3458A Gate array failures + various faults

[Oldtestgear]

In my recent haul of 3458A spares are 6 off 03458-66503 boards. All are marked "Gate Array Failure". Testing one does show this as the failure mode (reported by the DMM). My question is "do these fail regularly" or is this an unfortunate collection of dead boards? Another one is marked "dead ADC".

One of the 3458As I just collected has the dreaded "error 114 rundown convergence error " just to add to the mix of issues I need to work through.  At least I have enough boards to try swapping components once I have done some proper fault finding in the hope of getting one (two would be nice) of these boards working again.

Any comments are most welcome but please do not warn me that these are really difficult & expensive to fix. I know & am prepared to spend many hours to test the spares I bought & repair the faulty 3458A.


Cheers

Phil

[Dr. Frank]

Good Gracious!
You really collected a Zombie Army 

[aronake]

A good start would be to try all 03458-66503 (a.k.a. A3) boards and list all error messages here.

The gate array on the 03458-66503 quite rarely fails. The U180 chip very often fails and is often the cause to the convergence errors.

So if you are very lucky you have some (or all) of the boards with working and non drifty U180. You could than either buy a drifty A3 board to replace the U180 on that and may have a working A3 board, or sell the boards you have that have working U180 chips. The buyer would likely want to test the U180 on his A3 board before paying. But if it really works and is not drifty it may be worth 500 to 1000 USD each.

I find it quite likely that if a A3 board have gate array failure it would not go further in the self test could still also have a convergence error. And just because it does not show any error, it does not mean it is fully working. The test on service note 18 give some more indications if it actually works.

Its strange with so many gate array failure boards. Maybe they have been put in a 3458a that had some error that cascaded into the boards. Or maybe it actually is a more convergence error and someone just wrote gate array failure on them.

[Echo88]

Id recommend reading through the 3458A repair articles on xdevs: https://xdevs.com/
While i cant help in this case, there are quite a few guys on the xdevs IRC channel having extensive knowledge about 3458a repair.
Happy necromancing 

[IanJ]

Hi,

In 2018 I was repairing a 3458A and it had a faulty A3 board, i.e. U180 drift.
I bought a new A3 from Keysight UK, but they allowed trade in of the faulty A3.

Just wondering if you had several A3 drifty boards could you trade them all in for one new one and get a lot better discount?

Ian.

[Oldtestgear]

Keysight will not sell A3 boards now. It is a return for repair only option for these boards. As Keysight quote a fixed price repair for the 3458A it is tempting to return an instrument with a set of faulty boards to get value for money.

Phil

[mendip_discovery]

Is the cost of just sending the ones you have in + the price you bought them for = some profit if you sell them as newly refurbished.

[Oldtestgear]

Is the cost of just sending the ones you have in + the price you bought them for = some profit if you sell them as newly refurbished.

Even if Keysight was willing to refurbish the boards, which is very unlikely, the cost would be too high even if I had been given the boards.  As I paid good money for them this is not going to work for me, or any potential buyers.

I have a spare 3458a, so can test all of the boards to at least a pass/fail state. More careful testing will take extra time & may only be required for the analogue cards. Once I have finalised what I am doing I will be able to offer the boards here to anyone who wants them & then on eBay to a wider audience.  I hope that even faulty boards have some value to someone, plus there is a bunch of other spare parts including a couple of displays mounted on the front PCB.

Testing times ahead!

Phil

[coromonadalix]

if i recall, sending them to be repaired  implied an full calibration, you could not get out of it .. that why it was an easily 1200$ USD cost  in the past, i dont know the actual price now ...


and with theses u180 failures, was it a physical failure  ..  humidity entered in them ...  why  the drift is happening ??

@Oldtestgear   do check your fiber optic links, if they are clean and well inserted ... some have reported bad insertions when you "play too much in them"

i do hope you'll make one or more working    finger crossed

[aronake]

If one of the boards pass without any error you can do the service note 18 test on that to see if the U18o works.

Then desolder U180 of all other boards that have errors and replace in the working board and test each of the U180 for service note 18. You would then end up (hopefully) with a couple of U180 that are confirmed to be working, that people likely would be happy to pay quite some money for.

With a decent desolder gun desoldering a U180 is not too difficult.

[KT88]

I got a cost estimation of €3k last year for repair and cal from Keysight Böblingen.

[Oldtestgear]

I got a cost estimation of €3k last year for repair and cal from Keysight Böblingen.

That quote is very similar to the one I received from Keysight UK to repair one. This included calibration as well.

I hope that I will have a refurbished 3458a available for sale in the next couple of weeks. I have ordered the capacitors, FRAM, etc  to get it ready to sell. So far it is looking very stable....

Phil

[Oldtestgear]

Another question.
There are several members who have converted the Dallas NVRAM to FRAM with good results. My plan was to do the same but I have run into an issue that I did not expect.  Programming the FRAM using a TL866ii programmer is simple enough. That gets the original firmware into the DS1220 replacement  but what happens if the 3458a needs recalibration? The DS1220 programming algorithm does not work (I also tried a Dataman 48) meaning the 3458a cannot be recalibrated. Or have I got something wrong?

The same problem arises with the DS1230 replacements when used with a programmer.

So, the big question is what is happening in the 3458a? Is the read/write timing correct for both chips? Has anyone actually checked the timing to be sure?

I  am hoping for an explanation as I have already spent money & some time on this part of the project without success so far. Everyone has been happy with the outcome except me so have I gone wrong somewhere?

Thanks in advance

Phil

[IanJ]

Another question.
There are several members who have converted the Dallas NVRAM to FRAM with good results. My plan was to do the same but I have run into an issue that I did not expect.  Programming the FRAM using a TL866ii programmer is simple enough. That gets the original firmware into the DS1220 replacement  but what happens if the 3458a needs recalibration? The DS1220 programming algorithm does not work (I also tried a Dataman 48) meaning the 3458a cannot be recalibrated. Or have I got something wrong?

The same problem arises with the DS1230 replacements when used with a programmer.

So, the big question is what is happening in the 3458a? Is the read/write timing correct for both chips? Has anyone actually checked the timing to be sure?

I  am hoping for an explanation as I have already spent money & some time on this part of the project without success so far. Everyone has been happy with the outcome except me so have I gone wrong somewhere?

Thanks in advance

Phil

Whilst the algorithm doesn't work for the Dataman to recognize and characterise the *exact* device connected as a Dallas Non-Vol Ram it doesn't mean that the basic device IO in normal operation won't work, i.e. Address bus, data bus, WE, OE & CE pins as they are very much generic.
In normal operation it's the basic device IO that matters......the 3458A doesn't know it has a FRAM fitted, only that it is communicating using basic device IO.

When I converted my 3458A & 3245A both to FRAM (cal & settings) the first thing I tried was changing the GPIB address, powering off then on again.......and the new address recalled without issue.

To program a FRAM in your programmer you are always going to have the programmer set to a FRAM device type. The reasons for the algorithm's in the first place are for efficiency, speed, error checking, timing control, power management & optimization so that you can be sure your device is programmed 100% perfectly and as fast as possible in your programmer.

Here's a wee short I made...... https://youtube.com/shorts/Vz6c6m86O4c

Ian.

[Oldtestgear]

Hi Ian,
I was aware of the potential timing issue when swapping these devices & just wanted to be sure that I was not  setting myself up for potential issues.

Thanks

Phil

[Oldtestgear]

Another board failure.

I tested 4 off of the A9 reference board (3458-66059) and had one failure. Actually a control loop failure according to the 3458a display.

The LTZ1000 heater appears to have gone open circuit. Has anyone else experienced this failure mechanism? 

This is a failure I was not expecting as this is a simple board.

All comments are welcome.


Phil