low noise power supply - advice needed

[blana]

Hello,

Since my old modded ATX power supply got toasted while onstage, my next project is a low noise power supply, with a lot less parts involved.

I checked various older posts around the forum, some stabiliser schematics and pdfs around the net and came up with the solution presented in the link below. I tried to keep the part list to a minimum.

I am not very sure of this design since i'm a noob, so i decided to post this idea on the forum, maybe some of you guys can give me advice, or stop me before building an useless circuit...


https://www.dropbox.com/s/of265sk0z94kdes/toroidal%20%2B%20stabiliser.jpg


I plan on using a 12 V toroidal transformer and a LM 338 stabiliser. The capacitors will be big ones, the biggest i can scavange off damadged ATX sources, in the idea of filtering as much noise as possible. The toroidal transformer will be shielded from the rest of the circuit, to prevent hum induction (i'm not sure how yet, advice would be welcome) .



Thanks in advance.

[R.G.]

See "Power Supplies Basics" at geofex.com.

[blana]

Hey, thanks for your reply. I read the article once, but i'm gonna re-read it home more thoroughly. If i was supposed to see an error within my schematic afterwards, I'm sorry, but I didnt. It seems to me that my schematic and the one presented in the article are pretty close.

Thanks.

[Kesh]

a lot depends on how many amps you are after, but as you are using a 338 i'm guessing a lot?

put a cap across R2, maybe a 22u or 47u, or way less if a tantalum.

as you are using big caps put diodes across R1 and the top of the regulator to prevent them discharging back into the regulator.

both those should be in the datasheet (just making a lm350 based one myself and they are very similar)

put another cap in parallel to C1 with a high wattage very low ohm resistor between them, like one ohm. value depends on how much current you draw and how much voltage you can afford to lose. if you are going for 9V and your diodes are good you can probably afford to lose a few volts.

there's more you can do, snubbers, more caps and resistors, wiring the caps direct to the pins of the 338, little cap across the transformer secondary. but that's kind of hi-fi territory and not really necessary for a power supply for stage.

toroidals are low hum anyway, they don't give out magnetic fields much at all.

[ubersam]

The toroidal transformer will be shielded from the rest of the circuit, to prevent hum induction (i'm not sure how yet, advice would be welcome)

You could use Mu-metal sheets. I used Magnetic Shielding Foil and Magnet Shield on a Furman AR-1215 voltage reg to shield the other rack units from its emf. Worked really well.

[blana]

Hello again,

Thank you very much for your interest and help!

@Kesh: yeah i am aiming for a few amps, since i need to power 8 to 10 pedals, and i will be using multiple lm 338 to draw from the same toroidal (12v and 5 amps). Roughly, there will be 2 pedals per stabilizer, to avoid loading the stabilizer too much.

I think found the datasheet page you were talking about, it has the schematic with the protection diodes.  Does it look something like this?

https://www.dropbox.com/sh/3d94yguhhffibpd/juQNHnBzLC


Thank you once again for your help with this, its really a great improvement!

Regarding the hum - i used the toroidal a while ago in a cruder schematic, and the source had a LOT of noise. Unfortunately i dont remember exactly what i did, but i presumed the hum was induced bu the toroidal coils, thats why i was planning on shielding it from the rest of the circuit.

@ubersam: Thanks a lot for the info with the foil types! However, I have one question: am i supposed to just wrap the coil in the foil, similar to what they show on the site with the TV example? How exactly did you use it? I was thinking its supposed to work like a Faraday cage, and the ground needs to be connected to the cage ( to the foil in our case).


Thanks once again, you guys are great!

[Kesh]

Yeah, that's the diodes and cap across the ref voltage divider.

Pedals can draw as little as a couple of mA. Digital pedals maybe draw 50mA, maybe up to 150mA. See here. http://stinkfoot.se/power-list

So you will only need one LM338 or a lower current unit like a LM317. Even the LM317's 1.5A is probably over specified, but that's what most multi pedal PSU builders offer.

Here are some back of a beer mat calculations for a 1.5A supply from a 12V transformer:

The 12V transformer will give out about 16.9V peak. You should allow for a 10% loss as mains power isn't reliable. So 15.2V. Take away the rectifier voltage drop and you get maybe 13.7V peak depending on diode type. Your regulator will need at most 11.5 V to give out 9V. 13.7V - 11.5V = 2.2V spare.

There are two ways to go. Either allow the voltage to ripple by 2.2V by using a single cap, 4700u will probably be about enough at 60Hz supply, 6800u for 50Hz. Or you can put in a second RC circuit, as shown below, and have a voltage drop across the resistor. C4 should be the same 4700u to 6800u at least. This is what I'd do, as I'd rather have a smoothish 12.2 V going into my regulator than a rippling 13.7V.



To be honest If I really wanted 1.5A at 9V low noise PSU I'd probably use a 15V transformer as it gives you more leeway. Some bridge rectifiers drop more than 1.5V, and I'd rather use a bigger resistor. And maybe a 0R5 one before the first cap too. And bigger caps. But that's just me.

Oh yeah, perhaps use long life/high temp caps. High current ripple eats up caps for breakfast.

[blana]

Hey, thanks for all the great info!

I was under the impression that pedals eat up way more current, it seems i will not be needing that many LMs after all .

Thanks for the calculations also, they are very useful indeed. One thing i may have missed: what do you mean by a "0R5"? Is it meant to show that it is before R1 (just nomenclature)? And how low can i go with the resistances between caps? 1 ohm as you said in your previous answer?

Another idea that came to me speaking with a friend whose processor also got toasted while on stage  (yes, voltage varies a lot around here) - how feasible is it to add an oscillator after the stabilized voltage to obtain an AC source for more complex effects effects (like some processors, or the digitech whammy)? Or can this be achieved in a simpler way?

Thanks!

[ubersam]

How exactly did you use it? I was thinking its supposed to work like a Faraday cage, and the ground needs to be connected to the cage ( to the foil in our case).

For the AR-1215, the inside of the unit looked really cramped with the big toroid and the rest of the voltage regulating circuity. So instead of disassembling the unit to wrap the toroid in mu-meal, I just cut up the foil about an inch larger than the diameter of the toroid, bent the edges slightly and hot glued it on top of the toroid. Then for additional blocking, I cut up a sheet roughly the size of the top panel and spray glued it on. That was sufficient for that application.

Another application, i lined the inside of an aluminum box with the foil and housed an E-I xformer inside it.

[R.G.]

Regarding the hum - i used the toroidal a while ago in a cruder schematic, and the source had a LOT of noise. Unfortunately i dont remember exactly what i did, but i presumed the hum was induced bu the toroidal coils, thats why i was planning on shielding it from the rest of the circuit.

It's really, really unlikely that the toroidal transformer was the cause of the hum. It is vastly more likely that there was one or more problems with wiring, rectification or filtering. In most case, it is unnecessary to shield a toroidal from the rest of the circuit. There are exceptions, in ultra-sensitive stuff, of course, but pedalboards are generally not one of these.

[Kesh]

Hey, thanks for all the great info!

I was under the impression that pedals eat up way more current, it seems i will not be needing that many LMs after all .

Thanks for the calculations also, they are very useful indeed. One thing i may have missed: what do you mean by a "0R5"? Is it meant to show that it is before R1 (just nomenclature)? And how low can i go with the resistances between caps? 1 ohm as you said in your previous answer?

Thanks!

0R5 means half an ohm like 2k5 means two and a half thousand ohms.

Resistance between caps is an RC calculation 1/(2pi*R*C) balanced with an Ohms law calculation V/I = R

If you have 2.2V to play with at 1.5 amps ohms law tells you can have about 1R5 max , assuming that killed ALL ripple, which it won't, so call it 1 ohm.

1/(2pi*1*C) tells us that if C4 is 4700u, our -3dB corner begins at 34Hz. So we will be about 8dB down at 60 Hz and 14dB at 120Hz for the first harmonic hum, which is prominent.
This isn't great, but you are using a 12V transformer and salvaged caps. With a 15V transformer and 10,000u caps (or a bank of smaller caps) you could achieve much more.

But then some people might say all this is totally unnecessary because your regulator will reject most of the ripple anyway. And for stage work everything will be noisy anyway. I'd probably agree, but I also like to over engineer things if the cost is marginal. And it would be nice to have a PSU that was utterly silent in the studio.

You should also check out grounding topologies. And safety earths.

[blana]

Hey, once again, thanks to all of you for your help! I am feeling i am learning a lot in a very short time!

I discovered last night will diging up for the required parts that my toroidal was not 12v but 2x9 v, so i wired the coils in series so it outputs 18, so now i have more "leeway" as you said.

I bought the biggest 25v caps i could find - 6800u - and the smallest 5w resistors i could find - 2.7 ohms. I am guessing thats very good for the cutoff frequency which becomes around 8 hz at -3db and "minus a lot" for 50-60 hz.

The difference voltage will be around 6.5 volts, the resistance is 2.7 ohms, which leads to a current of 2.4 Amps. I am guessing thats too much. If i put two resistors in series i get a resistance of 5.4, leading to a current of 1.2 Amps at 6.5 volts, which i guess is safer.
Also, the cutoff frequency benefits from this becoming 4.3 something.

Is this correct? Did i mess something up?

You should also check out grounding topologies. And safety earths.

Oh, i totaly forgot about this problem: any advice here? What should be connected to the metal box? And what should not be connected to the metal box?

Thank you very much!

[Kesh]

Hey, once again, thanks to all of you for your help! I am feeling i am learning a lot in a very short time!

I discovered last night will diging up for the required parts that my toroidal was not 12v but 2x9 v, so i wired the coils in series so it outputs 18, so now i have more "leeway" as you said.

I bought the biggest 25v caps i could find - 6800u - and the smallest 5w resistors i could find - 2.7 ohms. I am guessing thats very good for the cutoff frequency which becomes around 8 hz at -3db and "minus a lot" for 50-60 hz.

The difference voltage will be around 6.5 volts, the resistance is 2.7 ohms, which leads to a current of 2.4 Amps. I am guessing thats too much. If i put two resistors in series i get a resistance of 5.4, leading to a current of 1.2 Amps at 6.5 volts, which i guess is safer.
Also, the cutoff frequency benefits from this becoming 4.3 something.

Is this correct? Did i mess something up?

18VAC will give 25.5 peak DC less 10% for bad mains supply is 23V, less 1.5V taken by the rectifier is 21.5 V. You want about 11.5 V into the regulator, so you have 10V spare to play with.

If you put the resistors in series you would get about a 10V voltage drop at just under 2 amps, but unfortunately 2 amps through that resistors comes to over 10W, so your power supply will only safely drive just under 1 amp, which is really all you need.

Remember most of the time your power supply will be drawing much less, so the drop across those resistors will be much less and up to 27V could be going into your regulator if mains is 10% over its rated voltage. This means your regulator will be coping with 18V more than the 9V it is outputting. Worse is when it is drawing substantial current but still coping with a big excess voltage. This is well within its limits, but it may need a heat sink. Try it without one and see, it will shut down safely if there's a problem. Just leave room to install one if necessary.

By the way, a common thing is to put a ballast resistor over your final power supply output, between 9V and 0V. Use one of about 1K5. It means you power supply is always driving 6mA when on. It helps stabilise it, and it is also useful for testing the thing without having to wire anything up to it.

Oh, i totaly forgot about this problem: any advice here? What should be connected to the metal box? And what should not be connected to the metal box?

Thank you very much!

There are legal safety issues about this, so I'm not going to recommend anything. But for purely academic purposes this is a very good resource about making safety earths that don't cause ground loop hums http://sound.westhost.com/earthing.htm

Another issue is fuses on the DC, some people use them, others don't. And a fuse on the AC side is possibly a legal requirement.

[blana]

Hey, thanks for all the infos!

Yeah, i forgot about the 10% variations, which was a very important variable. I think i can use 3 resistors in series, they are very cheap. I will surely use a heat-sink for the LM 338, i have scavenged a few from broken atx sources, so they are free.

Thanks once again, I think i'll get started!

[Jdansti]

Since you're asking questions about safety (a good thing to do) it's apparent that you're not an expert on working with mains voltages. That's ok, neither am I. It would be a good idea to find a local engineer who knows your local electric codes and have him/her help you with the proper grounding and other safety requirements. The last thing you'd want to do is be severely injured or killed, or cause your house or another building to burn down. 

[blana]

Hey, thanks for your concern! I will be very careful, i promise. I played with mains voltage before and the first rule is to never touch the circuit while its plugged. If it starts to smoke, just un-plug it. Monkey - electrician technique. 

My concern was more about the hum and ground loops.

[amptramp]

Hey, thanks for your concern! I will be very careful, i promise. I played with mains voltage before and the first rule is to never touch the circuit while its plugged. If it starts to smoke, just un-plug it. Monkey - electrician technique. 

My concern was more about the hum and ground loops.

One item I always use is a power bar with a circuit breaker on it.  If you plug the unit in and sparks and flame start, you will not be able to pull the plug.  I restore antique radios, so I use this trick with any new ones.  So far, two items of antique test equipment have tripped the breaker.

[blana]

Thanks for the tip, i just bought one.

[blana]

Hello again.

I just built the circuit, but it has some issues.

The resistances and the lm338 get hot fast. The output capacitor got swollen after i managed to do seme measurements. One note is that the output resistance (1.5k) got hot also, and i removed it in the idea of having no consumer and being able to measure the voltages without the circuit getting hot. It partly worked, meaning that it gets hot slower, but it still gets hot.

The current schematic is this:

https://www.dropbox.com/s/g7sdxzj4w4oxh0w/lm%20338%20source%20v2.jpg

The only diferences are: the LM 338, C4 - 470u at 200v and high temp, and the control resistances which are fixed.

What should i do? :S

[Jdansti]

Hello again.

I just built the circuit, but it has some issues.

The resistances and the lm338 get hot fast. The output capacitor got swollen after i managed to do seme measurements. One note is that the output resistance (1.5k) got hot also, and i removed it in the idea of having no consumer and being able to measure the voltages without the circuit getting hot. It partly worked, meaning that it gets hot slower, but it still gets hot.

The current schematic is this:

https://www.dropbox.com/s/g7sdxzj4w4oxh0w/lm%20338%20source%20v2.jpg

The only diferences are: the LM 338, C4 - 470u at 200v and high temp, and the control resistances which are fixed.

What should i do? :S

The LM338 is supposed to get hot but might need a heat sink to avoid overheating.

What did you have connected to the output as a load when the resistors got hot?  Are the resistors properly rated for the power that you expect them to have to dissipate?

Are the capacitors rated well above the voltages they'll see?