adding ESD protection

[circuitmangler]

I have a device that was manufactured in the early 90's that uses LS7166 quadrature decoder chips. The AB inputs of the LS7166s were connected directly to external rotary encoders using DIN connectors. After many years of use and repeated plugging and unplugging of the DIN connectors almost all of the AB input pins show signs of (presumably) ESD damage - they measure low resistance between the input pin and Vcc.

I would like to re-implement the device using a modern micro-controller. What measures should I take to protect the uC's input pins? Besides protection circuitry, is there a better type of connector to use?

Also, I'm curious to know if the LSI LS7166's chips are more or less prone to ESD damage than CMOS is.

Thanks!

 

[IRstuff]

Putting resistors in series and/or parallel to the inputs could help.

While it's generally not recommended for electronic circuits, connecting the circuit ground of the encoder to the ground of the controller will get rid of a LOT of apparent ESD problems. If your encoder is not powered by the controller, you could also be getting AC power-on transients that will also go through the inputs.

TTFN

FAQ731-376

 

[geekEE]

If you can put a series resistor in your signals, that will help. Cooper Bussmann makes some ESD protection devices that can go on each pin.

LS devices should be less prone to ESD damage than CMOS devices. I believe the geometries are larger and harder for a large voltage to break it down.

If you have a connector with a grounded shield, that may help some.

 

[IRstuff]

The inputs of an LS device are comparable in size to CMOS. Almost everything built since the late 80's use zener and field breakdown devices under the pads. In fact, some CMOS devices, by virtue of their additional semiconductor layers, incorporate SCR input protection that can handle ESD in the multi kV range.

Nonetheless, the devices are not designed to handle repeated and chronic ESD events. This has to be dealt with in the circuit configuration and usage.

So far, you have no evidence that it's ESD, and it might simply be a bad circuit configuration. We've had LS differential receivers that would routinely short because the receiver and transmitter were on different power supplies with different ground potentials. The different potentials would get equalized through the differential devices and the with 5 power cycles, they would die.

Another choice is to use optoisolated inputs. The input devices are isolated from the outputs and will be able to tolerate ground potential differences.

TTFN

FAQ731-376

 

[circuitmangler]

IRStuff - in my case the encoder is a simple led/phototransistor and is powered by the decoder device.

 

[CarbonWerkes]

Circuit-
Im not exactly clear on your topology, but for sensor IO I have had good luck with a part from Littelfuse called the SP720. It is a packaged diode/scr array wth 14 independent channels. The beauty of it is that the logic is based on the diode threshold to V+ and V-, so if for example you have a 5V MCU and use this same power supply for the SP720, you will see the device clamp any signal over 5.7V or so, which is normally not a problem for the MCU's IO pins. If it is, you can always use a lower V+ for the SP720 to clamp down at whatever level (I think it is rated 2V to 30V for operating supply). And it is very fast (2ns rise). Another benefit is that it has low impact on a circuit in terms of capacitance or leakage- but probably not an issue for you. So all you have to do is put the inputs in parallel with you signal lines, and perhaps have some series resistance before that parallel connection if there is risk of high currents.

Anyway, I have found this a very convenient and cost effective part.

Regards
R

 

[circuitmangler]

CarbonWorks - thanks for the lead. I'll look into it.

 

[GonzaloEE]

You may also use CeraDiodes from EPCOS to help clamping ESD at any I/O pin.

Good Luck!
Gonzalo

 

[IRstuff]

Almost all CMOS devices have those diodes built into every input.

TTFN

FAQ731-376

 

[GonzaloEE]

That's right, though most CMOS clamp diodes are unidirectional, so that any spike higher than VCC is routed to the power rail, likely affecting nearby ICs. CeraDiodes are bidirectional, and will clamp any +/- spike to ground.

Regards,
Gonzalo

 

[CarbonWerkes]

IR-

In my experience, some newer MCUs do not have protection on analog-capable pins- presumably because of the impact on the ADC. Strange, because most ADC chips seem to have this natively.

CM- I would just make sure the MCU option in question is protected fully before you make a decision on the need for supplemental ESD handling.

btw- an upside of the SP720 is that it will clamp significant current (2-3A) at a diode drop from VCC. This gives you a lot of flexibility as to where to place this device- you are not limited to sticking this post voltage divider in a 3V MCU application for example. The downside it, it needs to be powered to operate. Easy to manage with a supercap or whatever- the thing has near zero current draw.

Best,
R

 

[IRstuff]

Any leakage current can affect an analog measurement, while it wouldn't necessarily affect a digital input.

However, the quadrature decoding is a digital counting function.



TTFN

FAQ731-376