Super Regulator

Member
Joined 2021
Paid Member
Jan is going to be mad at me for asking such silly question, but I am running on empty these days…

I am about to build a CFA3 amp.
KG’s GRLV supply is the recommended supply for it, but I am fascinated by the Jung-Didden regulator.

I already have 2 of these transformers:

https://www.antekinc.com/as-1230-100va-30v-transformer/

Has anyone experience with a PCB that would provide rectification, filtering and maybe pre-regulation for a +/- 30ish V set-up?

I would have 1 complete supply per channel.
 
@Wellerman

I do have some 1611 I soldered up. Too bad I already ordered the AD817….

Would love to see how you implemented it to use on the Pearl. I assume the regulators on the Pearl are removed.

Any pictures of the build?
Just a little preview at page 115 of the Pearl 3 topic. I have yet to design the enclosures and order them at Modushop, so it will take some time before I can put everything together. It will be a CRCRC PSU followed by LM3x7 pre-regulators, and then Super Regulators for the final touch.
 
  • Like
Reactions: 1 user
Member
Joined 2011
Paid Member
You might consider adding either a cascade of "N" series connected power line filter modules which include common mode chokes, like (these) , upstream of your step down power transformer. Or you could add a 1-to-1 isolation transformer upstream, either by itself or in combination with power line filter modules. Either of these maneuvers will reduce AC common mode noise, plus reduce the amount of RFI pumped into the primary of your step down power transformer. If you want really, really severe overkill.
 
  • Like
Reactions: 1 user
Member
Joined 2021
Paid Member
Jan is going to be mad at me for asking such silly question, but I am running on empty these days…

I am about to build a CFA3 amp.
KG’s GRLV supply is the recommended supply for it, but I am fascinated by the Jung-Didden regulator.

I already have 2 of these transformers:

https://www.antekinc.com/as-1230-100va-30v-transformer/

Has anyone experience with a PCB that would provide rectification, filtering and maybe pre-regulation for a +/- 30ish V set-up?

I would have 1 complete supply per channel.

The RingNot PCB is mentioned in this thread.
 
You might consider adding either a cascade of "N" series connected power line filter modules which include common mode chokes, like (these) , upstream of your step down power transformer. Or you could add a 1-to-1 isolation transformer upstream, either by itself or in combination with power line filter modules. Either of these maneuvers will reduce AC common mode noise, plus reduce the amount of RFI pumped into the primary of your step down power transformer. If you want really, really severe overkill.
I forgot; the PSU starts with a Schaffner FN9244-1 IEC-inlet which has chokes. I like a bit of overkill ;-)
 
Member
Joined 2021
Paid Member
After some initial struggles with errors on my own PCB I now have a bunch of working super regulator boards. Very nice and thanks to Jan and others for their help here.

Reading through this thread I found the following interesting options to optimize. As I'm happy to have working SR boards now, I'm not looking to mess with them if there is a reasonable chance of breaking them by not really knowing what I'm doing. I'm quite the noob and really not an EE, so I'd much appreciate some guidance on any traps for young players. Or: whether they make sufficient sense to try out without having lab equipment (nor the expertise required if somehow I'd have access to a lab).

1. Choice of power transistor:
TTC004/TTA004 are an even better choice ;)
What is attractive about these compared to the D44H11? (Don't know if the suggestion was serious as I'm quite emoji-illiterate.) I found some comments on these Toshiba's in different places, but wouldn't know which specs of a transistor to pay attention to at this place in the SR circuit.

2. Choice of opamp. AD797 spooks me out because of all the authoritative advice to not use it because of the high risk of oscillations. But what about this one:
The ADA4625-1 is a great replacement for the AD797 and it's also a jfet op amp that allows Walt's upgrade to be used.
No increased risk of the opamp oscillating or any other pitfalls with this opamp as a drop-in replacement for the AD825? (All other things being equal.) It's around double the price of the AD825 - possibly worth it?

3. The aforementioned upgrade to the voltage reference's filter network:
More important fact is that many recent production capacitors will exceed requirements for a low leakage capacitor in voltage reference low pass filter, explained in the Walt Jung’s article: An_Improved_Reference_Filter_for_Audio_Regulators.pdf

(...)

So, without much headache in finding components, we can use 150 uF/4.7 kΩ RC filter in the super regulator. However, that will result in slowly ramping regulator output voltage, requiring some 3 seconds to reach 95% of nominal output voltage.
As this one is from WJ himself, I'm less hesitant about it.

Preferred choice of cap appears to vary, but Nichicon UKL looks good in terms of leakage spec on the data sheet. Judging by tombo56's tests of the filter published in this thread the Panasonic EEH-AZA is a good candidate too. (All ostensibly better in terms of leakage current than UKZ, a 'premium', 'audio' type suggested by WJ at the time.) So with either UKL or EEH-AZA I'd consider following tombo56's suggestion about the voltage ref 150µF/4k7 RC filter at the penalty of a longer ramp up to nominal output voltage.

Other than perhaps C3/C7 I'd leave all other caps (especially C4/C8) remain in accordance with the oft repeated instructions, i.e., general purpose electrolyics that are not low ESR.
 
  • Like
Reactions: 1 user
What is attractive about these compared to the D44H11? (Don't know if the suggestion was serious as I'm quite emoji-illiterate.) I found some comments on these Toshiba's in different places, but wouldn't know which specs of a transistor to pay attention to at this place in the SR circuit.
TTC/TTA004 are faster, have lower Cob, higher fT and hFE compared to D44H11/ D45H11. The BJT driver must also be low noise, low Cob, high hFE.
No increased risk of the opamp oscillating or any other pitfalls with this opamp as a drop-in replacement for the AD825? (All other things being equal.) It's around double the price of the AD825 - possibly worth it?
ADA4625 is the best op amp I have tried for this type of regulator, there is no problem with self-oscillating if everything is set up properly and with good decoupling. It also has the lowest noise of all jfet op amps. The plus is that it can work on 5V with 1V reference.
If you think it's too expensive, try ADA4610 or OPA1641 instead of AD825, both are better but I prefer ADA4610, even though it's worse on paper than OPA1641.
I don't know why but I have a feeling that AD op amps are better than TI op amps in this type of regulator. I had high hopes for the OPA1641(low noise, 5V compatible, jfet, good price) and in the end it disappointed me.

As for the capacitors themselves, I would stick to Walt's recommendations, I tested a lot of capacitors according to Walt's instructions and we skipped the low ESR capacitors(especially polymers,), because they can cause instability. Panasonic FC or Rubycon ZLJ (35V and higher) are good from the classic ones. I have now started using SMD tantalums with very low leakage currents, but they do not have a low ESR and high ripple current so they are an excellent choice.

You can achieve great progress with better references for 2.5V and 6.9V and the regulator itself can be further improved. My version now has 4 depletion mosfets and a total of 9 BJT transistors all because of better references.
 
  • Like
Reactions: 1 user
Thank you for clarifying. The GLED431 I was familiar with and I've been meaning to try out your SMD variation of it in the W. Jung shunt regulator thread. For the series SR I have not yet had a need for a Vref lower than 6.9V, but will of course use it when I do. Jung's Zener based 2.5V and 6.9V shunt Vref designs look great. Thanks for the PDF.

Can you post some links to the SMD tantalums that you are happy with?

One more thing: what do you mean by good decoupling as a condition for operating without self-oscillating? I see that the data sheet suggests adding 100nF decoupling caps on the ADA4625's power supplies. Something like a film cap? Or is more decoupling required?
 
Member
Joined 2011
Paid Member
Keep in mind that depletion-mode MOSFETs are only available as N-channel devices, not as P-channel devices. If you want to make a negative output regulator which is a mirror image of the positive output regulator (diodes and electrolytics reversed; NPNs and PNPs swapped; etc), this could be an issue.
 
Can you post some links to the SMD tantalums that you are happy with?
Vishay TL3 and KEMET T489 ,unfortunately Mouser does not have all the values but for output voltage up to 20V there is no problem with the choice.
One more thing: what do you mean by good decoupling as a condition for operating without self-oscillating? I see that the data sheet suggests adding 100nF decoupling caps on the ADA4625's power supplies. Something like a film cap? Or is more decoupling required?
That's right, DS suggests 100nF as close as possible to the op amp itself, but in some cases it's not enough, so now I've started using tantalum 10-22uF/16V or 25V. Otherwise, I use a 1uF film SMD for ADA4610, ADA4897 .In Walt's original article there are C5 and C13 of 120uF/25V FC for op amp decoupling.
 

Attachments

  • tl3.pdf
    492.4 KB · Views: 30
  • KEM_T2024_T489-1093560.pdf
    1.1 MB · Views: 25
Hello All,

I built, last week, 4 super regulators, 2 positives and two negatives. They work perfectly well.
But I had a problem with the focus because I hadn't read the instructions very carefully.
I wanted +15V and -15V output. the D2 and D7 zeners were 6.8V. The input was at 17.9V. It didn't work.
In the manual, it is written only once that there must be at least 5V of distance between the input and the output. That's why it wasn't working.
I set the input to 30V, D2 and D7 went to 15V and the super regulators worked the first time.
Thanks to the designer of this high-end power supply, which will power a line preamp
Alain
 
That's right, DS suggests 100nF as close as possible to the op amp itself, but in some cases it's not enough, so now I've started using tantalum 10-22uF/16V or 25V. Otherwise, I use a 1uF film SMD for ADA4610, ADA4897 .In Walt's original article there are C5 and C13 of 120uF/25V FC for op amp decoupling.
Yea, about that decoupling. This is where I start doubting if dropping in an ADA4625 is relatively low risk, as for me decoupling is still a can of worms. My understanding is that the devil is really in the details.

For example, all over the place (including on this forum, this post here is a great example) people warn against using different values caps in parallel as apparently that introduces the risk of ringing - exactly the problem we're trying to avoid. (Which a VNA may help display, which goes to show how the seemingly straightforward question of decoupling becomes complicated quickly.) My understanding is that a safe way is to either parallel multiple caps of the same value (e.g. Henry Ott), or to keep the difference in capacitance value low (say no more than 5x or 10x). 100nf next to 10uF is not exactly that. Then again, my understanding originates from reading stuff pretty randomly. And also, in the ADA4625 DS Analog Devices themselves suggest paralleling 10µF or 22µF. I imagine that the good folks over at AD know what they're doing when they say that. As you do the same thing it appears you're not worried about this risk of oscillations?

Also, adding 10µF or 22µF will already be further away from the ADA4625's supply pin. So it'll be less 'local' decoupling and more like adding another 10/22uF decoupling to the power supply rail of the board - in addition to the 100µF at the input and the 100µF at the output. How do you recommend placing this on the board relative to the opamp's PS pin and the 100nF?

What do you recommend for the 100nF? I have a small leaded radial Epcos MKT (the silver colored ones, 7mm lead spacing) that could work well, but that is already a bit big and clumsy for the size of the pins of the opamp. SMD MLCC would be better in that respect, but extremely low ESR. Not a problem if you use a 'regular ESR' cap on the output? (C4)

Vishay TL3 and KEMET T489 ,unfortunately Mouser does not have all the values but for output voltage up to 20V there is no problem with the choice.

Thank you for these. I see why the TL3 is an attractive cap for this board. I looked at the DS of these and apparently they need to be derated by as much as 50% even under 85 degrees C!? If this is true, then what is the point of rating them like they do? Here is a snip:

1709116849576.png


Footnote 5 appears to indicate that they can be operated at up to the rated voltage under 85 degrees, but the table seems to suggest that a 30V rail requires a 63V rated cap or placing more than one in series. Both aren't really doable (63V doesn't even exist in the TL3 range which only goes up to 50V and series placement would bring down the capacitance too far). Strange. I'm taking this seriously as there are plenty of posts on this and other forums warning for 'spectacular' failures of solid tantalums when operated without proper derating.

What about the T83? They are the same form factor, but more usable values are available (i.e., up to 63V, suitable for decoupling a 30V rail considering the above). They don't have the extremely low leakage of the TL3 but don't look too bad. And even as T83s are only available at Mouser and Digikey in the 'low' ESR variety (product designation ending on "L" instead of "S" for standard ESR), the ESR is probably not problematic.

Leaded tants could be an alternative, especially for the T83s: relatively low DCL and inductance spec but probably higher than SMDs.
 
You can achieve great progress with better references for 2.5V and 6.9V and the regulator itself can be further improved. My version now has 4 depletion mosfets and a total of 9 BJT transistors all because of better references.

For my home-made "PM329" 6.9V shunt voltage ref I want to build a small board preferably with SMD components. Have you figured out some SMD components to use for this like you did for the GLED431 in Jung's shunt regulator? If so, I'd be much obliged if you shared some suggestions for the 2N3906 PNP and PN2222A NPN transistor.

The Zener is available in SOT-23. The PN2222A is TO-92 only, but the 2N2222A is available in a hard to hand solder UA or UB package. I assume this is a drop-in replacement but is there anything more solder-iron friendly that can be used instead or is good ole TO-18 the best choice?
 
Member
Joined 2011
Paid Member
When I visit the parts search website www.octopart.com , and when I type "2222A" into its search box, octopart returns some surface mount transistors in the SOT-323 package and others in the SOT-23 package. MouserUSA has > 200K of them in stock and on the shelf, ready to dispatch.

Similarly, when I visit mouser.com and surf to their Discrete Semiconductors -> Transistors -> Bipolar Transistors chart, and then type "3906" into the "Search within results" box, I get lots of different options in surface mount packages. MouserUSA has > 1.6 million of them in stock and on the shelf, ready to dispatch.

A quick comparison of the surface mount transistor's datasheet, versus the thru hole transistor's datasheet, confirms or refutes the hypothesis that they are interchangeable (except for power dissipation!).

_
 

Attachments

  • capture.png
    capture.png
    7.8 KB · Views: 14