HP Agilent E3610A

[pgross]

Dear forum,

I've just used my HP E3610A supply on a analog amp project.

While measuring around I discovered, that the E3610A begins to self oscillate @ approximately 130 kHz while delivering  15 Volt.
The load was purely capacitive - a 2.2uF polyester cap connected to the power supply output with very short banana test leads.
My oscilliscope was hooked up in parallel with the cap, and a nice 30mV p-p 130 kHz sine showed up on the screen.

I know that HP/Agilent have released some service notes on this model. One of the service notes should eliminate oscillation when
the output voltage/load conditions are changed. The CR8 diode must be remove, and I can see that this diode i missing in my supply
also the C4 and C5 are missing - mentioned in another service note.

Does anybody know if the E3610A have stability issues with capacitive loads in general?

Best regards

Peter




 

[sync]

I just tested my E3610A from 0-17.5V with 1uF, 2.2uF and 4.2uF film caps. No oscillations.

Check that it's in the 2A/15V range. In the 3A/8V range it can output 15V with no load. But It's unstable.

Then check this. When it's oscillating add an electrolytic cap directly at the binding post. Try 100uF - 470uF. If the oscillation stops check the output cap (C3 470uF). Maybe it's bad.

[pgross]

Thanks for your info.

I will try the mentioned setup, and post my findings here asap.

Best regards

Peter

[pgross]

Follow up…

I have now checked my other HP branded E3610A supply, and it shows almost
the same oscillation phenomenon. The oscillation is measured with the range
switch in 2A and 3A position with no change in amplitude at all. Adding a 100uF
(low ESR) cap in parallel with the 2.2uF film capacitor defeats the oscillation.

I’ve checked the 470uF/50V C3 output cap, and it’s fine. Even if another 470uF
cap is added directly on C3, the oscillation amplitude is only slightly reduced.
If the same 470uF cap is placed right across the output protection diode CR3,
the oscillation is almost zero. So the pcb traces from the output pads to C3
certainly come into play here!

It should be mentioned, that the 2.2uF film cap is hooked up directly on the
binding posts. No oscillation is measured if the cap is moved to the end of two
banana test leads with the length of 1 meter each. I can live with that, but I must
say that I’m rather surprised that there is some kind of stability issue with this
power supply – or have I missed something? 

Here is a screenshot of the oscillation, the frequency is approximately 170 kHz.

[pgross]

Another follow up…

I decided to dig into this oscillation behavior a little further, and determined to see if it was possible to localize where the oscillation was generated. Probing around with my oscilloscope
It was pretty obvious, that the strongest osc. signal could be measured at the Voltage error amp. circuit.

Long story short, I tried to eliminate the oscillation with a capacitor in parallel with the resistor R6. And Tadaa! I managed to cancel the osc. utilizing a 220pF NP0 0805 cap mounted on the solder side of the board. See attached photo.

I have performed some tests on the power supply afterwards, but I’m not able to see any degradation in its performance after this modification. Please chime in if you have any comments or can tell me if there can be any obvious drawbacks with this hack.

Best Regards

Peter

[Huluvu]

Follow up…

I have now checked my other HP branded E3610A supply, and it shows almost
the same oscillation phenomenon. The oscillation is measured with the range
switch in 2A and 3A position with no change in amplitude at all. Adding a 100uF
(low ESR) cap in parallel with the 2.2uF film capacitor defeats the oscillation.

I’ve checked the 470uF/50V C3 output cap, and it’s fine. Even if another 470uF
cap is added directly on C3, the oscillation amplitude is only slightly reduced.
If the same 470uF cap is placed right across the output protection diode CR3,
the oscillation is almost zero. So the pcb traces from the output pads to C3
certainly come into play here!

It should be mentioned, that the 2.2uF film cap is hooked up directly on the
binding posts. No oscillation is measured if the cap is moved to the end of two
banana test leads with the length of 1 meter each. I can live with that, but I must
say that I’m rather surprised that there is some kind of stability issue with this
power supply – or have I missed something? 

Here is a screenshot of the oscillation, the frequency is approximately 170 kHz.

Hello Peter,

I can confirm this behavior also at my HP E3610A   
It looks like the low ESR of the Film Cap correspond with the regulation part of the PSU and starts oscillating (only my assumption)

Rgds Arne


 

[pgross]

Hi Arne,

Yes, that’s also my conclusion.

But as I mentioned in my previous post, I’m somewhat surprised to find that this high quality and otherwise rock solid power supply does exhibit stability issues.

I know it isn’t common to hook up a 2.2uF film cap directly on the binding posts, but anyway – at least in my book a high quality DC power supply must perform flawlessly with different kinds of loads, capacitive as well as inductive. Maybe I’m expecting too much ?

I’ve now performed exactly the same modification on my other E3610A supply with the same result – zero oscillation with capacitive loads. So far I haven’t discovered any new introduced faults after
these hacks.

Best Regards

Peter

[Huluvu]

Hm i'm am also a little bit disappointed...
The HP 3610A is the most used supply in my Lab and I like it very much.

What's the value of R6 ?

Rgds Arne

[ElectroIrradiator]

Interesting. At least some of the high end HP/Agilent PSUs with VFD has a quite large electrolytic cap connected right across the binding posts, on the back side of the front panel. This cap is in parallel with a 0.1uF film cap. Now it seems we know why.

[pgross]

The value of R6 is 1M5.

The E3610A does indeed have a rather large output capacitor - 470uF/50V - though it isn't bypassed with a film cap. Originally there were two series mounted 0.1uF ceramics – C4 and C5, at the output with ref to ground. Later, a service note (E3610A-04) suggested that these caps should be removed – because the output showed oscillation if either the + or – terminal was connected to ground!

Best Regards

Peter

[JimmyMz]

The E3610A does indeed have a rather large output capacitor - 470uF/50V, although it's not bypassed with a film cap. Originally, there were two series mounted 0.1uF ceramics (C4 and C5) at the output with ref to ground. Later, a service note (E3610A-04) suggested that these caps should be removed, because the output showed oscillation if either the + or – terminal was connected to ground!

This valuable information is news to me, due to me also owning this power supply. Do you have a link to the service notes you are quoting from Peter? Thanks 

[pgross]

Actually, the manual and service notes are obtainable through the Agilent hompage:

http://litstation.marketing.agilent.com/litapp/SearchSN.do?method=openExternalSNSearch&prodNum=E3610A

Best Regards

Peter

[JimmyMz]

http://litstation.marketing.agilent.com/litapp/SearchSN.do?method=openExternalSNSearch&prodNum=E3610A

I looked at a different website, which ironically didn't have service note 7 listed, but had all other service notes. This is precisely why I asked for a reference, since I had no idea of any HP/Agilent service note advising to remove CR8 (to eliminate output oscillations) http://cp.literature.agilent.com/litweb/pdf/E3610A-07.pdf
Peter, thank you for taking the time to link me to the notes. 

[Huluvu]

The value of R6 is 1M5.

ooops that's why the oscillation is possible 
I wonder what was the intend of the the HP Designers on that circuit .....hm

The E3610A does indeed have a rather large output capacitor - 470uF/50V - though it isn't bypassed with a film cap. Originally there were two series mounted 0.1uF ceramics – C4 and C5, at the output with ref to ground. Later, a service note (E3610A-04) suggested that these caps should be removed – because the output showed oscillation if either the + or – terminal was connected to ground!

As far as I remember I did all the Service Notes I found but we should really look again into the schematics if there is some space for improvement.
The (relative) big output cap should be avoided in a good PSU design, because you cannot control the stored energy.
But how too improve this without loosing the overall performance of the given design 

Rgds Arne
 

[sync]

Hi Peter

Thanks for the service notes link. My E3610A is from 1994 and doesn't had all mods. I done the missing ones and then my unit oscillates too.

It's caused by E3610A-05A. The 390 Ohm resistor between TP3 and TP4. Even without this mod it oscillates with a resistor and capacitor parallel at the output. HP, didn't you check this?

I done your mod with a 330p cap across R6. 220p was not enough (12V/21W bulb and 4.2uF). Now it doesn't oscillates any more. Thanks!

[pgross]

Hi Peter

Thanks for the service notes link. My E3610A is from 1994 and doesn't had all mods. I done the missing ones and then my unit oscillates too.

It's caused by E3610A-05A. The 390 Ohm resistor between TP3 and TP4. Even without this mod it oscillates with a resistor and capacitor parallel at the output. HP, didn't you check this?

I done your mod with a 330p cap across R6. 220p was not enough (12V/21W bulb and 4.2uF). Now it doesn't oscillates any more. Thanks!

You're welcome. I have done some experiments with the 390 Ohm resistor (Service note 05A), but I can't measure neiter any improvements or changes whatsoever. 220pF seems to be adequate to cancel out the oscillations - at least on my two supplies. I still can't figure out if there are any drawbacks utilizing this hack.

The value of R6 is 1M5.

ooops that's why the oscillation is possible 
I wonder what was the intend of the the HP Designers on that circuit .....hm

The E3610A does indeed have a rather large output capacitor - 470uF/50V - though it isn't bypassed with a film cap. Originally there were two series mounted 0.1uF ceramics – C4 and C5, at the output with ref to ground. Later, a service note (E3610A-04) suggested that these caps should be removed – because the output showed oscillation if either the + or – terminal was connected to ground!

As far as I remember I did all the Service Notes I found but we should really look again into the schematics if there is some space for improvement.
The (relative) big output cap should be avoided in a good PSU design, because you cannot control the stored energy.
But how too improve this without loosing the overall performance of the given design 

Rgds Arne
 

The output cap improves the pulse current capability of the power supply, so I dont think that it should be avoided. It is true that if the output capacitor is very large, the current limiting circuit can be defeated and in worst case a large current peak is able to damage load components. Therefore a bleeder resistor is necessary to discharge the cap fast enough to avoid catastrophic failures. The bleeder resistor, R46 in the E3610A supply is 390 Ohm.

Best Regards

Peter   

[sync]

220pF seems to be adequate to cancel out the oscillations - at least on my two supplies. I still can't figure out if there are any drawbacks utilizing this hack.

With the 220pF i had oscillations with a 4.2uF and a 12V/21W car light bulb as load. It stopped with 270pF. I use 330pF for a little safety margin.

The cap fix makes the control loop slower. I made a load transient test. From 1A to 2A at 15V and vice versa.

1A to 2A without the cap.


With  220pF.


With 330pF.


2A to 1A without the cap.


With 220pF.


With 330pF.

[pgross]

Hi Sync,

Thanks for your measurements - very interesting indeed.

It is easy to see the slower regulation response from the voltage amplifier with this cap mod.

I dont think that it will cause any problems when using the power supply in most situations, and it's a small penalty
to pay to get rid of the oscillation with capacitive loads - in fact the cap seems to get rid of the little overshoot too. 

Best Regards

Peter

[sync]

Yes, it's not a problem. The specs say 50us to recovery within 10mV. I measured about 4-8us. I think the overshoot indicates that the original compensation wasn't right.

[pietr]

Regarding service note 7, the removal of diode CR8: doesn't this damage the 2N6056 power transistors? If I understand the circuit correctly, removing CR8 will allow the emitter-base voltage to be up to 12V (when the base-emitter diode is in reverse, not conducting), while the 2N6056 datasheet lists a maximum emitter-base voltage of 5V.

[sync]

Yes. Good catch!
I just verified it with my E3610A. Without CR8 the emitter base voltage reaches 8.8V!

[alm]

This was actually a common failure mode of the older Harrison/HP power supplies: connect a cap to the output, turn off the unit or quickly turn down the voltage knob, and the B-E junction gets blown because of the reverse voltage (the base gets pulled down while the emitter stays at the same potential). An anti-parallel diode across the pass transistor B-E junction (and a series resistor between base and driving transistors) would be the fix .

[sync]

The CR8 method prevents it too. Without the current limiting resistor. Why does Aglient recommend to remove the diode? Have they lost their knowledge?

[rdl]

Heh, wonderful. I just bought a second one of these off ebay and now I am finding out they are not as good as people have been saying.  Ah well, probably still better than the cheap Chinese power supplies I could have gotten.