanche diodes.
Okay, gottcha.martin clark said:
About LEDs...lots of DIYers use them in various ckts and recommend them generically -- "green LED", "red LED" -- but not specifically in terms of type/brand/etc. I've found that type/brand/etc. can sometimes make a difference. To that end, do you recommend any specific type/brand/etc "green LED"?
Thx
-hm
There is an interesting thread about LED and Zener's noise.
Some noise measurements for LEDs and zener diodes
Green LED lower noise is not so clear.
Eric
Some noise measurements for LEDs and zener diodes
Green LED lower noise is not so clear.
Eric
No, that's true enough, but here they produce a convenient forward voltage and have very low dynamic impedance (10-20ohms I find, vs. 10-20x that for a true zener at reasonable bias currents). Those ar the reasons I use them.
Anytime you want a low (even lower)-noise reference, it's usually trivial to add some RC filtering.
Anytime you want a low (even lower)-noise reference, it's usually trivial to add some RC filtering.
Super Ultra-Bright LED Lamp
Here's what I use. I used this same LED for a Flea clock regulator; it seems to work just fine.
Eric Juaneda said:
Here's what I use. I used this same LED for a Flea clock regulator; it seems to work just fine.
An externally hosted image should be here but it was not working when we last tested it.
Re: Listening LM317 filtering
The info on this page, to me anyway, is somewhat confusing...
The "First listening test" seems to indicate (AFAICS) that having no Cout and no Cadj is best for subjective sound. This despite the o'scope noise you noted on this page.
Fast-forwarding to the end of the article the following note and schematic comes out of nowhere (i.e., no description, analysis, raison d'etre, etc.) ...
"Here is an example of good subjective sounding solution."
The use of Zeners in the R2 position is similar to Martin's suggestion, so I'm assuming you used the same reasoning. Are subjective tests to follow for this ckt?
Eric Juaneda said:
The info on this page, to me anyway, is somewhat confusing...
The "First listening test" seems to indicate (AFAICS) that having no Cout and no Cadj is best for subjective sound. This despite the o'scope noise you noted on this page.
Fast-forwarding to the end of the article the following note and schematic comes out of nowhere (i.e., no description, analysis, raison d'etre, etc.) ...
"Here is an example of good subjective sounding solution."
The use of Zeners in the R2 position is similar to Martin's suggestion, so I'm assuming you used the same reasoning. Are subjective tests to follow for this ckt?
I quote from the site:
"With regulator, we think at first noise rejection. The LM317 datasheets from National semiconductor gives decoupling solution. What about sound and musicality?"
Thus, the point wasn't "finding the best combination of caps, out of ANY combination"... But let's look on the "bright side": at least no "fantabulous" / overhyped caps were tested
(ie. BlackGates and other marketing-spawned stuff) 
"With regulator, we think at first noise rejection. The LM317 datasheets from National semiconductor gives decoupling solution. What about sound and musicality?"
Thus, the point wasn't "finding the best combination of caps, out of ANY combination"... But let's look on the "bright side": at least no "fantabulous" / overhyped caps were tested
(ie. BlackGates and other marketing-spawned stuff) Khron said:I quote from the site:
"With regulator, we think at first noise rejection. The LM317 datasheets from National semiconductor gives decoupling solution. What about sound and musicality?"
Thus, the point wasn't "finding the best combination of caps, out of ANY combination"... But let's look on the "bright side": at least no "fantabulous" / overhyped caps were tested
OK. That's a 'plus', at least. ;-)
But I was mainly trying to make the point that since it's typically not a good idea to use low-ESR capacitors at the output of a three-terminal regulator, it seems like a 'listening comparison test' should at least include what might be the 'best case' scenario, at least if the main objective of the test is to find something that 'sounds better'. Or, maybe I misunderstood the objective. At any rate, I would like to have seen the results of a comparison between doing it 'the right way' and what I felt obligated to point out were probably all sub-optimal cases (about which I could be wrong, of course).
Ironically, maybe, If one reads the stuff at the other link, that was in a message nearby, i.e. at http://www.acoustica.org.uk/t/3pin_reg_notes1.html , and one then also follows the link, there, to http://www.tnt-audio.com/clinica/regulators2_impedance1_e.html , one can see an explanation of
why a low-ESR capacitor at the output of a typical three-terminal regulator is (typically, at least) 'a bad thing'. (I could be wrong, but that seems to me to have been corroborated by many other sources, including the regulator manufacturers, themselves.)
Eric (and all): I am sorry if it seemed like I was 'piling on', in a negative or inappropriate way. I do admit that I am biased, in the sense that I am less comfortable with subjective 'testing' than I am with engineering methods and measurements, which, I also admit, sometimes does cause me to have a 'less positive' attitude toward some of the information and ideas that I perceive to have come from the 'subjectivist' camp. (However, FWIW, I do actually believe that 'listener impressions' testing can, ideally at least, provide extremely-important and useful data. This is "audio", after all.)
(Not to put too fine a point on it, but) I probably should have crafted my message so that it would have been perceived more as an attempt to try to provide helpful or educational information, rather than as mere criticism. As an engineer, I usually welcome and appreciate blunt technical criticism, but very often forget that most other types of people might not be accustomed to that style of communication.
about article
Hi everyone,
Here is some response about previous questions.
I wrote my article to understand relation between regulator configuration and listening result. You see that problem is not simple. There is no configuration where you can say: This configuration is better then all other one. You can use graphic to improve one listening parameter at the expense of another one. The best you can obtain is an equilibrate configuration, not absolute one.
Theses graphs are not limited to LM317 regulator. Even with my model 4 preamplifier discrete open loop regulator, graphs stills true.
Along some threads, you can read that this solution sounds better than this one. Problem is not so easy.
Eric
Hi everyone,
Here is some response about previous questions.
I wrote my article to understand relation between regulator configuration and listening result. You see that problem is not simple. There is no configuration where you can say: This configuration is better then all other one. You can use graphic to improve one listening parameter at the expense of another one. The best you can obtain is an equilibrate configuration, not absolute one.
Theses graphs are not limited to LM317 regulator. Even with my model 4 preamplifier discrete open loop regulator, graphs stills true.
Along some threads, you can read that this solution sounds better than this one. Problem is not so easy.
Eric
For better sound, you can shunt LM317. In my model 3 preamplifier I used resistor. Actually I use FET (BF245C) as current generator. Rejection level is less degraded with FET than resistor.
Eric
Eric
Attachments
Eric, I saw on your website you use a two-diode plus one transformer primary coil as a first part of your psu's. I did not find any explanations for this setup, did you elaborate anywhere on this? If not, it would be quite interesting if you'd share your reasoning.
BTW, I like your style of questioning the persumed certainties...
Rüdiger
BTW, I like your style of questioning the persumed certainties...
Rüdiger
Hi Onvinyl,
Compared to a standard PSU using only C smoothing or PI filter CRC, it greatly improves sound. The serial inductor (24H + 50 Ohm) improves smoothing, like adding capacitors without defaults. On the inductor's output, tension is sinusoidal. That mean during half time, LM317 is supplied by inductor, and half time LM317 is supplied by capacitor. For a complete discussion about this see Shared Current Power Supply.
I implement this filtering on my model 4 preamplifier.
Eric
To perform my test I used this schema. It was not necessary for LM317 testing.Onvinyl said:
Compared to a standard PSU using only C smoothing or PI filter CRC, it greatly improves sound. The serial inductor (24H + 50 Ohm) improves smoothing, like adding capacitors without defaults. On the inductor's output, tension is sinusoidal. That mean during half time, LM317 is supplied by inductor, and half time LM317 is supplied by capacitor. For a complete discussion about this see Shared Current Power Supply.
I implement this filtering on my model 4 preamplifier.
Eric
Low C, Hi L P/Supplies
Hi Eric,
Have come recently to your website - interesting and also quite fascinating.
I can see the similarities to valve power supplies (ie. as seen on "TubeCad" site, etc) but initially, found it difficult to ignore the use of large C in supplies - the "Shared Current P/S" article is very helpful and seeing that 24H (not mH!) choke as a standard transformer makes it much less intimidating, although the absense of an o/p electro cap looks a bit "naked"!.
Traditional "mind set", I'm afraid - not easy to change.
As I'm in the middle of building a "Marchello" headamp, this is a good a time to try this out, although current is much higher at 250 mA/ch. I wonder if adding a simple Cmultiplier before the reg. to reduce hum and series Z will also reduce performance?
A silly idea came to mind - what would happen if (wait for this one!), if you were to use a little bit of voltage from one of those sec windings of that "choke" transformer to introduce a little bit of "anti-hum" ripple to the zener/transistor base of regulator cct (superimpose some out-of-phase ripple)?
A bit weird, but ....
Hi Eric,
Have come recently to your website - interesting and also quite fascinating.
I can see the similarities to valve power supplies (ie. as seen on "TubeCad" site, etc) but initially, found it difficult to ignore the use of large C in supplies - the "Shared Current P/S" article is very helpful and seeing that 24H (not mH!) choke as a standard transformer makes it much less intimidating, although the absense of an o/p electro cap looks a bit "naked"!.
Traditional "mind set", I'm afraid - not easy to change.
As I'm in the middle of building a "Marchello" headamp, this is a good a time to try this out, although current is much higher at 250 mA/ch. I wonder if adding a simple Cmultiplier before the reg. to reduce hum and series Z will also reduce performance?
A silly idea came to mind - what would happen if (wait for this one!), if you were to use a little bit of voltage from one of those sec windings of that "choke" transformer to introduce a little bit of "anti-hum" ripple to the zener/transistor base of regulator cct (superimpose some out-of-phase ripple)?
A bit weird, but ....
Re: Low C, Hi L P/Supplies
Hi jameshillj,
I design regulator like linear amplifier: constant tension on variable load. So, I don't think that putting high cap on output would be a good thing. Regulator's quality is certainly as important as linear stage. Everything you ear comes from power supply.
I think to another idea. You could put an unused secondary to an active stage to create a 'multi phased' tension. Inductor’s output current could be constant.
Eric
Hi jameshillj,
A good Cmultiplier is probably a good thing. The post regulator could hightly degrade general sound! What kind of regulator do you want to use?jameshillj said:
I design regulator like linear amplifier: constant tension on variable load. So, I don't think that putting high cap on output would be a good thing. Regulator's quality is certainly as important as linear stage. Everything you ear comes from power supply.
Interesting idea, it is not easy to implement.
I think to another idea. You could put an unused secondary to an active stage to create a 'multi phased' tension. Inductor’s output current could be constant.
Eric
Eric,
Your reg is a radical departure from the way I usually look at p.supplies and to understand them properly, it's better for me to build one and find out.
I generally use high current, low Z source, CCS + shunt systems with large C's with bipass caps (snubber units) - have had good success with simple, high voltage Cmultipliers (tubes).
I prefer to avoid most active, high feedback systems (super regs,etc) - I just don't seem able to get them to sound "right".
Your low Cap system is quite different to me, and quite intriguing.
I had in mind to build your pre3 reg for the headamp, as you have already done all the hard work, and I wonder if the higher current demand would cause any Cct changes.
It may also be "just the thing" to give the MOX based Xover a better result.
My thinking about the Cmultiplier was to use it replace one of the 100uF caps in the pre4 p/supply but, as you say, it's possibly just the wrong thing to do.
About your pre4 p.supply, where are all the missing components?
(Just "kidding" you!).
It looks to be, too simple!
.... and How's That for a REALLY Stupid thing to say!!
Those unused secondary windings DO look inviting, and there is certainly enough voltage to work with.
A phase compensated, servo driven, constant current choke, .... hmmm!
Look what you've started, and it's only Friday!
Actually, there appeared on the TubeCad site last year an inductor p.supply with feedforward/feedback regulation, and this also looks to be a too simple but extremely clever design, just like the Aikido system - a very smart mind, has John Broskie.
Thank you for your articles.
Your reg is a radical departure from the way I usually look at p.supplies and to understand them properly, it's better for me to build one and find out.
I generally use high current, low Z source, CCS + shunt systems with large C's with bipass caps (snubber units) - have had good success with simple, high voltage Cmultipliers (tubes).
I prefer to avoid most active, high feedback systems (super regs,etc) - I just don't seem able to get them to sound "right".
Your low Cap system is quite different to me, and quite intriguing.
I had in mind to build your pre3 reg for the headamp, as you have already done all the hard work, and I wonder if the higher current demand would cause any Cct changes.
It may also be "just the thing" to give the MOX based Xover a better result.
My thinking about the Cmultiplier was to use it replace one of the 100uF caps in the pre4 p/supply but, as you say, it's possibly just the wrong thing to do.
About your pre4 p.supply, where are all the missing components?
(Just "kidding" you!).
It looks to be, too simple!
.... and How's That for a REALLY Stupid thing to say!!
Those unused secondary windings DO look inviting, and there is certainly enough voltage to work with.
A phase compensated, servo driven, constant current choke, .... hmmm!
Look what you've started, and it's only Friday!
Actually, there appeared on the TubeCad site last year an inductor p.supply with feedforward/feedback regulation, and this also looks to be a too simple but extremely clever design, just like the Aikido system - a very smart mind, has John Broskie.
Thank you for your articles.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.