New Rigol DC PSU's

[Monkeh]

In this "High-End" PSU.

Is it rust in this picture?

Your personal crusade to defend the honour of products you sell is rather tiresome.

[jancumps]

In the driver's manual of my 1989 Peugeot 205 they tackled the issue by mentioning that:

Rust is a natural phenomena

[olsenn]

Hoping Rigol will have something to say about it first - but I tend to doubt it.  The real fix is recall/redesign/rework but no way are they doing that based on a bunch of tech-nerds whinging that something isn't as good as it could/should be (but hasn't broken.... yet).

They will certainly fix this issue in future board revisions; however, if you've already purchased one, good luck getting them to do a trade.

[mcinque]

ISSUE HAS BEEN REPORTED
I've talked with my local representative after sending him the description of the issue, all the photos and the flir images: he have called bejing and told me that in Rigol this issue is known and has been corrected in the last revision (he don't know exactly how).

Rigol says that this is due to a defective batch of the linear regulator, not to a bad engineering  but he suggest to wait for the next week, when he will have a meeting with a Rigol Engineer, because the Chinglish language spoken by the phone operator in Beijing was not easy to understand.

He will make us know about any news regarding the issue.

He has also offered to ship to me freely the components to solve the issue as soon they are available and that at the moment Rigol does not have planned any official press announcement about this problem.

Let's wait and see.

[Rufus]

Rigol says that this is due to a defective batch of the linear regulator, not to a bad engineering

A faulty batch of LM317s that work but get steaming hot? Don't see how that is possible. Maybe Dave's investigation will shed some light on the issue.

[Fsck]

Rigol says that this is due to a defective batch of the linear regulator, not to a bad engineering

A faulty batch of LM317s that work but get steaming hot? Don't see how that is possible. Maybe Dave's investigation will shed some light on the issue.

dissect it! the heat capacity of the encapsulating material could be painfully low, heating up insanely for very little energy. etc. etc.
tons of possibilities, only an autopsy will tell.

[mcinque]

A faulty batch of LM317s that work but get steaming hot? Don't see how that is possible.

That it's their reply, possibly not the true.

Maybe Dave's investigation will shed some light on the issue.

I trust in him. Or in someone that will break the warranty label, desolder that regulator and install a brand new one to see if something changes. 

[ElectroIrradiator]

If a faulty LM317 is able to maintain the specified output voltage, like 3.3 or 5V, then I cannot see how it is able to overheat. The power dissipated is (V_in-V_out)*I_load, and with V_out at its rated value, the remaining parameters are fixed beyond the control of the LM317.

The IC cannot even dump a fault current through the regulator pin, as that has a high resistance to ground in your typical LM317 schematic.

If the IC die is improperly attached to the heat sink tab, then the actual heat sink wouldn't show an elevated temperature, as we have seen demonstrated.

So if the load survives, what fault can physically happen in a LM317, which causes it to overheat? Oscillations? Where does the power go?

----
Edit: Silly typo in formula corrected.

[Electro Fan]

Does a LM317 ever do anything that could contribute to/allow/enable a turn-on spike?

[Corporate666]

If a faulty LM317 is able to maintain the specified output voltage, like 3.3 or 5V, then I cannot see how it is able to overheat. The power dissipated is (V_in-V_out)*I_load, and with V_out at its rated value, the remaining parameters are fixed beyond the control of the LM317.

The IC cannot even dump a fault current through the regulator pin, as that has a high resistance to ground in your typical LM317 schematic.

If the IC die is improperly attached to the heat sink tab, then the actual heat sink wouldn't show an elevated temperature, as we have seen demonstrated.

So if the load survives, what fault can physically happen in a LM317, which causes it to overheat? Oscillations? Where does the power go?

----
Edit: Silly typo in formula corrected.

I think most of us would probably agree that blaming this issue on a bad batch of LM317's is a diversion... but... if that is indeed the company position on the matter, then it would seem they are accepting that these units contain faulty parts and would need fixed.  I can see both sides in this... impatient owners will say Rigol should send a new PSU in advance with a return label for the old one, but that is not realistic.  On the other hand, I'd rather not have to send mine to Rigol and wait (weeks?) for them to fix it.  I use this thing often. 

If a self-repair kit is made available, that would be fair, but I can see Rigol not wanting people soldering around inside their PSU's.  I also don't think a simple regulator change can solve this.  It's a linear regulator so dissipation is outside it's control.  Maybe something else on the board is drawing a lot more power than it should?  I've made those mistakes before... like driving an array of transistors with 330 ohm base resistors instead of 3.3k resistors... leading to 10x more current draw than necessary.  But whatever the issue is, I just hope a reasonable solution can be found.

[ElectroIrradiator]

This may sound kind of backwards, but somehow I would prefer if there was a simple external explanation for the problem. A bad batch of components causing problems in the field isn't unheard of, and is sort of excusable if true. Almost any other explanation I can immediately think of, ought to have been caught during design verification/prototype testing: High load current (not entering sleep in CPU during idle?), inadequate heat sink size, wrong component values or inadequate/improper air flow, just to mention a few possibilities.

Think I will wait for Dave's video, and see what he has discovered, before speculating any further.

[TerminalJack505]

Could Rigol have got hold of counterfeit LM317s?  Are people even counterfeiting LM317s?  I don't know that it would be worthwhile, personally.  I've heard of high-margin parts such as power amplifiers being counterfeited.  People decap them and the die is really puny as compared to the genuine part.

Whatever the problem, I'm Glad to hear they've fixed in in the latest revisions.  It's obviously a fairly critical problem since Dave's unit was rebooting itself because of it.

[Mr Simpleton]

It's not rocket science  Simple LM317 linear regulator running hot!
It comes down to voltage drop and current consumpton. Or a parasitic oscillation which increases the current.
So should be pretty simple to measure the voltage drop, and maybe use a hacked DSO2072 to see if there is an oscillaton. My bet is that they somehow have connected the input of the LM317 to the wrong rail, having too large voltage drop   

Still this holds me off buying this until it has been resolved...

[riconette]

Whatever the problem, I'm Glad to hear they've fixed in in the latest revisions.  It's obviously a fairly critical problem since Dave's unit was rebooting itself because of it.

question is: how to identify "later versions"? is the unit i'll get delivered in a few days a "later version"? do i have to open it to actually see if there is a difference, including voiding my warranty?

and what about dave and all the others definately having an "old version"?

[marshallh]

Both ON and ST are making 78xx parts with new cut down TO220 packages. The metal tab is very thin and shorter. They did this to the same part number and I know ST offers the old leadframe as a new part no. So if you want to keep using the same part, you need to change your BOM

[mcinque]

question is: how to identify "later versions"? is the unit i'll get delivered in a few days a "later version"? do i have to open it to actually see if there is a difference, including voiding my warranty?
and what about dave and all the others definately having an "old version"?

I think that we don't know how at the moment. Of course we'll be able to identify the defective units just by serial numbers, but not until Rigol publicly admits the issue and list the serial numbers ranges affected.

I think that next thursday, after the meeting with the Rigol's engineer, my local representative should give me a better explaination. And I hope a smart solution.

The fable of a defective component sounds really strange also to me, it's an evident bad pcb and heatsink design, but... we should admit that Rigol CAN effectively do a good engineering (just look at other Rigol'ls teardown videos); an oversight like that is so stupid that I cannot (or I won't) believe it's a faulty design.

[EEVblog]

It's not rocket science  Simple LM317 linear regulator running hot!
It comes down to voltage drop and current consumpton. Or a parasitic oscillation which increases the current.

Bottom line. The LM317 is dissipating about 5W at 240V (will change with line voltage of course). It is dropping out of regulation when it overheats.
It is almost certainly running above the recommended max die temp, even with the airflow.

[mickpah]

It's not rocket science  Simple LM317 linear regulator running hot!
It comes down to voltage drop and current consumpton. Or a parasitic oscillation which increases the current.

Bottom line. The LM317 is dissipating about 5W at 240V (will change with line voltage of course). It is dropping out of regulation when it overheats.
It is almost certainly running above the recommended max die temp, even with the airflow.

Even if they recall/rework what a major pain in the arse. time & cost of packaging shipping alone to get it back to them is enough to make me rethink any future Rigols purchases. I was in the market for a function gen, used Agilent is looking better than new Rigol now
AND I got rid of the box as I have no room to store. Feeling burnt here.

[mcinque]

time & cost of packaging shipping alone to get it back to them is enough to make me rethink any future Rigols purchases.

You read my mind. The right thing they should do is recall al the defective units, giving a new unit to the customer while taking back the defective ones. A fairytale.

I was in the market for a function gen, used Agilent is looking better than new Rigol now

The funny thing is that many Agilent's devices are manufactured by Rigol...

However, they are NOT idiots at PCB design. All the Rigol's spectrum analyzers and DSO on the market are well engineered. The point is that the project of 832 PSUs seems to have been done hastily.

[mickpah]

However, they are NOT idiots at PCB design. All the Rigol's spectrum analyzers and DSO on the market are well engineered. The point is that the project of 832 PSUs seems to have been done hastily.

and yet the all the board are marked rev 2.0 !! thank goodness they didn't ship rev 1 then.

[EEVblog]

The funny thing is that many Agilent's devices are manufactured by Rigol...

The 1000 series scope is, but apart from that I have not heard of any other Rigol OEM Agilent gear?

[echen1024]

The funny thing is that many Agilent's devices are manufactured by Rigol...

The 1000 series scope is, but apart from that I have not heard of any other Rigol OEM Agilent gear?

Most of Agilent equipment is not OEM'd. Also, try a new LM317. I seriously think this is just Rigol PR speak  to stall for time, while they can think of the actual issue, and think of a way to admit it. You might want to investigate the output oscillations of the LM317, and try putting it away from the electrolytics.

[Mr Simpleton]

So it seems the heat sink is a bit to small and/or the LM317 attached to heat sink is not properly done!
Maybe operating the PS at 240 causes the unstabilized input to be a bit too much. Not sure but China should have 220 and 230/240 in Austraila?

Still not enough safety margin. Bigger heat sink should take care of any problem.

[Electro Fan]

It's not rocket science  Simple LM317 linear regulator running hot!
It comes down to voltage drop and current consumpton. Or a parasitic oscillation which increases the current.

Bottom line. The LM317 is dissipating about 5W at 240V (will change with line voltage of course). It is dropping out of regulation when it overheats.
It is almost certainly running above the recommended max die temp, even with the airflow.

Does toggling the unit on and off give it a chance to cool down with each toggle or does on/off toggling increase the heat?  (Seems somewhat unlikely the two are connected but just curious to see if there is any potential relationship between the heat issue and the turn-on spike issue.)

[EEVblog]

Dave, how would you feel about replacing your LM317? Your warranty is already void and it would confirm or disprove Rigol's claim.

As far as I know, Rigol have not claimed anything?
They have simply admitted the problem exists and they have a hardware solution for it with a bigger heatsink.