LM317/LM337 accuracy

[Stuey]

Hi,

I built a simple power supply as per the one on Nuuk's website which has a 50VA 18-0-18 transformer, 4700uF smoothing caps and a series of paralleled polypropylene bypass caps before the above regulators. It's built on stripboard.

http://myweb.tiscali.co.uk/nuukspot/decdun/gainclonepre.html

The idea was to have a two stage reg in which this first stage would put out ~+-21V and the second stage reg to +-15V.

However, after carefully building the (simple) first stage, the damn thing was putting out -22.2V and +21.8. The unloaded DC voltage before the regs was a steady 27.7V both sides.

I'm using 220R and 3K6 1% metal film resistors to set the voltage.

I checked the resistors' values by disconnecting one end - they were within spec. I then bought two new LM regs, and replaced these and the resistors after matching the new resistors with my meter.

Now the voltage is -21.5V and +22.0V...

I'm fairly sure the regs are proper ONsemi devices.

Any tips on what could be the issue? I know it's hard over the net, but I'm stumped. I've even checked for minor current leakage around the regs through rosin between tracks - there's nothing wrong.

Cheers

Stuey

 

[Christer]

The internal voltage reference of the regulators have a tolerance of almost +/- 5 %. LT makes a version with +/- 1 % tolerance, if I remember correctly. However, it is cheaper and simpler to just adjust the resistor values so you get the desired voltage.

 

[zacster]

If you use a pot to set the resistance, you can get the voltage perfect and it will hold steady.

And the bear is a white polar bear at the north pole.

 

[peranders]

Notice also that an absolute voltage isn't necessary, only a stable and noise free one.

 

[Stuey]

Ahhh, thanks. Three replies giving the three possible answers I'd thought about...!

I just assumed that they'd give the same voltage out if the Vin was equal side to side.

Thanks people.

Cheers

Stu

 

[richie00boy]

Also that the +ve and -ve regs have different Iadj so by using larger than the datasheet recommended 120R you increase the error due to the voltage developed due to this current.

But don't worry, as said a bit either way won't matter.

 

[Stuey]

Thanks for that additional info Richie.