Hi all,

I'm looking help to calculate the best input filtering times for my application. I am using an incremental encoder which generates 1000 pulses per revolution. this is attached to a shaft which ,when counting, runs at 17.5 RPM. this works out at roughly 0.292 revolutions per second resulting in a switching frequency of 292 Hz. Is this correct? and if so should i increase my input filter rather than lower it. I used to have a problem that when I ran the encoder at speed it wouldn't count, then I reduce the input filter and it did count. but when i ask it to make a move of 2400 mm, it only moves to 2100 mm (approximately). Does this mean it is picking up noise from somewhere?

any help is much appreciated,

Attached is a couple screen shots of the current config.

Hi, I don't know if it is picking noise from anywhere. It is loosing pulses.

What kind of acquisition card do you have?

In general, if the base frequency of a square wave is 292Hz, the signal of that square wave have much higher frequencies. How much higher? It depends on how many higher frequency sinusoids are ok for your board not to loose pulses and the application to run smoothly. Let's say, rule of the thumb, at least 10 times the base frequency (so 2920Hz). This just to give you an indication.

Your input board will have a threshold to evaluate if the signal is 0 or 1. If you set your frequency too low, you loose pulses count.

SP2607

Hi all,

I'm looking help to calculate the best input filtering times for my application. I am using an incremental encoder which generates 1000 pulses per revolution. this is attached to a shaft which ,when counting, runs at 17.5 RPM. this works out at roughly 0.292 revolutions per second resulting in a switching frequency of 292 Hz. Is this correct? and if so should i increase my input filter rather than lower it. I used to have a problem that when I ran the encoder at speed it wouldn't count, then I reduce the input filter and it did count. but when i ask it to make a move of 2400 mm, it only moves to 2100 mm (approximately). Does this mean it is picking up noise from somewhere?

any help is much appreciated,

Attached is a couple screen shots of the current config.

Hello.

(1) Filter time

The filtering time is shorter than what is needed. I calculate 85µs should be okay. See point 2 for a detail you need to know.

Starting with the time setting you have there should not be a problem. Make VERY sure that the hardware configuration is downloaded. The CPU in the project diagram should be highlighted to ensure the hardware is loaded.

At times it is needed to use rebuild in the compile option. This can be followed through by download-and-reset option.

(2) Pulse per revolution

The specification says 1000 pulse per revolution for the encoder.
The PLC is using A/B pulse measuring method.
The PLC registers (1000 x 4 = 4000) triggers for each revolution.
The unit distance should be referenced to the 4000.

(3) Movement takes time to come to a standstill

Did you check what encoder count has been asked and how much you actually get after the movement comes to a standstill? This is important. 

Some mathematical calculation may be required to ensure that the movement comes to a stop at the point you need it to stop. (I assume that is what you are aiming for.

See what you can do with these hints for now.
Please provide feedback on what the progress is / isn't.

Regards.

Hi B. Edulis,

Thanks for the response. I have connected  a signal board 6ES7 221-3BD30-0XB0 to accommodate for high speed signals. I am assuming when you are talking about a threshold, you mean the input filtering? or is there another setting im missing?

qwazee

Hello.

(1) Filter time

The filtering time is shorter than what is needed. I calculate 85µs should be okay. See point 2 for a detail you need to know.

Starting with the time setting you have there should not be a problem. Make VERY sure that the hardware configuration is downloaded. The CPU in the project diagram should be highlighted to ensure the hardware is loaded.

At times it is needed to use rebuild in the compile option. This can be followed through by download-and-reset option.

(2) Pulse per revolution

The specification says 1000 pulse per revolution for the encoder.
The PLC is using A/B pulse measuring method.
The PLC registers (1000 x 4 = 4000) triggers for each revolution.
The unit distance should be referenced to the 4000.

Hi  qwazee

Thanks for the response and I will try these points out and get back to you. Just a few questions on point 2 though, Is it that every unit the %ID1000 presents is 1 of the 1000 pulses in a revolution or is it part of the 4000 pulses you are detailing? My measurement for length per pulse would have been based on that value receiving 1000 pulses per revolution and not 4000, but the samples that i have cut so far have not been that far wrong, only losing accuracy over an increased length.

SP2607

Hi  qwazee

Thanks for the response and I will try these points out and get back to you. Just a few questions on point 2 though, Is it that every unit the %ID1000 presents is 1 of the 1000 pulses in a revolution or is it part of the 4000 pulses you are detailing? My measurement for length per pulse would have been based on that value receiving 1000 pulses per revolution and not 4000, but the samples that i have cut so far have not been that far wrong, only losing accuracy over an increased length.

Remove the doubt and guesswork. Test it.
Turn the encoder one turn and see how many pulses you get.

I see you zero your encoder count.
Be VERY careful how / when you do it.
Doing so while the movement is still on progress will give you many problems.

Consider not restarting the count.
Alternative: Capture the count at that moment and use it as base.
countNow = countEncoder - countCaptured

~

 No, I meant an inner threshold for the logic of the board to acquire the signal as "1" or "0". Nothing you can set up directly, but you have to consider while setting up the filter time.

In fact, if you filter out the square wave too much, then the square wave will loose its "squareness" and become smoother and not so well defined, with more ripple, expecially around transition points (from 0 to 1 or from 1 to 0). This "uncertain" wave shape may fool the sampling of the board. The filter acts before the sampling.

SP2607

I am assuming when you are talking about a threshold, you mean the input filtering?

Hi Both,

After a bit of working, trail and error on the project I have finally fund the route cause of the problem. In the program I was calculating the distance per pulse using the sprocket diameter. Although this initially appeared to work. Upon further investigation I found that this was incorrect. I setup a test routine in which I looked for the Z mark from the encoder. made a mark on the conveyor frame, then ran the the conveyor for 500 pulses, then stopped and made another mark. this gave me my distance per revolution, and is the value i use to calculate the distance per pulse instead. I repeated this test several times just to increase the accuracy of the result.

Many thanks for both of your contributions. they helped a lot.