24V transient spike

[requim]

Any ideas what could be causing a 24V transient spike when powering on/off an Agilent e3610a power supply? Bad transformer? caps? resistors?

[amspire]

Any ideas what could be causing a 24V transient spike when powering on/off an Agilent e3610a power supply? Bad transformer? caps? resistors?

Just had a look at the service manual http://128.238.9.201/~kurt/manuals/manuals/HP%20Agilent/HP%20E3610A,%2011A,%2012A%20Operating%20&%20Service.pdf

It has no output switch so you have to rely on the power switch - I see your problem.

I will take a stab at this. May be wrong.

The +/- 12V regulator has a 330uF cap before the +12V regulator (C7) and a 470uF cap (C13) before the -12V regulator.  I believe the idea is that the +12V rail has to collapse faster then the -12V rail on power down, but I think in your case the +12V is staying up longer then the -12V.

So I would check C13 and replace if necessary. It may now be a lot less then 470uF.

Alternatively, if R1 has gone open circuit, then the volts to the regulator transistors will not be discharged as quick as it should, and that could cause a similar problem.


Richard.

[requim]

Thanks for your input.  I just received the power supply and cleaned it up.  I've opened up the power supply and it looks pretty basic on the inside which gives me hope that I have a fighting chance of fixing this thing.  I'm waiting for my order of resistors and capacitors to arrive before I dive into it. 

I'm a total beginner in over my head so please forgive me for asking what I am sure are basic questions.  In order to test, do I need to use an oscilloscope or can I use my multimeter?

[amspire]

A multimeter is all you really need.

Just check R1 is the correct value (something like 1K - I forget exactly what it should be).

If your multimeter has a capacitance range, you can use that to capacitors directly.

If not, if you put the multimeter on a fixed high resistance range, you will see that when you test a capacitor, the voltage will rise at a fixed rate until you get the "OL" on the screen.  The rate it rises is proportional to the capacitance.

The voltage needs to rise on C13 should be about 40% slower then for C7.  If C13 has low capacitance, it may rise quicker then C7 and that is not good.

Richard

[requim]

Another simple and silly question -- Am I testing this while the unit is powered?  And for the record i'm using a Fluke 87 V so I believe it can do the cap testing referred to in the earlier post.

[amspire]

No. No power. and check that there is no voltage across any of the capacitors before testing. If you are lucky, you can measure in-circuit. Otherwise, you may need to lift one leg of the capacitors before testing.

Richard

[requim]

Okay so C13 the 330uf 35V cap measures at 307uf and C7 the 470uf 35V cap measures at 471uf so both seem to be within tolerance.  Any other ideas?  Also, I noticed if I reversed the probes I got higher readings.  Why is that?

[amspire]

Well it is pretty important that R1 is 2K and not an open circuit.

If it has gone open circuit, then the main capacitor will still have a voltage after the regulator IC's stop working during shutdown.

I would assume that the regulator IC's need to keep working until after the main capacitor is discharged.

How long after you turn the supply off do you get the 24V spike? Also can you see if C2 is 1000uF o 10,000F?

Richard

[requim]

Okay.. got a new battery since it died last night before I realized I also needed to measure R1.

So after 5+ minutes and switching to manual ranging I got a reading of 2.1k ohms on R1.  I did the reading in circuit so I'm not sure if that's what you wanted or if you wanted me to desolder it and test it.

the spike turns out to be about 26V lasts for several seconds and then starts to trickle down.

As for C2 it is a Nichicon 10,000 uF 50WV (what is WV?) 105 degree cap.  I tried to measure it but I assume it is out of the range of my DMM.  It jumps up to around 9300 uF and then shows 0L on the screen.

[amspire]

As for C2 it is a Nichicon 10,000 uF 50WV (what is WV?) 105 degree cap.  I tried to measure it but I assume it is out of the range of my DMM.  It jumps up to around 9300 uF and then shows 0L on the screen.

I thought the 1000uF in the service manual had to be a misprint.  10,000uF means the main cap can take minutes to get close to a full discharge, so to eliminate output glitches, the op-amps have to keep working till the +12v is discharged.

Now I think I made a mistake in assuming that of course HP had carefully designed this power supply so it shut down perfectly - especially as they had no output switch. I could almost bet my house on it.

Well, luckily I didn't, or I would be on the streets by now.

Here are the service notes:

http://litstation.marketing.agilent.com/litapp/SearchSN.do?method=openExternalSNSearch&prodNum=e3610a

There initially was an overshoot condition on power down when both the +/-12 supply capacitors were 330uF. So they increased the -12V one to 470uF.

The scary bit in the service note is this:

If the readings increase after the power supply is switched “off” by more than 5%
then the power supply has “excess turn-off overshoot”.

So 5% overshoot on turn off is fine! 

Is this really HP? I just find it hard to believe!

So if your supply is fully working (voltage works over the full range, current limit works, output clean and all the other 4 service notes have been looked at), it is possible that HP's bodgy fix of C13 is not goon enough. It is not just that they have the overshoot, but when you turn a supply off, you want it to immediately turn off. Not sometime in the distant future -- which is the way it works now.

My suggestion is to replace C13 with 1000uF 50V and see if that cures the overshoot problem.

Also if it was me, I would drill a hole in the front panel and add a toggle switch so I could switch the load on and off.  Then you will have a MUCH better power supply. 

Until you do that, the safe way to use this power supply is to unplug the load before switching off.

Other then the switch off fiasco, it is a great supply.

Richard

[requim]

So I think I figured out what is wrong.  So the parts list indicates C7 should be 330uF 35V 20% AL-ELECT and C13 should be 470uF 50V 20% AL-ELEC, however looking at the board it definitely looks like someone replaced both caps and what they did was put on C7 a 470uF 35V cap and on C13 they put a 330uF 35V cap. 

I'm assuming that was a mistake on the part of the person who performed the repairs/replacement.  So if I'm not mistaken I should order the correct parts and replace them, correct?

Also, would it be a good idea to perform the service replacements on some of the other service notes?  Specifically #3, 4, and 5? The unit I have appears to have been made in 1992, which from what I understand puts it in the first year of production.

One other thing, while I don't think service note 1 needs to be performed (replace the three output terminals) I am thinking of buying a green output terminal (like they use now for ground on the same model) and pulling the red - terminal and putting the black ground terminal in its place and then putting the new green terminal in the ground position.  Is that going to cause me any problems?  (not the colors, but the replacement/swapping of terminals)

Oh, and there is one other problem with the unit I've noticed is that the current knob seems to be off by 1.22 amps or so.  What i mean is it's range seems to be from 1.22 - 4.22 instead of from 0 - 3.0.  Any suggestions?

Thanks again

Jon

[amspire]

So I think I figured out what is wrong.  So the parts list indicates C7 should be 330uF 35V 20% AL-ELECT and C13 should be 470uF 50V 20% AL-ELEC, however looking at the board it definitely looks like someone replaced both caps and what they did was put on C7 a 470uF 35V cap and on C13 they put a 330uF 35V cap. 

[\quote]

Hey, that was my first thing I sid to check. Absolutely if the caps have been swapped, it will cause this problem.

I still think it is bad that the supply does not turn off immediately, so I would still add a output switch myself.

I'm assuming that was a mistake on the part of the person who performed the repairs/replacement.  So if I'm not mistaken I should order the correct parts and replace them, correct?

Also, would it be a good idea to perform the service replacements on some of the other service notes?  Specifically #3, 4, and 5? The unit I have appears to have been made in 1992, which from what I understand puts it in the first year of production.


[\quote]

You will find your supply already has most of the service note fixes, but if not, then yes do them.

One other thing, while I don't think service note 1 needs to be performed (replace the three output terminals) I am thinking of buying a green output terminal (like they use now for ground on the same model) and pulling the red - terminal and putting the black ground terminal in its place and then putting the new green terminal in the ground position.  Is that going to cause me any problems?  (not the colors, but the replacement/swapping of terminals)


[\quote]

Go for it.

Oh, and there is one other problem with the unit I've noticed is that the current knob seems to be off by 1.22 amps or so.  What i mean is it's range seems to be from 1.22 - 4.22 instead of from 0 - 3.0.  Any suggestions?

Thanks again

Jon

R27 is meant to be 1 ohm. If it is some larger value , replace with 1 ohm.

Then just do the calibration on page A5. You don't need the 0.1 ohm resistor - you just need to measure the output current on your multimeter.

When you adjust the peak current down from 4.33A, the minimum current will probably drop too.

You must adjust R31 before R32.

Richard

[requim]

Oddly enough I gave the correct values when first tested them, however when I said they were within tolerance I was referring to the labels on the caps and had apparently mixed them up when comparing them to your note.  C'est la vie.

Anyway, I measured R27 and it started at 2.5 ohms and oscillates between 2.1 - 2.2 ohms, which I assume means it needs to be replaced, correct?


I've included a photo of the board because I have a couple more questions:  One, why are the traces exposed where they are on the board?  And two, it looks as though there is some corrosion around the solder joints, how should I remove it?

Thanks again

[requim]

So I looked up the resistor specified on digikey:  1 ohm 1% .125W TF, but I can't find a match, well correction, I found one match for 5000 minimum of a non-stock item.  Can I substitute something else or should I look at a different vendor?

Also, does TF = Thick Film?

[amspire]

R27 is probably oscillating because the leads are slightly oxidised rather then the resistor faulty. 

It would need to be further out to cause the error you are seeing. Try soldering a bit of clean wire on the ends and measuring again. If it is genuinely faulty, it should be replaced, but it is probably OK.

I think what has happened is that you will find R31 has been turned to one end. Perhaps someone thought it was great getting an extra amp out of the supply.

Richard

[requim]

I'll check into it.  Thanks.  Any ideas on my other questions?

[amspire]

Not sure what has occured. It could easily be that the the original capacitor in the 470uF position exploded, and the residue was never cleaned up properly.

The chemical in an electrolytic capacitor is acidic, so I would make a weak solution of sodium bicarbonate (bicarb) and clean the board with something like cottenbuds.

Afterwards, clean well with pure water, and let make sure the board is really dry.  It doesn't matter too much if some tracks are exposed - it looks like they might be gold plated.

With the messy joints, use solder wick or a sucker to remove the old solder, and resolder.

About the resistor, it is not critical in any way, but metal film is the common type of resistor to use. As I said, I don't think you will need it, unless the resistor has signs of acid damage.

Richard.

[requim]

When you say that I don't need the 0.1 ohm 0.1% 10 watt resistor but can just use my DMM can you please explain how to do it?  Because I honestly have no idea how to do it without the resistor. I've read through the calibration and since I'm a total novice I can't understand how not having the resistor will allow me to correctly calibrate the unit.

[amspire]

Turn the supply voltage and current controls down to zero. Plug the multimeter leads between the 20A input and the common input. Put the meter on to the DC Amps range or 20A range, or whatever the meter uses for current.

Now put the meter leads between the + and - power supply outputs.

Turn the power supply volts and current up until the current on the meter reaches a maximum. The supply will now be in current limit mode. It should hopefully be similar the the reading on the supplies current meter, but trust the Fluke 87V multimeter readings over the power supply meter's readings.

Inside the multimeter, there an accurate resistor of perhaps 0.01 ohms and a voltmeter across it. So the reason you don't need an external 0.1 ohm resistor is that your multimeter already has a perfectly good one inside it.  The value of 0.1 ohm never mattered - it could be 1 ohm, or 0.001 ohm. HP obviously uses a 0.1 ohm resistor in its factory. As long as it is a known resistance, and it can handle the full power supply current.

So just use the meter to measure the current, and get that supplies current ranges properly calibrated.

Richard