period measurement for calculating acceleration - high speed counter resolution

I want to measure the period of a 60Hz signal with .001 Hz accuracy resolution. The Schneider M430 high speed counter BMX EHC 0200 has a resolution of 1 us (so 1MHZ counter clock), which yields a resolution of 0.0036 Hz. This signal always varies by +- 0.0036 Hz. Using a filter with time constant of 33ms (two periods) averages out this noise and provides a decent signal, but doesn't achieve .001 Hz and I would prefer to add as little delay as possible.

I know there is an allen bradley high speed counter with resolution of 0.25 us so 4 MHz counter clock, and this would provide frequency measuring resolution of 0.0009 Hz, so I can expect the the measured frequency to constantly vary by +- 0.9 milliHz, so I still have 1 milliHz noise.

What PLC has the fastest counter for measuring the period of a square wave? Do I have to move away from PLC type devices for counters approaching 10MHz?

is this something you must do all the time or just once. There is test equipment that can do what you want easily.

Is the 60Hz signal a sine wave? Why do you then mention square wave. I am confused.
What are you really trying to do?

Peter,
This is measuring the speed for turbine control. The signal is 0-120V AC run through a signal conditioner such as http://www.actuationtestequipment.com/PTInterface/PT_Interface_Page.html to make it 5 or 24V square wave.

I want to use a double derivative PID and for this require a high resolution low noise speed signal.

This will be a permanent installation in a control panel in an industrial type environment.

Maybe I need to rephrase the question, what PLC has the fastest counter clock that anybody knows of and does period measurement? So far I know AB at 4 MHz, but I have heard that maybe a koyo has 6 MHz but can't find this model yet.

thanks for any hints

OK, now that I know what you are doing is there anyway to put an encoder on the turbine?

I wouldn't use that signal conditioner if you want fast response. There is a 3 pole Butterworth filter that will delay the true RPM. You may get what looks like 0.001Hz resolution but it won't be current. You did mention having little delay.

I know our motion controller is very fast compared to a PLC and we have a MDT interface that counts gated or PWM signals with a 240 Mhz counter but I wouldn't trust that signal conditioner to be fast. I think your best chances would be to do AtoD conversions on the attenuated +/- 8-10 volt sine wave directly and do some math.

In addition, our controller has a second or double derivative gain built in.

0.001Hz is pretty fine. One would need to look at where the voltage crossed 0. It would probably cross 0 between two samples. A linear interpolation could indicate when the voltage crossed 0. A trickier interpolation could be tried. For instances using two voltage samples before and after the zero crossing may provide better results.

Take a look at AB 1734-VHSC. It has a 5Mhz internal clock. It can be set-up in "Continuous/Rate" mode and will give you frequency of the input from your signal conditioner. Frequency resolution feedback can be set to .001 or higher.

Input signal must be square wave with 50% duty cycle (equal on/off time). Don't ask how I know.

The turbines are equipped with a 'toothed wheel' or gear wheel, say 60 teeth on 360 RPM turbine yielding a 60Hz signal square wave with proximity sensor or sine wave with magnetic pickup. Another one in memory was 14 teeth at 900 rpm so a 240 Hz signal. The problem is the gears are of varying quality, some will feature a double tooth, some will be out of round, some will have varying depths on the teeth, etc. Usually they are mounted on the end of the shaft and have a certain wobble back and forth as well as up and down.

Rather than have the turbine control solution depend on this unknown gear wheel I believe many electrical/control engineers revert to using the voltage signal from the generator as it is much more consistent.

The signal conditioners are indeed zero crossing detectors. Some filtering is required to remove power system harmonics, but yes that does add delay. The next time I have one on my desk I will measure this.

bce123 that looks interesting, thank you for the idea. I will explore the possibilities that piece of hardware allows for.

SC3128 thank you for the part number, it sounds promising. "Continuous" mode makes it sound like it is a free running counter. Many period measuring devices have been known to measure every other period, or reset the count every cycle, thereby throwing out half a count.

thanks all for the input

Omron CP1H (shoebox) has 4 axes 100k counter built in plus 10k outputs and PWM - also has 64 bit floating point maths if you need it.

VON_hydro, what are you doing with these readings? Even reading data across the back plane of a PLC will be slow and have delays. You are using the frequency to do what?

the most demanding application is control the speed of a turbine and generator by switching electrical load on and off using SCRs. the number of elements to be on or off is decided by the PLC, and then the elements are switched on or off at the zero crossing by external circuitry.

The PLC only has to calculate how many elements should be on for the next cycle every 16ms and then actuate the outputs.

I would like to have one go-to frequency measurement setup that will work for this application, control of power output every cycle, as well as traditional control of the turbine speed by adjusting the power input, which is slower and less precise.

The little Omron box has the counter speed you require. It also has interrupts both digital and programmed and is pretty quick. I use the cheaper one (CP1L) a lot for small jobs. A fairly full program for a swimming pool filtration plant scans in 3-4ms so I doubt 16ms would be a problem (1/2 of course).
Transistor outputs would fire SCRs pretty quickly too.

thanks bobb. another great recommendation. the woodward panel is looking really good btw

the most demanding application is control the speed of a turbine and generator by switching electrical load on and off using SCRs.

Click to expand...

So this is a speed control application. How fast can you switch the SCRs?

If I were trying to control the speed of the turbine I would first try to do the easy thing and use an encoder. If not I would generate a target sine wave for the turbine to follow. The difference between target sine wave minus and the actual sine wave from the turbine is an error that is used to compute the output. This could be done at 4Khz which much faster than using signal conditioners and transferring data across the back plane of a PLC.
All that other stuff just gets in the way and causes phase delay.

Originaally posted by Peter Nachtwey:

The difference between target sine wave minus and the actual sine wave from the turbine is an error that is used to compute the output.

Click to expand...

Peter, wouldn't that define the phase of the signals and not the frequency deviation? Or is the idea that you assume you can stay within a given phase range and the control will ultimately settle on an output that stabilizes phase error?

Keith