Definition of a short circuit, orders of magnitude

[mellowtron]

Hi all,

I work at a company that makes a certain piece of test equipment, and it has a custom "main board". I recently was asked to investigate a series of failures with the main board and work with the manufacturer to solve the problem. A boost converter was getting very hot. I found that there were faulty components (the boost converter) because there was a short circuit (a few ohms up to 150 ohms depending on the board) between the one of the inductor connector pins and Ground.

I was asked to do testing to find the failure rate. I was able to isolate the bad manufacturer lots, but that's not the question here.

As I measured the resistance using a Keithley 2100, between the inductor pin and GND, if there was no short circuit, then the resistance would slowly (over 5-10 seconds) reach about 35 k Ohms. That seemed to be the "normal" value for boards without this failure.

But then a few days passed, and I started testing another lot. The same resistance that was 35 k Ohms was now showing up as 50 k Ohms and then slowly rising, never seeming to fully settle.

I understand that there are effects that cause this. Parallel resistances, parasitic capacitance and inductance, stray current paths due to contamination, other things that sound like bullshit when you try to explain to your boss.

What is up with these changing readings? Have you experienced this? Do you have any insight into how to explain this to other people when I hardly can explain it to myself? Any comments would be welcome.

[schmitt trigger]

Many moons ago, we had a failure of a board that controlled the proportional hydraulic valves used in huge construction machinery. The failure would cause unintended movement.

Unintended movement in a powerful machine is extremely dangerous. In a split second can easily maim or kill an unwary person, or cause extensive damage.

As such, a deep evaluation was performed. Long story short, a dendritic growth between an internal power plane and a via carrying a signal.
The investigation took a lot of time and involved using all sorts of technical analysis. Was expensive too, and required assistance from two external labs.

Just an example. YMMV.

[mtwieg]

If this node is connected to the inductor of a boost converter, that means that this node is also DC-shorted to the input supply voltage of the boost converter. And I'm guessing this net is connected to many components, including resistors, caps, diodes, and complex ICs.

In that case, it's not surprising that the measured resistance is highly inconsistent. It will likely depend on temperature, and on the test current/voltage used by the DMM (so different DMMs may give different answers). The polarity of the DMM's leads could also change the measurement greatly.

But 35-50K is still far more than the 0-150ohm you see with a bad board. So such inconsistency might not be an issue if you just want to identify damaged board.

[Manul]

I prefer measuring node signatures with IV curve tracer. Measuring resistance is not consistent and not as informative. Using a DMM diode test mode (both polarities) is also a useful technique if full IV signature is not an option.

[Rerouter]

Also for IC outputs and inputs, using diode mode can sometimes be a very good metric of good / bad

An overcurrent event may not kill the pin, but it does change things.

It depends on how far into the weeds you want to go, but an I/V can still tell a clear story on a complex net, as each components effects usually have different thresholds.

[harerod]

... IV curve tracer ...

What a coinicidence, I talked about this with an EMS partner just yesterday. Out of curiosity - what tool do you use? Thanks.

[schmitt trigger]

I used the Huntron Tracker 2000. It has been out of production for a while.

https://huntron.com/products/tracker2000.htm

[mtwieg]

Curve tracer would probably be much more useful, but only if its sweep rate can be made very very slow.

[temperance]

As I measured the resistance using a Keithley 2100, between the inductor pin and GND, if there was no short circuit, then the resistance would slowly (over 5-10 seconds) reach about 35 k Ohms. That seemed to be the "normal" value for boards without this failure.

The current injected by the DMM in resistance mode is slowly charging the input capacitor of the boost regulator.

The variation in readings between now and two days ago has something to do with those capacitors not being completely discharged. What was the start value? The start value was probably also different from the value measured two days ago.

Me thinks you should have been able to work this out on your own if your job is troubleshooting electronics.

[zapta]

I used the Huntron Tracker 2000. It has been out of production for a while.

https://huntron.com/products/tracker2000.htm

Will this do the job?

https://www.hackster.io/diyguyChris/revolutionize-troubleshooting-build-your-own-circuit-tester-fa6aa9

[Gyro]

Will this do the job?
...

Not if the circuit has significant filter capacitance (likely as it is a boost regulator). The basic Octopus circuit makes a handy component tester but is not so good for circuits, or in-circuit.