LM317 load capacitance

[dotneck335]

I just found the schematic for the LF353-assisted LM317/337 voltage regulator board:
LM317 LM337 Active Servo Adjustable Regulated Power Supply Bare PCB Board LF353 | eBay
It appears that the LF353 is being used in AC-coupled inverting mode feeding back to the adjustment leg of the LM regulator. I guess this is to reduce noise? Comments?

 

[infinia]

Comments?

What are the sellers claims, does he have any data?
why do you need to add band-aids on a 3T Vreg for line level analog stuff ?
If there is a problem fix the problem 1st, don't add complexity on top of basic circuits , or PCB layout problems. Some of these products are dreadful.

 

[mchambin]

+1
Until more information about this mod.
IMO, this is BS.

 

[Shinja]

Yes it is. I saw circuits like this on Matsushita/Technics products but using 78xx regulators.
.I think LF353 is not suitable here because it is not so low noise part.Typical bipoler opamps may be good enough. Power on transient is another concern both bipoler or LF353. LF353 may have phase reversal symptom often being old FET opamps which may cause something bad at power on/off?

 

[traderbam]

The op-amp appears to be adding 30dB of negative feedback to the LM317 at ac over a wide bandwidth. I see what's intended but is it stable? Having those 10nF caps across the output isn't optimal (as has been discussed).

 

[mchambin]

+1
At best a tradoff improving ripple or load regulation for less stability.

 

[flyingfishtw]

I found a really interesting paper on this subject ,
http://www.calex.com/pdf/3power_impedance.pdf

Originally I thought the cap value, esr, output impedance, were just dark magic.

 

[dotneck335]

Yes it is. I saw circuits like this on Matsushita/Technics products but using 78xx regulators.
.I think LF353 is not suitable here because it is not so low noise part.Typical bipoler opamps may be good enough. Power on transient is another concern both bipoler or LF353. LF353 may have phase reversal symptom often being old FET opamps which may cause something bad at power on/off?

Yes, I agree that the LF353 is not the ideal selection here. I think an OPA1652 would be a better choice---much less noise, lower input bias/offset current and lower offset voltage.

 

[dotneck335]

Having those 10nF caps across the output isn't optimal (as has been discussed).

Sorry, I missed that part of the discussion. Please enlighten me! I also noted the lack of the datasheet-recommended protection diodes across the in/out of the regulator IC and the reference resistor.

 

[dotneck335]

What are the sellers claims, does he have any data?
why do you need to add band-aids on a 3T Vreg for line level analog stuff ?
If there is a problem fix the problem 1st, don't add complexity on top of basic circuits , or PCB layout problems. Some of these products are dreadful.

This is not necessarily used for line level---could be for a microphone, guitar, or phono preamp with fairly large amounts of gain. Lowering the noise of the supply could be quite beneficial, and it's really pretty simple, not complex at all.

 

[ammel68]

or phono preamp with fairly large amounts of gain.

Really?? I certainly wouldn't use LM3xx regulators for that application with so many better options out there from T.I. and Linear Technology.
I find their performance mediocre at best for low gain, discreet line stage/buffer applications.

They work okay with most 6dB gain op-amp circuits I've tried. Then again, I haven't found many regulators that didn't work in those applications given op-amp's very good PSRR.

 

[infinia]

Lowering the noise of the supply could be quite beneficial, and it's really pretty simple, not complex at all.

OK lets say it is easy. What could be easier than making out a wish list. What ripple and noise specs would you need. AND would you be able to verify it once I put the thing in front of you.
a data point for your files...A low wattage lab.grade DC linear supply is spec'd @ PARD < 200 uV RMS (BW= 10MHz) which can cost hundreds of dollars and weight over 10 lbs.
Periodic and Random Deviation (PARD)
Noise and ripple voltage superimposed on a converter’s DC output. Typically specified at full load, it is expressed in peak-to-peak or RMS volts over a given bandwidth.

 

[dotneck335]

Well, I don't know....but the Jung-Didden regulator and others show less than 1uV noise at < 100 KHz on this graph....unless I'm reading it wrong:
https://cdn.shopify.com/s/files/1/1006/5046/files/V4_JW_F7.pdf

 

[infinia]

LM317 from the graph I get 1.8 mV BW=10MHz

1.8E-7 V sqrtHz * 10E6 Hz


I measured a fully bypassed TL431c biased at 15mA~ 330uV RMS

 

[dotneck335]

Why is noise at 10MHz relevant for audio?

 

[ammel68]

Why is noise at 10MHz relevant for audio?

Dotneck...the Jung/Didden regulators are simply "Da Bomb" for lower level circuits! You may question their noise level at different frequencies for whatever reasons, but apparently you don't get that these are some of the best performing regulators on the planet??
Just remember...in the end it's ALL about the SQ and that's where the Jung/Didden regulators REALLY deliver.

In the unlikely event you have a LM3xx circuit that will outperform Walter Jung's Super Regulator, PLEASE post it here as I want to see it.

Merry Christmas and Happy New Year!

 

[mchambin]

Noise at 10 Mhz. ROFL.

 

[dotneck335]

Dotneck...the Jung/Didden regulators are simply "Da Bomb" for lower level circuits! You may question their noise level at different frequencies for whatever reasons, but apparently you don't get that these are some of the best performing regulators on the planet?? Just remember...in the end it's ALL about the SQ and that's where the Jung/Didden regulators REALLY deliver.
In the unlikely event you have a LM3xx circuit that will outperform Walter Jung's Super Regulator, PLEASE post it here as I want to see it.
Merry Christmas and Happy New Year!

Whoa, whoa...you're misunderstanding me. NO DOUBT that the Jung-Didden regulator is the BEST! I'm not questioning their noise specs; I'm asking infinia why he is quoting noise specs at 10 MHz?? 10 MHz noise does not seem relevant to audio. The Jung-Didden regulator IS the best, but if you total up the cost of parts on the B.O.M.(including the ~$20 for the PCB), it's over $40!! The regulator board I found (LM317 / LM337 +/-1.5V~37V Adjustable Dual Voltage Regulator Power Board | eBay) is only $10 total!! So I am wondering how close it would come for being 25% of the cost--that's all! If the noise contributed by the regulator is 100 db down, I doubt that I (or anybody else except Max Headroom) can hear that.

 

[traderbam]

Why is noise at 10MHz relevant for audio?

Please don't use common sense; it has a tendency to cause panic in the forum.

 

[jackinnj]

The Jung-Didden regulator IS the best, but if you total up the cost of parts on the B.O.M.(including the ~$20 for the PCB), it's over $40!! The regulator board I found (LM317 / LM337 +/-1.5V~37V Adjustable Dual Voltage Regulator Power Board | eBay) is only $10 total!! So I am wondering how close it would come for being 25% of the cost--that's all! If the noise contributed by the regulator is 100 db down, I doubt that I (or anybody else except Max Headroom) can hear that.

How much is your time worth. Would you buy an Armani suit and put on dirty trow underneath?

Fortunately, you can hear the difference -- at least we could when all the regulators shown in the graph were used to power the Borbely All-JFET Line Amp. Mind you, the Borbely pre, very low noise and vanishingly low THD, only has 40dB of PSRR so any problem on the supply rails makes itself manifest when it gets to your ears!

The error amplifier of the LM317/LM337 is slow by comparison, the result being harmonics impinging upon the supply rails.

If I had to use an off-the-shelf OEM regulator I would prefer the LT1963A.