Agilent / HP 33120A AWG repair

[TheSteve]

I bought an Agilent 33120A off ebay that was supposed to be working. Sadly it failed the full self test with several errors. I do believe I now have it working 100% - here it the story, maybe it will help someone else.

The full self test failed with error 603 - Waveform RAM readback failed and error 608 - SYNC signal detection, Elliptical filter path.

The first thing I did was replace the waveform ram - U404 and U405. They are only a few dollars each and are easy to get from Digikey, Mouser etc. After swapping the ram the unit powered up and passed all self tests and worked great for several days.

A few days later I powered it up and it wasn't working properly, sure enough the errors were back. If I powered the unit on enough times it would eventually start working. I double checked the soldering on the ram but that made no difference. After some time looking at the schematic and probing with a scope I found one data line from the waveform ASIC U206 was dead. I lifted the pin and verified the problem was with U206 and that the line wasn't being loaded down etc.  Oddly enough if you bang on the chip a bunch of times it would sometimes start working. I figure a bond wire inside had failed perhaps.
Now U206 is a PLCC 68 ASIC made by TI, part # F107563FN. Not a part stocked by any parts house. I eventually decided to give seekic.com a try and got replies from 15 or so chinese vendors offering to sell me the part. After weeding out the most suspect I replied to a few of them. Only 2 or 3 would accept paypal, there was no way I was going to risk a bank transfer. The price for a single piece varied from 45 to 100 USD. I wasn't sure I wanted to commit that much plus shipping/paypal fees etc but after a month decided to give it a go. The total was 79.00 USD with DHL shipping from the vendor which scared me the least. They gave me a tracking number within 48 hours and 2 days after that I had the chip. It was brand new and looked real enough. The silkscreen and molded in markings matched the real part top and bottom so I decided to solder it in.
So far so good, the 33120A is once again working 100% - I guess I'll need to burn it in for a week or two but I suspect it will keep working now.
The vendor I got the part from told me there was only a single supplier with the part and that they only had 3 of them(two left now) - so if anyone else needs U206 to fix a 33120A now is probably the time to buy it.
This was the first time I tried buying a part like this direct from china and it is nice that the result is positive so far.

I did get the 33120A for a good price to begin with and got money back from the seller when it didn't work. So in the end even with the extra 79 USD it was still a good deal, and has option 001 - external ref in/out

Here is a pic with the replacement U206 installed.

[HighVoltage]

Interesting story.
How did you remove and install U206?
Heatgun?

[TheSteve]

Absolutely - I used a Hakko 850 clone hot air station(43.69 shipped in the US off the 'bay) to remove the ram and the waveform ASIC U206. After cleaning the pads with solder wick I hand soldered the IC's back in place. Using kapton tape and metal shields to protect anything sensitive the PCB can remain in the chassis which makes it a pretty quick job.

[HighVoltage]

That is impressive that you did this repair inside the chassis, I think I would have take it out.

First thing I thought, when reading your story was, that there is a bad solder spot on the ASIC and not necessary a bond wire inside the ASIC.
 

[TheSteve]

I really wish that had been the case - but I lifted the suspect pin off the PCB and soldered a wire directly to it. I also spent a lot of time verifying and resoldering connections before I lifted the pin.

This thing had been an evil thorn in my side for too long. Everything else broken in my shop had been fixed, so letting it beat me was not a pleasant thought.

[HighVoltage]

I think you can be lucky that you even found a fitting ASIC. I had a similar situation with an Agilent instrument and there was no way to get the part and I had to give up on it. I used to have instruments professionally repaired, until I started to do it myself. Most Agilent gear actually is easily repaired but some - as you said - require a lot of attention.