ESD protection for I/O of Shift Registor

Hi,

there are non conductive solutions, like RF, IR, fiber optic...

I don´t want to search for the datasheets (If you want us to use them, please provide links) .. but I guess you need a couple of signals.
--> We don´t know whether one direction only or both direction signals.
--> We also don´t know any timing / frequency / data rate...
--> Why not UART? (to reduce wire count)
--> we don´t know whther there is just one cable, and just one receiver board with 80 relays, or ar there 10 cables with ten receiver boards with 8 relays each...
--> we don´t know whether the cable is fixed or are there connectors at the cable
--> we don´t know whether there is only signals or also powee supply in the cable. The whole power supply system may cause new problems

If it was my application (wired) I´d probably go for UART with RS485/RS422 differential signaling over one single pair

Why:
* If you have clock and data signals you need to take care of timing with respect to each other. UART does not suffer form this.
* UART could be simply used with IR or fiber
* differential signaling does not suffer from different noise problems
* RS485/RS422 drivers are more rugged aginst ESD than simple microcontroller I/O.
* fixed cable, shielded (wihtout connector) does not need ESD protection .. as long as the signals on PCBs and wires can not be touched.

So a lot of "we don´t know" ... I come back if there is enough information for a more detailed discussion. I expect a draft to show the whole picture of signals and power supply.

Klaus

Logic IC have a certain degree of built-in ESD protection. As stated in manufaturer documents, it's primarily intended to handle ESD exposure during assembly and doesn't fully protect against ESD in the application environment, other surges and particularly not against overvoltage events with simultaneously applied supply voltage.

External circuit interfaces be should generally equipped with additional protection. In case of your remote logic interface, you can probably avoid that ESD is directly applied to interface nodes because they are not exposed. But you can hardly avoid that voltage transients caused by ESD to the cable screen or device enclosure affects logic operation. At least temporary false switching can be expected.

A "dumb" remote shift register interface can't be considered reliable in presence of electrical noise. Under circumstances, you can get away with sufficient fast update rate in combination with relay switching delay. I would prefer a fail-safe protocol ignoring corrupted switching commands. This requires however additional logic or a microcontroller on the remote board.

Hi KlausST,
I would like to thank that you have studied my question from top to Bottom. Here I would like to give some more details ypou needed

KlausST said:

I don´t want to search for the datasheets (If you want us to use them, please provide links) .. but I guess you need a couple of signals.

Click to expand...

Here is the link of the Data sheet -CD4094.

KlausST said:

--> We don´t know whether one direction only or both direction signals.

Click to expand...

It is One way communication! Microcontroller sends data and it will not receive / wait for any reply/feedback

KlausST said:

--> Why not UART? (to reduce wire count)

Click to expand...

Honest answer is to save 4-5$ in each PCBs. Board space will be little larger then have to add RS232/RS485 bridge ICs. I believed Sending data to drive shift register will be so simple and neat!

KlausST said:

--> we don´t know whther there is just one cable, and just one receiver board with 80 relays, or ar there 10 cables with ten receiver boards with 8 relays each...

Click to expand...

Only One received board with multiple relays, Say 80 Numbers approximately.

KlausST said:

--> we don´t know whether the cable is fixed or are there connectors at the cable

Click to expand...

Dsub -15 Connector is used on the both the sides of the cable to connect Transmitter Board (MCU ) and Relay driver board(CD4094 and ULN2803)
Before send/receive 74HC541 iC used to kill ESD

KlausST said:

--> we don´t know whether there is only signals or also powee supply in the cable. The whole power supply system may cause new problems

Click to expand...

Yes. You are right. 2 Number of Cables carryinh power for the Transmitter Board (MCU )
. One wire belongs to +15V and other one is 0

KlausST said:

If it was my application (wired) I´d probably go for UART with RS485/RS422 differential signaling over one single pair

Click to expand...

I could understand RS485 having advantages. But I strongly believ implementing shift register concept is so easy & cheapest solution

KlausST said:

--> We also don´t know any timing / frequency / data rate...

Click to expand...

For every 0.5 Second We need to send 10bytes of data.(10 x 8 = 80 Relays)

KlausST said:

Why:
* If you have clock and data signals you need to take care of timing with respect to each other. UART does not suffer form this.
* UART could be simply used with IR or fiber
* differential signaling does not suffer from different noise problems
* RS485/RS422 drivers are more rugged aginst ESD than simple microcontroller I/O.
* fixed cable, shielded (wihtout connector) does not need ESD protection .. as long as the signals on PCBs and wires can not be touched.

Click to expand...

I clearly understand RS485 is the best solution here!! I am curious to know what else could disrupt the data flow and potentially cause permanent damage to the hardware.. Please advice me the hardware parameters to be taken care other than RS485.

FvM said:

Logic IC have a certain degree of built-in ESD protection. As stated in manufaturer documents, it's primarily intended to handle ESD exposure during assembly and doesn't fully protect against ESD in the application environment, other surges and particularly not against overvoltage events with simultaneously applied supply voltage.

External circuit interfaces be should generally equipped with additional protection. In case of your remote logic interface, you can probably avoid that ESD is directly applied to interface nodes because they are not exposed. But you can hardly avoid that voltage transients caused by ESD to the cable screen or device enclosure affects logic operation. At least temporary false switching can be expected.

A "dumb" remote shift register interface can't be considered reliable in presence of electrical noise. Under circumstances, you can get away with sufficient fast update rate in combination with relay switching delay. I would prefer a fail-safe protocol ignoring corrupted switching commands. This requires however additional logic or a microcontroller on the remote board.

Click to expand...

Dear FvM.. Thanks for giving me a new knowledge about the ESD.. So happy to speaking with worlds finest people

FvM said:

I would prefer a fail-safe protocol ignoring corrupted switching commands

Click to expand...

Can you give a glimpse about the fail-safe protocol. During the communication actual clock/data will be transformed to anything due to external noise!! How we can idendify that data is not the actual one?

Some standards like RS-422 have "failsafe" modes but only for
spwcific hard fault conditions (open RX inputs). The PHY parts
only have to address likely physical-layer problems.

At 5m you could expect >10MHz operation per the RS-422 spec.

If you want "packet-level" checking, that's then some far-end
intelligence (parity, checksum, ???) and near-end resend on
command, I guess. Seems not likely to be worthwhile, unless
this desired integrity is part of the contract or you have other
reasons for a local brain to be sniffing the serial link.

I think I would be nervous following this approach. There may be noise between grounds at the sending and receiving end and there will almost certainly be a lot of noise as relays are turned on and off. Using a simple serial packet risks either the clock of data becoming corrupted and that is likely to cause wrong relays to be selected or deselected. Additionally, the whole data for all the relays has to be sent at each update, they won't be individually addressable

If you insist on this method, use differential signals for the data and clock. The differential driver and receivers (for example MAX488) have ESD protection and far better ground voltage/noise tolerance than simple buffers.

There is nothing technically wrong with your idea but those of us with gray hair (my avatar picture is almost 20 years old!) and real world experience in industry would appreciate how much can go wrong with this proposal.

Brian.

Prabhakarankft said:

Dsub -15 Connector is used on the both the sides of the cable to connect Transmitter Board (MCU ) and Relay driver board(CD4094 and ULN2803)
Before send/receive 74HC541 iC used to kill ESD

Click to expand...

Why send/receive? I thought it is just one way.

And HC541 is an on board driver, not for cables. No ESD protection, no cable impedance driver.
You can do what you want, but I see no benefit in using a HC541 at all.
If you use UART, a single RS485 interface, single pair cable (instead of multiple not_paired wires) it may be even cheaper, but much much more reliable.

If I´m not mistaken then you already made your decision, you just want to find someone to support your (in my eyes unreliable) idea.

Klaus

With 74HC series driver being approx 50 ohms and 200 ohm twisted unterminated like RS485 (which is) it is going to overshoot. Maybe close to 300mV absolute max limit. But damping the source by adding 100 ohm series will strike a balance between rise time and overshoot with 100 pF/m cable load will improve results.

If you can find 2T+2R versions of the 26C3x RS-422 parts, or
possibly better RS-485, those could let you throw high noise
margin differential data and clock, and loop them back to the
host for real time inspection of link integrity. What to do if bad,
that's on you. But you could do stuff like, inspect the readback
and if OK, send a latch "command word" before any relay
action (your SR outputs should not be direct to the coils, but
broadside latched after full word is delivered); then relay actuation
ought to have no consequence (you might even choose to gate
the clock of the latch and the shift register, with a one-shot from
the coil drive signal to prevent false clocking - if -that- can be
made noise-robust, itself.

D.A.(Tony)Stewart said:

With 74HC series driver being approx 50 ohms and 200 ohm twisted unterminated like RS485 (which is) it is going to overshoot

Click to expand...

I really thankful for giving extra knowledge about the 74HC series drivers! Sure I will add 100E Resistor in series..

KlausST said:

If I´m not mistaken then you already made your decision, you just want to find someone to support your (in my eyes unreliable) idea.

Click to expand...

I have to thank all the suggestions from you! I have changed my mind set to go with the RS-485 / RS422.

Prabhakarankft said:

I have to thank all the suggestions from you! I have changed my mind set to go with the RS-485 / RS422.

Click to expand...

O.K.

May I ask: Do you use SPI with 3x (4x) RS485/RS422 drivers and 3x(4x) pairs cable
or UART with 1x driiver and 1x pair cable?

***
Power supply wiring has big impact on whether it will work or not. ...How big the ESD/surge voltages and currents are. How big the GND_shift will be and the whole GND loop problem.

Klaus

KlausST said:

May I ask: Do you use SPI with 3x (4x) RS485/RS422 drivers and 3x(4x) pairs cable
or UART with 1x driiver and 1x pair cable?

Click to expand...

My actual plan is to use 6 wires. 4 for RS422 (A,B,Y,Z) and 2 wires used to carry the power supply for receiver(Relay side). Current consumption is very less about 200mA.. Power supply is used to power up the Logical ICs only( 15V --> 5V convertor (LM2576 IC used)

5 Metres of 6 Core Shielded Cables are used and both the ends are having Dsub Connectors . Sheild is connected to Metal parts(Outer Body) of the Dsub connectors!! Please correct me if anything goes wrong!

My long terms question is What are all the external noises can be in the application area? How we can study about all?

I am sure ESD can be injected whenever Human touch is possible! That is not possible in my application! Other than what will will the HW or will change behaviour of electronics?

Your 1st step is write design specs to fill a page for all important assumptions on environment, inputs, and output. {Electrical, Mechanical, Optical, Climatic) with the purpose, intended application.
Then detail all performance requirements.
Response time to address all 120 Relays. Max Data Rate. Single Relay toggle addressable or batch ? Error detection/correction method. Parity? Hamming code or CRC/ECC. Packet length ? 2 way communication for status?? Power fail detection? Power Recovery method all off or saved state? Backup battery ? Battery status? Failure detection? User interface ?

I did this 45 yrs ago for a Blank Brandt payload launch control with 15A 30A relays using extra contacts to give feedback (P&B brand). Then I used an HP9825 computer on LDDS over 1 mile to another to control the box with HP rack mount ADC/DAC/R-programmable Lambda PSUs etc. over HPIB (IEEE 488) There was one flaw in design. Sense contacts < 2A were not gold plated so when sequencing all Relays as fast as possible, it sounded like a misfiring gattling gun. So I put 10 uF Tanatulm caps across each sense contact that had pullup R, and problem solved. Then after I left company I found out the LD data set (LDDS) ( RS485) were causing havoc whenever SW radio from control tower was used. That EMI could have been fixed with CM choke e.g. Balun Ferrite clamps and pF suppression caps.

The only time I have experience RS232 problems was in the late 80's with 8 computers sharing a printer Mux box with ground & PSU noise from one PC.

I think as long as there is only one SMPS in the network, no issues. But with two, ground noise needs PE gnd to divert from shield. Ethernet uses a differential and common mode PHY to overcome these issues. You should have no problems with +/-5V RS-232 but include parity as a minimum and run at max baud rate. I would use RJ jacks instead if you have a lot, but DB-9 is OK Are these daisy chained to multiple stations? or just 1 box.

You should expect EMI from any/all power relay contacts. So consider STP wire or UTP and define reactive loads. L will cause switch off arcs and and C loads will cause surge ON impulse noise. Arrange cable at right angles to RS232 to isolate noise or either end of box. Then consider snubbers RC, RD, RZ type , depending on release time speed needs. (diodes slow turn off, so Diode +R or D+ Zener is faster) Or just use R or film C snubbers but cause leakage current on AC power contacts so keep < 1mA.

Have fun. But make sure you define your plan with all specs or expected performance 1st and list all possible interference tests to verify. This is your Design Spec and DVT Design Verification Test DVT plan.; a priori !! Make sure you think of Design for cost, reliability, testability, manufacturabilty and maintainability. Get local mentors to review your specs. Have a design Review and repeat as often as needed. Budget your time, cost and performance. Meet at least 2 of 3 on your 1st project. Be conservative and expect Murphy's Law to be happen.

D.A.(Tony)Stewart said:

I did this 45 yrs ago for a Blank Brandt payload launch control with 15A 30A relays using extra contacts to give feedback (P&B brand).

Click to expand...

Hi.. I am really a lucky person to hear your 45 years R&D history! Too many things are very new for me! I will understand all and I come back if I have any other questions!!

D.A.(Tony)Stewart said:

So I put 10 uF Tanatulm caps across each sense contact that had pullup R, and problem solved.

Click to expand...

Do you mean, You have placed across the Common and NO(Relay outputs) or across the relay's Coil?? Pullup resistor is bit confusing.. Possible share a hand sketch. I can gain extra knowledge and I am always eager to learn new stuff.. Sorry that discussion is moving to another direction! Since I have used n-number of relays might have to face other issues related to relay! So Iam preparing myself.

D.A.(Tony)Stewart said:

Be conservative and expect Murphy's Law to be happen.

Click to expand...

Murphy's Law is an interesting one to read!! Mistakes may happen and I am ready to accept. Taking the advice of those who know the answers to test everything can eliminate half of the mistakes

All my Relays had small signal parallel contacts to verify the primary contacts were powered ON or OFF at the same time. (useful for remote sensing when the application varies from AC to DC in voltage.) But 50 yrs ago they had not yet enforced the rule that all relays < 2A with silver alloy contacts MUST BE GOLD (usually flash finish) plated for ratings to prevent oxidation (rust). So my cap burnt off the nano-layer oxidation rust every time the sense contacts closed. The cap would last long than the relay from my calculation using MIL STD HandBook 883-B It must be H or Z by now.

Prabhakarankft said:

My actual plan is to use 6 wires. 4 for RS422 (A,B,Y,Z) and 2 wires

Click to expand...

Makes no sense. Neither for SPI nor for single direction UART.

Prabhakarankft said:

Power supply is used to power up the Logical ICs only( 15V --> 5V convertor (LM2576 IC used)

Click to expand...

Makes no sense either. I don´t think you can power your circuitry and 80 relays from 15V 200mA.
What voltage drop do you allow on the 5m cable?

Prabhakarankft said:

My long terms question is What are all the external noises can be in the application area? How we can study about all?

Click to expand...

To answer your questions you need to give the requested informations first.

Prabhakarankft said:

I am sure ESD can be injected whenever Human touch is possible! That is not possible in my application!

Click to expand...

You use DSUB connectors.... and you think ESD is not a problem? --> It´s not only the "high impedance" human body ESD .. but it´s the low impedance machine model ESD! Much higher currents should be expected.

Klaus

KlausST said:

Makes no sense. Neither for SPI nor for single direction UART.


Makes no sense either. I don´t think you can power your circuitry and 80 relays from 15V 200mA.
What voltage drop do you allow on the 5m cable?


To answer your questions you need to give the requested informations first.


You use DSUB connectors.... and you think ESD is not a problem? --> It´s not only the "high impedance" human body ESD .. but it´s the low impedance machine model ESD! Much higher currents should be expected.

Klaus

Click to expand...

We may be right or wrong if we made incorrect assumptions. DSR, DTR, Tx, gnd or Tx, Gnd, RTS, CTS or CD, Gnd, 15V, -3V lol

Maybe we shud expect students to list all design specs and list all assumptions and planned environmental stress tests.

Purpose, All I/O's, speed, functions, problems, Environment etc.

KlausST said:

Makes no sense. Neither for SPI nor for single direction UART.

Click to expand...

Yes It is not a Single Direction UART.. For Debugging purpose We have moved with duplex communication! May We can use in future

KlausST said:

Makes no sense either. I don´t think you can power your circuitry and 80 relays from 15V 200mA.
What voltage drop do you allow on the 5m cable?

Click to expand...

Separate Power Supply used to Power Up the Relays. We need to just pass the signals via Opto-couplers. So no connections in between relay circuits and this 15V power supply! it is purely to power up the Logical ICs and MCUs

KlausST said:

You use DSUB connectors.... and you think ESD is not a problem? --> It´s not only the "high impedance" human body ESD .. but it´s the low impedance machine model ESD! Much higher currents should be expected.

Click to expand...

I will look into MM ESD and will check the possiblities to arrest!! These terms are very new for me!!

D.A.(Tony)Stewart said:

We may be right or wrong if we made incorrect assumptions. DSR, DTR, Tx, gnd or Tx, Gnd, RTS, CTS or CD, Gnd, 15V, -3V lol

Click to expand...

According to Me 4 wires are for A,B,Y & Z( Dedicated for RS422) and rest of the wires are for carrying the Power supply +15V and GND

Prabhakarankft said:

We need to just pass the signals via Opto-couplers.

Click to expand...

Now there are optocouplers! The first time ... we are at post#18 and you refuse to draw a simple sketch. But bring new informations piece by piece.

I´m not playing this game. I´m out.

Good luck with your project.

Klaus

You have a duty to disclose all design specs before choosing a method then retrace all steps and assumptions to describe any problem if you want expert help.

Otherwise it's a pita. Don't be vague.


a priori

1. design specs.
2. Environmental stress and interference
3. processes
4. components

Don't choose #4 first . That leads to sub-optimal designs.


A posterpriori.

Tony