HP/Agilent/Keysight E4403B repair

[Bicurico]

Hello,

I received a broken Keysight E4403B spectrum analyzer from my friend, to see if I can repair it for him.

*** Disclaimer ***
This is not a business, I don't receive any money for my repairs, some devices I purchase myself, others are loaned to me. All help I request in my repair threads are for my own learning and not to take anyones business away. I have a propper job in something completely different (CAD/CAM/CAE/RP/metrology), this is just a hobby. I felt like saying this, as suddenly I find myself repairing devices where the difference between broken and working can mean a substancial profit for some people, but this is not the case here.

Day 1 - First assessment

The device is indeed broken, it does not start up.

In fact it is completely dead! I read the service manual and the only test they mention is to see if there is any power on the probe power connector on the front panel, when the power cable is inserted. There is no power on neither pin.

If i measure the voltage on the 20A 32V DC fuse, I will get aprox. 10V, which is apparently correct. So something is at least converting 220V AC into 10V DC.

I opened and cleaned the power supply module and found no visual fault. I did find a loose DC fuse inside!!!  God knows what this may have caused.

Looking at the power supply and the microcontroller board fitted into it, I realize this might be a really difficult job for me.

So here is my first help requests:

1) What should happen on a working E4403B, once you insert the power cable? Should the fan rotate for like a second (as the R&S CRTU) or does nothing happen? Should there be any LED going on?
2) Could it be the frontend module? But then, shouldn't I have power on the probe power connector?
3) Are there any test points on the power supply?
4) Does anyone have some tipps to get me started? Like common power supply failures for this device?

Thanks,
Vitor

[razberik]

https://www.eevblog.com/forum/repair/agilent-e4406a-repair/

Perhaps it is similar ?

[Bicurico]

Hi,

Unfortunatly the linked thread is not about the power supply, but yes, the models are fairly similar and probaly use the same power supply type.

Regards,
Vitor

[Marcel G]

1 Power on button
2 white screen
3 black screen
4 black screen with Agilent logo and type number , firmware version.

From point 2 is de blower running.

17.05 VDC on de fuse at the backside. (and upper point at the DC input).

Regards,
Marcel

[Bicurico]

I just measured again and I get 0V on either pin (upper DC pin, DC fuse and probe power on front panel).

I don't knwo how I measured any voltage yesterday, but then it was really late at night and I might have had the meter in mV or so.

This is actually a good sign, because NO voltage means that something is probably wrong right at the input...

Regards,
Vitor

[Bicurico]

After further testing, I noticed that:

a) There is only AC168V inside the transformer, but moving on in the circuit there is suddenly no voltage at all.
b) After a new visual inspection, I noticed that the ETD34 transformer looks not so good on the side: I can see some white oxidation/something pushed outside the protective layers (sorry, can't explain it better).

It is now >2 a.m., so I will call it a day, but I suspect that this transformer might have a cut wire somewhere. Will measure it ASAP. Can anyone elaborate on the ETD34 transformer type? I found only geometric datasheets, but non explaining the input/output voltages.

Will upload photos this weekend.

Thanks,
Vitor

[Bicurico]

Hi,

Today I removed the ETD34 transformer which had some odd looking oxidation peeking out of the windings, but after measuring it, it seems fine. I soldered it back onto the board.

Also, meanwhile I figured out that ETD34 is a standard for the cylindrical core and case of the transformer - the end use dictates how the coils are then applied. This means that without further documentation, I would not be able to rebuild such a transformer... Initially I thought ETD34 was the standard of the complete transformer with coils.

To make things worse, I looked superficially on the Yahoo User Group for HP/Agilent and the few relevant threads mention that this particular E4401 power supply dates from the HP days and has been reused under Agilent and Keysight. No service guide seems to have ever been published. 

Attached is a picture of the open power supply. Notice the addon board that slides in like a DIMM memory. I guess it has a microcontroller to switch the power on/off with relevant delays and to monitor voltages, etc.

The picture shows the areas where I was able to measure AC voltage and where it seems completely dead (the unmarked areas). Note that this is just a rough sketch. I could measure 185V and 90V AC on some parts of the board connected to live power line, but the remaining regions of the board were totally dead. Also, I measured A with a clamp meter and there is no relevant power consumtion. I assume this tells me there is no short.

I still hope to find some failed component, which, when replaced, gets life back into this spectrum analyser.

Again, any opinions are welcome on how to tackle this.

Regards,
Vitor

[bitseeker]

Hi Vitor,

If you don't have any obvious shorts, check the transistors, diodes, and opto-isolators. Apart from capacitors, they're common fault sources in supplies. For example, a UPS I was fixing a little while ago had a fried power MOSFET, two dead opto-isolators, and a toasted current-shunt resistor in its power supply.

[SMB784]

One thing I will say is that you have the dreaded clear-yellow RIFA safety capacitors.  Those things are prone to failure at a moments notice (they develop hairline cracks in the plastic that allows moisture in), even with the device switched off (they can fail merely when the power cord is plugged in).  It's possible that they have already "silently" failed (they can fail loudly, and the results are often quite alarming and smelly), causing your device not to function.

You should replace those anywhere you see them, just as a matter of habit.  They are a ticking timebomb waiting to destroy your equipment.  Hopefully they are the issue!  If not, keep probing around component by component to see where the failure is.

EDIT: I have attached an edited form of your image with the offending capacitors circled in yellow.

[Neganur]

I don’t know this unit too well but if the E4403B has a soft-power ON switch, test the 5V rail in the PSU. Bad caps in this circuit will prevent it from starting up.

[Bicurico]

Hi,

Thanks for the reply. There is no DC power anywhere inside the PSU... Only AC in the first few sections of the PSU.

Regards,
Vitor

[Neganur]

Is this the PSU of the e4403b ?

[SeanB]

In the blue area there are 2 small electrolytic capacitors. In any non functioning supply I would regard those as suspect, and remove them to see the value and ESR. Of course once you remove them it is often easier to replace with a new one first, and not to solder the old ones back in. Seeing as one lives in a hot space next to a heatsink, and probably is the startup power supply that is the first thing to change.

[Bicurico]

It looks exactly the same, but the sticker says:

Agilent (2001.07)
E4401-60186
S/N: KRxxxxxxxx
Made in Korea

I will replaces the capacitors for sure...

Also, I made aquaintance with the RIFA capacitors in another repair... They tend to burst open.

Regards,
Vitor

[Bicurico]

No luck. The capacitors are OK, apparently. Removed one and measured it externally: was within specs. The other I measured in place and even so, they show reasonbable readings.

It *might* be the RIFA capacitors. I tried to measure them (in place) and got nonsense readings. Not sure if I can measure them like this, though.

I will try to order replacement ones. Especially the orange one 330nF looks slightly suspicious, because it has cracks in the plastic case. But then, these cracks are just on the exterior.

The struggle goes on...

Regards,
Vitor

[Neganur]

Actually, co-reading on the Yahoo mailing list, it would have been useful if you had mentioned what and how you are measuring the capacitors and deem them "good".

The UNI-T 61E is not an ESR meter so whatever you measured (the capacitance?) is not a good indicator for the state of the component. It may have high ESR and hence be broken but still show capacitance on the DMM.

[Bicurico]

You are right.

I do have a chinese low cost capacitor/inductor meter. I will use that on my next attempt.

I was/am still kind of mentally locked in the assumption that some part must have fully failed for no DC voltage to be measurable anywhere inside the PSU.

I assume(d) that a capacitor failure would produce boot problems, bad filtering, etc., but at least SOMETHING would show up voltage wise?

And please make no mistake: I am still learning the art of repair, so please bear with my mistakes. Constructive critizism is welcome!

Regards,
Vitor

[vtp]

Right in the middle of the board there's a silk screened text "CUKonverter".

So it looks like the converter topology is a CUK and the supply came from TESLAco. There is plenty of literature available (use google) to figure out how CUK power supplies work. Do some reading first instead of trying to stumble blindly around. Never mind if the schematics is available or not, once the topology is known it is not hard to figure out what function each component has.

Ooh I so wish I had this under repair and see all the Cuk magic applied...

Good luck.

[Bicurico]

Hi,

The PSU is repaired and the E4403B alive.

The repair was a nightmare and was carried out by a friend with infinitely more knowledge than myself.

On the first stage of the PSU there was a burned trace which I overlooked. Also, a MOSFET was dead and with it an IC.

Repairing HP/Agilent/Keysight power supplies is a nightmare and a huge deal:

- no documentation or schematics whatsoever
- lots of custom compoments (aka unobtainable)
- built-in microcontroller that switches supply on/off, verifies status, etc. - if this gets broken, good bye to the power supply, as you won't find the software inside the controller...
- dangerous due to switching power supply
- very complex PSU with lots of sections

HP/Agilent/Keysight does not want any information regarding the PSU to leak.

Bear this in mind when hunting for second hand units of the E44xxA/B series!

Don't believe me? Try to find second hand, refurbished or new power supplies for this series. You won't find any for reasonable money.

Regards,
Vitor

[Bicurico]

By the way: anyone familiar with any hack to activate extra options... 

[bitseeker]

Check the silkscreen or any stickers on the power supply for a manufacturer name. Often, the power supply is not made by HPAK.

[Bicurico]

That's the point: the PSU was developed by CUK and has been in use by HP/Agilent/Keysight without any public datasheet or info. Not by any coincidence, but on purpose.

As I said, it is repaired now, but at great efforts.

Regards,
Vitor