EEVblog #512 - Rigol DP832 Bad Design Investigation

[EEVblog]

An LM317 needs 3v across it so that's 8v, add on the ripple and line input range and a bit less than 12v input average at the high end of the line range isn't far out.

A nominal 9V RMS tap would give 12.7V peak. Subtract 1.2V for full bridge diode drops and say 1V for ripple, still leaves you with 10.5V.
BTW, I'm reading 10.5V RMS on the tap. It's marked 10V on the transformer label.

[edpalmer42]

Dave,

When you were trying to trigger the scope on the problem, you missed the biggest, boldest trigger signal possible - the output voltage!  Use channel 2 on your scope to monitor the output voltage and trigger a single shot capture when the output disappears.  Set the trigger to capture in the middle of the memory so that you've got the channel 1 signal before and after the output disappears.

As for whether the regulator caused the reset, you appeared to be using slow sweep speeds.  Depending on how the LM317 responded to the overloads, it could be causing a very narrow outage glitch.  You wouldn't have seen it at slow sweep speeds.

Ed

[Corporate666]

The obvious easy fix is to replace the reg with a switcher - there are plenty of drop-in 7805 replacements.

Maybe that explains how this happened in the first place.  Someone designed it with a switcher in mind, and a bean counter came along and said "wait, that part is 5 times the price of this other regulator - let's go with the cheaper one!"

[Corporate666]

If Rigol wanted to be as cheap as possible then, as a hotfix, they could probably send all of the current users one of those 20W 8R power resistors that they use for testing audio systems.  Just stick that sucker inline between the +12V output and the LM317's input and it would probably dissipate enough heat to solve the problem.

I was thinking about the fix and the more I think about it, the more of a real shit-show this is for Rigol.  I don't think there is any way a company like Rigol would ever offer up a fix where the user is soldering anything.  Actually, if Dave said that heat sink is PCB-mounted with tabs, I don't even see how they could offer up a replacement heat sink because that would require soldering as well.  Not to mention they would have to agree to waive the "Warranty void" labels being broken in future warranty claims, and on top of that what if someone breaks their unit while attempting the fix? 

Really, all these units will need to go back to Rigol.  I have no idea how many they sell or have sold of the 832 series, but a recall is going to be unheard of for a test equipment manufacturer, I'd think.  I think Rigol is going to be extremely hesitant to do that, especially in light of the fact that we don't know any have actually failed.  On the other hand, we know it's an ST LM317 with a max junction temp of 125C and we know it's running above that, so Rigol can't really say the design is fine either.

I really hope they don't do what Dave suggested might happen, and distribute a bullshit firmware that runs the fan balls-out to compensate.

They really need to do a recall on this.  Sucks to have to do it, but what can ya do...  actually, a recall and a coupon for a healthy discount on another piece of Rigol gear bought factory direct might work out well...  they could make up the cost of the recall with that.  I'd love to get a DG4062 for $200 off, let's say

[talvor]

Is it normal to use 12v and regulate it down to 5?

Nope. Normally you pick something like a 9V tap for that.

This is a newbie design question.

Dave said that it is not normal to use a 12v input to regulate down to 5v.  Would a voltage divider work to get the 9v out of the 12v input?

[johnnyfp]

You still have to dissipate the drop of 12v to 9v. Your now just doing it in the Resistor instead.

[jon_joy_1999]

105c caps right next to a device that reaches 111c, smells like planned obsolescence (and boiled electrolyte) to me!

[TerminalJack505]

I was thinking a dropper resistor would be a workable kludge too but if that same rail is expected to power any connected USB devices as well then it probably wont. 

Like someone else mentioned, some kind of active voltage dropper would have to be used.  It would need its own heat sink and the LM317 still might need a bigger heat sink for the worst-case scenario. 

[pickle9000]

I have seen heat sinks doubled up, in one case a much larger one so the fins all came out in the same direction. No soldering required at least.

[johnnyfp]

Ah hell, let's just replace it with water cooling 

[EEVblog]

They really need to do a recall on this.  Sucks to have to do it, but what can ya do...

My guess is they will quietly fix the issue in new production units, and then honour the 3 year warranty claims as and when they come up, and hope it's not too many.

[Monkeh]

The thin tab is indeed perfectly normal.



Both of those are genuine ST parts, and the right hand, single-gauge one came from Farnell. Cost almost twice as much as the other, too.

[Bored@Work]

Really, all these units will need to go back to Rigol.

Maybe they are lucky. The power is coming to the board via some connector. If the power from that connector is only going into the regulator they could device a fix that is supposed to be plugged in between the cable and the board connector. Like a pre-reglator, dropper resistor, etc. If they are lucky no soldering required.

But my bet is on Dave's suggestion. They might do nothing and just treat broken stuff under warranty, hoping there won't be too many. Hoping people won't find their receipt after a year or two, and probably playing the stupid game they played in the past: Service, support and warranty only for units bought via official distributors, screwing everyone who bought on ebay, or did use some other way of doing a gray import.

[Laidukas]

Does this video mean that higher end model that shares the hardware also has this issue?

Damn, I trusted RIGOL so much, I have a lot if RIGOL equipment and now they spoiled their reputation.

[EEVblog]

I have confirmation that the issue has already been fixed in a new board revision.
They changed the board to fit a larger heatsink, but it's not a huge amount bigger. See photo.
They measured the temp at 76C in a 28C ambient , but I'm not sure if this is free standing out of the case, or in the case with airflow.
They will be replacing all Australian units, I don't know about other countries.

I would not call this a proper belt'n'braces fix, as I think the heatsink still isn't big enough, but it probably now puts it into the (just) acceptable category.
They have also apparently made a few other fixes.

[AndyC_772]

It's rather sad that they didn't take the opportunity to increase the gap between the heat sink and those two electrolytic caps a bit further, it's not as though there's no room

[riconette]

I have confirmation that the issue has already been fixed in a new board revision.
They changed the board to fit a larger heatsink,…

so much for the claim that "defective parts" have been used.

[EEVblog]

It's rather sad that they didn't take the opportunity to increase the gap between the heat sink and those two electrolytic caps a bit further, it's not as though there's no room

Technically they are arse backwards. The airflow will come from the side vents, over the heatsink, and then out the back over the caps. The heatsink should have been toward the rear (left in image).

[AndyC_772]

Actually I take it back... I've just seen how bad they were on the first rev!

[mickpah]

They will be replacing all Australian units, I don't know about other countries.

I would not call this a proper belt'n'braces fix, as I think the heatsink still isn't big enough, but it probably now puts it into the (just) acceptable category.
They have also apparently made a few other fixes.

this restores my confidence a bit, but I think I owe you a thanks Dave.
Call me cynical but I don't think there would have been any action without your BS call

[Monkeh]

They will be replacing all Australian units, I don't know about other countries.

I would not call this a proper belt'n'braces fix, as I think the heatsink still isn't big enough, but it probably now puts it into the (just) acceptable category.
They have also apparently made a few other fixes.

this restores my confidence a bit, but I think I owe you a thanks Dave.
Call me cynical but I don't think there would have been any action without your BS call

I really don't think they sourced a heatsink, modified the board, and had it produced and tested in just two days..

[AndyC_772]

They might have decided whose units get replaced with new ones, though.

[EEVblog]

Call me cynical but I don't think there would have been any action without your BS call

I can't take credit.
They obviously knew about this and had this fix in place for some time.
How they found out, I don't know.
And I don't know if these new units are on shelves yet or not.

[Redtailed]

From the forum on video #509, I said:

I remain confident that if Dave opened up a Rigol product that reeked of oversight, bad design, or dangerous flaws, he'd have a field day showing them in all their glory and ask the manufacturer to respond.

I swear I didn't predict this.

Please no email asking for lottery ticket numbers.

[firewalker]

Why a mosfet as a series pass element?

Alexander.