Re: Re: Re: Re: Improvements
I have tested this circuit with 4 BC337 and no resistors between the base-emitter. In room temperature the leakage is about 0,3 mA and if I warm up the transistors, boom, lots of current. This was with shorted base at the first transistor. If the base isn't shortened the transistors saturate of your shear presence. Real sesnsitve circuit when unconnected, different thing when connected though.
Seriously Elso, I don't see that any capacitance is multiplied. the 431 "ruins" this. I see only a normal (burt slow) shunt regulator with buffered output.
peranders said:
I have tested this circuit with 4 BC337 and no resistors between the base-emitter. In room temperature the leakage is about 0,3 mA and if I warm up the transistors, boom, lots of current. This was with shorted base at the first transistor. If the base isn't shortened the transistors saturate of your shear presence. Real sesnsitve circuit when unconnected, different thing when connected though.
Seriously Elso, I don't see that any capacitance is multiplied. the 431 "ruins" this. I see only a normal (burt slow) shunt regulator with buffered output.
Why etc
Peranders, I agree with you, this is not a cap multiplier. It would be if the 431 was replaced by a resistor, in such a way that the base of the first transistor is a fraction of the unregulated input and is "following" this input.
Also, there really should be resistors from the base of each transistor to its emitter or possibly another point to control the leakage. Look at it as the load resistor for the previous transistor: the voltage on this resistor becomes the input voltage for the next stage. If you leave it off you have no idea what is going on in the circuit and you might as well randomly throw in a handful of parts and hope it will work better.....
Jan Didden
Peranders, I agree with you, this is not a cap multiplier. It would be if the 431 was replaced by a resistor, in such a way that the base of the first transistor is a fraction of the unregulated input and is "following" this input.
Also, there really should be resistors from the base of each transistor to its emitter or possibly another point to control the leakage. Look at it as the load resistor for the previous transistor: the voltage on this resistor becomes the input voltage for the next stage. If you leave it off you have no idea what is going on in the circuit and you might as well randomly throw in a handful of parts and hope it will work better.....
Jan Didden
I think a Darlington with a current gain of > 1000 (maybe 10000-30000) is more than sufficient in order to buffering. The DAC current is maybe 50-100 mA (I don't know exactly) so the base current with a Darlington will be 10 µA or so.
I have seen "tripleton" in huge power transistors (> 300 A).
I have seen "tripleton" in huge power transistors (> 300 A).
Improved (?) Circuit
I did throw some parts together again in the reaction vessel and this came out. Couldn' t resist it. Improved circuit ? I dunno. Slight of thin in the bass.
This circuit also works a a pure <B>capacity multiplier</B> when you replace R7 by 2M (two 1M in series) and the LM329 replaced by 1M. [You do read carefully Peranders?] Yes now the voltage is a fraction of the raw supply voltage and follows the variations in the mains volatge. But this is also a <B>working</B> circuit. Voltage after the inductor is 17.9V in my supply. Am I listening to euphonic colorations?
Hi Jan,janneman said:Peranders, I agree with you, this is not a cap multiplier. It would be if the 431 was replaced by a resistor, in such a way that the base of the first transistor is a fraction of the unregulated input and is "following" this input.
Also, there really should be resistors from the base of each transistor to its emitter or possibly another point to control the leakage. Look at it as the load resistor for the previous transistor: the voltage on this resistor becomes the input voltage for the next stage. If you leave it off you have no idea what is going on in the circuit and you might as well randomly throw in a handful of parts and hope it will work better.....
Jan Didden
I did throw some parts together again in the reaction vessel and this came out. Couldn' t resist it. Improved circuit ? I dunno. Slight of thin in the bass.
This circuit also works a a pure <B>capacity multiplier</B> when you replace R7 by 2M (two 1M in series) and the LM329 replaced by 1M. [You do read carefully Peranders?] Yes now the voltage is a fraction of the raw supply voltage and follows the variations in the mains volatge. But this is also a <B>working</B> circuit. Voltage after the inductor is 17.9V in my supply. Am I listening to euphonic colorations?
Attachments
Elso, Per-Anders, Jan,
Since you are arguing about whether the circuit is a capacitance
multiplier or not, perhaps you should all define what you mean
by this term. I have a strong feeling you mean different things
with the term so the argument is rather about this term than
about the circuit.
Since you are arguing about whether the circuit is a capacitance
multiplier or not, perhaps you should all define what you mean
by this term. I have a strong feeling you mean different things
with the term so the argument is rather about this term than
about the circuit.
Cap Multiplier(?)
Hi Christer,
I wil stop arguing about the term cap multiplier as Jan is right: the circuit with the resistors as I indicated is the original cap-multiplier circuit in the Low-Noise Electronic Design book.
Wether my circuit with the TL431 or LM329 is a cap multiplier or not I find in fact totally uninteresting. What counts for me is how it <B>sounds</B> and if it <B>works</B> in conjunction with the TDA1543 as a NON-OS DAC. Just in search for a better sounding supply regulator for the TDA1543, triggered by the finding of Patwen and the remark by Fred that quote: <B><I>The Jung regulators can be problematic with digital and D to A circuits because this resonances are more easily excited by the large high frequency currents. </b></I> (First page of this thread)
Hi Christer,
I wil stop arguing about the term cap multiplier as Jan is right: the circuit with the resistors as I indicated is the original cap-multiplier circuit in the Low-Noise Electronic Design book.
Wether my circuit with the TL431 or LM329 is a cap multiplier or not I find in fact totally uninteresting. What counts for me is how it <B>sounds</B> and if it <B>works</B> in conjunction with the TDA1543 as a NON-OS DAC. Just in search for a better sounding supply regulator for the TDA1543, triggered by the finding of Patwen and the remark by Fred that quote: <B><I>The Jung regulators can be problematic with digital and D to A circuits because this resonances are more easily excited by the large high frequency currents. </b></I> (First page of this thread)
Why etc
Elso,
I just picked up on your quote from Fred and I do not completely agree.
The Jung regulators should get a capacitor at the output with a not-too-low ESR. Walter (and I, for that matter) have on various occasions (also IIRC in the AA article) explained that those very low ESR film caps everybody likes for coupling are NOT GOOD for any regulated supply bypass.
Every professional supply designer will select bypass caps with a close look at the ESR. If you keep yourself up to date on the modern low-dropout designs you will see that the capability to work stably with low-ESR caps is a big issue, because it requires specific design adaptions.
Unfortunately, most audio affectionados are so obsessed with very high quality low ESR coupling caps that it is automatically assumed that they are good for supplies also. Not so!.
As they say: "Imagination gives you wings, but knowledge gives you feet"...
Jan Didden
PS Can I borrow your reaction vessel? I have this box of junk I want to transform into a new car...
Elso,
I just picked up on your quote from Fred and I do not completely agree.
The Jung regulators should get a capacitor at the output with a not-too-low ESR. Walter (and I, for that matter) have on various occasions (also IIRC in the AA article) explained that those very low ESR film caps everybody likes for coupling are NOT GOOD for any regulated supply bypass.
Every professional supply designer will select bypass caps with a close look at the ESR. If you keep yourself up to date on the modern low-dropout designs you will see that the capability to work stably with low-ESR caps is a big issue, because it requires specific design adaptions.
Unfortunately, most audio affectionados are so obsessed with very high quality low ESR coupling caps that it is automatically assumed that they are good for supplies also. Not so!.
As they say: "Imagination gives you wings, but knowledge gives you feet"...
Jan Didden
PS Can I borrow your reaction vessel? I have this box of junk I want to transform into a new car...
Jan,
since you brought up the topic of low-dropuout regulators and
stability, are they really better than ordinary 3X7 regulators?
I haven't read all threads on this topic, but it seems frequently
recommended to use low-dropout regulators 1085 etc. instead
of traditional 317s and alike. However, there is hardly any need
for the low dropout in audio circuitry, since they are usually not
battery powered. Further, according to several application notes
the traditional regulators are stable for arbitrarily high load
capacitance (and not sensitive to ESR, I think), while the low-
dropout ones are stable only for a certain range of capacitance,
not too low and not too high, and sensitive to ESR as you say.
It seems mostly to be asking for a lot of problems to use low-
droput regs. and capacitance is stability seems often discussed
in the threads on these regulators. Do they have any other
advantages that make it worthwhile to take the extra cost
and trouble of using them, rather than using 317/337?
since you brought up the topic of low-dropuout regulators and
stability, are they really better than ordinary 3X7 regulators?
I haven't read all threads on this topic, but it seems frequently
recommended to use low-dropout regulators 1085 etc. instead
of traditional 317s and alike. However, there is hardly any need
for the low dropout in audio circuitry, since they are usually not
battery powered. Further, according to several application notes
the traditional regulators are stable for arbitrarily high load
capacitance (and not sensitive to ESR, I think), while the low-
dropout ones are stable only for a certain range of capacitance,
not too low and not too high, and sensitive to ESR as you say.
It seems mostly to be asking for a lot of problems to use low-
droput regs. and capacitance is stability seems often discussed
in the threads on these regulators. Do they have any other
advantages that make it worthwhile to take the extra cost
and trouble of using them, rather than using 317/337?
Jung Regulators
Sorry Jan,
You completely missed my point.
The Jung like regulators I used were NOT with a low ESR cap at the ouput just a ordinary elctrolytic and I did not mention any low ESR caps in my posts. Also just as Patwen these were NOT oscillating (flat line on the scope).
Furthermore I noticed a difference in sound between a MC7805 and a LM1086 just as Patwen.
I also tried once a L1021-5 and a opamp for the supply circuit of my clock circuit as suggested by Wildmonkeysects and this gave a disappointing result. I my preamp I am using Jung like regulators and here they gave a unexpected improvement compared to LM317/337 with bypassed adjustment pin.
The suggestions made by Fred made me to design a NON-negative feedback circuit, though I started WITH negative feedback in order to evaluate the difference...
Fred mentioned a simple follower and the detrimental effect of large amounts of negative feedback. I usually think hard about Fred's hints and mostly very early in the morning come with very good ideas.
You also mentioned low dropout regulation but I am totally uninterested in low drop regulators as I can have a dropout of 15V. Lowdropoutregulators have got a lot of attention by the industry as these are important to get maximum battery life in portable equipment like cellphones etc. The whole low-drop outregulator concept complicates things and I don't need it.
Concluding I admit my circuit is not much different from the one Guido Tent , Heiligers et. al. are using. And Guido quite honestly states that it was borrowed from the Audionote DAC-3 supply. After all I only replaced the zenerdiode with a TL431 or LM329, and some minor mods that go unnoticed and I will not spell out.
Sorry Jan,
You completely missed my point.
The Jung like regulators I used were NOT with a low ESR cap at the ouput just a ordinary elctrolytic and I did not mention any low ESR caps in my posts. Also just as Patwen these were NOT oscillating (flat line on the scope).
Furthermore I noticed a difference in sound between a MC7805 and a LM1086 just as Patwen.
I also tried once a L1021-5 and a opamp for the supply circuit of my clock circuit as suggested by Wildmonkeysects and this gave a disappointing result. I my preamp I am using Jung like regulators and here they gave a unexpected improvement compared to LM317/337 with bypassed adjustment pin.
The suggestions made by Fred made me to design a NON-negative feedback circuit, though I started WITH negative feedback in order to evaluate the difference...
Fred mentioned a simple follower and the detrimental effect of large amounts of negative feedback. I usually think hard about Fred's hints and mostly very early in the morning come with very good ideas.
You also mentioned low dropout regulation but I am totally uninterested in low drop regulators as I can have a dropout of 15V. Lowdropoutregulators have got a lot of attention by the industry as these are important to get maximum battery life in portable equipment like cellphones etc. The whole low-drop outregulator concept complicates things and I don't need it.
Concluding I admit my circuit is not much different from the one Guido Tent , Heiligers et. al. are using. And Guido quite honestly states that it was borrowed from the Audionote DAC-3 supply. After all I only replaced the zenerdiode with a TL431 or LM329, and some minor mods that go unnoticed and I will not spell out.
Why etc
Elso and Christer,
Sorry if I was not clear enough. I did not want to suggest that LDO's are better than "ordinary" regulators. I was merely reaction on Fred's statement that the (un)stability of the Jung regulators would give rise to oscillations through the digital load currents. What I meant was that Jung type regulators are often unstable because they are bypassed with low ESR caps, and that then also gives rise to susceptibility to high speed load currents.
But with care to this I don't think there is an inherent problem. In fact, ANY regulator has an inductive output resistance because of the simple fact that the output resistance increases with freq. And what is a component that has increasing impedance with rising freq? Yes, an inductor. Since all regulators have bypass caps on the output, they all have the same problem.
My ref to LDO's was just meant to remind that bypass caps need specific attention. LDO's have PNP pass transistors (mostly) which normally have a smaller bandwidth than NPNs, which makes the problem more acute. You see often adverts in journals like EDN saying ..."worlds lowest dropout reg stable with any bypass cap!" In fact, there is a manufacturer who has trademarked "AnyCap LDO" (I think it is LTC). The LDO data sheets I have seen do not suggest that they are well suited to audio, but I must confess that I do not follow it closely.
But in one area we agree Elso: I also have my best ideas early in the morning. If the ideas can win the competition with last night's scotch, that is.
Jan Didden
Elso and Christer,
Sorry if I was not clear enough. I did not want to suggest that LDO's are better than "ordinary" regulators. I was merely reaction on Fred's statement that the (un)stability of the Jung regulators would give rise to oscillations through the digital load currents. What I meant was that Jung type regulators are often unstable because they are bypassed with low ESR caps, and that then also gives rise to susceptibility to high speed load currents.
But with care to this I don't think there is an inherent problem. In fact, ANY regulator has an inductive output resistance because of the simple fact that the output resistance increases with freq. And what is a component that has increasing impedance with rising freq? Yes, an inductor. Since all regulators have bypass caps on the output, they all have the same problem.
My ref to LDO's was just meant to remind that bypass caps need specific attention. LDO's have PNP pass transistors (mostly) which normally have a smaller bandwidth than NPNs, which makes the problem more acute. You see often adverts in journals like EDN saying ..."worlds lowest dropout reg stable with any bypass cap!" In fact, there is a manufacturer who has trademarked "AnyCap LDO" (I think it is LTC). The LDO data sheets I have seen do not suggest that they are well suited to audio, but I must confess that I do not follow it closely.
But in one area we agree Elso: I also have my best ideas early in the morning. If the ideas can win the competition with last night's scotch, that is.
Jan Didden
Jan,
I did not read into your posting whether you preferred LDO or
not, I just decided to bring up a subject I have been thinking
about and which was related to your comment. I have looked
around at a lot of old threads in the archive and it seems that
the good old 317 are very seldom discussed, and it often seems
that the 108X and similar are the only three-pin regulators
people consider. Hence I assume that a lot of (or even most)
people think these are better for one reason or another. I have
not found any clear statements of why they think so, and since
there seems to be some drawbacks with respect to stability I
wonder if there is any reason to prefer them to the 317.
There are also variant of 317, of course. LT317 is usually
recommended instead of the usual ones, although LT337 seems
to be discontinued, unfortunately. There are also the A versions of
317 and 337.
I did not read into your posting whether you preferred LDO or
not, I just decided to bring up a subject I have been thinking
about and which was related to your comment. I have looked
around at a lot of old threads in the archive and it seems that
the good old 317 are very seldom discussed, and it often seems
that the 108X and similar are the only three-pin regulators
people consider. Hence I assume that a lot of (or even most)
people think these are better for one reason or another. I have
not found any clear statements of why they think so, and since
there seems to be some drawbacks with respect to stability I
wonder if there is any reason to prefer them to the 317.
There are also variant of 317, of course. LT317 is usually
recommended instead of the usual ones, although LT337 seems
to be discontinued, unfortunately. There are also the A versions of
317 and 337.
Regulators and capacitors
You guys are converging on some interesting points. A regulator's
rising output inpedance due to the falling open loop gain of a regulator makes the output impedence look inductive. Hang a capacitor on this and you have a resonant circuit at some point. If you have significant load currents at this point (like a DAC for example) you can have problems.... Discreet transistor regulators with flat output impeadances that don't resonate with the output caps might be well worth looking at. A small resistor of 0.1 to 1 ohm between the output of the regulator an the load capacitor is also worth experimenting with. The load would be across the output cap. The bottom line is that the transient response of the regulator needs further investigation even if the ringing is outside the audio band.
there are a number of Linear micropower regs
Such as?
You guys are converging on some interesting points. A regulator's
rising output inpedance due to the falling open loop gain of a regulator makes the output impedence look inductive. Hang a capacitor on this and you have a resonant circuit at some point. If you have significant load currents at this point (like a DAC for example) you can have problems.... Discreet transistor regulators with flat output impeadances that don't resonate with the output caps might be well worth looking at. A small resistor of 0.1 to 1 ohm between the output of the regulator an the load capacitor is also worth experimenting with. The load would be across the output cap. The bottom line is that the transient response of the regulator needs further investigation even if the ringing is outside the audio band.
there are a number of Linear micropower regs
Such as?
Re: Regulators and capacitors
Fred, no offense meant, but this really is nothing new. And as I said, ALL regulators exhibit rising output imp with freq, thus are inductive, and thus can resonate with the bypass cap. Yes, also your discrete regulator. It's either just an emitter follower (cathode/source follower) or a common emitter amp with heavy feedback, and they all show this.
The only difference is in the freq range where it occurs.
Jan Didden
Fred Dieckmann said:
Fred, no offense meant, but this really is nothing new. And as I said, ALL regulators exhibit rising output imp with freq, thus are inductive, and thus can resonate with the bypass cap. Yes, also your discrete regulator. It's either just an emitter follower (cathode/source follower) or a common emitter amp with heavy feedback, and they all show this.
The only difference is in the freq range where it occurs.
Jan Didden
Christer said:
Well, you know how it goes. Someone is very convincing on a specific part, and suddenly everyone wants it. The reason why is often lost in the mists of time, if there ever was a good reason to start with. I do think though that at least data-sheet wise the 108x has an edge over the 317, but I have no experience how audible it would be.
Jan Didden
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.