advertisement


Howto: TeddyReg

T

teddy_pardo

Trade: Teddy Pardo
Here is an attempt to summarize the original long (and maybe confusing) TeddyReg thread. My preferred implementation is with a Fetlington (thanks to pigletsdad for the idea, and to agent_cooper79 for taking the initiative and providing the transistors).


I'll start with the diagram for the positive circuit:

NewRegFETlington.jpg


Resistors R1 and R2 are used to set the voltage at the output of the linear regulator LR1.

The output is

Vlr1 = 1.25 x (1 + R1/R2)

R2 is anything between 100R and 1000R

The dropout on the gyrator is determined by the ratio of R3 and R4 plus the dropout on T1 and T2. The resistor dropout is a percentage of Vlr1 and is equal to Vlr1 x R3 / (R3+R4). The dropout on T1 and T2 is constant and equal to around 1V on load.

The list of components for a positive 25V regulator is:

R1 = 3.4K
R2 = 150R
R3 = 150K
R4 = 1M
R5 = 100K
R6 = 100-250R

C1 = 0.1uF X7R Ceramic
C2 = 10-20uF Tantalum
C3 = 10-20uF Tantalum
C4 = 20-33uF Tantalum
C5 = 0.1-0.3uF X7R Ceramic
C6 = 10-20uF Tantalum/Oscon or 100uF Silmic, or even better 1-10uF film capacitor (MMK/MKS)

T1 = 2SK117 - GR
T2 = D44H11
LR1 = LM317/LT1086

For the negative version

T1 = 2N5460 (note that the pin layout is different)
T2 = D45H11
LR1 = LM337 (note that the pin layout is different)
Obviously the capacitor polarity should be inverted.

I would recommend to divide the dropout between LR1 and the gyrator evenly. As an example, if the input voltage is 33V and you want the output to be 25V, set LR1 to 29V.

Notes:

1. The no-load voltage is around 0.5-1V higher than the on-load voltage. It is due to the fact that when no current passes through the transistors there is no dropout on them. It can be avoided by adding a 1K - 3K resistor at the output of the regulator, but it's usually not needed.

2. With high values of R3 (50K and above), it takes several seconds until the output voltage rise to the desired value. If you need faster settling use smaller resistors, but the results will not be as good.

3. If you don't want to bother with FETs which are sometimes hard to find, low noise bipolar transistors BC550C/BC560C provide very good results as well.

4. C1 is optional and required only if the regulator is far from the power supply.



Here are some pictures of a positive and negative regulators, mounted on a strip board. The board needs to be cut in one single place only, below the FET.

IMG_3745.jpg


IMG_3746.jpg


IMG_3747.jpg


IMG_3748.jpg


Hope that helps.

Enjoy, Teddy
 
Teddy,

Very useful. I think you probably mean R2 is anything between 100R and 1000R.

Us battery powered types like to save juice, so 1000R when we can be sure that the following circuit will draw 10mA or more, and 100R when we can't, because LT1086/LM317 only operate correctly when supplying 10mA+. 100R will cause 12mA+ to flow out of the reg and down R2/R1.

Nice board you are using.

David
 
Thanks David, I've fixed it on the original message to avoid confusion

BTW, this board will soon become a three way active crossover... :)
 
Hi Teddy, so is this the follow on from " The best Reg I've ever built " thread, if so thankyou for the neg supply.
Teddy I have some very nice axial Solid Tants akin to the Vishay types, these are made by Siemens & ITT, would these be a good replacement for the Tant Beads??
Thankyou for your help
Paul
 
Pulse Studio said:
I have some very nice axial Solid Tants akin to the Vishay types, these are made by Siemens & ITT, would these be a good replacement for the Tant Beads??
Click to expand...

Do you mean wet or solid Tants? Axial are usually wet tantatlum and they work very well. I don't know about axial solid, but I guess that any tantalum will work well here.
 
Thanks Teddy, I believe they are solid types, they are damn heavy for the physical size, I have up to 180uF but only 6.3V, but I do have a couple of 33uF 25V and 47uF 25V, and some 100uF 20V, I also have some others 3.3uF 63V that are Filmcap TFI BD35A, they were all ex millitary caps.
Cheers
Paul
 
It sounds like you have wet tantalum caps. Yes, they are very good, you can also use them for coupling and feedback.
 
Depending on the dropout and the heat-sink, it can withstand 1.5A or even more. However, due to it's output impedance I wouldn't take it that far, especially if you have non-constant current loads. It is best used for preamplifiers, CD player analogue stages, DACs, crossovers, etc. that is several hundrads of mA, with constant current. For higher currents look at the PowerReg
 
Some questions on the choices for the JFETs.

The higher the transconductance of the JFET, the lower the output impedance of the TeddyReg - am I right?

A 2SK117-BL will have a higher transductance than a 2SK117-GR, assuming the same gate pinch off voltage, and thus be a better choice - am I right?
 
The difference between the BL and GR is their Idss. This IMO will mainly have an impact on the initial no-load dropout, but I don't think that it will affect the output impedance. I have some GR and BL types, so I'll try to do some measurements and report.
 
Teddy,
what motivated you to choose 20-33uF Tant and 0.1-0.3uF X7R for C4 and C5? Did they measure better, sound better, or are they in fact better adapted to the circuit (for RF filtering)? Have you tried these same value/type caps in an ALWSR-based VBE, and were these caps suitable there as well? I tried modest-cost X2 caps you suggested in ALWSR-based VBE and they work very well...
 
I may have missed this in the archives ...

In theory or practice ... How does the TeddyReg, in its current form, compare to ALWSR regulator for the Flea clock specifically. By this I mean, if one removes the ALWSR regulator part of the Flea and replaces it with a TeddyReg, what performance gains/losses may be encountered?
 
The ALWSR regulator and the Flea are two entirely separate projects.

The Flea is a low-power, low-noise regulator that can deliver around a few 10s of mA whereas the ALWSR is a low-noise regulator that can deliver in excess of 1A. Neither depends on the other and neither is a part of the other, so it isn't possible to [remove the] "ALWSR regulator part of the Flea and replaces it with a TeddyReg" because that part doesn't exist.

They both share a similarity in that they use opamps (AD815 for the ALWSR and AD797 for the Flea) as error amplifiers, meaning that they are active regulators. However, for the purposes of comparison it's only fair to look at the TeddyReg and ALWSR because of the limited current delivery of the Flea. That's not to detract from the Flea: for its intended purpose, it's probably very hard to beat it in terms of performance and noise measurements.

In comparison then, Teddy's reg is purely passive and does not use any active error correction. Its design is based around a simple linear regulator and capacitance multiplier with steep RC filtering. The ALWSR is, on the other hand, a highly sophisticated active error-correcting regulator based on Walt Jung's designs.

As for performance gains/losses, this hasn't been measured scientifically - with the exception of output impedance; the TeddyReg has an output impedance which is orders of magnitude higher than the better-performing ALWSR.

Having said that, anecdotal reports suggest that those who have listened to circuits powered by TeddyRegs prefer the sound to circuits powered by ALWSRs, but there hasn't been wide-spread testing of this yet. I'd expect a lot of people to start building and testing these over the next year.

I hope that goes some way to clearing things up.

Kind regards,
Carl
 
Don't do it... This regulator needs about 10 seconds to reach the desired voltage. Enough time to make you CD spin very fast without clock control...
 
Thx for your detailed reply.

hacker said:
They both share a similarity in that they use opamps (AD815 for the ALWSR and AD797 for the Flea) as error amplifiers, meaning that they are active regulators.
Click to expand...

But this ALWSR_rev2.9 document...

http://www.alw.audio.dsl.pipex.com/Manuals/ALWSR_rev2.9_iss009.pdf

... and this Flea doc....

http://www.acoustica.org.uk/t/naim/35clockreg.html

... state AD797 for BOTH regs. Or might I be looking at old version of either/both regs.?

BTW, IMO these regulators need better "publicity" than the kind of helter-skelter way they are scattered on multiple sites (and multiple formats: pdf, HTML). This makes locating/comparing a challenge for novices like me. I think some kind dedicated page or a forum "sticky" that has 'em all -- and the latest/updated version of each (including the Teddy stuff) -- would go a long way.
 
Transconductance is measured by the change in Idss divided by the change in gate to source voltage.

BL grade has a higher Idss at a given gate to source voltage, as compare to the GR grade Toshiba JFETs, and thus a higher transconducter and lower output impedance when used in a TeddyReg.

Am I correct to assume the above? Thanks
 
You must log in or register to reply here.


advertisement


Back
Top