Testing ICT Backdrive Circuitry

[Singapura]

backdrive test

Can anyone advice or point me to websites that teaches how to test in-circuit tester's backdrive circuitry. I am experiencing false failures while testing certain PCBs even after replacing the called out ICs, but when I run self-test on the tester there was no problem.

I suspect that the driver circuits of the tester can no longer backdrive as well as it used to. The backdrive specifications are:

Logic High source current: 500mA
Logic Low sink current: 300mA

Any suggestion is greatly appreciated. Thanks!

 

[TurboPC]

Hi,

I don't have any web site to suggest, but I worked many years with ICT testers and I have some insight I want to share. First thing, I am pretty sure that your ICT tester is working perfectly, since there are many reasons for the problems you are experiencing. I will asume here that you test has been properly designed/customized and is working.

1- Always disable surrounding components that may drive at the same time as the tester. This is the first rule. Don't think that the ICT backdrive is always 100% efficient, especially at fast speed. You may try to change the test vector speed, but there is no guaranty of success. Remember that upstream component may sink a lot of current in trying to beat the ICT backdrive.

If you can't disable upstream components, think about making them drive '1' instead of '0'. A logic '1' is easier to overdrive by the ICT than a '0'.

Otherwise, perform a cluster test by trying to access nodes that can be easily disabled or overdriven.


2- Make sure that the ICT is driving/receiving the proper logic levels of the device under test. Sometimes, you may want to drive higher (like 4.5V instead of 4V for a 5V device) just to overcome upstream devices, or you may change the logical '1' and logical '0' receiving threshold.

Long wires in the ICT fixture may cause ringing because of their inductive nature. It gets worst with heavy overdriving. Again, changing the speed of the test may have some effects.

3- Make sure the proper fixture probes have been selected for each type of pads or contact on the PCB. Make sure probes are not too old, oxydated or bent causing unreliable contact. I've seen proble contact in the order of 1, 10 or 20 ohms that may affect the tests, especially in case of heavy backdrive.


There may be some ways to pinpoint the problem. What is your ICT tester? What are the UUT and the upstream components? Can you share the schematics or the test? I'll be happy to have a look...

TurboPC

 

[Singapura]

Hi TurboPC

Thanks for your reply and suggestions, which are very helpful indeed. Like yourself, I worked with ATEs for many years too, and some of the methods you pointed out I've tried before.

The ATE I am having problem with at present is a Schlumberger S750. It is quite a piece of machine, though a little aged (over 10 years in use). Why I suspected the tester is because the PCB I am working on passed all the small ICs consistently and easily 2 months ago, but after the spares came and I replaced the defective IC, some small ICs started failing. My colleage has two such PCBs on hand and I borrowed them to help isolate the cause. They too failed the same ICs as on my PCB. The ICT fixture is in pretty good condition due to low usage, so naturally my suspicion rests on the tester.

In fact, about 2 years ago, the power supply (5V 120A) that drives the tester's test electronics died and we spent over two weeks recovering it. Since then, we have occasional hiccups when the tester seems to behave strangely. I have called on my maintenance support contractor who performed self-test on the tester but each time no problem was reported. As I see it, the self-test fixture uses a straight through loop-back wiring to test the driver/receiver circuitries, and therefore if there is a degradation on those drivers which cannot provide sufficient backdriving on heavily loaded nodes, the in-circuit routines will most likely fail.

I have given quite some thoughts as to how to do a realistic self-test that will isolate those 'weak' drivers, but as it is company's asset, I need to know the correct technique so as not to damage the tester.

Most PCBs that I work on are military related, so it is not convenient to release (sorry about that). But I will be glad to share insights with you on anything that will be helpful to you in return (but I assume you have more years of experience than me as regard to testing).

Still, glad to know you bother to take the time to help. Thanks!

 

[TurboPC]

My god! A Schlumberger S750! Wow! When you talked about diagnostic mode, I thought about an HP3070 (which is kind of old also) or something newer where diagnostic was pretty advanced...

I played with one (S750) at least 8 years ago, and I thought it was much older than that. I can't offer much support for this one (only a 3-month experience), but I tend to agree with you that something could be wrong with the drivers. We had problem with this type of machine as well.

In any cases, keep us informed of your tests, maybe someone will have an idea or in last resort, it will improve our culture...

Good luck!

Just a note: have tou tried to use a good logic analyser over the DUT and check the waveform/logic level while the test is performed? Can you see the backdrive problem? Maybe a digital scope can be used to look at some suspicious node and see ringing or strange logic level... Just a thought...