advertisement


LM117/317 datasheet tracking pre-regulator

Christian,

Yes its essentially the same circuit. I Joined pfm in Sept 2004 and prior to this pfm member Andrew Weekes had designed a PCB implementing the 2000 Jung version. There were several group buys and lots of discussion. Caps were a hot topic. The regulator as implemented by Andrew was known here as the ALWSR for short.

This link is to the manual, a more recent copy than mine:
http://www.connect-audio.co.uk/Regulator/ALWSR.pdf

There was a commercial site selling boards but it has closed:
http://www.at-view.co.uk/alwsr.htm

I think Andrew withdrew from this forum around 2006, possibly due to some ill health. I would have to go searching for threads, so you'll probably be more satisfied with your efforts than mine. On pfm he has user name Andrew L Weekes and on diyaudio ALW.

He appears in the long thread on DIYaudio : 3 terminal Jung Super Regulator Kit

David
 
SPArts155,

Good point on the elcap leakage current. I would be as a child trying to find a new op-amp. You probably know that Walt found that the AD825 sounded better than his first choice of AD797. What chance have mere mortals got? Analog and Natsemi have new ranges of op-amps with improved processes since the original articles, so maybe there is a better device out there.

I mentioned silver oxide batteries because they have an almost constant output voltage until they expire. With any kind of load, alkaline output voltage drops and so will the regulator output. In fact for the output to drop from 27V to 24V is a small loss in sound quality compared to the gain from a better Vref, so perhaps I should stick with them.

Its a given that new gain setting resistors will be needed. Sad part is that I matched my original input impedances to 0.25%.

David
 
I have recently implemented TPR feeding a 5V DAC chip, as per acoustica's schematic.

I was not quite satisfied with it.

After a week or so I have found it as overly warm, lacking the midrange and high frequency transparency to some degree, but surprise was quite a good bass, big and bold (low output impedance?).

It would be great if someone could post a layout just to check if my (and those of others) approach was adequate.

Baki
 
Baki said:
I have recently implemented TPR feeding a 5V DAC chip, as per acoustica's schematic.

I was not quite satisfied with it.

After a week or so I have found it as overly warm, lacking the midrange and high frequency transparency to some degree, but surprise was quite a good bass, big and bold (low output impedance?).
Click to expand...
Baki: Thx for your feedback.
Which version of the TPR did you build? What other regs are you comparing the TPR to?(see my note below)
The earlier version of Martin's TPR (ie. the version with C_adj I built at the start of this thread; not the latter "experimental" version) was an improvement -- over a much-tweaked single-317-based reg (tweaks include snubber, etc) -- in the the qualities you've noted. I think it takes a few dozen hours for components to properly "break/burn in". So give it a bit more time.
Martin: The SPICE sims, suggested by some, are IMO not that important. On your acoustica page, with earlier version, you conducted far more important o'scope tests. So if you could 'scope out the new TPR, that's a better use of your time than SPICE-ing.

On the use of LED ... I've heard that room fluorescent lights can create distortions in the LE-diode. So it may be a good idea to keep the new TPR "under the hood" ;)

EDIT:

Not sure that I, myself, have clearly and comprehensively spelled out what I've compared/contrasted the TPR to/with/"against". So, FWIW ...

- Snubbered/tweaked in-line regulators (i.e, tweaked on the equipment's PCB, component or solder side)
- Tangentsoft.net's TREAD and STEPS (Note: As of mid-2010, Tangent removed his PSUs from his site w/o much notice [dunno why: mesuspects the competiton was too intense and/or his PSUs were not all that great to begin with ... see this DIYAUDIO post ]. Complete pages of the Tangent PSUs are still avail at Archive.org as noted in the just-given links). Both the Tangent PSUs also have rectification stages (the STEPS has an xformer and elaborate chokes) -- but they are merely basic single-317 datasheet schemas. Nothing more. To equalize the comparison between the STEPS and TPR, I used the STEPS rectifier to feed the TPR.

I have built a PFM Flea clock, but I have not built/tried any of the other "exotic" regs ... including ... ALW, TeddyReg, SuperPower, Paul Hynes Design, Tent Labs, etc.

Misc. Notes:

Quantitative metrics via Fluke 87-mk3 DMM, both under- and no-load condit. To my chagrin: my Tek 'scope is still on the blink and no SPICE modeling, either!!!

Used in external PSUs: A NAIM-like external job for Philips/Magnavox CD-650 (and other TDA1514A/SAA7220/etc./-based CDPs). Also used for headphone amp external PSU.

I haven't yet built the Notes 4 (LED) TPR -- only the prev. versions, with C_adj, and some experimentation with addit. bypassing and gyrators (the gyrator -- at least my implementation of it -- made SQ worse [see earlier posts in this tread].
 
I have implemented the latest version - Notes 4 (LED) TPR posted on acoustica.org. I have put two green leds for 5V and additionally 10uF CAdj.

It is very compact so I suppose the build is not an issue.

It feeded two WM8741 dacs so total ripple is around 100mA.

I have compared it with a stock NJM78M05 regulator (80 dB of ripple rejection) and home made first stage (first half) of the SuperTeddyReg, because of the space consideration.

TPR, as well as the first stage of the SuperTeddyReg have already played for 20 to 30 hours each...

Comparing to the stock regulator TPR was a bit better in some areas, but unpleasantly warm and uninvolving. It was not a real pleasure to listen to but, as I said before, bass was surprisingly good.

The first stage of the SuperTeddyReg, which has also played for a whole week, was better in every aspect except the bass. The bass was weak, not fully developed.

During this weekend I have managed somehow to implement a "complete" SuperTeddyReg (I ordered a diy pcb) so I would wait for another couple of days. Seems that the bass could also be an issue but to the less degree. Will see.
 
Baki: I don't know what you mean by "It feeded two WM8741 dacs so total ripple is around 100mA." IME, "riipple" is usually reported in mV (milli-volts).
You may be correct about "warm" TPR sound. I find its extreme low bass to be v. good and detailed.

What about PRaT? The TPR does this at least decently, so IMO, this is "involving". Not sure about Teddy (no experience).

BTW ... a correction from my last msg. I do have two Flea regs (themselves cascaded down from 15V TPRs) powering low-power digital-logic ICs in a CDP
 
13mh13 said:
Baki: I don't know what you mean by "It feeded two WM8741 dacs so total ripple is around 100mA." IME, "riipple" is usually reported in mV (milli-volts).
You may be correct about "warm" TPR sound. I find its extreme low bass to be v. good and detailed.

What about PRaT? The TPR does this at least decently, so IMO, this is "involving". Not sure about Teddy (no experience).
Click to expand...

I meant to say "load current" instead of a "ripple". As we increase the load the output impedance of the LM317 drops. That could be the reason of a very good bass of the TPR.

As for the PRaT I would say that, after three listening sessions (3-4 hours each), SuperTeddyReg is better and not by a margine. Actually, almost everything is better. I would only like to have a TPR's very good bass reproduction...

DIY version of the SuperTeddyReg (pcb + shipping is 10 €, parts up to 10€) is worth every cent (penny).
 
Baki: thx for reporting back.
PRaT is important, and if STR really is superior (for this attribute) by a wide margin, that's important. The Wolfson DACs are known for their v. good innate bass perf., IAC, so STR may be a much better choice (match) for them despite superior bass of TPR.
About STR ... never built it ... but even at DIY pricing, it's still quite $$. I'm not surprised it performs as well as it does. The development thread for it, here at PFM, is long and extensive and a lot of trial/error and crowdsourced effort seem to have gone into its evolution.
I pointed to the Belleson Superpower reg. a few days ago (funny how the prefix "Super" keeps appearing in the lexicon!).
Belleson SEEMS to conduct a lot of "scientific" comparisons against some other popular reg ... including ... "Superpower compared to SuperTeddyReg":
http://www.belleson.com/comparesuperted.php
 
Baki/anyone:

What are you feeding your TPR (or other reg.) output to? E.g., DAC (D/A processor), CDP, amplifier, pre-amp, etc.

What are you feeding your TPR (or other reg) with? I.e., rectifier, transformer, SMPS, etc. How much overhead voltage are you using -- have you found an "optimal" amount (e.g., post-rectifier/pre-rectifier V_dc = 10 for a 5.0 and 3.3 V_dc apps, as in the Wolfson/other DACs, SRCs, logic ICs, etc.)?

"I'll show you mine, you show me yours!"

As noted before, I'm using these regulator projects mostly for playing around in experimental units; mostly older Philips CDPs, some recent Musical Fidelity DACs, Asus Xonar XT soundcard, etc. Believe it or not, 95% of my day-to-day listening is via sub-optimal PC audio or iPod Touch. ALL my listening is headphones/IEMs-only. Haven't used loudspeakers in years!

Images of Philip CD-650 and DIY external PSU project below:

IMG_6381_090326_013204_crp_sm.jpg

left: External PS
center: heavily modified CD-650
right: DIY headphone amp; Sennheiser HD-650


IMG_6376_090326_012056_crp_sm.jpg


IMG_6378_090326_012132_crp_sm.jpg

External BALANCED-pwr PSU: Three DIN V_dc out (two on front; one rear). 120VAC (balanced pwr out); Corcom EMI/EMF filter; Inrush current suppression (soft start)

IMG_6379_090326_012320_crp_sm.jpg

Philips (Magnavox) CD-650 (1986 model); Kwack clock; DEM reclocking for TDA1541A; inter-IC re-clocking; double-Flea (far upper right) for certain logic ckts.
 
Martin (or anyone) ... can you comment on this:

Those 12v, 2A SMPS adapters (also sp. 'adaptor') are ubiquitous ... e.g., used for external hard-drives.
309720404_322.jpg


I've got several lying around and want to adapt them to a decent-quality 9-10v adapter for audio use (specifically, headphone amp that normally runs on 9v batts.).

I have a Tangent TREAD reg., which is based on a single LM317, and pretty much goes off the datasheet. I thought putting TREAD after the 12v, 2A SMPS would give me okay pwr (haven't built anything yet!). But Tangent says this is not a good idea because:

It's probably not isolated:

Power Supply Matters

Even if you happen to get lucky that way, it probably has too high a switching frequency for the LM317 to be very effective. The LM317 is best at handling low-frequency noise.

If you absolutely must linear-regulate HF noise, you either need a wide-band regulator like the Jung/Sulzer types, or you need lots of passive filtering (CLC and such).
Click to expand...
So what do you folks think? ... any way to QUICKLY/CHEAPLY ... uh ... 'adapt' the noted SMPS adapters for clean audio pwr ... or give up and go back to linear (transformer) -- non-SMPS -- routine?

EDIT: Some putatively better (Medical-grade) SMPS adaptors are avail. from, e.g., Mascot.no, ELPAC, etc:
http://iccnexergy.com/elpac/medical-power-supplies/external/mw12/
http://www.farnell.com/datasheets/31423.pdf
http://www.mascot.no/?CatID=1186
Dunno how good they are WRT noise, ripple, isolation, etc.


NOTE: Tangent's TREAD is only avail on Archive.org
v1.1-assembled.jpg
 
Medical grade means low chance of giving somebody a shock; it has little to do with noise on the output.
 
PigletsDad said:
Medical grade means low chance of giving somebody a shock; it has little to do with noise on the output.
Click to expand...
The term "Medical-Grade" is also loosely packaged with Hospital-Grade ... IAC, it can ALSO mean less EMF/EMI, so as not to interfere w/sensitive bio-medical equip. This means less noise/cleaner pwr. Not just radio, but hash back into the pwr lines (i'll look for some refs on this)
As far as the adapters I noted ... not sure which std. they comply to?
 
Hello.
I'll just resurrect this thread as I'm looking to build this psu with a lm338 or lt1084.
Now, I didn't understand how to calculate the R1/R3/R4 resistors for any output voltage. I'm looking for 13V and 16V output (I'll make two of them for small amplifiers). The thing is that I already have a 16VAC transformer thinking that I'm going to get 16VDC on the output using the normal regulator application but the notes on this application state that I need about 6V headroom. Doing the math I can work out to about 15.5V maximum output from this? The maximum current drawn would be in the range of 2-3A maximum.
 
For the TPRs you could go the zener route with 12v and 15v zeners which should give you 13.25 and 16.25 as Vout. If your traffo gives 16v AC then you should get 16x1.414=22.6v DC
 
I'd like to use the LED version, with resistors.
16VACx1.414 -1.1V from the bridge I get 21.5V into the regulator. If I can drop 5 to 5.5V max that would land me in the sweet spot for my amp.
 
Then calulate R3 / R4 for about 16v out. From nearest standard ressitor values R3=240ohms, R4 =2700ohms would give you about 15.8v out.

R1 you can just leave at anything convenient 100-470ohms regardless of output voltage setting.

The problem here is your likely current draw - your range of '2-3A' is way outside the capability of the LM315. LM338 would work, as will LM1084; but I'd implement these just as a straight regulators first (ie no tracking pre-reg). You can also add a large PNP external transistor to an LM317 to get to these or greater current outputs, but again, as a single regulator - making it a tracking or pre-regulated reg becomes complex and wasteful (heat) very quickly.

For such higher current regs splitting your reservoir cap into two and adding a small impedance between the two (CRC or CLC) is a much tidier approach with good benefits to reducing HF noise from the supply.
 
I saw that both lm338 and lt1084 have this appliction in their datasheets.
I have 3 paralleled filtering caps and was using a single regulator in my design but someone recommended me that I make a version with a tracking preregulator. I think R1 should be 120 ohm for lm338/lt1084. I know how to set the output voltage for single regulator so if it's done the same way I will give it a go. Thank you.
 
I stand corrected. On the lt108x datasheet there's no tracking preregulator application. Only on lm317/lm338.
Another thing, if I add a tracking preregulator, how much would would the last regulator have to deal with? Could I leave it with no heatsink at 3A output?
 
You must log in or register to reply here.


advertisement


Back
Top