EEVblog #512 - Rigol DP832 Bad Design Investigation

[c4757p]

For what you need 500mA on that PSU to power USB devices?

It's very simple: you should not be able to thermally stress and possibly damage, or crash, any device with a USB port by plugging in a perfectly valid USB device. Ports are external, they should be able to take a bit of abuse.

[brainwash]

Reminds me of my cheap car stereo that reboots if I plug in the phone but works fine with a stick. Hmm... I might try to do a mod on that.

[rs20]

How about replacing the linear reg with one of these switching-regulators-in-a-TO-220-footprint devices?:

http://www.murata-ps.com/data/power/oki-78sr.pdf

Less than US/AU$10 from digikey. Now of course, switching noise is always a concern with any switching regulator, but I get the impression that nothing sensitive/analog is being powered off this line?

[sync]

If you look at Dave's video you can see there is actually plenty of ripple after the LM317; the ripple seemed to increase with temperature.

Sure, you can stick a larger TO220 and heatsink to the whole thing but any added consumer (like USB) will force the part some more and result in high inefficiency. I would still stick a switched preregulator before the LM317.

A adequate cooled LM317 can handle 1.5A @ 7V dropout with out problems. With low ripple and noise. It's just 10W.

This is a linear power supply. It's not about efficiency. It's wasting 10s or 100s watts anyway. Saving 3W with a switching preregulator doesn't change that.

[brainwash]

It's not about power consumption nor ripple: the LM317 now takes raw rectified input anyway so a switching preregulator will probably reduce the ripple.
That additional 3-7 watts count towards the thermal characteristic of the entire unit causing the fan to rev a lot of the time and reduce unit life by a few percents. Plus, the bigger heatsink is not free so that money is better spent on something more efficient.

[nigel53]

The LM317 doesn't have particularly good ripple rejection though, so you've still got the wideband noise problem that a cheap switcher will give.

I believe the datasheet says something about 70db rejection and that should be pretty good.

Look at datasheet, 70dB will not be at typical switching frequencies!

[sync]

It's not about power consumption nor ripple: the LM317 now takes raw rectified input anyway so a switching preregulator will probably reduce the ripple.

Unlikely. A proper used LM317 has orders of magnitude lower mains ripple that needed. But a switching regulator is a RF transmitter and a LM317 is not good at filtering RF. It doesn't make sense to use a preregulator. Just replace the LM317 with a proper switcher then.

That additional 3-7 watts count towards the thermal characteristic of the entire unit causing the fan to rev a lot of the time and reduce unit life by a few percents. Plus, the bigger heatsink is not free so that money is better spent on something more efficient.

A better thermal design will give you more than a few percents more life time. Just bold the dammed LM317 on the main heatsink - done. If this makes thermal problems then the design is broken anyway.

[Steffen]

The power supply should not reboot if the USB port is loaded. Like a computer. What do you think of a computer which will crash when a USB port is loaded/overloaded?

No joke, i've seen computers rebooting when a USB device has been plugged in. Reason: Broken USB connector on the motherboard and shorted out pins when plugging sth. in. I found such computers in my school.
For those cellphone chargers: I've also seen measurement devices in the 4-5 digit price sector with USB ports specified for max. 100mA. This was written in the spec. and in the manual. For thumb drives and mice/keyboards this is totally ok. If you need more ports than the typically 1-2 you get, there is always the possibility to use an active usb hub.

[sync]

For those cellphone chargers: I've also seen measurement devices in the 4-5 digit price sector with USB ports specified for max. 100mA. This was written in the spec. and in the manual. For thumb drives and mice/keyboards this is totally ok. If you need more ports than the typically 1-2 you get, there is always the possibility to use an active usb hub.

It's ok when an USB port limits the current to 100mA. But surely the poor LM317 in the Rigol doesn't withstand additional 100mA.

[FrankBuss]

It's ok when an USB port limits the current to 100mA. But surely the poor LM317 in the Rigol doesn't withstand additional 100mA.

The USB 2.0 specification is very clear in chapter 7.2.1: "Systems that obtain operating power externally, either AC or DC, must supply at least five unit loads to each port. Such ports are called high-power ports."

Battery-powered systems are allowed to supply only one unit load (a "unit load" is 100 mA) and for bus-powered hubs, up to five unit loads are allowed after configuration (the device requests the amount of power it needs and the hub allows it only, if the sum is not greater than the five unit loads from upstream). So it is not an USB standard-conforming host, if the current is limited to 100mA.

[nigel53]

The fitting of a switching regulator, either as a complete replacement or pre-regulator, must be carried out with caution. It is not just a question of hf noise on the output, but hf noise introduced back onto the input supply; this could have impact on the EM compliance of the unit. If I had one of these, then I'd be looking at fitting a discrete linear pre-regulator (any npn TO-220 device with suitable specs) onto the main heatsink, leaving the LM317 as is, but handling less volt drop.

[sync]

If I had one of these, then I'd be looking at fitting a discrete linear pre-regulator (any npn TO-220 device with suitable specs) onto the main heatsink, leaving the LM317 as is, but handling less volt drop.

Then you can just move the LM317 to the main heatsink. Much easier. You will have the isolation issues with both solutions. But it's not really a problem.

[nigel53]

If I had one of these, then I'd be looking at fitting a discrete linear pre-regulator (any npn TO-220 device with suitable specs) onto the main heatsink, leaving the LM317 as is, but handling less volt drop.

Then you can just move the LM317 to the main heatsink. Much easier. You will have the isolation issues with both solutions. But it's not really a problem.

Very true, but as stated 'if I had one of these' then I would over-engineer (always been sad like that!), total dissipation split between 2 devices will always be more reliable than 1 device

[rsjsouza]

I was lurking at this thread for a while and I can't help but think that, if this is the first power supply designed by Rigol, it may be suffering from the well known "Rev 1.0" syndrome... Or maybe it is their major first redesign aiming to cut costs, as they have a much higher cost power supply that yields less than half the output power and similar noise/ripple/regulation specs... (I did a quick glance at both specs).

In any case, it becomes a hard sell after all this flak... I wonder if they will leave the model as it is or will slap a "Rev B" or "Mk II" or something else to differentiate the previous faulty products...

[nigel53]

Looking at the price/performance of this unit are we getting a bit paranoid! The switch-on pulse, only occurs with the 'hardware' power switch, is this really a problem? I have a TTI Electronic Load, that has a slow-start facility, this does not function with the 'hardware' power switch. I am now aware of this and ensure the unit is powered on before enabling any 'soft' controls, not a problem.
Yes, the LM317 issue could be a long term problem, only time will tell, but in normal use appears to be at the moment a non-issue. Now we are aware of it, then we could take steps to improve the overall reliability if we really felt the need. This is not the first time, and will not be the last, that a piece of equipment (of any manufacturer) has been produced with an obvious design flaw. To state that the DP832 is a faulty device, is at the moment rather an exaggeration.

I use a Farnell LT-30-2 PSU, big analogue meters, heavy duty mechanical input/output switches & not a digital device in sight!!! Any damage caused by using this is my stupid fault only!!

[edpalmer42]

Hi

A better fix for the voltage regulator would be to change the transformer winding, from the hunt for the problem there looks to be +8v a drop across the regulator, no reason to have so much voltage drop, 4 volts would give lots of margin while dropping the power dissipation in half. With the larger heat sink things would be be just right.

If you mean lowering the currently used transformer tap output, that would mean re-winding the transformer (basically, making a new one). Hardly an optimal solution. If you mean using another tap which is already there used for the main output, well, the point is to have the logic supply (and the interface ports) separate and galvanically isolated from the main outputs. Anyway, most fans are rated 12V and I strongly assume this one is 12V as well.

If a buck winding will fit on the transformer without any problems physically, it should do the trick. Plus an extension on the heat sink of course.

I'm surprised that this idea hasn't had any more discussion.  It's potentially the cleanest and easiest DIY solution to this problem.  If the center area of the transformer is open, you might only need 10 to 15 turns of wire to make a buck winding to drop the voltage to the proper level.  You might even be able to thread the new wire in without disconnecting all the other windings.

Has anyone checked this out?

Ed

[nigel53]

Hi

A better fix for the voltage regulator would be to change the transformer winding, from the hunt for the problem there looks to be +8v a drop across the regulator, no reason to have so much voltage drop, 4 volts would give lots of margin while dropping the power dissipation in half. With the larger heat sink things would be be just right.

If you mean lowering the currently used transformer tap output, that would mean re-winding the transformer (basically, making a new one). Hardly an optimal solution. If you mean using another tap which is already there used for the main output, well, the point is to have the logic supply (and the interface ports) separate and galvanically isolated from the main outputs. Anyway, most fans are rated 12V and I strongly assume this one is 12V as well.

If a buck winding will fit on the transformer without any problems physically, it should do the trick. Plus an extension on the heat sink of course.

I'm surprised that this idea hasn't had any more discussion.  It's potentially the cleanest and easiest DIY solution to this problem.  If the center area of the transformer is open, you might only need 10 to 15 turns of wire to make a buck winding to drop the voltage to the proper level.  You might even be able to thread the new wire in without disconnecting all the other windings.

Has anyone checked this out?

Ed

If you have access around the core, looking at the mounting in Dave's video, I doubt it. You then need the specification for the toroidal fitted (Max temp. rise above ambient, max ambient temp etc.) to even spec. the added winding correctly for temperature, will it pass the insulation tests as per original? I wouldn't class this as the easiest DIY solution, any modifications involving any items connected to the mains supply are fraught with potential hazards

[Dread]

Yes; it is pretty clear to me!  That's my opinion, in your world is there no room for people to have opinions that differ from yours?

I am not attacking you at all - but I call it like I see it, and when I see bullshit, I call it bullshit.

I don't know what you mean "tracking your posts"?  I just remember that you and that Citizen guy seemed to have a beef with the supply (and/or Dave) and have been grinding out that beef in the threads on it.  You overstate every negative aspect of the unit and downplay every positive aspect to fit your conclusion that it's shit and Agilent is better.  You don't even own one of these PSU's and have never used one but you are arguing with folks that own both, and you seemed to get very offended when people criticized Agilent.

One of my other hobbies is racing cars, and the same thing happens on all the car forums.  The Mustang guys spend eons arguing with the Corvette guys that their car is better.  The Corvette guys argue the same to the Viper guys.  The Viper guys argue the same to the Ferrari guys.  And the Ferrari guys argue the same to the McLaren guys.   

There are some who just appreciate all the cars (or power supplies), and when those folks see someone waging a war against a product, it's pretty clear.  As it is in this case.  You don't like this PSU - we get it, so don't buy one.  But feel free to keep telling us all how it's total shit and only an Agilent is worth buying and every feature the Agilent lacks is unnecessary and every feature the Rigol has is unnecessary 

Unfortunately once again you don't get it.  I don't have any particular love of Agilent as a matter of fact I jokingly rubbed Dave when I said he uses that Agilent meter instead of the Fluke when he's taking certain measurements, why don't you go back and read my original post again.

The point you keep missing is that I am just saying to people to first make sure that anything you buy is a piece of gear that does it's core function precisely and accurately and then look at Ergonomics and build.  If that checks out then look at all the bells and whistles because none of the other stuff matters if it's not precise, reliable and comfortable to use.

And BTW I am also into Racing, in this case Rally Car racing. I use to sponsor a rally car and am very familiar with the wars, especially the ones between the Mitsubishi drivers and the Toyota drivers.  That's also a good example of this problem because the Mitsubishi EVO 8 & 10 guys keep pointing out how great their Cars are and how much HP they have.  The EVO Guys always start out the rally with the best run times but by the third run they always keep blowing up the engines and losing the rally to the Toyota's. (And yes the car I sponsored was a Mitsubishi VIII)
BTW If you think this hobby is expensive try paying $7000 USD for a set of shock absorbers.

[Dread]

This is just my opinion.   I admire the fact that so many of you guys want to fix it yourselves but any such action will most likely void your warranty.  I know the pain some of you must be feeling, when I bought my scope for $800 I wanted to return it but I had already paid Fedex $150 in shipping along with $300 in customs duty so it was not an option.  In this case your problem is a factory fault you should just write to Rigol, file a complaint and then just use the PS as normal until Rigol comes up with a revised board.  Then follow up and make sure they ship you a upgraded board.  It's not like the thing is unusable in it's current state!   If Dave had not done a teardown most of you would have merrily kept on using it feeling very happy.

Just my 2 cents.

[Dave]

This design reminds me very much of the Agilent E3631A. The distribution of components throughout the case, the case itself, the way they arranged the heatsinks, how things are connected,...
It's like they took E3631A's design, made it super ugly and messed up the circuit in every way imaginable.

I would definitely like to see Rigol's other power supplies (DP1116 and DP1308) torn down. Shahriar from The Signal Path blog already made an in-depth review of them a while ago, but unfortunately there was no teardown. Any chance you could get your hands on those, Dave?

[edpalmer42]

Hi

A better fix for the voltage regulator would be to change the transformer winding, from the hunt for the problem there looks to be +8v a drop across the regulator, no reason to have so much voltage drop, 4 volts would give lots of margin while dropping the power dissipation in half. With the larger heat sink things would be be just right.

If you mean lowering the currently used transformer tap output, that would mean re-winding the transformer (basically, making a new one). Hardly an optimal solution. If you mean using another tap which is already there used for the main output, well, the point is to have the logic supply (and the interface ports) separate and galvanically isolated from the main outputs. Anyway, most fans are rated 12V and I strongly assume this one is 12V as well.

If a buck winding will fit on the transformer without any problems physically, it should do the trick. Plus an extension on the heat sink of course.

I'm surprised that this idea hasn't had any more discussion.  It's potentially the cleanest and easiest DIY solution to this problem.  If the center area of the transformer is open, you might only need 10 to 15 turns of wire to make a buck winding to drop the voltage to the proper level.  You might even be able to thread the new wire in without disconnecting all the other windings.

Has anyone checked this out?

Ed

If you have access around the core, looking at the mounting in Dave's video, I doubt it. You then need the specification for the toroidal fitted (Max temp. rise above ambient, max ambient temp etc.) to even spec. the added winding correctly for temperature, will it pass the insulation tests as per original? I wouldn't class this as the easiest DIY solution, any modifications involving any items connected to the mains supply are fraught with potential hazards

That's a surprisingly cautious attitude on this forum!    You can't tell from the pictures if there's room or not.  Someone would have to undue the mounting nut and look.  Temperature rise is a non-issue.  We're only talking about a few watts on a transformer that's rated for 400 or 500 watts.

Insulation testing is also a non-issue.  The transformer has all the necessary insulation.  The extra winding will be outside everything else and so, protected by the existing transformer insulation.  Even though it will be connected in series with one of the existing windings, that winding already has the necessary ratings.

Ed

[Corporate666]

Unfortunately once again you don't get it.  I don't have any particular love of Agilent as a matter of fact I jokingly rubbed Dave when I said he uses that Agilent meter instead of the Fluke when he's taking certain measurements, why don't you go back and read my original post again.

The point you keep missing is that I am just saying to people to first make sure that anything you buy is a piece of gear that does it's core function precisely and accurately and then look at Ergonomics and build.  If that checks out then look at all the bells and whistles because none of the other stuff matters if it's not precise, reliable and comfortable to use.

And BTW I am also into Racing, in this case Rally Car racing. I use to sponsor a rally car and am very familiar with the wars, especially the ones between the Mitsubishi drivers and the Toyota drivers.  That's also a good example of this problem because the Mitsubishi EVO 8 & 10 guys keep pointing out how great their Cars are and how much HP they have.  The EVO Guys always start out the rally with the best run times but by the third run they always keep blowing up the engines and losing the rally to the Toyota's. (And yes the car I sponsored was a Mitsubishi VIII)
BTW If you think this hobby is expensive try paying $7000 USD for a set of shock absorbers.

I get what you are saying just fine.  I am simply saying it's bullshit.  You were complaining about this power supply before the thermal issue was discovered, and your basis for that complaining was the feel of the rotary knob, the layout of the keypad, having a big color screen, etc, etc.  In other words, the "ergonomics and build" which you just claimed are unimportant next to basic functionality.

So like I said, you are drastically overstating the issues of this supply as being dealbreakers, and ignoring it's positive aspects as being irrelevant and unnecessary in a power supply.

And you don't even own one.  Apparently a lot of people disagree with your take because many of us have bought one and enjoy many of the features that you say are crap and we don't need. 

The great thing about free markets is that you are not required to buy one of these PSU's.  But don't think you should be immune from criticism when you want to lord your opinions over us all and tell us that a power supply you don't own and have never used is crap and we bought them because we were fooled by useless features that we don't need, and we should be more like you and buy good equipment that is lacking in features and overpriced but has things you consider more important.

I think after your many posts about this power supply that you don't own, your stance on the matter has been made perfectly clear.  Maybe we can now get back to discussing the relevant issue in this thread and look for a fix... or are you not done yet telling us all how shit this supply is and how it's essentially a paperweight in comparison to the real stuff you buy?   





Back OT.  I had gotten a call from Chris @ Rigol a couple of weeks back after we had bought a few pieces of Rigol equipment to ask what our plans were going forward for test equipment and such.  He gave me his contact info and said to keep in touch.  So I did!  I contacted him and asked him to chime in on this issue.  I hope we get a reply.  I noticed TEquipments return policy is 7 days, so this is something that will need to be fixed by Rigol directly, and I intend to hold them to fixing it - especially considering it's an issue they knew about (as evidenced by them having already made a version 3 board while they were *still selling f-ing defective units knowingly*).

[AlfBaz]

So, playing the devils advocate for a minute, I'm curious about the train of thought that the unit is defective, I mean, is it really?

I'm working on fixing an old crt DSO right now and I have about 10 minutes before the crt overheats as I have it I pieces on the bench to fault find, so the air flow from the fan is not getting to where it's supposed to.

People seem to be advocating (including the actual rigol fix) to install a larger heatsink but isn't the unit shunting enough heat away from the regulator with the current air flow?

[Stonent]

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

Absolutely! I'd consider placing electrolytic caps directly besides a heatsink (>30-40 °C) in a lab grade device a show stopper. It's built to fail.

I guess if you come into your lab one day and your power supply is off and it smells like cat pee, you'll know. 

[Stonent]

Meanwhile, i am wondering about something toally different....

Why put in an lm317 where a 7805 would have done ....

I thought the same thing, 7805 would be cheaper, no?