Adjustable power supply for your bench

[Kevin Mitchell]

I've got quite a few projects I need to wrap up but really should get this one out of the way. I've got some voltmeter modules on the way and plan to box this one up so I can stop stealing my Pedal Power from the pedal board.

Features 4 adjustable supplies - 3 positive and 1 negative.
Should output something like 1 to 15 VDC per channel.
Circuit is per the LM317 & LM337 datasheets


Voltmeter modules

-KM

[amptramp]

It looks like you only have halfwave rectification for each polarity.  The result is you have current going in both directions through the transformer but these currents are likely to be unequal and a bridge would give fullwave rectification which offers cancellation of the 60 Hz waveform at the output.  This would require a centre-tapped transformer.

Also, where are the electrolytic filter capacitors?  You need one at the input to each polarity of rectifiers and it looks like you only have ceramics there.  The output theoretically can do without but usually you add an electrolytic at each output as well.

Draw up an actual schematic because it is hard to trace out a circuit from a layout diagram and I may be missing something.

[Kevin Mitchell]

Thank you for your input. Indeed I did leave out filter caps! For the time being at least. Regarding the input voltage - I had not considered any sort of cancellation since I've used this configuration in other bipolar power supplies where the current pull was typically far from equal between the rails. But that was with L78XX, L79XX type regulators. I'm not sure if that makes a difference.

Here's a snap of the datasheet;



If I goofed up the rectifying I'll just use a CT transformer 

-KM

[PRR]

> Here's a snap of the datasheet

This *assumes* the guy in the other department who designs raw DC supplies put proper fat caps on the end of his work. The chip-maker tells you, the regulator-man, that unless you are "near" those big caps, you "must" have little caps AT the chips to swamp inductance in the possibly long raw-DC feed wires.

In this case I would extend the left end of the board an inch and have two 1,000uFd caps wedged-in between. Then the 0.1u caps may not be essential, but are too cheap/small to scrimp.

[Rob Strand]

LM317 & LM337 datasheets

Those are good choices form simple adjustable supplies.
You don't *have* to use a 240 ohm resistor, a 220 ohm or 270 ohm is fine although it will change the maximum output voltage a small amount.

In this case I would extend the left end of the board an inch and have two 1,000uFd caps wedged-in between. Then the 0.1u caps may not be essential, but are too cheap/small to scrimp.

Yes, it's not good to skimp on those.    Especially today where the parts don't quite behave like the originals.

An interesting thing about the input caps, not just for the LM317 but for others, is the distance from input caps is sometimes not specified and sometimes is it is.  Like the National semiconductor datasheet indicates the input cap isn't required when the distance is less than 6 inches from the input caps, whereas On-Semi and Fairchild only say an "appreciable distance".

Another finer point is the negative regulators often have different input and output  cap requirements than their positive counterpart.  I haven't checked but in the back of my mind I'm quite sure the LM337 is like that.  You should check it.

[antonis]

Good work, Kevin..!!

Some strictly personal notes:
In case of pedal build/test/troubleshoot it should be more convenient to use standard (fix output) regulators, like 78XX/79XX, to save pots, resistors and voltmeter modules..

Of course, an adjustable benctop PS is more flexible but to also look (and be..) more "professional" it should exibit some extra abilities, like current limiting (a MUST for experimental circuits..) and "all down to zero volts" output..
(brute versions of both the above abilities could be easily obtained with rather simple circuits..)

P.S.
Lugs 1 & 2 of all pots should be shorted..

[Kevin Mitchell]

Okay okay! I've revised the layout considering the comments.
-Filter caps added
-Input caps placed closer to regulators
-(not mentioned) added space for heatsinks

The datasheet doesn't mention placing the input caps close but does suggest this for the 240 ohm current set resistor    But I've seen these supplies all around where this was not done so I'm not sure how crucial it would be. Perhaps the layouts are small enough for it to suffice.

I'll reconfigure the AC input if necessary. If my drawn method doesn't work out I'll do what amptramp suggested - a CT transformer and diode bridge.



Better? 
-KM

[Rob Strand]

The datasheet doesn't mention placing the input caps close

It doesn't spell it out but by implication of them being required when far from the caps is they should be close.  Usually the closer the better.

The LM337 should have an output cap.  The datasheet says 1uF tant or 10uF electro (upto 1000uF electro).  It also specs an input cap of 1uF tant.  Tants on the input side tend to fail so an electro say 10uF will be better.

but does suggest this for the 240 ohm current set resistor    But I've seen these supplies all around where this was not done so I'm not sure how crucial it would be. Perhaps the layouts are small enough for it to suffice.

You can put a wide range of values there.  There is a small current coming out of the adjust pin.  The smaller the output resistor the more current flowing through it.  If the current is large it swamps the small adjust current so the adjust current can't affect the output voltage.  The small current isn't constant it can have a small hum component (actually input voltage dependency).  The adjust current varies with temperature which affects drift, which is not important for your case.

One thing about your layout is the ground track on the top left is bad.  It is connecting the regulator ground to a noisy ground pretty much adding hum to the output of that channel.  You don't want any connections to the line between the common AC input and where the two electros join.  So for example flip the top cap so the common AC line goes between the two caps first.  This puts the AC connections on the left.  Ideally the wire between the caps is fat.  Then the circuit ground connects on the right.  You can see the idea in the PCB layout at the bottom of this page.   You will have to move the ground connection for the top left regulator.

https://xtronic.org/circuit/amplifier/symmetrical-power-supply-for-potency-audio-amplifiers/

[antonis]

https://xtronic.org/circuit/amplifier/symmetrical-power-supply-for-potency-audio-amplifiers/

<off topic on>
Plain enough, Rob but, IMHO, unacceptable use of Watts (instead of VA) for transformer power rating.. 
<off topiv off>

[Rob Strand]

Plain enough, Rob but, IMHO, unacceptable use of Watts (instead of VA) for transformer power rating..

To be honest I didn't read that page at all.  I only wanted a pic showing the idea of separating the input and output power routing.

[Kevin Mitchell]

The LM337 should have an output cap.

It does. It is lower on the board than the other ones but shouldn't be a problem moving it closer to the wire.

One thing about your layout is the ground track on the top left is bad.

That's a good point - I definitely overlooked the ground at that spot. I've moved it and now everything connected to ground happens after the filter caps.



                                 


-KM

[ElectricDruid]

I built a "single channel" version of this circuit to enable me to test pedal circuits at lower voltages and also up to 20V or so. Very handy.

For bonus points, I bought a cheap 2 wire digital voltmeter on eBay and stuck that on the output so I could see what I was getting. It works down to about 3.5V or 4V before it gives up, but that was probably lower than I needed for a pedal power supply anyway, so that was fine. Definitely worth the couple of quid it cost to add. This type of thing:

https://www.ebay.co.uk/itm/0-56-Color-DC-AC-5V-30V-500V-Digital-Voltmeter-Voltage-Panel-Meter-LED-Display/273406755547?hash=item3fa85002db:m:mYh6JcO0qZTGzVc9Qu7vMTw

[Kevin Mitchell]

For bonus points, I bought a cheap 2 wire digital voltmeter on eBay and stuck that on the output so I could see what I was getting. It works down to about 3.5V or 4V before it gives up,

That's why I went with the 3 wire ones 

-KM

[ElectricDruid]

For bonus points, I bought a cheap 2 wire digital voltmeter on eBay and stuck that on the output so I could see what I was getting. It works down to about 3.5V or 4V before it gives up,

That's why I went with the 3 wire ones 

Fair enough. I might next time too. But as I said, for a pedal supply it doesn't matter much since I don't need voltages below what it can measure.

Which voltage did you use for the meter supply? Did you have to include a another regulator for that? (like LM78L05 or something?)

[Kevin Mitchell]

Which voltage did you use for the meter supply? Did you have to include a another regulator for that? (like LM78L05 or something?)

The specs on the listing has some conflicting information. The title says 0-100v but the description suggests staying under 30 to avoid damaging the module.

For power 3 to 30 volts should work. Less than 3 volts the LED will start to dim.
Though I wont need low voltages often I'd rather have the option than not. Other than that the 2 wired ones are probably just as well. The link to my modules are in the first post.

-KM

[Rob Strand]

If you want to get to zero volts there's some nasty tricks like connecting the adjust pot to a negative bias voltage (1.25V).  Perhaps leave the adjust cap going to ground.

Here's the basic idea,
https://www.circuitlab.com/circuit/7k69b4/screenshot/540x405/

But I don't recommend the method with the diodes for quiet supplies as some input ripple appears across the diodes.  You could improve this with a two stage shunt regulator.   Ideally you would use another LM337 as the -ve "bias" for the LM317's.

While I know this scheme works I can't guarantee the reliability and freedom from output glitches when you turn on the power.   Power glitches would be more prone with the doubler type input rectifier you are using since one input rail will *always* come-up before the other.   A small amount of insurance would be to put a 100 ohm load on the output of LM337 bias regulator so it pins the output if the negative rail isn't up.

Interesting trick anyway.

[printer2]

The EBay volt-amp meter with three wires needs a separate ps if you want to measure below 3.5V. It needs between 3.5-30V to run the module, hooked up to the red and black wire. The yellow/white wire  is the one you hook up to your voltage you are measuring, which could be up to 100V.

[mozz]

Much easier to pick up a HP supply at a hamfest for $5, they will last another 50 years.

[Kevin Mitchell]

The yellow/white wire  is the one you hook up to your voltage you are measuring, which could be up to 100V.

Notes:
1. When the voltage is lower than 3V, the display brightness will darken but it will not affect the correct measurement.
2. Measuring voltage higher than 30V may damage this product.
Included:
10 x Panel Voltage Meter 3-Wire Lead Connector

To note - they did not come with the panel mount as the picture indicates. Those bastards.

-KM

[Kevin Mitchell]

Keep forgetting to post here. I did eventually throw a small variant of the circuit together. A simple ajustable bipolar supply on vero. There's a couple jumpers hidden underneath. I'll post the layout later.