Agilent 33220A teardown and quick repair

[Chipguy]

Hi,

welcome to teardown... ok it's just a photo gallery but also a quick repair.
My Agilent 33220A 20MHz Arbitrary Signal Generator failed

This is the failure it showed, so the PLL unlocked. Not only this, the unit froze after a few seconds displaying this.
It can only be turned off, the soft power switch still worked surprisingly.



Opening the unit is easy. After taking off the protection rubbers and the handle there are 2 screws at the back (TX20) to remove the plastic holder and then you can take the housing cover off. Well made unit.



To get into the Agilent the inner top cover needs to be removed by unscrewing 2 screws.
The switchmode PSU is attached to the cover. Its a 12V/40W open frame module from Artesyn.
All screws are of same length, very nice.



After a while looking around (see gallery) I started measuring the voltages and found that the entire power supply started to drop all voltages after a few minutes into operation.
So I started tapping tantalum caps to check if any of them gets hot.
I found these on the bottom of the board, and the 2nd from the upper line right was a bit darker.
Well what can I say, that bastard of pure yellow evil burned my finger.
Those tantalum caps are manufacturered by Kemet, according to the service manual (1st edition) they are called "T491..."



I have ordered all 8 of them but they didn't arrive before the weekend.
Since I need the sucker I had to do a quick fix:



This version with 8 x 47µF tantalum capacitors has only been built from around 2003 to 2005.
They changed the design to 10µF ceramic caps at those locations and added 2x 220µF electrolytic caps to those +15/-15V rails.

This is now the 3rd time a Agilent/HP unit fails on me because of those damn tantalums.
I call them "pure yellow evil", first they give you a hassle by making ur unit fail and then they burn your finger if you are not careful.

For those who want to play along here are the links to the service manuals:
Link to Edition 1 Service Manual http://cp.literature.agilent.com/litweb/pdf/33220-90010.pdf
Link to Edition 4 Service Manual http://cp.literature.agilent.com/litweb/pdf/33220-90012.pdf

Here is the gallery of high resolution pictures, enjoy:

[SeanB]

You could probaly just leave that 47uF in there and stick it down with a drop of flexible polyurethane single component adhesive, or some acetoxy free silicone sealer.

Military boards we used that to hold down the replacement electrolytics, after persuading them to come free from the death grip of the original conformal coating, most common being Mesopoulet K which was pretty much immune to any solvent that we used , often the GRP resin under it would dissolve first. But you could solder through it, and just stay away from the cyanide fumes. Fan was very handy to blow them away, and did double duty to cool the DUT as well normally, or to keep you cooler in the heat.

[tautech]

Well what can I say, that bastard of pure yellow evil burned my finger.

    

I'm amazed at your restraint. 

At least yours was easy to find, which is not always the case.
Tantalum failures are one of the MOST frustrating things IMO.
Why do people use these bloody things. 

Nice thread thanks

[Chipguy]

You could probaly just leave that 47uF in there and stick it down with a drop of flexible polyurethane single component adhesive, or some acetoxy free silicone sealer.

What you can't see on the photo that I only had 22µF at hand.
I am going to replace all of those 8  yellowdevils  anyway, once the replacements arrive. The ones on the board are wave soldered, i.e. glued on the bottom of the board with the usual red gunk. = double evilness   

It is good to know that they got rid of those in 2005 once better stuff was available 

[Chipguy]

...
Why do people use these bloody things. 

Nice thread thanks

Thanks.
Once (2003) there was a time when there was nothing else available when you wanted 47µF with low ESR (0.15 Ohms).
Even 1µF/1206 ceramic capacitors did not exist back then, at least not for a price you were willing to pay.

Today I stay as far as I can using those.
When in doubt-> C4... (oh no that was Jamie from Mythbusters) ... no... when in doubt I use those polymer capacitors, only good brand ones.
They are expensive but they should last a long long time.

[SeanB]

What you can't see on the photo that I only had 22µF at hand.
I am going to replace all of those 8  yellowdevils  anyway, once the replacements arrive. The ones on the board are wave soldered, i.e. glued on the bottom of the board with the usual red gunk. = double evilness   

It is good to know that they got rid of those in 2005 once better stuff was available 

Not that the capacitance is an issue, more the low ESR, providing the supply is capable of providing the current to the board. The cap is there to handle local current so it only needs to be over a few uF, and lower in ESR than a certain amount. Easy way to remove them is to place a hot soldering iron on top with a small bit of solder on the tip and leave it there until the capacitor is hot enough to reflow the leads, then simply flip it off the board as the body will have broken free of the red dot. Terminal for the capacitor, but does not damage the board at all. Even works for a QFP chip if you use a big tip and get enough heat in.

[Chipguy]

Not that the capacitance is an issue, more the low ESR, providing the supply is capable of providing the current to the board. The cap is there to handle local current so it only needs to be over a few uF, and lower in ESR than a certain amount. Easy way to remove them is to place a hot soldering iron on top with a small bit of solder on the tip and leave it there until the capacitor is hot enough to reflow the leads, then simply flip it off the board as the body will have broken free of the red dot. Terminal for the capacitor, but does not damage the board at all. Even works for a QFP chip if you use a big tip and get enough heat in.

I had looked that up before. The original one had 0.15 Ohms.
The one I put in has about the same.
Bit lower at 120 Hz, but a bit higher at higher frequencies.
It works great atm. Has to hold until sometime next week.

[jpb]

On the subject of tantalum capacitors - I'm about to replace all the electrolytic capacitors on my recently bought 15 year old Keithley 2015THD meter. It is working ok so this is just a precaution and in the vague hope that it will make the rather dim display a bit brighter.

The unit also has dozens of surface mount tantalum capacitors - is it a good idea to replace these or should I just leave them be as they've functioned ok for 15 years or so?

If I get enthused enough to replace them, what should I replace them with? More of the same spec tantalums?

The general engineering adage I was brought up on was "if it ain't broke don't fix it" but with capacitors and aged equipment a certain amount of precautionary actions seem sensible.

[HighVoltage]

Your 33220A probably has a Serial Number starting with 43
Agilent fixed this problem only with those instruments, that have a Serial Number that starts with 44

I had several of these 33220A over the years and some failed in the first few weeks of operation.
Of course, Agilent fixed them under warranty.
But then it is a surprise that such a problem arises on a high quality instrument.

[Chipguy]

Your 33220A probably has a Serial Number starting with 43
Agilent fixed this problem only with those instruments, that have a Serial Number that starts with 44

You are right, it starts with "MY43..."
I wonder if the tantalums they used in the production are really KEMET T491 like mentioned in the Service Manual.

[Chipguy]

On the subject of tantalum capacitors - I'm about to replace all the electrolytic capacitors on my recently bought 15 year old Keithley 2015THD meter. It is working ok so this is just a precaution and in the vague hope that it will make the rather dim display a bit brighter.

The unit also has dozens of surface mount tantalum capacitors - is it a good idea to replace these or should I just leave them be as they've functioned ok for 15 years or so?
[...]
The general engineering adage I was brought up on was "if it ain't broke don't fix it" but with capacitors and aged equipment a certain amount of precautionary actions seem sensible.

I would say, keep that rule up. However once one of them fails, replace at least every single one of same capacity and voltage rating.
That's what I decided to do. Since I don't want to open it 8 times for every blown tantalum that will come up pretty soon.

[VK5RC]

I just had a tantalum fail short on a HP 34401and it burnt the board and took out a zener voltage regulator as well! PCB trace still intact (just).
I replaced all tants of the same make and value.
Photo of the burnt offering below, C551 and C553 needed to be removed for access.

[jpb]

I would say, keep that rule up. However once one of them fails, replace at least every single one of same capacity and voltage rating.
That's what I decided to do. Since I don't want to open it 8 times for every blown tantalum that will come up pretty soon.

That seems sensible, though reading these forums has given me a great fear of exploding tantalums destroying the circuit board -  - but then replacing them with new tantalums doesn't guarantee the new one won't explode anyway!

[Chipguy]

Playing with tantalums is like playing with Murphy's law.
It will always get you.

When in fear that tantalums blow up, you could replace them with fused ones.
I saw those on the Farnell product list when I was searching for spare parts.
But beware: They are expensive.

[tautech]

I would say, keep that rule up. However once one of them fails, replace at least every single one of same capacity and voltage rating.
That's what I decided to do. Since I don't want to open it 8 times for every blown tantalum that will come up pretty soon.

That seems sensible, though reading these forums has given me a great fear of exploding tantalums destroying the circuit board -  - but then replacing them with new tantalums doesn't guarantee the new one won't explode anyway!

2 things....
In the vast majority of cases they are used for de-coupling and therefore precise value is not critical, so if values are available , why not use multilayer ceramic?

Most of the failures I have seen have been bead tantalums with a 35V or less rating.
I have always replaced with 50V versions.

As we all know the general quality of componentry has improved, but are there still failures with modern manufacture of these little yellow devils? 

[JaspaJami]

What are the real selling prices for these 33220A:s? I have been offered one that seems to be in good condition and serial is MY44xxxxxxx for 200€ (235USD, 314AUD, 175GBD).

[Chipguy]

Seems to be a good price.
A few yearsback they were still about 1000 EUR.

Edit: YES it is a good price. These days they sell from 400 to 800 EUR.

[JaspaJami]

Ok, bought it. It also comes with 54652B ( https://www.keysight.com/en/pd-1000000563%3Aepsg%3Apro-pn-54652B/rs-232-and-parallel-interface-module ), have to find someone with 54600-series scope who would have use for that.

[HighVoltage]

You got a really good price, as long as the instrument is working well.

[JaspaJami]

Orginally i was buying Agilent E3630A power supply for 120€. Seller uploaded wrong picture and item that he sold for me was Agilent 34401A Multimeter. Well i did get that for same 120€ so im happy because i needed anyway both and its easier to get power supplys. And now i bought this 33220A for 200€

All of them seem to be very good condition and dont give any errors.

[Flanbix]

I plugged back my old Agilent 33220A after probably 4 years of storage.
I seems like the screen backlight is taking ages to start up. I cannoy remember if it was already doing that before but I suppose it should start straight away.
Does the screen have a known aging problem on these devices?

[HighVoltage]

This is my "old" 33220A from 2014 and it is used almost every day.

The display still looks (to me at least) the same as in 2014 and it comes on right away without any significant delay.