What is actually fixed in Rigol dp832(a) Rev3?

[Sparky]

It's beginning to look like the larger heatsink might not have made it into the board updates?!? 

Well, in the post you link to, John South mentions the larger heatsink versions will be leaving China around Oct. 17th - so still in the future. Perhaps it will be another 3-4 months or so before that size is found as standard.

Yep, so it's definitely the updated version, I was just expecting to see even larger heat sinks (like shown in the links above)...so, bit of a surprise...

Good point marmad!  Not so surprising after all

[leafi]

I just got my DP832. I will have to take a look. It is still in the box and just arrived today.

[iceisfun]

Where is the best place to order from to make sure you get the newest revision hardware?

I have been wanting one of these PSU units

Right now I have a ebay Mastech PSU and it works great, but its time to upgrade it. For $200 the Mastech unit was a work horse and never really failed me.

[grego]

Where is the best place to order from to make sure you get the newest revision hardware?

I have been wanting one of these PSU units

Right now I have a ebay Mastech PSU and it works great, but its time to upgrade it. For $200 the Mastech unit was a work horse and never really failed me.

If you are in the US tequipment.net has the Rev 2.10 boards as I pictured above.

[centon1]

TEquipment
DP832 F/W 00.01.06
S/N DP8C1532xxxxx
Board revision V02.10 (2013-03-26)

Programmable Triple Output DC Power Supply
Less than $ 500.00 CAD. Half the price of a decent smartphone.
Something you can fix yourself, should you have to.
Works for me. I'm in!

[grego]

Well I can confirm they definitely didn't fix the output spike on powerup with the V2.10 motherboard.

I've attached scope shots from all three channels (sequential) measuring on powerup.

[rbola35618]

Grego,

Any chance of measuring the current of the spike by placing a 1 ohms resistor. If there is not current, then it might not be an issue.

Robert

[grego]

Grego,

Any chance of measuring the current of the spike by placing a 1 ohms resistor. If there is not current, then it might not be an issue.

Robert

Sure I can set that up.

Regardless, it's a problem.  There should be nothing on the outputs when the supply is turned on - especially since all the outputs default to off (as they should).

[Electro Fan]

Any chance someone with any power supply can show us some screen shot's similar to Grego's (in terms of amplitude and time) that shows their power supply exhibiting a more flat response during turn-on (with the output turned off)?  Maybe someone with an Agilent or other highly regarded power supply?  Thanks

[Svuppe]

Any chance someone with any power supply can show us some screen shot's similar to Grego's (in terms of amplitude and time) that shows their power supply exhibiting a more flat response during turn-on (with the output turned off)?

Well, I guess that is my cue to register on this forum, after being lurking around here for a while.
This is a screen capture of my Agilent U8002A turning on:

Mikael

[grego]

Any chance someone with any power supply can show us some screen shot's similar to Grego's (in terms of amplitude and time) that shows their power supply exhibiting a more flat response during turn-on (with the output turned off)?

Well, I guess that is my cue to register on this forum, after being lurking around here for a while.
This is a screen capture of my Agilent U8002A turning on:

Mikael

Wow - transient spikes there too.

[Svuppe]

Here is the same measurement of my U8002A, but with a 680 ohm resistor across the output (couldn't find any smaller).

[tnt]

[quote vauthor=grego link=topic=20841.msg312679#msg312679 date=1382134812]
Regardless, it's a problem.  There should be nothing on the outputs when the supply is turned on - especially since all the outputs default to off (as they should).
[/quote]

Well, without load and just connected to the 1M input of a scope, anything could cause this, just induction in the wires, or even capacitive coupling between two disconnected relay contacts could realistically transfer such spikes.

But all those have no currents behind them, as soon as you connect them to something real, a real load, they'll be crushed ...

[alm]

Depends on what that something real is. What if it's connected to an MCU that represents a load of a few mA, or even a high-impedance input of an MCU of FPGA? Bench supplies are not just used for driving motors, sometimes they might just be used as reference voltage for an ADC.

That said, this spike might very well be just a common-mode spike (unless someone verified this).

[Rufus]

Wow - transient spikes there too.

What do you expect when you switch mains with a contact that is going to arc and probably bounce onto almost any piece of equipment?

The transient will likely be mostly common mode coming through stray capacitance between the mains side and isolated power side. Your scope will see some of it as differential because of the huge impedance difference between the probe ground and tip and the scope is connected to the same earth as the power supply though its mains lead.

I doubt you will find a power supply with a real mains switch that doesn't appear to produce similar transients.

I thought the problem that Dave noticed was some real voltage lasting for 10s of ms on the output indicating the power supply regulator circuit or control circuit has a power up transient problem and some real voltage could do some real damage to a connected circuit.

When a circuit has multiple power supplies that may appear at different times or rise at different rates it is difficult to predict what will happen during power up and power down transients.

[arvidj]

I want to order an DP832 from TEquipment and would like to know what to expect when I open the box.

I've read this thread and am trying to summarize what it says. Feel free to provide corrections.

Original issues:

1) the overheating heatsink / voltage regulator

This appears to have been an issue with the Rigol version 2.0 boards. Rigol responded to issue with a revision 2.1 board which appears to be what is used in the units that are currently being delivered.

John South suggests Rigol will have "yet another revision" ... Re: EEVblog #512 - Rigol DP832 Bad Design Investigation Reply #433 that would have begun shipping on October 17^th but no one has reported having received a unit with this board installed.

There have been several "mitigate the risk yourself" solutions suggested including replacing the LM317 with a switching style three terminal equivalent, putting a bigger heatsink on the existing LM317 and removal of the LM317 from the board and remotely mounting it on a much larger heatsink.

It has been suggested that the "larger\remote heat sink" solution ... be it by the DP832 owner or Rigol ... may not be a good long term solution as the localized temperature of the LM317 may get quite high even with a large heat sink. The "correct" solution would be to reduce the amount of power the LM317 is being asked to dissipate by lowering the input voltage, a solution that is best accomplished by Rigol. 

2) the messed up ammeter readout when current is under 12mA or so

I have not been able to find any reports of this issue having been resolved.

3) the output voltage spike when turning on the unit

I have not been able to find any reports of this issue having been resolved but if I understand the discussion in theses posts ... Reply #38,  Reply #39 and  Reply #40 there seems to be some thoughts that the current behavior [no pun intended] is satisfactory.

The "mitigate is yourself" solutions would seem to be "do not have anything plugged into the supply" when you turn it on or an outboard switch between the channel terminals and you device ... both of which seem to be fraught with human error possibilities ... especially if you are old like me

4) the current sense wires for channel 2 and 3

This was discovered in Re: Another issue with the Rigol DP832 power supply: sense wires carrying 3 amps! but I have not seen any information related to Rigol addressing the issue.

The "mitigate the risk yourself" suggestions are to take the front panel off and putting a heavy piece of wire between the channel 2 and 3 negative binding post connections ... though no one has reported actually doing that ... or doing the 'heavy but ugly wire' between the negative channel 2 and 3 posts on the outside of the front panel.

Corrections to this summary are encouraged.

[Electro Fan]

Any chance someone with any power supply can show us some screen shot's similar to Grego's (in terms of amplitude and time) that shows their power supply exhibiting a more flat response during turn-on (with the output turned off)?

Well, I guess that is my cue to register on this forum, after being lurking around here for a while.
This is a screen capture of my Agilent U8002A turning on:

Mikael

Mikael, Thanks for posting the image(s).  Looks like your power supply might be the leader in terms of having only two spikes (under no load) both with relatively low amplitudes and short durations. 

Maybe someone can post images from a power supply with an even cleaner turn-on?

Or maybe there is no such thing as a perfectly clean turn-on (with no load) with a typical hobbiest to medium price power supply (or even a high end power supply)?

Maybe a clean turn-on can only happen with at least some token load attached?

Looks like the jury is still out on this stuff?

EF 

[Svuppe]

I just tried again, but this time with probe and ground both connected to the negative terminal of the psu. And I got pretty much the same picture. I believe Rufus is right in the "spikes" being common mode bouncing.
I should have tried a differential measurement with two probes, but that will have to wait until tomorrow. I am on the way to bed now.

[Electro Fan]

From the Agilent U8001A / U8002A User’s and Service Guide - notice the comments about "Wait for a few seconds for the output of the power supply to settle."  Not sure if this is relevant to the overall discussion of turn-on spikes.  (Also note the instruction to add 0.01 Ohm current shunt monitoring resistor across the output to be tested.)

Constant Current (CC) Verification
Current Programming and Readback Accuracy
This test is to verify that the current programming and
readback accuracy are within published specifications. The
accuracy of the current monitoring resistor must be 0.1% or
better.
Procedures:
1 Power- off the power supply and connect a 0.01 ? current
shunt monitoring resistor (RM) across the output to be
tested and a DMM across the current monitoring resistor
(RM).
2 Power on the power supply.
3 When the power supply is in limit mode, program the
output voltage to maximum programmable value and the
current to 0 A.
4 Enable the output.
5 Wait for a few seconds for the output of the power supply
to settle. Make sure that the power supply is in CC mode.
6 Divide the voltage drop (DMM reading) across the current
monitoring resistor (RM) by its resistance to convert to
amps and record this value (IO). This value should be
within the limit of 0 A ±20 mA.
7 When the power supply is in meter mode, record the
current reading displayed on the front display of the
power supply. This value should be within the limit of
IO ±20 mA.
8 Disable the output.
9 When the power supply is in limit mode, program the
output current to full rated value, i.e. 3 A for U8001A and
5 A for U8002A.
10 Enable the output.
11 Wait for a few seconds for the output of the power supply
to settle. Make sure that the power supply is in CC mode.
Service Guide 5
U8001A/U8002A User’s and Service Guide 77
12 Divide the voltage drop (DMM reading) across the current
monitoring resistor (RM) by its resistance to convert to
amps and record this value (IO). This value should be
within the limit of:
• U8001A: 3 A ±30.5 mA
• U8002A: 5 A ±37.5 mA
13 When the power supply is in meter mode, record the
current reading displayed on the front display of the
power supply. This value should be within the limit of:
• U8001A: IO ±30.5 mA
• U8002A: IO ±37.5 mA

[Svuppe]

I should have tried a differential measurement with two probes, but that will have to wait until tomorrow.

Now I've done the two probe test. One probe to the positive output and the other to the negative. Both referenced to ground.
The math (bottom) line is the difference between the two channels. Same scale.

[LaurenceW]

Well, I think that proves the point, exactly. All of the screen shots I am seeing here are not "real" voltages appearing across the output terminals of the power supply, but simply electromagnetically coupled noise from the mains switch. I can recreate that, with my 'scope probe simply lying unconnected on my bench, while switching on other kit!

A few screenshots from my own power supply - an old school analogue TTI PL330P. All single-ended measurements.

• First shot shows the output under a 100R load, when set to 5V and the mains input is switched on. Good, clean rise in the output over 10mS, which is only one half of a mains cylce! The little step at 1V moves about, depending on when in the mains cycle the AC switch is thrown. This is not significant.

• Second shot shows the same conditions, but set for 24V. A similar result, as expected. No dramas.

• Final shot shows same 24V output setting, but this time enabled by the DC output switch (AC mains inout on, throughout this time). On my supply, the DC output switch is REAL, not electronic. There are a variety of turn-on spikes, but I would see the same result from a 24V battery, to be honest! Different results, each time. Notice the time duration, here. It is all over in a few uS. This is a mixture of switch bounce, and AC currents flowing into the inductive and capacity load of assorted bench leads. Can look exciting, but it is really no issue. If you have ciruit loads that might be disturbed by this sort of power supply, you've got a bigger design problem with your circuit! The power supply itself is not a problem, in this case.

Under no-load conditions, the output voltage is reached, somewhat quicker.

I do go back to an earlier observation about the Rigol power supply design. Dissipating - was it around 7 Watts? - great chunks of power in an internal 5V regulator is just bad Karma. Sticking a whopping great heat sink on it in the hope that nobody will notice, is not good Karma.

Finally, I think it was Bill Gates who once said, "Never buy a piece of software until it reaches at least V3.0" And if any of you remember Windows 2, you'll know he was right.

[grego]

Well, I think that proves the point, exactly. All of the screen shots I am seeing here are not "real" voltages appearing across the output terminals of the power supply, but simply electromagnetically coupled noise from the mains switch. I can recreate that, with my 'scope probe simply lying unconnected on my bench, while switching on other kit!

A few screenshots from my own power supply - an old school analogue TTI PL330P. All single-ended measurements.

• First shot shows the output under a 100R load, when set to 5V and the mains input is switched on. Good, clean rise in the output over 10mS, which is only one half of a mains cylce! The little step at 1V moves about, depending on when in the mains cycle the AC switch is thrown. This is not significant.

• Second shot shows the same conditions, but set for 24V. A similar result, as expected. No dramas.

• Final shot shows same 24V output setting, but this time enabled by the DC output switch (AC mains inout on, throughout this time). On my supply, the DC output switch is REAL, not electronic. There are a variety of turn-on spikes, but I would see the same result from a 24V battery, to be honest! Different results, each time. Notice the time duration, here. It is all over in a few uS. This is a mixture of switch bounce, and AC currents flowing into the inductive and capacity load of assorted bench leads. Can look exciting, but it is really no issue. If you have ciruit loads that might be disturbed by this sort of power supply, you've got a bigger design problem with your circuit! The power supply itself is not a problem, in this case.

Under no-load conditions, the output voltage is reached, somewhat quicker.

I do go back to an earlier observation about the Rigol power supply design. Dissipating - was it around 7 Watts? - great chunks of power in an internal 5V regulator is just bad Karma. Sticking a whopping great heat sink on it in the hope that nobody will notice, is not good Karma.

Finally, I think it was Bill Gates who once said, "Never buy a piece of software until it reaches at least V3.0" And if any of you remember Windows 2, you'll know he was right.

I agree entirely.  I think it's significantly less of an issue for us in the US since we're one 110 vs 240 but it's still a bad design if it's supposed to support all the voltages worldwide as an input.

[alm]

Now I've done the two probe test. One probe to the positive output and the other to the negative. Both referenced to ground.
The math (bottom) line is the difference between the two channels. Same scale.

Note that the residual signal may very well be due to the fairly limited CMRR of this setup, especially at higher frequencies.

[arvidj]

As a noob I would like clarification of the previous "turn on voltage" discussion as it related to the Rigol DP832.

If I am understanding several of the posts related to the Agilent U8002A and TTI PL330P there is no real "turn on voltage" issue. What is seen on the scope is simply electromagnetically coupled noise and (a) not caused by the power supply and (b) is not an issue.

Can this conclusion be extended to the behavior exhibited by the Rigol DP832 or is additional testing required?

As a totally separate issue, I have revised my "summary" to indicate that simply putting a larger heat sink on the LM317 is a questionable solution as the localized temperature of the LM317 may be extremely high even with a very large heat sink. The correct solution is to reduce the total amount of power the LM317 is being asked to dissipate.

As usual, corrections to my naive and inexperienced conclusions are appreciated.

Arvid