[randytsuch]

Just in case someone wanted a good voltage regulator

In the past, ALW sold his version of the jung as a kit, but it was rather expensive.  There were some group buys at diy places for reasonable prices, I bought some from a diyaudio group buy, but if you missed the buy, you were sol.

But now someone is selling just the pwb's, or assembled boards for a reasonable price
http://at-view.co.uk/alwsr.htm

About $12 for a pwb, and $130 for a built/tested module.

With the right heatsinking, you can get at least 1 amp out of these.

I think I have collected all the parts to build these, and somewhere I have a spreadsheet with the information, if anyone is interested.  There is a different version for negative voltages, and a slightly different PL for voltages less than 5V.

Randy

[tonyptony]

May not be enough for a SB3. A thread a while back on the Slimdevices website forum indicated that the SB3 is capable of drawing >1A at various points in its operation. Standard engineering practice in a case like this is to derate around 50% beyond what you'd need. A supply capable of providing at least 2A would be a safer bet.

[randytsuch]

May not be enough for a SB3. A thread a while back on the Slimdevices website forum indicated that the SB3 is capable of drawing >1A at various points in its operation. Standard engineering practice in a case like this is to derate around 50% beyond what you'd need. A supply capable of providing at least 2A would be a safer bet.

I decide to doublecheck, and found I made a mistake 

In the manual, Andy said he guranteed 200ma, but most would put out 500+ ma.  I used a couple in a dynahi class A headphone amp, which was supposed to use 600ma per rail (I never measured it).

For other reasons, I have not used this amp much, but the jungs did not have a problem supplying this much power, but I did use some hefty heatsinks on it.

For a SB3, there are a couple options.  Use a jung to replace the 3.3 reg, which I am pretty sure it could handle.  Or, take some stuff of the 5V input, like the display power, and have a seperate supply for that stuff.   Not a simple drop in kind of thing.

Randy

[art]

For a SB3, there are a couple options.  Use a jung to replace the 3.3 reg, which I am pretty sure it could handle.  Or, take some stuff of the 5V input, like the display power, and have a seperate supply for that stuff.   Not a simple drop in kind of thing.

Randy

There you go! Try making one just for the chip that handles the SPDIF out and oscillators. Keep that chip's supply as far away from the rest of the of the unit.

Pat

[Paul Hynes]

What you will hear with the Jung type regulators is that 3 terminal regulators are not very good for audio use. The Jung regulator is a noticeable improvement because it offers better supply line rejection and lower impedance over a wider bandwidth than the original 3 terminal regulators. However it is not speed optimised for digital circuit load current swings with clocking speeds of tens of MHz. You have to be damn fast to keep up with these fast current changes. Most 3 terminal regulators have lost the plot at frequencies over 1 MHz (in fact you could argue this for much lower frequencies). The way to look at this is the better the power supply performance the better the sonic performance and it’s all a question of degree. You have to consider what the load current changes do to the regulator output and how the resultant reactions will affect the signal processing circuits.

The error amplifiers currently specified for use in the ALW version of the Jung topology have an open loop bandwidth of around ten times that of the typical three terminal regulator which is heading in the right direction but not far enough. The transient response settling time and slew rate are quite reasonable for low frequency analogue circuit supply but these error amplifiers could be quieter for low-level analogue use. Unfortunately the error amp originally specified well on noise but was rather slow, and from all reports, suffered stability difficulties.

Regulator speed and settling time is of the essence with digital products like the CD PRO2, Squeezebox and the associated DAC products often partnered with these items. You need to go further with regulator performance to get the best sonic performance from these digital products.

A word of warning for any DIY enthusiast thinking of replacing the error amplifiers in the ALW/Jung circuit for something much faster. If you don’t know what you are doing and/or don’t have wide bandwidth test equipment, don’t just throw a high-speed error amplifier into the circuit and expect to automatically get an improvement. It is very easy to end up with an oscillator rather than a regulator and a budget 20 MHz scope won’t notice oscillations of a few hundred MHz.

Regards
Paul

[JoshK]

Thanks Paul for the summary info.  This sort of big picture view is helpful for those of us tinkering DIY'ers.

My naive assumption would be why bother going to all the trouble with super regs in front of the SB when they just feed into switchers to various other portions of the circuit.  I would rather rid of the switchers and build regulators for each section, something I want to do (when I get a round tuit).  Paul's comments means I have to chew on the digital psu some more.  I'd probably go over to diyhifi.org and see what those guys are doing and probably end up buying a kit from tentlabs or something similar.

[art]

C'mon, Paul...you are being generous.

Yes, it makes little sense to put an exotic outboard supply to a SB2/2, and leave the internal regs "as is".

The Jung-style regs are very sensitive to the error amp used. I do not advise the average DIYer futzing with changing them. You are asking for problems 99% of the time if you do.

Of course, if you don't have a good 'scope you may not know how deep the doo-doo you just stepped into really is.

As for "high-speed digital" supplies:

No supply is doing anything at the really high frequencies. It is all the bypass caps. The supply keeps a tight control on operating point, which is necessary to prevent jitter from decision point shifting. But, regardless of the design, they all crap out at high frequencies. There is simply not enough loop feedback available to do anything.

Pat

[randytsuch]

Thanks for the advice guys.

Now that my original thread is off topic, I guess a little more won't hurt.

I just looked at the tent labs shunts, and he only claims a bandwidth of >250kHz for it, but it may be that he is unwilling to play the specsmanship games that companies play.

In any case, I actually have two SB's, a SB2 and a SB3.  I think I will take one of them, and build up a 3.3V Jung, and a 5V jung.  Planning to use the 5V for the DAC, and the 3.3V for the oscillator and SPDIF driver.  I guess I could build another, simplier reg for the FPGA, not sure if I will be motivated enough to do that.

BTW, I know my limitations (at least some of them), and I have read about oscillation issues with Jungs, so I am not going to change anything from ALW's implmentation of the Jung to try to speed it up.  We do have a 1.5 gHz Tek scope here at work, but I have not figured out how to use it yet (I migrated to the dark side, management  )

Randy

[Paul Hynes]

You will gain some very useful benefits from dealing with the short-comings of the master supply. However, you guys are right about the need to deal with the local regulators. This is where you can make the most profound improvements, right at the business end. As Pat pointed out, the SPDIF and oscillator sections as well as the DAC itself are usually quite sensitive to poor power supplies. To do the job properly you have to deal with all the supply regulators and the power distribution. Until you do this you aren’t hearing anywhere near the full potential of your equipment. It’s rather like the dance of the seven veils. As you peel a veil away you reveal more each time.

The Tent Labs regulators have relatively slow error amplifiers.

Pat, I’m a generous person by nature. I stopped using three terminal regulators 27 years ago because I didn’t like them then. Not a lot has changed in the world of three terminal regulators since then.

Sooner or later every error amplifier will run out of loop gain and cease to function. However it is certainly possible to achieve good regulation at the clocking frequencies in use in the current crop of digital audio electronics. The question is where does the error amp crap out. The more extended the error amplifier’s regulation ability, with frequency, and the faster it’s load transient settling time the less it will inflict supply induced jitter on the data stream. Bypass caps are part of the equation and need to be chosen and applied just as carefully as the error amplifier.   

Sorry Randy I didn’t mean to take you off topic. There is a lot of interest in power supplies in general. Perhaps the circle could do with a specific section devoted to all things power supply related.

Regards
Paul

[randytsuch]

Sorry Randy I didn’t mean to take you off topic. There is a lot of interest in power supplies in general.
Regards
Paul

No problem.  This is much more interesting to me anyway, I just started the thread in case anyone was wanted to build a jung.  I have a few boards in my project drawer, waiting to be stuffed.

Over at diyaudio, and originally from pink fish, there is buzz over something called the "flea".

I will figure out how to post the schematic later.  They had kits for awhile, looks like they sold them out.

Randy

[randytsuch]

OK, here is the flea schematic

[ezkcdude]

Hey, Randy. I know Ray (6h5c) over at diyaudio has built the Flea to bypass the local clock on my ezDAC. He'd be a good person to ask about that, I think.

http://home.quicknet.nl/qn/prive/ra.vdsteen/flea_en.html

[Paul Hynes]

The AD797, used in the Flea and the earlier version of the AKLW/Jung regulator, may be quiet but it is not suitable for use as a wideband high performance regulator. It may be an improvement over a three terminal regulator but it’s too slow for high performance use with digital circuits clocking in the tens of MHz area. To be frank I wouldn’t use it for regulating analogue circuit power supplies either, as its impedance curve falls through most of the audio band and it’s settling time is barely better than a three terminal regulator.

Regards
Paul

[art]

Not much has low-Z at tens of MHz. Other than bypass caps.

Agreed that impedance that is not flat across the audio band is the best approach. Some will disagree, which is fine.

Pat

[randytsuch]

Not much has low-Z at tens of MHz. Other than bypass caps.

Agreed that impedance that is not flat across the audio band is the best approach. Some will disagree, which is fine.

Pat

Which brings up the question I was going to ask.  Why can't you depend on capacitance for the MHz range, and use something like a Jung to get up to where the cap can take over?  Maybe you could get better performance if you had both a regulator that could work into the 100's of Mhz range, as well as good caps, but I for one, am not sure how to make a regulator with that kind of performance.

My current thoughts are to spend some time studying Walt Jung's articles from his website
http://www.waltjung.org/Regs.html
Then add in a jung or two into my SB.

May not be the perfect solution, but I am not sure how to do better.

Randy

[Builder Brad]

Hi Randy,

thnx for the info on the Super regs. Most of the good deals here seem to be in the US, or Aus, so its nice to be able to deal locally!

FYI

the SB2 uses an internal switcher for the 3.3v rails and the SB3 was upgraded with a linear reg for its 3.3v.

It would be interesting to know if the SB2 sounds different to the SB3, using the digital output, in standard guise, in theory the SB3 should be better here.

You could then tweak the SB2s 3.3v power with the super reg, and see if you can better the SB3.

Brad

[Builder Brad]

For anyone looking for a Super Reg with higher power capabilities, Its worth looking at someting like the JSR-04 as per the P-A web site:

http://sjostromaudio.com/_unsql/hifi/index.html

Peranders is very approachable and helped me work with one of his standard PCBs

I used a modded JSR-04 to give +5 and +12 volts for my SB2 with good results. This SB2 was originally modded by Red Wine Audio with the Battery upgrade and the JSR-04 power supply sounded considerably nicer that the battery.

I have a second SB2 with similar mods using only a single 5v 3 amp Power One supply that I cannot get to sound as good the SB2 with the split supply.

I will use one of the Jung modules that this thread refers to for a clean 3.3v supply in my 1st SB2, luckily I have a 12v rail to work from so I can get the full pre-reg benefits minus the concerns re. voltage headroom

Brad

[randytsuch]

Hi Brad
Thanks for the info, I did know that Peranders has a jung version too, but I have not tried his.  I guess SM would not be a big deal, there aren't that many parts.  I wonder how his SM version compares to ALW's through hole Jung?

But, I already have the ALW boards, and just verified that I have all the parts, I just need to put them together now.

BTW, there are some things you can get over there, that I am having trouble finding.  I was looking for Elna Starget caps, ALW is recommending them now, for a couple of the caps, and they are hard to find over here.  I already have some Rubycon ZA's, so I may just go with those.

For your SB2 versus SB3 question, it would be interesting, but I am not good at doing comparisions like that.  I know  better than to say I will do it, because most likely, it will never happen.

Randy

[peranders]

Hi Brad
Thanks for the info, I did know that Peranders has a jung version too, but I have not tried his.  I guess SM would not be a big deal, there aren't that many parts.  I wonder how his SM version compares to ALW's through hole Jung?

This is an old thread, yes I know, but my latest 4-layer board super regulator got tested by jackinnj and I must say that I got very happy to see that my pcb was even better than the Jung/Didden version although the difference was small. The result of the noise measurements was 921 nV vs. 1.46 uV Jung/Didden.

Thread here

Randy, did you build anything?