what is the best way to protect sinking inputs

Thread Starter

YazeedM

Joined Jan 21, 2016
6
I have a controller with sinking inputs ( should be grounded to activate) , I will use the system in a noisy environment, I want to protect these inputs from EMI and voltage spikes, what is the best way to do that ? optocoupler ? LPF ? Schmitt triggers? any advice will be appreciated
 

ErnieM

Joined Apr 24, 2011
8,377
An opto coupler should keep any hash isolated from your secret circuit, but this of course exposes the opto coupler to the same hash.

For any other ideas please disclose your secret circuit and the expected noise.

And welcome to the forums!
 

AnalogKid

Joined Aug 1, 2013
10,986
Your list of possibilities combines circuit protection with signal conditioning techniques.
What is the input voltage range?
What is the input impedance?
What is the signal source?
What is the source impedance?
What is the input pulse width?
What is the connection wire length?
What is the source of the radiated noise?

ak
 

Thread Starter

YazeedM

Joined Jan 21, 2016
6
An opto coupler should keep any hash isolated from your secret circuit, but this of course exposes the opto coupler to the same hash.

For any other ideas please disclose your secret circuit and the expected noise.

And welcome to the forums!
LOL ErnieM, its not a secrect circuit, what I have here is Galil DMC-21x3 motion controller based on motorola 32bit microprocessor, this controller has digital IOs and limit switches inputs. My real concern is about the limit switches, they are TTL (0-5 Volts) pulled up internally to 5VDC through 4.7kOhm resistor. They should be Grounded to change their state.

So there is no real circuit here, its just the inputs from the controller, I need to make this filter/insolation/protection cuicuit in order to connect the Inputs to this curcuit before feeding them to the controller.
 
Last edited:

Alec_t

Joined Sep 17, 2013
14,280
A starting point could be the use of reverse-biased Schottky diodes to clamp the input signal to the 5V and ground rails.
 

Thread Starter

YazeedM

Joined Jan 21, 2016
6
Your list of possibilities combines circuit protection with signal conditioning techniques.
What is the input voltage range?
What is the input impedance?
What is the signal source?
What is the source impedance?
What is the input pulse width?
What is the connection wire length?
What is the source of the radiated noise?

ak
The inputs are TTL (0-5 Volts) pulled up internally to 5VDC through 4.7kOhm resistor. They should be Grounded to change their state. and they could be carried on a up to 30 meters cable in a very noisy environment where there is a lot of machinery that operates on strongly varying currents, plus it is a motion controller so there are many stepper motors.
 

hp1729

Joined Nov 23, 2015
2,304
A starting point could be the use of reverse-biased Schottky diodes to clamp the input signal to the 5V and ground rails.
That (diodes) sounds like the easiest way. Optoisolators would require the input side be powered, not just a switch to ground. But optos are a superior result. Power glitches and static discharges just can't get through at all. Environment? Carpeted floor might suggest optos. Cement floor suggests diodes.
 

OBW0549

Joined Mar 2, 2015
3,566
The inputs are TTL (0-5 Volts) pulled up internally to 5VDC through 4.7kOhm resistor. They should be Grounded to change their state. and they could be carried on a up to 30 meters cable in a very noisy environment where there is a lot of machinery that operates on strongly varying currents, plus it is a motion controller so there are many stepper motors.
I can think of five possible things you could do to minimize the noise problem:

1) Wherever possible, keep your input signal cables as far away as you can from AC power wiring and wiring carrying switched high currents, such as the drive signals powering your stepper motors. Run them in separate cable trays, if you can.

2) If you can, run your input signals through shielded cables rather than through unshielded wiring.

3) Make your motion controller inputs less sensitive to high-frequency noise by passing the input signals through lowpass filters located close to the controller. Even something as simple as a single capacitor (something between 0.1 μF and 10 μF, I'd guess) from input to ground might help. Don't let your inputs respond any faster than they need to.

4) Make the inputs more resistant to interference by making them harder to drive. Shunting the existing 4.7 kΩ resistors with external 470 Ω pullup resistors, for example, will increase the input contact closure current from 1 mA to 10 mA, but it will also make it more difficult for coupled interference to give you false inputs.

5) Consider powering all machinery, and your motion controller, through separate line filters. In some environments, this can help quite a bit.

I'm not big on optocouplers as a noise-control measure; they'll give you galvanic isolation when that is needed, of course, but for noise control they tend to just move the noise susceptibility to a different location-- i.e., the optocoupler input. Diodes will provide clamping, which might help when combined with (3) and (4) above.
 

AnalogKid

Joined Aug 1, 2013
10,986
As noted above, noise and transient control are two different things. For transients, a 5.1 V zener diode across each input will have the same effect as two Shottkey diodes, one to each rail. Can the controller respond to the individual transitions in switch bounce, or does it already filter them in either hardware or software?

ak
 

Thread Starter

YazeedM

Joined Jan 21, 2016
6
Thank you all ! :) I didn't expect that number of responses :).

First:
Yes I'm using the same exact cable to carry the motor phases current (up to 5 Amps, 20 KHz PWM) from the stepper motor driver and the limit switches signals from the mechanical switches to the controller box ( contain the controller and eight motor drivers) why ? because I don't want a messy cabling around the place. the cable is an 8 core 18AWG cable, twisted pairs, individual shielding (100% foil coverage and 90% tin braid coverage) and an overall shielding (100% foil coverage and 90% tin braid coverage). attached is a photo of the cable drawing. So there will be 8 cables going out of each box running on the same cable tray. The floor is epoxy, though no cables will be on the floor.

Second:
There is no any kind of filtering performed by the controller. I only can change the polarity of the limit switch, active H / active L
 

Attachments

Thread Starter

YazeedM

Joined Jan 21, 2016
6
I just recevied this answer from one of the engineers at Galil MC( the controller manufacturer) :

"The optos are really current-sensing inputs, as opposed to voltage. EMI disturbances tend to be low-current events, whereas it's not uncommon to see voltage spikes of 100V or more. Also, the opto-isoaltion is "slow" so any very narrow noise spikes are effectively filtered out. Nothing is perfect, but optos do a pretty good job of protecting the controller's expensive components and avoid false IO triggers. "
 

hp1729

Joined Nov 23, 2015
2,304
Thank you all ! :) I didn't expect that number of responses :).

First:
Yes I'm using the same exact cable to carry the motor phases current (up to 5 Amps, 20 KHz PWM) from the stepper motor driver and the limit switches signals from the mechanical switches to the controller box ( contain the controller and eight motor drivers) why ? because I don't want a messy cabling around the place. the cable is an 8 core 18AWG cable, twisted pairs, individual shielding (100% foil coverage and 90% tin braid coverage) and an overall shielding (100% foil coverage and 90% tin braid coverage). attached is a photo of the cable drawing. So there will be 8 cables going out of each box running on the same cable tray. The floor is epoxy, though no cables will be on the floor.

Second:
There is no any kind of filtering performed by the controller. I only can change the polarity of the limit switch, active H / active L
Cables look good.
 

nsaspook

Joined Aug 27, 2009
13,081
I just recevied this answer from one of the engineers at Galil MC( the controller manufacturer) :

"The optos are really current-sensing inputs, as opposed to voltage. EMI disturbances tend to be low-current events, whereas it's not uncommon to see voltage spikes of 100V or more. Also, the opto-isoaltion is "slow" so any very narrow noise spikes are effectively filtered out. Nothing is perfect, but optos do a pretty good job of protecting the controller's expensive components and avoid false IO triggers. "
Very good. I've used these types of devices in many high energy EMI isolation designs for standard low speed switching inputs.
http://www.vishay.com/docs/83646/ild1.pdf

I usually also socket this part as I've seen them smoke internally while still protecting the controller inputs from high energy events.
 

Thread Starter

YazeedM

Joined Jan 21, 2016
6
Howe come you left it for 12 posts before mentioning Galil??
I have used Galil products for <20yrs now!
Is this to a Motion controller or their PLC?
Max.
Hahaha Max, I have said that it is Galil in the 4th post. Anyway it is Galil DMC-21X3 motion controller. :)
 
Top