An update on the Rigol DP832 lab power supply fix.
An update on the Rigol DP832 lab power supply fix.
IMO, 
+1
I'm still not convinced about this design.
So, they've fixed the over heating problem. This issue shouldn't have made it off the drawing board, how did it get through testing and development?
Ditto the software bug - didn't they bother to test this design before putting it into production?
"Precision" ? Where? This design doesn't have proper sense terminals. As you've just demonstrated in the video, it's easy to lose 20mV down a few cm of wire when you start pushing current through it - that's twice the claimed basic resolution! It makes a mockery of the 90 bucks you paid to check that the 1mV software switch worked.
Is there a cap missing on the new board? When you have them side by side at around 7 minutes, there are two small electrolytics just to the right and above the three main caps on the old board, but I can see only one on the new. There is some odd perspective difference, in that a lot of the components visible on the new board to the immediate right of the big heatsink are almost entirely hidden on the old, but I don't think that explains the cap.
After cancelling my original unit and waiting for the latest revision of the board I ended up with a DP832A board V02.10 dated 2013-03-26, with the the 'short' black anodized heat sink.
As Dave encourages I opened it up. I lifted the warranty sticker which had already been lifted on one side by whom ever upgraded it to the 'latest?'
Once in for the penny I opted to replace the heat sink with a much larger one mounted on stand-offs fastened to the side and re-located the LM317 from the board to it similar to an earlier post in the forum.
For me as a 'hobbyist' even with its shortfalls it has much more than I may require and for the cost of the DP832 I think it is a great deal.
So I am happy.
I could see how 'Lab' procurements & installations might be 'upset' if the higher priced units are based on the same design though.
"Precision" ? Where? This design doesn't have proper sense terminals. As you've just demonstrated in the video, it's easy to lose 20mV down a few cm of wire when you start pushing current through it - that's twice the claimed basic resolution! It makes a mockery of the 90 bucks you paid to check that the 1mV software switch worked.
The unit easily meets in specs when used in the normal ground return mode.
Yes, I don't like it any more than you do, but it is far from being a mockery, or useless. I now know the limitation and can avoid that if I require the precision.
Does that mean that the entire return current when using the "wrong" jack is going through the sense wire? Is it of sufficient gauge to handle that without issue? usually sense wires are not very thick.
It doesn't seem that unusual of a case to come up in practice as I often use my current supply to give +12 and +5 to the same device that share a common ground with some pretty beefy current going through the 12V line, I'd hate to think I could blow an internal sense wire by choosing the wrong ground plug.
Every "multiple output" PS I've ever used has had completely independent/isolated returns ("the black binding post") such that the outputs could be wired in series. I have and have used split ("bipolar" or "+/-" or whatever you wanna call 'em) supplies that have a common return with positive-referenced and negative referenced outputs. But this DP832 is just plain weird.
AFAIC, that 5V output is NOT a third "multiple" output; it's a "split" output with the second...
(Me smell chop phooey!) And, the true test of a "multiple output" PS is if its outputs can all be wired in series. So me no buy DP832!
For V3.0, fix the returns, and make the "5V" output adjustable and/or with presets (1.8V, 3.3V, etc.)
Every "multiple output" PS I've ever used has had completely independent/isolated returns ("the black binding post") such that the outputs could be wired in series.
It's not uncommon at all to have supplies with a shared common output like this. Especially a lower voltage rail like this one uses. It's a design/cost tradeoff.
It's not uncommon at all to have supplies with a shared common output... It's a design/cost tradeoff.
I have a couple, but they're "split supplies" not "multiple output supplies." (Maybe I'm "spoiled" in the PS department? Anyway...)
The 5V output is adjustable.
Just checkin' to see if you paytention...
(I gotta go crash... It's bloody midnight already!
)
Cheers!
Is there a cap missing on the new board?
Yes, they removed one smaller cap.
Edit:
Regarding that current sensing in the ground connection, shame on Rigol. Every serious article about power supply design tells you not to do it on the ground connection, unless the channel is completely isolated. With that current sensing their claim to a precision power supply is just hogwash.
With that current sensing their claim to a precision power supply is just hogwash.
Only for the usage case demonstrated. Otherwise it
is a precision supply and meets its spec.
So in the 4 possible connection usage cases:
1) Single supply = meets precision spec
2) Dual isolated supply = meets precision spec
3) Dual isolated + 3rd channel (separate return grounding) = meets precision spec
4) Dual isolated + 3rd channel (shared return ground) = may not meet it's precision spec (depending upon output current)
With that current sensing their claim to a precision power supply is just hogwash.
Only for the usage case demonstrated. Otherwise it is a precision supply and meets its spec.
So in the 4 possible connection usage cases:
1) Single supply = meets precision spec
2) Dual isolated supply = meets precision spec
3) Dual isolated + 3rd channel (separate return grounding) = meets precision spec
4) Dual isolated + 3rd channel (shared return ground) = may not meet it's precision spec (depending upon output current)
I guess Rigol also needs to update their advertising/brochure with a small footnote ....
"Specification on the precision is not guaranteed and not applicable at certain connection"
Regarding that current sensing in the ground connection, shame on Rigol. Every serious article about power supply design tells you not to do it on the ground connection, unless the channel is completely isolated. With that current sensing their claim to a precision power supply is just hogwash.
Except putting the sense resistors low here has caused no problem, and nor would I expect it to. It'd work just fine if they strongly tied the two common terminals together, and just had one voltage sense wire sampling the potential at this "one true ground". The thing that I find really interesting and original (to my newbie eyes) about this schematic is that CH2 and CH3 are completely isolated, except for the one connected ground sense point (let's just pretend they commoned them properly, not via sense wires). In particular, they have completely separate coils on the main transformer, separate rectifiers, etc etc, all the way out to the outputs,
and all the way back again. That makes the location of the current sense resistors completely irrelevant according to your basic Kirchoff's Voltage Law. Rigol just put them low because it's easier to measure there.
I'm a bit tired and what I wrote above might not make sense, but let's at least agree on this point:
if they moved the current sense resistors to the high side, but left the messed up commoning-via-sense-wires, then the voltage error with "incorrect connections" would still exist. Nothing to do with that problem at all.
Except putting the sense resistors low here has caused no problem, and nor would I expect it to. It'd work just fine if they strongly tied the two common terminals together, and just had one voltage sense wire sampling the potential at this "one true ground".
Indeed, the effect of which I showed in the video.
I'll probably mod my own unit to fix that.
You are going soft Dave
, this is a lazy Claytons fix. As you said the correct solution would have been to reduce the voltage coming from the tranny. I guess Rigol already have a stock of trannies they need to shift. IMHO there is still far too much power going into heating the lab. As you would say, you can fly to the moon on a few hundred milliwatts. Have you considered adding a contra winding the secondary to reduce the tranny output voltage? Yes it's a dirty hack but an effective hack all the same.
Well have to say, it's looking more attractive every day, but I do not like the round keypad.
Al
You are going soft Dave , this is a lazy Claytons fix. As you said the correct solution would have been to reduce the voltage coming from the tranny. I guess Rigol already have a stock of trannies they need to shift. IMHO there is still far too much power going into heating the lab. As you would say, you can fly to the moon on a few hundred milliwatts. Have you considered adding a contra winding the secondary to reduce the tranny output voltage? Yes it's a dirty hack but an effective hack all the same.
Nah bro, add contra windings to the
primary!! Then
all of the secondary side circuitry will think it's in the USA
(modulo frequency)
You are going soft Dave , this is a lazy Claytons fix. As you said the correct solution would have been to reduce the voltage coming from the tranny.
There also may be a margin issue on worldwide voltages at play, perhaps there isn't much to be gained here. I have not run the numbers...
IMHO there is still far too much power going into heating the lab.
I agree, but this is a linear lab supply, perhaps the most inefficient bit of gear in any lab. So pissing away a few watts here isn't a big deal.
Have you considered adding a contra winding the secondary to reduce the tranny output voltage?
I'm not that fussed to bother with doing that.
Hmm considering that so far all people have got black anodized heatsinks on their PSU, would it be possible that they just put in custom (much larger) heatsink just for you?
Hmm considering that so far all people have got black anodized heatsinks on their PSU, would it be possible that they just put in custom (much larger) heatsink just for you?
Nope. All the boards shipped to Emona had that larger heatsink.
Don't ask me how, but I've got a couple pictures of Rigol's newest power supply, code named the DP92:
Don't ask me how, but I've got a couple pictures of Rigol's newest power supply, code named the DP92:
I would have guessed they would just use some water cooling on that next time.
But seriously talking how can they fail 2 times in a row with such... quite easy task. Okay they got it wrong the first time around... no big deal, they fixed it and then it still wasn't enough and they had to fix the fix. (Reminds me one of Dave's teardowns where he found bodges on bodges)
Who would have guessed that they fail to calculate the size requirement for that heatsink, twice.
Any news on the reset and the power-on (voltage spikes while output is disabled) issues?