I made a pcb layout for a jung super regulator*. I am planning on putting the regulators on the analog side of my pcm1792 dac project.
Here are pictures of:
schematic that i used
top layer of pcb
bottom layer of pcb
The footprint of the super regulator is 2" x 1".
As for the use of the super regulator, I figured that since the super regulator will physically be right next to the chip, I tied the sense lines to the output lines. Also, since I already have capacitors on the pcb for decoupling, I removed the 100uF capacitor on the output. Also, the jung regulator, in it's current state will be part of the analog ground plane, with the input from my power supply having the ground seperate from the analog ground plane. (the analog ground plane will be wired seperately back to the central ground).
Are there any issues with my initial implementation? Any improvements that can be done? I tried to keep the wiring as short as possible. Any suggestions/comments/criticisms/ are greatly appreciated.
* The underlying circuit used in this layout is based on the Walt Jung regulator, and primarily based on a modified version by ALW. I claim no intellectual ownership of this design 🙂
--
Brian
Here are pictures of:
schematic that i used
top layer of pcb
bottom layer of pcb
The footprint of the super regulator is 2" x 1".
As for the use of the super regulator, I figured that since the super regulator will physically be right next to the chip, I tied the sense lines to the output lines. Also, since I already have capacitors on the pcb for decoupling, I removed the 100uF capacitor on the output. Also, the jung regulator, in it's current state will be part of the analog ground plane, with the input from my power supply having the ground seperate from the analog ground plane. (the analog ground plane will be wired seperately back to the central ground).
Are there any issues with my initial implementation? Any improvements that can be done? I tried to keep the wiring as short as possible. Any suggestions/comments/criticisms/ are greatly appreciated.
* The underlying circuit used in this layout is based on the Walt Jung regulator, and primarily based on a modified version by ALW. I claim no intellectual ownership of this design 🙂
--
Brian
peranders said:
I was going to have the analog ground run back to the central ground. This is what ALW did in his schematic. Is this a bad idea?
--
Brian
Long loops create inductance (not good) and the possiblity to pick up noise. Yes, not very good idea. Check one more time how Andy has done.BrianGT said:
Nice layout!
The only thing I would add is a small decoupling cap right across the supply pins of the error amp to avoid supply noise and coupling that could cause ringing or even oscillations.
On the grounding, this is not so simple without knowing the total layout. For example, the agnd from the feedback network and the ref network could be connected together near the amp input, and then connected to central gnd. The current through this lead to central ground can cause unwanted signals, but since they are common to both the reference and the feedback, they will be a common mode input to the errror amp and thus the effect on the final output voltage would be greatly reduced. You really would have to look in this sense to the other wiring as well.
Jan Didden
The only thing I would add is a small decoupling cap right across the supply pins of the error amp to avoid supply noise and coupling that could cause ringing or even oscillations.
On the grounding, this is not so simple without knowing the total layout. For example, the agnd from the feedback network and the ref network could be connected together near the amp input, and then connected to central gnd. The current through this lead to central ground can cause unwanted signals, but since they are common to both the reference and the feedback, they will be a common mode input to the errror amp and thus the effect on the final output voltage would be greatly reduced. You really would have to look in this sense to the other wiring as well.
Jan Didden
"I removed the 100uF capacitor on the output."
I wouldn't do that! You have the real potential for an unstable or marginally stable regulator. The circuit is really designed NOT to use high Q (very low ESR) capacitors. The problem is that this regulator is more complicated than it looks. The output capacitor type and location is critical for it to work correctly and sound the best. I am sure that Andy will have some further insight and you may want to email him Brian.
"The only thing I would add is a small decoupling cap right across the supply pins of the error amp to avoid supply noise and coupling that could cause ringing or even oscillations."
Not without some resistive decoupling
of the op amp PS terminal from the regulator output I wouldn't. I believe this was discussed in one of the Jung articles and I will try to find it.
I wouldn't do that! You have the real potential for an unstable or marginally stable regulator. The circuit is really designed NOT to use high Q (very low ESR) capacitors. The problem is that this regulator is more complicated than it looks. The output capacitor type and location is critical for it to work correctly and sound the best. I am sure that Andy will have some further insight and you may want to email him Brian.
"The only thing I would add is a small decoupling cap right across the supply pins of the error amp to avoid supply noise and coupling that could cause ringing or even oscillations."
Not without some resistive decoupling
of the op amp PS terminal from the regulator output I wouldn't. I believe this was discussed in one of the Jung articles and I will try to find it.
pre-regulator?
From the comment in your schematic I understand that you will connect Vgnd to Agnd somewhere. I'm wondering if you pre-regulator shouldn't have it's Vgnd connected to Vout? That way, the LM317 will put Vprereg approximately 2.5V above Vout, instead of 2.5V above Vgnd. (See ALW's page for details under The Tracking Pre-Regulator.)
Cheers,
Alfred
BrianGT said:
From the comment in your schematic I understand that you will connect Vgnd to Agnd somewhere. I'm wondering if you pre-regulator shouldn't have it's Vgnd connected to Vout? That way, the LM317 will put Vprereg approximately 2.5V above Vout, instead of 2.5V above Vgnd. (See ALW's page for details under The Tracking Pre-Regulator.)
Cheers,
Alfred
What me worry?
From the comment in your schematic I understand that you will connect Vgnd to Agnd somewhere. I'm wondering if you pre-regulator shouldn't have it's Vgnd connected to Vout? That way, the LM317 will put Vprereg approximately 2.5V above Vout, instead of 2.5V above Vgnd. (See ALW's page for details under The Tracking Pre-Regulator.)
Ummmm...... How is the regulator going to work with 2.5 volts on the input. Go read
the LM317 datasheet for how to program the output voltage for the LM317.
http://www.national.com/ds.cgi/LM/LM117.pdf
From the comment in your schematic I understand that you will connect Vgnd to Agnd somewhere. I'm wondering if you pre-regulator shouldn't have it's Vgnd connected to Vout? That way, the LM317 will put Vprereg approximately 2.5V above Vout, instead of 2.5V above Vgnd. (See ALW's page for details under The Tracking Pre-Regulator.)
Ummmm...... How is the regulator going to work with 2.5 volts on the input. Go read
the LM317 datasheet for how to program the output voltage for the LM317.
http://www.national.com/ds.cgi/LM/LM117.pdf
Attachments
Re: What me worry?
Is this smart youngster with the waving ears a juvenile picture of you???
BTW I found the Jung regulator better sounding without the preregulator, but who am I??😕
(Just some free advice to confuse the forum members
)
Hi Fred,Fred Dieckmann said:
Is this smart youngster with the waving ears a juvenile picture of you???
BTW I found the Jung regulator better sounding without the preregulator, but who am I??😕
(Just some free advice to confuse the forum members
)pcb layout
Brian, you must do wider traces in " main current rail ", you get lower output impedance.
Brian, you must do wider traces in " main current rail ", you get lower output impedance.
That's wierd
Elso,
Mine sound better with it in place and that's not just my opinion, but that of everyone who's tried it 😉
Brian,
You're pre-reg is connected incorrectly as the schematic currently stands. Think about it, it's producing 2.5V at it's output relative to it's Kelvin sense point. You either need to raise Vout, or connect it differently, with the Kelvin sense point to the super-reg output, not agnd, thereby produce a voltage 2.5V above Vout.
Andy.
Elso,
Mine sound better with it in place and that's not just my opinion, but that of everyone who's tried it 😉
Brian,
You're pre-reg is connected incorrectly as the schematic currently stands. Think about it, it's producing 2.5V at it's output relative to it's Kelvin sense point. You either need to raise Vout, or connect it differently, with the Kelvin sense point to the super-reg output, not agnd, thereby produce a voltage 2.5V above Vout.
Andy.
Re: That's weird
Hi Andy,
Except me of course
And remove/omit C7 and ....... no more hints from me
(Another free advice to further confuse the forum members )
ALW said:
Hi Andy,
Except me of course
And remove/omit C7 and ....... no more hints from me
(Another free advice to further confuse the forum members
)Rail the 7Cs
C7 lowers the output impedance by a factor of Vo / 6.9. You do want to use a very good cap here. You can use a film cap and scale R8 and R9 to much higher value but the same ratio for the desired output voltage. You could leave it out and have a higher output impedance as well. I believe Andy and I talked about this a couple of months ago and he has commented on doing the same thing I outlined here, with C4 and R7 a little while back on the forum. These are pretty obvious mods for the experienced amplifier tweaker. The original Sulzer article talked about the cap across the feedback resistor making a significant audible difference, due to the lowered output impedance from the increased loop feedback. C7 will also reduce the amplification of voltage noise from the voltage reference. This where use of very good parts versus the change in objective measurements can call for design or part changes based on listening test. There could be sonic improvements from C7s removal but a large part of development for the circuit was focused on reducing the output impedance, and the audible improvements resulting from this. The measurement of PSRR, output impedance, and output noise will show worse numbers with C7 removed.
There is a third approach which achieves both the removal of C7 and a slight increase but not doubling the output impedance for a 15 Volt supply output voltage for the same current draw and with the same op amp. Anybody want to guess? Hint: It requires knowledge of the common mode characteristics of the op amp. Andy.... you're disqualified since you already know.
C7 lowers the output impedance by a factor of Vo / 6.9. You do want to use a very good cap here. You can use a film cap and scale R8 and R9 to much higher value but the same ratio for the desired output voltage. You could leave it out and have a higher output impedance as well. I believe Andy and I talked about this a couple of months ago and he has commented on doing the same thing I outlined here, with C4 and R7 a little while back on the forum. These are pretty obvious mods for the experienced amplifier tweaker. The original Sulzer article talked about the cap across the feedback resistor making a significant audible difference, due to the lowered output impedance from the increased loop feedback. C7 will also reduce the amplification of voltage noise from the voltage reference. This where use of very good parts versus the change in objective measurements can call for design or part changes based on listening test. There could be sonic improvements from C7s removal but a large part of development for the circuit was focused on reducing the output impedance, and the audible improvements resulting from this. The measurement of PSRR, output impedance, and output noise will show worse numbers with C7 removed.
There is a third approach which achieves both the removal of C7 and a slight increase but not doubling the output impedance for a 15 Volt supply output voltage for the same current draw and with the same op amp. Anybody want to guess? Hint: It requires knowledge of the common mode characteristics of the op amp. Andy.... you're disqualified since you already know.
same links as previous post:
Here are pictures of:
schematic that i used
top layer of pcb
bottom layer of pcb
I updated the pictures and schematic. As for the changes that I made, I hopefully fixed the schematic, with regards to the pre-reg. I initially misunderstood the schematic.
I also made the traces a bit fatter (20mils -> 30 mils), and increased the sizes of the pads for the regulator and mosfet, in order to handle heat dissipation better.
As for C7, I am thinking about just leaving it for this, as I want to get my pcbs made next week. For the next revision, i can play around with this.
In order to make this work for negative voltages, what do I have to change? I should have the version for +5v and +15v worked out, now I just need -15v.
(note, I probably haven't specified the correct resistors for the +15 in the schematic, but this will be changed later.)
Let me know if you have anymore comments. I appreciate all of your input thus far.
--
Brian
Here are pictures of:
schematic that i used
top layer of pcb
bottom layer of pcb
I updated the pictures and schematic. As for the changes that I made, I hopefully fixed the schematic, with regards to the pre-reg. I initially misunderstood the schematic.
I also made the traces a bit fatter (20mils -> 30 mils), and increased the sizes of the pads for the regulator and mosfet, in order to handle heat dissipation better.
As for C7, I am thinking about just leaving it for this, as I want to get my pcbs made next week. For the next revision, i can play around with this.
In order to make this work for negative voltages, what do I have to change? I should have the version for +5v and +15v worked out, now I just need -15v.
(note, I probably haven't specified the correct resistors for the +15 in the schematic, but this will be changed later.)
Let me know if you have anymore comments. I appreciate all of your input thus far.
--
Brian
Re: Rail the 7Cs
Hi Fred,
Adding "C7" to my regulator caused a mild oscillation of the +15V supply and no change in soundquality. 😱
See also this thread:
http://db.audioasylum.com/cgi/m.pl?forum=tweaks&n=32609&highlight=elso+carlos&r=&session=
and
http://headwize2.powerpill.org/projects/showproj.php?file=gilmore3_prj.htm
😎
Fred Dieckmann said:
Hi Fred,
Adding "C7" to my regulator caused a mild oscillation of the +15V supply and no change in soundquality. 😱
See also this thread:
http://db.audioasylum.com/cgi/m.pl?forum=tweaks&n=32609&highlight=elso+carlos&r=&session=
and
http://headwize2.powerpill.org/projects/showproj.php?file=gilmore3_prj.htm
😎
you could add a small decouplecap to the input pin on the prereg - say 100nF - film or ceramic, it won't hurt......😉
Re:See also
See also this paper by Hank Zumbahlen of Analog Devices:
http://www.analog.com/UploadedFiles/Application_Notes/14452667AN394.pdf
Fig. 26 on page 11.😎
Elso Kwak said:
See also this paper by Hank Zumbahlen of Analog Devices:
http://www.analog.com/UploadedFiles/Application_Notes/14452667AN394.pdf
Fig. 26 on page 11.😎
Fred Dieckmann said:
Fred,
I don't remember that this was discussed specifically, but it could be. Anyway, the (small) resistor between reg output to the opamp supply is a good idea. The only reason I can think of when it is REQUIRED would be if the cap across the opamp would be to "good" (low esr etc) to cause oscillations. But I think that danger is low because there would always be at least an inch or more between reg output and the cap, which would be enough to prevent instability.
Jan Didden
But maybe a good idea.
'The only reason I can think of when it is REQUIRED would be if the cap across the opamp"
I have found quite a few audio designs that use parts fast enough to get into some twitchy behavior at RF frequencies. Parasitics such as the self resonance of electrolytic capacitors and even worse, their interaction with film bypass caps, can make for some unanticipated RF instability. One of the main reasons for decoupling is to remove the op amps supply from the effect of load currents drawn from the supply, particularly at high frequencies. While the Jung bootstrap supply is a great approach at audio frequencies and especially in terms of power supply rejection for the unreglated supply feeding the regulator, the potential for undesired coupling at RF frequencies is very real for a circuit which is regulating its own supply. Even very small value resistor would isolate the op amp supply from the output circuit at frequencies where the op amp and output device have gone through enough phase shift to make phase margins rather small. I have found that paying attention to good RF decoupling in audio circuit often results in better sonic results, and is one of the most overlooked areas in audio circuit design. High Q low ESL and ESR electrolytics are very often used now for audio circuit power supplies. These can make the problem of resonant behavior at high frequencies even more of a design challenge to deal with. A small resistor is good way to improve the RF isolation with very little effect on the audio frequency performance. I believe even values even as low as a few ohms can be beneficial to RF decoupling and stability.
'The only reason I can think of when it is REQUIRED would be if the cap across the opamp"
I have found quite a few audio designs that use parts fast enough to get into some twitchy behavior at RF frequencies. Parasitics such as the self resonance of electrolytic capacitors and even worse, their interaction with film bypass caps, can make for some unanticipated RF instability. One of the main reasons for decoupling is to remove the op amps supply from the effect of load currents drawn from the supply, particularly at high frequencies. While the Jung bootstrap supply is a great approach at audio frequencies and especially in terms of power supply rejection for the unreglated supply feeding the regulator, the potential for undesired coupling at RF frequencies is very real for a circuit which is regulating its own supply. Even very small value resistor would isolate the op amp supply from the output circuit at frequencies where the op amp and output device have gone through enough phase shift to make phase margins rather small. I have found that paying attention to good RF decoupling in audio circuit often results in better sonic results, and is one of the most overlooked areas in audio circuit design. High Q low ESL and ESR electrolytics are very often used now for audio circuit power supplies. These can make the problem of resonant behavior at high frequencies even more of a design challenge to deal with. A small resistor is good way to improve the RF isolation with very little effect on the audio frequency performance. I believe even values even as low as a few ohms can be beneficial to RF decoupling and stability.
For making the negative version, with these components changes:
D44H11 (npn) -> D45H11 (pnp)
ZTX751 (pnp) -> ZTX651 (npn)
LM317 (+reg) -> LM337 (-reg)
If I swap to these components, and flip over my caps and diodes, would this be correct for a negative regulator?
--
Brian
D44H11 (npn) -> D45H11 (pnp)
ZTX751 (pnp) -> ZTX651 (npn)
LM317 (+reg) -> LM337 (-reg)
If I swap to these components, and flip over my caps and diodes, would this be correct for a negative regulator?
--
Brian
- Status
- Not open for further replies.