Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

How to diagnose ESD chip damage?

Status
Not open for further replies.

Cicero

Active Member
Wondering if any of you folk have some suggestions for me:

We use(d) a Linear Tech LTC485 for our RS485 comms, and over a few months have had several board failures, and a few field failures. Swap the chip and its all good again - the majority of the time. We use the DIP8 package, and so have placed an IC holder because it happens so frequently. It is the only part we have these problems with as well.
More recently though, the failure rate has progressed passed the point of thinking its just coincidence/normal wear-and-tear, and I had to get involved to find out why, and resolve it.

My gut feel was that its super sensitive to ESD, however its purely an 'educated' gut feel. I'd like to have some decent data to back myself up, but lack a test process to get this data. Most of our failures are during the manufacturing process, and picked up during first test of the 485 comms. Field failures are minimal, but still happen.

We have taken some additional measures to protect against ESD in the manufacturing and handling process, and have also purchased a different manufacturers part which can withstand harsher conditions. So far so good on that, and no failures, so people are happy.

But I've still got that nagging feeling that I want to know if it is in fact ESD or not, and I'd like to investigate further.
I was wondering what any of you have done in a similar situation?
 
Can you use a sacrificial chip ot two and deliberately abuse them with controlled static to see the effect?
Have you ruled out heat as a cause of failure?
 
Can you use a sacrificial chip ot two and deliberately abuse them with controlled static to see the effect?
Have you ruled out heat as a cause of failure?
Hi Alec, yeah we have about 200 of em in stock, and we dont want to use them for any products. So they're available for abuse.

It can't be heat because there's no overvoltage conditions/prolonged heat time even. We can literally pull a new one out a tube, place it into the board and it'll malfunction. Out the tube failure.

So do you suggest just testing a load of them in a controlled fashion? Maybe hire an ESD gun or something.
 
Sure:
LTC485.png
 
Yeah true, that does make me uneasy, usually I'd put on tranzorbs and polyfuses on the inputs, but this is a legacy board and worked perfectly for ages with the ST485 (now obsolete). Change to the LTC and we have these issues.

Not sure exactly when it blows in the process, but since its a comms device, we only notice its blown when connected and attempt comms.

I've scoped the TTL output from a few of them, some fail to nothing coming out, others just can't keep the output latched high...so you get no high bits. When it's meant to be high for a few microsec's, it only keeps it high for 300ns. So inconsistent failures as well.
 
You could make a daughter board with connectors that fit the DIP socket, and add additional protection for the chips. Like you said, some transorbs or transils, a bit of capacitance etc. should give you a better protection than a naked chip connected straight to long cable runs.
 
Have you contacted LT and asked for their input? I have found them to be quite useful in cases like this. Otherwise, getting an ESD gun and comparing LT with the other manufacturer would be my next step.
 
this is a legacy board and worked perfectly for ages with the ST485 (now obsolete). Change to the LTC and we have these issues.
That's a move from a bipolar chip to a CMOS chip. Although the LTC may have TTL compatible inputs, won't they be higher impedance than the old chip and hence more susceptible to noise pickup?
 
You could make a daughter board with connectors that fit the DIP socket, and add additional protection for the chips. Like you said, some transorbs or transils, a bit of capacitance etc. should give you a better protection than a naked chip connected straight to long cable runs.
Yeah worth a shot. I'm gonna make a few units up with standard protection and without. Trial them for a while and see how it goes.
Have you contacted LT and asked for their input? I have found them to be quite useful in cases like this. Otherwise, getting an ESD gun and comparing LT with the other manufacturer would be my next step.
I haven't yet. It might be something I'll pursue after I've done more investigations.
That's a move from a bipolar chip to a CMOS chip. Although the LTC may have TTL compatible inputs, won't they be higher impedance than the old chip and hence more susceptible to noise pickup?
I think this may be the crux of the issue. It seems the move to CMOS reduces the ESD resilience to what they mention as though a feature, of 'protection of up to 2000V'. As far as ESD is concerned...thats pretty poor in my opinion. I'm still concerned it (seems) so susceptible to handling though, and if so, how come we're the only people who seem to have issues with it?
 
Looks like my guess on higher input resistance for the LTC was wrong. According to the datasheets and app notes, "ST485 has a Rin greater than 40kOhm" and LTC485 has a receiver input resistance of 12kΩ (6k for the -M version).
 
FWIW (somewhat related): I did have one issue with a commercial product that controlled a stepper motor interface from a PC to an X-ray machine.The manufacturer wouldn't provide any info except the RS232 chips blow and they are socketed, so you can replace them. I added opto-isolation and the problem went totally away.

For the very few Rs485 system, we never did a lot of hot-plugging.

It's not going to help you

I think I'd agree. Contact LT. Could you test setup be causing failures?

Remove those 100K resistors. Protection is supposedly built in if the inputs are open. The IC is supposedly protected against ESD, but the 100K resistors might actually negate that. The next think I would try would be reverse biased Shockley diodes for the 100 K resistors which is typical OP amp protection.
 
KISS, how would the 100k negate any sort of ESD protection?
Also, i don´t think Shockley diodes are readily available any more. Schottky perhaps?

Cicero, the 2000V ESD protection is probably true, but note that this is usually only related to the HBM - human body model, and 2kV is just a tiny voltage that may appear when handling the part, definitely not anything like the energies coming from long cables. See https://en.wikipedia.org/wiki/Human-body_model and https://www.onsemi.com/pub_link/Collateral/TND410-D.PDF for reference.
 
KISS, how would the 100k negate any sort of ESD protection?
Also, i don´t think Shockley diodes are readily available any more. Schottky perhaps?

Cicero, the 2000V ESD protection is probably true, but note that this is usually only related to the HBM - human body model, and 2kV is just a tiny voltage that may appear when handling the part, definitely not anything like the energies coming from long cables. See https://en.wikipedia.org/wiki/Human-body_model and https://www.onsemi.com/pub_link/Collateral/TND410-D.PDF for reference.
Yeah exactly, 2kV is piss poor!
 
Yeah exactly, 2kV is piss poor!
Not sure we understand each other, the 2kV HBM is only relevant to handling in the factory as the chip is soldered to the board. It is very much irellevant to any real-world use of the part.
 
Not sure we understand each other, the 2kV HBM is only relevant to handling in the factory as the chip is soldered to the board. It is very much irellevant to any real-world use of the part.
Not sure why you think we're misunderstanding each other, I'm agreeing with you, in my opinion it shouldn't even a be listed as a feature of the device.
 
This thread has gone from diagnose ESD damage to protection. Depending on the devise I X-RAY the component to determine if any internal components are subject to damage. This doesn't always identify the damage and I know sectioning has also been used.

Jason.
 
Hi Cicero,

The LTC485 is a bit on the delicate side; the drivers and receivers maximum input voltages are only +-14V. And the LTC485 has no ESD rating, unless I have missed it.

The LTC2862 RS485 transceivers in the datasheet below are a bit more manly; the drivers and receivers maximum input voltages are +-60V, which is more practical. And the ESD rating is 15KV on the RS485 line pins.

The LTC2862 is pin compatible with the LTC485.

But, it is unwise to connect any chip to the outside world without appropriate protection- Vishay TranSorbs are normally the best protectors. With the LTC2862 a single 50v bidirectional TranSorb from each RS485 line pin to the chip 0V pin should do the job nicely.

If you stick with the LTC485 do the same thing but use 13V TranSorbs- this is just a very rough assessment- you would need to do a full analysis, because the operating voltage window is so close to the not-to-exceed voltage window.

Remember though that a considerable current may flow, momentarily, when the TranSorbs are doing their job, so the wiring and PCB races need to be substantial.

One last point, there needs to be a ceramic decoupling capacitor (X7R) directly across the supply pins of the transceiver chip

spec

UPLOADS
 

Attachments

  • LT_LTC2862&3&4&5_RS485_transeiver_60V_protected.pdf
    381.4 KB · Views: 118
  • Vish_diode_protection_selector.pdf
    1 MB · Views: 142
  • LT_LTC485_RS485_transeiver_low_power.pdf
    543.5 KB · Views: 126
Last edited:
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top