LM317 based PSU for 48v troubleshoot

[ohm]

Hello 

I've been trying to build a 48v supply for phantom using a modified 24v lm317 based circuit board. Schematic #1 below.

The 48v circuit is based around schematic #2 below

To create the voltage doubler I replaced the D2 and D3 (schematic #1) with 2 1000uf 63v capacitors as I do not have any 470uf 50v caps at hand

I was also unsure of the 'series resistor' on the 317 input value (schematic #2) as this is not stated so used 68R 1W as saw another diagram somewhere (cant remember where ) using a 47r 1W

Also cut the track on the cct board between D4 and the input of the LM317 and soldered the resistor here

Added diode accross 317 in/out

Omitted C4 and C1 from the original

Used the 4k7 trim pot instead of 10k

Also there is a 1000uf 63v across the output replacing the 330uf 63v (schematic #2)

Here is the problems I have -

The circuit puts out 70v DC

The 4k7 pot has been smoked

Any fixed value resistor put in place of the pot also starts smoking

The voltage doubler is putting out around 70v DC  before it gets to the lm317 :-\

why would it give 70v DC when the transformer is only putting out 27v AC

Are my oversized capacitors causing issue?

And would this smoke the pot?

This is causing much frustration as it should be a simple circuit and is the last thing standing between me and pumping 48v to any mics attached to these beauties  :-[

Any help appreciated  8)

 

[mjrippe]

Check the orientation of the LM317 and the diode across it.  Also check your circuit for solder bridges and cold joints.

 

[ohm]

just double checked the 317 and the diode, they're in as they should be... also no solder bridges or cold joints

Without the 317 in circuit the voltage doubler is still putting out 70v DC from 27v AC

Any ideas why?

Thanks

 

[Brian Roth]

"In theory", a voltage doubler will produce a DC voltage twice the PEAK value of the AC source.

In this case, the Peak AC voltage would be 27 * 1.414, or approx 38.178V.  Double that and you have 76.356 VDC.

In The Real World, an actual measured voltage of 70 VDC sounds in the ballpark.  That is a decent "raw" voltage for the input of a phantom regulator.

As for the regulator problem described, it is possible that the LM317 might have failed, especially since it really isn't rated for these sorts of voltage if something accidentally shorts the output of the regulator to ground/0V.

Bri

 

[Harpo]

just double checked the 317 and the diode, they're in as they should be

For 48V out, circuit is missing a diode across R1 to protect from C3 discharging. With 1000uF cap at output this cap will look like a short until it is charged, far exceeding the max.allowed i/o differential across the LM317 (a LM317-HV would have a better chance to survive). You could use a clamping zener diode in parallel to the diode across the vreg (or better use a TL783 vreg in the 1st.place). Not knowing the fitted value of R1, if it is 120R, the 4K7 pot/trimmer/rheostat would be dialed in to 4K488 for +48VDC out, dissipating 0.49W, if R1 is 220R, the 10K trimmer would be dialed in to 8K228 for +48VDC out, dissipating 0.27W. If your trimmer has a shape outlined on the Velleman pcb, these for usual could take max.0.2W and you probably want the needed wattage doubled up.

 

[PRR]

> why would it give 70v DC when the transformer is only putting out 27v AC

I think that is the "right answer"??

As Brian said:

27V AC is RMS. The power line is Sine. The peak of a Sine is 1.414 times the RMS.

A single peak-catch diode and capacitor will try to charge-up to the peak. 27V*1.414= 38V.

The Voltage-Doubler scheme does this twice. 38V+38V= 76V.

No rectifier is perfect. You can bet on 10% droop, often more. 90% of the ideal number is about 69V.

1,000uFd at the reg output is probably a Poor Idea. At turn-on, the cap voltage is *zero*. The raw DC pops up to +69V. The regulator feels that full 69V for an instant. It is only rated 35V or so. It may be OK *after* the output comes up to 48V; then it only feels 21V. But it died in the first milliSecond, trying to lift-up voltage on that huge cap.

Try 0.1uFd there. We still get 69V across the LM317 but for a much shorter time. It may not die right away. (Also the input seldom pops-up the full 69V instantly; sometimes it ramps-up over 16mS and the '317 may be able to charge 0.1uFd that fast.)

I would NOT trust it in real life. It may work a few times then die on the 13th or 113th turn-on.

There are other ways to build a 70V-48V regulator. Jack-strapped '317 (with external transistor), simple Zener plus pass-transistor, etc. IAC, be using 100V parts not 35V parts in a 70V system.

 

[Brian Roth]

Years ago, I learned the hard way that a 317 is dicey for a phantom regulator.  I find it odd that many/most of the kits currently being sold (typically producing +/- something as well as 48 VDC) continue to use a 317 for the phantom rail.

Hence, I always use  TL783 for the task.

http://www.ti.com/litv/slvs036m

Bri

 

[ohm]

Thanks everyone... your input has helped explain a lot

I was looking at the 783 or 317HVT...

looks like the 783 would be a better option

From looking at the datasheet it should just slot into the existing circuit

How do I determine the values of R1 / R2 and Co ?

(equations arn't my strong point and the text on data sheet doesnt't make much sense

thanks so far the help... trying to find decent info on building a 48v supply has been the most difficult

 

[mattamatta]

Years ago, I learned the hard way that a 317 is dicey for a phantom regulator.  I find it odd that many/most of the kits currently being sold (typically producing +/- something as well as 48 VDC) continue to use a 317 for the phantom rail.

Hence, I always use  TL783 for the task.

http://www.ti.com/litv/slvs036m

Bri

Thank you for that, Brian.  I wasn't familiar with those, but now I'm going to have to get a hold of some and try them out.

 

[ruffrecords]

Here's the circuit of a PSU I designed several years ago. Phantom supply is at the bottom.

http://www.ianbell.ukfsn.org/EzTubeMixer/docs/TwinChannel/PSUcct.jpg


Cheers

Ian

 

[abbey road d enfer]

Years ago, I learned the hard way that a 317 is dicey for a phantom regulator.  I find it odd that many/most of the kits currently being sold (typically producing +/- something as well as 48 VDC) continue to use a 317 for the phantom rail.

Hence, I always use  TL783 for the task.

http://www.ti.com/litv/slvs036m

Bri

+1.
LM317 for phantom power is just a time bomb. Considering the low price and good availability of the TL783, I don't understand why some insist on using a 317, at the cost of adding all sorts of protections. I guess they think they're smart. This is not directed to you, ohm, I understand you didn't know about the 783.

 

[moamps]

How do I determine the values of R1 / R2 and Co ?

Hi,
Vref=1,25V typ
choose R1 (I like 150ohm)
from: Vout=Vref(1+R2/R1) goes R2=R1((Vout/Vref)-1) so:
R2=150((48/1,25)-1)=5610 ohm, use standard value 5600ohm,
I=Vout/(R1+R2), P(R2)=I2R2=0,4W, so use 5k6/1W or 2W for R2

Any elco between 10/63V and 100uF/63V will be fine for Co.

Regards,
Milan

 

[bockaudio]

Agreed I've got tons of 783's doing 48v out there no problem never even considered the 317. 783 is a no brainer.
If you want a really fun project try an IRF830 as a shunt 48v.

 

[ohm]

Thanks Milan! Great help

was using 317s as this is what I had lying around and saw so many other (i guess now, badly designed) PSUs using them

Kinda glad it needed troubleshooting based on the great input here looks like I'll have a more robust 48v supply in the end..

Thanks.. just waiting on a couple of 783's to arrive now... (one spare just in case

 

[ohm]

Ok 783s have landed 

I have a few 1w to 2w resistors, but none at 5k6

There is a 9k 2w resistor

Based on the calculations from Milan and Ians diagram I could use the 9k for R2 with 240R for R1

Does this sound right or should I wait and order a lower value than 9k?

thanks so far

 

[abbey road d enfer]

Ok 783s have landed 

I have a few 1w to 2w resistors, but none at 5k6

There is a 9k 2w resistor

Based on the calculations from Milan and Ians diagram I could use the 9k for R2 with 240R for R1

Does this sound right or should I wait and order a lower value than 9k?

thanks so far

Perfect. 240r sets the current in the resistor chain at 5mA.

 

[ruffrecords]

Ok 783s have landed 

I have a few 1w to 2w resistors, but none at 5k6

There is a 9k 2w resistor

Based on the calculations from Milan and Ians diagram I could use the 9k for R2 with 240R for R1

Does this sound right or should I wait and order a lower value than 9k?

thanks so far

Perfect. 240r sets the current in the resistor chain at 5mA.

Personally I would go for a lower value than 240R. The reason is that the TL783 needs a minimum current to regulate properly. It is 15mA with 125V across it but the current drops as the voltage across it drops. From the data sheet it looks like it drops to close to 5mA with 20V across it. So 240R and 5mA is within spec but only just.

Cheers

Ian

 

[abbey road d enfer]

Ok 783s have landed 

I have a few 1w to 2w resistors, but none at 5k6

There is a 9k 2w resistor

Based on the calculations from Milan and Ians diagram I could use the 9k for R2 with 240R for R1

Does this sound right or should I wait and order a lower value than 9k?

thanks so far

Perfect. 240r sets the current in the resistor chain at 5mA.

Personally I would go for a lower value than 240R. The reason is that the TL783 needs a minimum current to regulate properly. It is 15mA with 125V across it but the current drops as the voltage across it drops. From the data sheet it looks like it drops to close to 5mA with 20V across it. So 240R and 5mA is within spec but only just.

Cheers

Ian

I never had any problems with that. I know that spec sheets recommend a lower value (higher current), but I think it's in order to escape liability, in case someone didn't read the datasheet (although I know it never happens ;D)

 

[ohm]

A couple more questions before I fire this up (and hopefully not smoke the circuit again...)

Should I lower R1 to 230 or 220 then?

Ian your diagram has 100k across the DC inputs. Is this needed?

Is this heatsink too small?

Actually another question related to the 24v Velleman circuit

Another one of these kits is being used to supply 24v to the preamps.

As it uses the lm317 should I be adding in the protection diodes like below (D12 / D11)

(guess the TL783 doesn't need the protection diodes?)

Also should I add in C1 and C2 like below (i'm guessing for ramping the 48v to prevent clicks... is this right?)

 

[abbey road d enfer]

Should I lower R1 to 230 or 220 then?

The difference with 240r would be minimal.

Is this heatsink too small?

We could answer this question if we knew the current draw. If it's for a couple of channels, it should be largely enough.

As it uses the lm317 should I be adding in the protection diodes like below (D12 / D11)

They are needed only when the unregulated voltage is higher than 37V. You can add them, though, they can do no harm.

guess the TL783 doesn't need the protection diodes?

Only if the unregulated voltage is higher than 125V.

Also should I add in C1 and C2 like below (i'm guessing for ramping the 48v to prevent clicks... is this right?)

C1 & C2 are part of the voltage-doubler circuit, they have no role in ramping the voltage.
You don't need them AT ALL in the +/- rails.