EEVblog #511 - Rigol DP832 Power Supply Teardown

[ElectroIrradiator]

My Mk. I eyeball claims the analog regulator schematic superficially appear similar to the standard ones use in many Agilent PSUs, which is probably not a bad thing. TL072+LM393? That plus the 3A output current limit is probably the reason for the 1000uF caps across the output rails, many Agilent supplies have those as well (with similar potentially negative effects, as mentioned by Dave).

The MCU on the top board looks like it has part of a Cypress logo in the upper LH corner, but I have no clue as to which one it might be.

If somebody else makes a tear down, then it might be interesting to get the physical dimensions of the toroid transformer, without mounting hardware and connecting wires. It looks quite large and the dimensions should give us an idea about the VA rating, which should be around 350-450VA or so.

----
Edit: Dave provided the answer to my question in a different thread. The transformer is 110 mm (dia) x 105 mm (height).

[Oracle]

Rust in a factory new equipment gear is wrong considering the amount of money you give... but maybe is not rigol fault: maybe the equipment was previously stored in a humid place by the local distributor.

[Dread]

It seems all he does is complain, so I'll no longer respond to his comments directly.
Of course, he conveniently forgets the fact that my previous teardown video (the UPS) combined whiteboard stuff with the teardown which made it quite popular. But no, that doesn't fit with his opinion that my teardowns are just the same and useless, so he ignores those ones.
This one was obviously for those who wanted to see inside the Rigol supply, and nothing more. I never claimed it be anything more.

Now that video was excellent. It had broad appeal and was an educational treat.  IMO the best teardown videos will almost always be the simple ones that can be presented in a complete dissection that demonstrates exactly whats going on in sub sections that one can use in other designs. The APC proprietary hybrid design was well worth the admission fee

[andete]

Thanks Dave for this excellent informative tear-down!

I'm just wondering, are the large output caps really a problem in practice? Would it be interesting to test?

[dougg]

Is someone making a Dave English -> Amurrican English dictionary? I'm not quite sure what to make of "It's the duck's guts." Is that a good thing?

Dave has some int-teres-ting pronunciations, even to Australian ears. Perhaps he spent some time in the land of the long white cloud ... And expressions like "in like Flynn" (referring to the long dead actor Errol Flynn's love life) are not common currency, at least not when I drove a cab in Sydney when I was a EE student. But it breaks both ways, the first time I read "dove under the table" in Canada I thought about pigeons ...

I make do with an Agilent U8001A PS which is fine for my purposes, apart from a damn loud fan. It gets around power on voltage spikes by having an output relay that is disconnected at power up until an activate button is pressed on the front panel; and I can check the voltage and current settings before pressing the activate. Perhaps Dave could test the 832's constant current performance give that large capacitor across its output.


PS Flynn was born in Hobart, Tasmania so that is the Australian connection. He died in 1959 (before Dave was born?) after a Hollywood career more noted for its off-screen exploits.

[rs20]

I've asked Rigol support, but had worse than useless (i.e., clearly completely wrong answers), so I'm going to ask this where I should have asked the whole time: eevblog forum!

Would it be safe to parallel channels two and three together, and set channel two to 30V? Clearly the channel three, maxing out at 5V, can't source any current, so it's a pointless thing to do (I'll get back to this), but what I'm wondering is would channel three actually sink current or get damaged? I see what might be considered protection diodes http://www.flickr.com/photos/eevblog/9607810872/#in/set-72157635251122253, but they appear to be parallel to the load, presumably preventing the output from being reverse biased. <aside> Also, the current shunts are on the ground side, which is a little bit funky. Does that mean that the supposedly common negative terminals have 40 milliohms between them? I just went to check, and indeed, pumping one amp through one negative terminal and out the other gives you a drop of 38mV. </aside>

The reason I'm asking is I want a supply that supplies 1 volt @ 9 amps, and then seamlessly transitions to providing 30 volts @ 6 amps.

[Christe4nM]

Thanks for doing the teardown Dave. I agree with free_electron and others that there's always things you can learn. I for one like to look at board layout, choices they've made at system level etc.

Two questions for those that like to share their knowledge:

1) If I remember correctly I thought that the E-I laminated transformers were preferred in PSUs as they provide better isolation (less unwanted coupling) between primary and secondaries compared to toroidal.

2) Can anyone explain why they need a full applications processor in this thing, alike the Agilent DMM? From my (inexperienced) point of view that seems way overkill for either instrument. Although I get that running 'real-time' datalogging, USB, Ethernet and the color screen may require a fast processor, it seems to me that the current top end microcontrollers (for example the Cortex-M3/M4) are capable enough? Am I overlooking something that I don't know about?

[Rufus]

I'm just wondering, are the large output caps really a problem in practice? Would it be interesting to test?

Yes they are a problem especially if you don't expect them to be that large. A current limit is often set to try to provide some protection for a circuit. For example you could set a 10mA current limit to test an LED and if the voltage is set too high you could kill the LED with the energy stored in the output capacitors regardless of the 10mA current limit.

[EEVblog]

I've asked Rigol support, but had worse than useless (i.e., clearly completely wrong answers), so I'm going to ask this where I should have asked the whole time: eevblog forum!

Would it be safe to parallel channels two and three together, and set channel two to 30V? Clearly the channel three, maxing out at 5V, can't source any current, so it's a pointless thing to do (I'll get back to this), but what I'm wondering is would channel three actually sink current or get damaged?

Get damaged most likely.
For a PSU to "sink" current is has to be a 4 quadrant type, which very few bench supplies are. This one certainly isn't just a regular output series pass transistor.

[RobB]

Any thoughts on the complexity of this PSU.
I am of the opinion that PSU's are among the most used and abused bits of equipment and therefore most likely to be damaged in service. I'd hate to have to trouble shoot this thing, too many proprietry components like the flex cable, unidentified components etc. I appreciate this is a precision PSU with lots of nice user features but I'd trade that for something more mundane but repairable.
Finally in the words of Niel Young. Rust never sleeps. Some cold galv paint on those raw cut edges should help.

[Citizen]

I don't know, maybe it is just me...

What's the point, Citizen. You carefully put this comment both on the blog andd youtube comments.
For a few video's already.
The obligatory Citizen rant.
Blogger envy?

I only stated it for this video, here and on youtube, because there are two dieferent communities.
It is not obligitary rant, it is my opinion. I am not envy, why should i? Maybe you mean my tutorials?Cmon, i am not blogger, and i  not planning to be one at all.

This community reminds my a temple, where everyone worship the perfect Dave.
 Really look how agressive  users get to my  actually constructive oppinion:

Good , hopefully you stay humble and humbly understand and agree that your taste do NOT represent majority of the viewers

Meh .. just look at the "attitude" at his videos, always proclaims it proudly.Definitely not even the same league as Dave's

--ROFL, sure i am no match for  great Dave!:))
And there some angry fanatics on the youtube. I dont like this kind of cumminity.

 But to me Dave's videos are getting  less educative and more entertaining, this happens to many TV-show, even to Discovery Channel. Before you could  learn lots  from his videos: how to solder, desolder, measure, repair,layout, design, test, program, more detailed  mailbags, more unusual equipment for TT. For me the last interesting   and educative video was the JTAG FF.

I think i will just leave eevblog if anything will change. Sure not a big deal for Dave, since he has  thousands of  others  followers. But consider,  few monthes ago i was looking Dave videos with much more interests, and always gave him positive feedbacks. So something has changed since then.

[rs20]

I've asked Rigol support, but had worse than useless (i.e., clearly completely wrong answers), so I'm going to ask this where I should have asked the whole time: eevblog forum!

Would it be safe to parallel channels two and three together, and set channel two to 30V? Clearly the channel three, maxing out at 5V, can't source any current, so it's a pointless thing to do (I'll get back to this), but what I'm wondering is would channel three actually sink current or get damaged?

Get damaged most likely.
For a PSU to "sink" current is has to be a 4 quadrant type, which very few bench supplies are. This one certainly isn't just a regular output series pass transistor.

Thanks for responding, Dave! To clarify, I wasn't wanting it to sink current, was just hoping/expecting it would have a diode protecting its output and not get damanged. On your advice, I won't take that chance -- I'll get myself a schmicko Schottky and connect it from CH3+ to CH2+ and take my power from there. Cheers!

[anotherlin]

2) Can anyone explain why they need a full applications processor in this thing, alike the Agilent DMM? From my (inexperienced) point of view that seems way overkill for either instrument. Although I get that running 'real-time' datalogging, USB, Ethernet and the color screen may require a fast processor, it seems to me that the current top end microcontrollers (for example the Cortex-M3/M4) are capable enough? Am I overlooking something that I don't know about?

Yes, there wouldn't be any problem to do all the mentionned stuff  needed using a high-end microcontroller.
But my guess, developing the software is much easier with a bigger faster processor, and this "full application" processor is not much more expensive compared to a high-end microcontroller.

[jancumps]

... and maybe they needed it to drive the flashy display of the model A, and for the optional add-ons.

[baljemmett]

But to me Dave's videos are getting  less educative and more entertaining,

"But it's all right now, I learned my lesson well,
You can't please everybody, so you got to please yourself."

... or perhaps you'd rather Dave went back to driving a truck?

[anotherlin]

My Mk. I eyeball claims the analog regulator schematic superficially appear similar to the standard ones use in many Agilent PSUs, which is probably not a bad thing. TL072+LM393? That plus the 3A output current limit is probably the reason for the 1000uF caps across the output rails, many Agilent supplies have those as well (with similar potentially negative effects, as mentioned by Dave).

Agilent's PSUs are basically textbook: triac or SCR pre-regulator + series-pass transistor regulator with feedback control using opamps.
Except using a switch moded pre-regulator, or not having one, there's not much other way to design a linear supply (using a shunt style regulator maybe).
It's no suprise Rigol's PSUs look "superficially similar".

[Citizen]

But to me Dave's videos are getting  less educative and more entertaining,

"But it's all right now, I learned my lesson well,
You can't please everybody, so you got to please yourself."

... or perhaps you'd rather Dave went back to driving a truck?

But the educative content was the reason why many ppl,including me, joined EEVBLOG  few years ago. This is maybe the answer , why I started to give Dave criticism and neg. votes for last few monthes. But sure if you want to run a blog for living, you should consider the majority, and change yourself to hold the most viewers.
This happens everywhere: Discovery Channels was once very interesting and educative, but now it is switched to reality shows like Jungle/Alaska/Bering Sea Gold ,
same with BBC.

[Cdngreybeard]

The rigolna.com DP800 site has two application notes that discuss active loads and series parallel connections.

The notes confirm that the DP800 series are first quadrant supplies and that for the DP832 series channel 1 and channel 2 or 3 may be connected in series or parallel.   

The notes are an interesting read.  Check them out.

[free_electron]

Just watched the video. Here is the interesting bits my trained eye spotted

- they use flat flex from front panel to top board , then simple loose-wire connectors from top board to bottom board.... Anyone care to guess why ? Citizen maybe ? (I know why)
- milled slot around the reference diode for temperature stabilisation(thermally induced mechanical stress)
-guarded trace from the opamp to the cpu, thats more than 16 bit there ...
-that cpu is a Silicon Labs device , judging by the font of the stuff they didnt scrape off. It would match the guarded trace. Silabs has some cpus with 24 bit a/d ...
-mosfet as pass regulator element
- the big caps don't need 'goop' to tack em down as those use snap-in pins.
-look at the routing between the main cpu and its ram.. Lots of curly wurly stuff. This is ddr3 !
-noticed the text 'batteryv' close to the cpu.... This thing can have battery backup apparently... I wonder if this board may also be used in other instruments... Like their signal generators.

Stuff i learned from the video:

- there is now apprently special triac driver ics in sureface mount. Need to look that up. Last time i used a triac was 20 years ago. And probably will not use one in the foreseeable future but still the driver is nice to know. I am aware of the moc3061 zero cross optocoupler l and how to drive a triac using a small pulse transformer. Wonder what this driver ic does...

-the flatflex gave me an idea for a project i am working on (the flatflex is that white semi-rigid cable from the front panel to the top board. This is essentially a plastic tape with slats of flat copper adhered to it. Almost a flexible pcb. )

- toroids are not necesarily flat pancakes... This toroid is much taller than its diameter ... Never seen one with that aspect ratio.

So in this 'another psu' video there is plenty to be learned ! You just have to know where to look !
If i walk away learning 1 thing from a video or article, that's time well spent ! Plenty of educational value ! And i learned two things from this video (triac driver, toroid aspect). That says enough....

[BravoV]

..<snip>...

I dont like this kind of cumminity. community

..<snip>...

I think i will just leave eevblog if anything will change.

..<snip>...

Anyone care to guess why ? Citizen maybe ?

..<snip>...

If i walk away learning 1 thing from a video or article, that's time well spent ! Plenty of educational value ! And i learned two things from this video (triac driver, toroid aspect). That says enough....

Save your breath, he is leaving and that is his last post here.

[mrflibble]

Save your breath, he is leaving and that is his last post here.

Maybe to join a self-help forum, with a focus on accepting criticism without getting defensive? 

[Blaffetuur]

Thanks Dave for this excellent informative tear-down!

I'm just wondering, are the large output caps really a problem in practice? Would it be interesting to test?

would like to see a video about that !

[Mikey]

Great teardown, only thing I dont like about it, is that the device is one that I will never get my hands on

The cheap end models that are total crap compared to this is more useful, I have already used some to find what equipment to get.

[mrflibble]

Just watched it, and there was a good bit of inspiration & assorted tidbits in there. Thanks for a nice teardown Dave. 

[ElectroIrradiator]

My Mk. I eyeball claims the analog regulator schematic superficially appear similar to the standard ones use in many Agilent PSUs, which is probably not a bad thing. TL072+LM393? That plus the 3A output current limit is probably the reason for the 1000uF caps across the output rails, many Agilent supplies have those as well (with similar potentially negative effects, as mentioned by Dave).

Agilent's PSUs are basically textbook: triac or SCR pre-regulator + series-pass transistor regulator with feedback control using opamps.
Except using a switch moded pre-regulator, or not having one, there's not much other way to design a linear supply (using a shunt style regulator maybe).
It's no suprise Rigol's PSUs look "superficially similar".

 

I'm talking about the actual schematic for the series regulator, built with discrete components, not the overall method of building a linear PSU. Have a look at the schematic and PCB layout for, say, the triple rail Agilent E3631A.

For instance the Triacs are not used as pre-regulators, nor are using SCRs an option here. The Triacs are used as bipolar switches to select taps on the transformer secondaries, to limit power dissipation in the series pass device at low output voltages. In the E3631A they are triggered using optically coupled Diacs, which I suspect may be the SMDs close to the Triacs.