Hi Guys,

I'm designing an industrial system with a camera and a LED array and I'm having some trouble with the quality of the signal. I add a picture of the signal from the scope. The "A" is the Trig IN (entering the camera from an optocoupler) and "B" is the LED array that flashes when the Camera take the picture.

I have a microcontroller that send a signal to an optocoupler that triggers the camera. From the optocoupler to the camera it's a not shielded, not twisted cable (about 30 cm). Everything is on the same reference, which is grounded in a metal cage, which is grounded with the plant(factory) ground.

I've heard alot of things about impedance matching to "terminate" my transmission line... Adding LC filter to reduce Slew Rate that cause the high frequency signals (rising edge) to reflects in the wire.

I'm quite lost in all of this.

Any help would be appreciated !

Thanks,

-Yannick.

 

I can see the leading edge of the 10 millisecond pulse overshooting to -6V
or is that 1v below zero? Can't find the zero reference. The black rectangles?

I can only think of a hardware fix. Add a clamping diode so the negative going pulse can't go more than one diode drop below ground.

Anybody else got an idea?
Some sort of drawing of the circuit might help. It's difficult to spot wiring mistakes on a scope display.

edit: am I barking at what you consider to be the problem?
edit: Please post your location as "barking" means something entirely different in British.

 

I think you are describing glitches from switching the LEDs getting into your video signal. This reply is based on that understanding. Most likely, at the frequencies you are interested in, impedance matching won't make a difference, but keeping the energy out of the camera will.

Here are some things to consider. In addition to #12's good advice, you may need to just do one or two of these, and you might have to do all of them in order to obtain satisfactory results.

1. Make sure that the the current from the LED does not return through the chassis -that it only flows through the connecting wires. To keep the current within those conductors:

A) Make sure that the LED circuit only connect to ground at the power supply decoupling capacitor on the power supply or control unit (the LED does have its own decoupling capacitor doesn't it?).

B) Make sure the LED circuit power supply decoupling capacitor has some impedance or resistance between it and the rest of the circuitry, and make sure that there is a good high frequency decoupling capacitor on the other side of that impedance (such as a good quality 0.01 uf multilayer ceramic capacitor + an electrolytic capacitor in parallel).

C) Yes, definitely twist those wires, even though they are only 30 cm long.

D) Put a ferrite bead around the cable, preferably near the end of the cable nearest the circuit that makes the pulses.

E) If the above measure's don't solve the problem, you can try using a shielded twisted pair and grounding only one end of the shield to the chassis (experiment to find out which end works best). Adding the shield is likely to make the situation worse or at the least, more complicated, so try this last.

2. You can control the rise time, as you suggested. This is best done by placing an inductance or resistance in series with the pulse as close to the pulse generator as possible, then placing a small capacitor from the output of the inductance or resistance to the ground on the power supply decoupling capacitor in the pulse generator. The values depend a lot of the how much you have to "roll off" the signal to make the glitch acceptable, how the pulse generator reacts to this additional load, and how this low pass filter affects the behavior of the LED -if the LED is being strobed, then too much filtering would reduce the effectiveness.

3. Change the timing of the switching pulse. I had a similar problem, when a microcontroller updated some indicator lights on a broadcast camera once a field. which caused a visible glitch in the camera's output. Changing the firmare so that the indicators were updated during vertical blanking solved the problem. If you camera has a blanking interval, try changing the timing of the LED switching signals so that they LED drive changes during blanking time(s).

4. Stop turning the LEDs on and off. If they LEDs are being strobed to control total exposure, change to either linear DC drive or heavily filtered pulse width modulation instead of driving the LEDs with pulses.

 

Thanks alot for that reply !

I still have some questions about your reply.

A) Make sure that the LED circuit only connect to ground at the power supply decoupling capacitor on the power supply or control unit (the LED does have its own decoupling capacitor doesn't it?).

If everything is on the same ground, which is the chasis (metal box), it's ok ?

B) Make sure the LED circuit power supply decoupling capacitor has some impedance or resistance between it and the rest of the circuitry, and make sure that there is a good high frequency decoupling capacitor on the other side of that impedance (such as a good quality 0.01 uf multilayer ceramic capacitor + an electrolytic capacitor in parallel).

You mean that I must have a decoupling capacitor for my LED circuit and that I must have another capacitor between my LED circuit and the rest of my circuit ?

C) Yes, definitely twist those wires, even though they are only 30 cm long.

When you twist, you twist your differential signal, right ? If I do have 5 Sinking Output in my cable (single-ended signal), how do I twist them ?

D) Put a ferrite bead around the cable, preferably near the end of the cable nearest the circuit that makes the pulses.

About that ferrite, how can I know what to use ?

2. You can control the rise time, as you suggested. This is best done by placing an inductance or resistance in series with the pulse as close to the pulse generator as possible, then placing a small capacitor from the output of the inductance or resistance to the ground on the power supply decoupling capacitor in the pulse generator. The values depend a lot of the how much you have to "roll off" the signal to make the glitch acceptable, how the pulse generator reacts to this additional load, and how this low pass filter affects the behavior of the LED -if the LED is being strobed, then too much filtering would reduce the effectiveness.

For that filter, if I use resistance I'll "round" my Squarre signal and if I use inductance it will be linear, right ?
If I have a rising time of 4us without any filter and I want to increase that time, to reduce high frequency signal (to reduce reflection in wire) how can I "calculate" that value of the capacitor. What the link between Low Pass Filter in (Hz) and my rising time. (Fourrier ?)

Thanks for eveything !

I know it's quite over my competence, but isn't the best way to learn ?

-Yannick.

 

On programming forums they work with posted code. We're an electronics forum, so we speak the language of schematics. Please post yours.

 

Hey !

I'll try to get the schematics, or to "draw" them. Before I provide those schematics, if anyone could answer the general questions, it would be nice...

- Is it ok to ground everything on the chasis of the metal box. (Microcontroller / Camera / StrobeController / LEDs / ...)
- Can I mix AC / DC grounds ? (Put every grounds and every V- on my chasis)


Thanks,

-Yannick.

 

Hey !

- Is it ok to ground everything on the chasis of the metal box. (Microcontroller / Camera / StrobeController / LEDs / ...)
- Can I mix AC / DC grounds ? (Put every grounds and every V- on my chasis)


Thanks,

-Yannick.

Normally I would say yes but this is precisely why we want a schematic. We don't know what you mean by (see Bold Red) this. Do you mean signal Gnd, AC mains Gnd.....We see a lot of crazy stuff here, so it's reassuring to know when we're in sync with you.

By the way, since a short length of coax is cheap, why use twisted wire? UTP technology is balanced. Once you unbalance it (by grounding one conductor) it is no longer a transmission line. It's just a length of wire. This length of wire can take on the properties of inductors and capacitors. For fast rising edges and high frequencies, it can look like an antenna.

 

I'm trying to get the schematic for the circuit, I'll post them later today.

For the ground, I'm talking about GND (from power supply) and V-. For every devices. Is it ok to take the GND from my PowerSupply and put it on the Chasis and all the V- from my devices and put it to the chasis, and only take the V+ from my PowerSupply.

Thanks.

 

No, I don't see any problem with a common ground.