New Rigol DC PSU's

[ElectroIrradiator]

The LM317 has a remarkably high R_th,JC at 17^oK/W, so how hot is too hot depends highly on the actual device dissipation.

[Rufus]

The LM317 has a remarkably high R_th,JC at 17^oK/W, so how hot is too hot depends highly on the actual device dissipation.

Data sheets from 3 different manufacturers say 4, 5, and 5, C/W for the LM317 in TO220.

[EEVblog]

Thanks for the confirmation.
IMO this is just way too high, and bad design oversight. There wouldn't be any margin in this at all. And at that temp right next to the filter caps is just begging for premature failure.
If anyone else can get some heatsink contact data as well that would be great.

[ElectroIrradiator]

Data sheets from 3 different manufacturers say 4, 5, and 5, C/W for the LM317 in TO220.

TI says 17 K/W for TO-220, Linear says 4 K/W typical, but no maximum for 'T' package for standard chips (not LM317A). Haven't checked any other brands.

[gilbjd]

I'm getting up to 104 degrees Celsius on the heatsink Dave. The unit has been powered up for an hour or so with no load.  Mine has a lot of adhesive between the two capacitors. I can't actually see the heatsink so I have to bend the thermocouple wire a bit and slide it into position to make contact.

[Rufus]

Data sheets from 3 different manufacturers say 4, 5, and 5, C/W for the LM317 in TO220.

TI says 17 K/W for TO-220,

That data sheet also says 19 C/W for JA and 3 C/W for JP. The JC value being resistance to the plastic of the case, JP to the metal tab, and JA the whole case to ambient is the only interpretation that makes sense.

[EEVblog]

I'm getting up to 104 degrees Celsius on the heatsink Dave.

Well, that's just ridiculous now. This is a massive design oversight 

[dr.diesel]

I'm getting up to 104 degrees Celsius on the heatsink Dave.

Well, that's just ridiculous now. This is a massive design oversight 

Assuming this carries across all units, it's more than ridiculous.  More like a downright embarrassment to Rigol's PS design team, and cause for a complete recall of all units, IMO.

[ElectroIrradiator]

That data sheet also says 19 C/W for JA and 3 C/W for JP. The JC value being resistance to the plastic of the case, JP to the metal tab, and JA the whole case to ambient is the only interpretation that makes sense.

Yes, you are probably right, considering the second footnote. However R_th,JC is usually taken to mean between the die and the heat sink contact surface, so they ought to rework that table a bit.

[Rufus]

Assuming this carries across all units, it's more than ridiculous.  More like a downright embarrassment to Rigol's PS design team, and cause for a complete recall of all units, IMO.

They will fix it in software by turning up the fan speed

[olsenn]

Why would this happen with no load? Or is it just a internal digital proccessor's power that overheats?

[wizzy]

Was about to purchase one today... Is there room for a larger Heatsink on it?

[EEVblog]

Why would this happen with no load? Or is it just a internal digital proccessor's power that overheats?

It has nothing to do with the load, it is just powering the digital control circuitry.

[EEVblog]

Was about to purchase one today... Is there room for a larger Heatsink on it?

Technically, yes. But with the TO-220 freestanding like that (from memory I think the HS is not attached to the PCB), it would not be be advised.
It's tempting to mount it on the diode bridge heatsink and hand wire it back in. But then you have isolation insulation issues.

[mcinque]

And at that temp right next to the filter caps is just begging for premature failure.

Honestly I'm quite disappointed by this stupid design. Even a noob knows that caps hate high temperatures. Why the hell they don't take it in consideration?

[dr.diesel]

Is there room for a larger Heatsink on it?

Like Dave said, you'd have to isolate it, which is not undoable.  Point is though you shouldn't have to, and if they missed something this simple, what else did they dick up?

[mcinque]

if they missed something this simple, what else did they dick up?

You're right... 

[Rigby]

And at that temp right next to the filter caps is just begging for premature failure.

Honestly I'm quite disappointed by this stupid design. Even a noob knows that caps hate high temperatures. Why the hell they don't take it in consideration?

It was deemed an acceptable risk, probably.  I've seen hot caps last a very long time, and I've seen hot caps last a mere week.  Presumably they've done some testing and determined that they'll last beyond the warranty period.

I'm still an electronics newbie, but it seems that you could add a bit of thermal insulation between the heat source and the heat sensitive components, giving the heat sensitive components a longer life.  Am I wrong on that?

[robrenz]

I'm still an electronics newbie, but it seems that you could add a bit of thermal insulation between the heat source and the heat sensitive components, giving the heat sensitive components a longer life.  Am I wrong on that?

Maybe some Aerogel

[ElectroIrradiator]

Maybe some Aerogel

Would protect the caps, but further heat up the LM317 due to the reduced air flow.

[sha256]

After about an hour I'm getting 100C.  I'm thinking it's epoxy (to solve Dave's unattached heatsink concern) and bigger heatsink time (to solve some dumb ass mistake).

[Rigby]

Maybe some Aerogel

Would protect the caps, but further heat up the LM317 due to the reduced air flow.

Perhaps some simple ducting or a fan with higher CFM would keep things cool enough.

(My naivete took me in a lot of places that make no sense before I came to that conclusion.  Maybe that one makes no sense, either, but it makes more sense than my peltier idea by a wide margin.)

[Hydrawerk]

Why would this happen with no load? Or is it just a internal digital proccessor's power that overheats?

It (LM317) has nothing to do with the load, it (LM317) is just powering the digital control circuitry.

So the overheated LM317 is going to blow up in two years. Then the digital control circuitry will be destroyed by too high voltage (maybe 12V?).
Well, i don't think the Rigol engineers are stupid. They want me to buy a new Rigol PSU every two years.
Only Tektronix stuff will last for decades, (but their scopes are obsolete and not being innovated.)

[Fsck]

from the picture of the hand + thermocouple against the side of the case, it looks like it might be possible for you to punch a large hole in the side of the case and mount a 40-60mm fan.
or you could just attach a 12cm fan to the outside without any physical alterations or something.

[Mr Simpleton]

Why not measure the voltage drop over the LM317 and then current... put in a power resistor in series to reduce disipated power from the LM317, or why not squeeze in a bigger heat sink?

Ultimate solution would be finding a lower voltage source, from within the 832, that could drive the digital board, and have the LM317 run cooler.