PI simulator

[green]

Been working on an excel simulator to compare with the bench controller with ground balance. I manually enter the calculated times from I10 into L5, can someone tell me how to auto enter them when I change the delay or sample times? Was going to include the zip program but I forgot how, if anyone wants it remind me how to do it.

[green]

Charted some data using the simulator. Wanted to see the effect of varying delay and sample times for low TC targets (small nuggets). I adjusted GB delay time with a 100usec GB sample for GB. For the low TC targets GB didn't effect the data. Maybe ground does.

[6666]

Was going to include the zip program but I forgot how, if anyone wants it remind me how to do it.

Hit reply
advanced
Manage Attachments:

[green]

zip file, hope someone can show how to auto load the row data.

[green]

I did the simulator in excel. Maybe someone with math skills could write some formulas to enter the data that would be easier to use. Think knowing what should happen would make it easier to troubleshoot the hardware.

[WM6]

Are this sites of any help?:

https://strokescribe.com/en/read-ser...xcel-2007.html

http://www.windmill.co.uk/mettler.html

http://www.instructables.com/id/How-...or-Scale-to-a/

[green]

Are this sites of any help?:

https://strokescribe.com/en/read-ser...xcel-2007.html

http://www.windmill.co.uk/mettler.html

http://www.instructables.com/id/How-...or-Scale-to-a/

Thanks for the sites. I'm trying to simulate what happens with my hardware controller. I forgot about adding the target and ground effect for the EF sample to the simulator. Can effect the output. Integrate amplifier out normal and inverted. Normal during target and EF sample time and inverted during GB sample time. Need to make all the sample times adjustable for experimenting. EF time= GB time-target time. Whether it's done with excel or formulas doesn't matter. Maybe adding the EF sample to the simulator would be worth while. The controller shows fixed times for all but GB delay time, a guess at what I thought might work. All of the times are adjustable with my bench controller. Want to compare bench results using different coils and amplifiers with simulator. Mostly trying to learn something.

[green]

Maybe I've been looking at changing target sample time wrong. Was thinking it doesn't have much effect, but changing sample time changes integrator gain. If I change the feedback resistor to keep the gain constant the signal amplitude decreases instead of increasing with increasing sample time. What is a good maximum gain for the integrator? Anything over 1 the integrator could saturate before the amplifier. Maybe I'm still looking at it wrong?

[green]

Thanks for the sites. I'm trying to simulate what happens with my hardware controller. I forgot about adding the target and ground effect for the EF sample to the simulator. Can effect the output. Integrate amplifier out normal and inverted. Normal during target and EF sample time and inverted during GB sample time. Need to make all the sample times adjustable for experimenting. EF time= GB time-target time. Whether it's done with excel or formulas doesn't matter. Maybe adding the EF sample to the simulator would be worth while. The controller shows fixed times for all but GB delay time, a guess at what I thought might work. All of the times are adjustable with my bench controller. Want to compare bench results using different coils and amplifiers with simulator. Mostly trying to learn something.

Using the bench controller I see one thing I did wrong. The EF sample time shouldn't be adjustable. Should equal GB time-target time, auto adjust when changing GB or target sample time. Maybe other obvious things I missed.

[green]

Modified the controller. Switch to enable or disable GB. EF time equals GB time-target time. Charted decay for California clay, US nickel and a US quarter(target recording-no target recording). Set times after coil turn off. 6usec delay, 6usec target sample,10.6usec delay, 78.4usec GB sample. The 10.6usec delay is what it took to GB the clay while leaving the target and GB times fixed. Entered the times in the PI simulator. Didn't GB. Turns out I've been reading the GB slope wrong. After entering the correct slope the simulator did GB. Looked at integrator out with and without GB for the nickel and quarter. With GB the nickel is less but the quarter is higher. Times aren't optimized, just what I started with.
.

[green]

Charted some data. Adjusted GB sample time for a minimum change when placing a bag of clay close to the coil with the controller on GB. Recorded integrator change, no coins and coins at 2 inches with and with out GB. Entered the times in the simulator. The quarter charted close to a straight line on log-log with a slope of -.74 so I used it for the simulator slope and 10usec TC for the US nickel. The percent change with the simulator and bench controller were in the ballpark. The simulator didn't quite balance the clay with the controller times. Integrator out is a lot more stable since I added EF. Got in a hurry to take data before, should have known better.

[Davor]

If you pay a close attention to the whole EF balancing deal, you'd notice it effectively turns a PI Rx into a chopper-stabilized (auto-zero) amplifier that is effective in eliminating drift and 1/f noise in general.

[green]

If you pay a close attention to the whole EF balancing deal, you'd notice it effectively turns a PI Rx into a chopper-stabilized (auto-zero) amplifier that is effective in eliminating drift and 1/f noise in general.

The way I'm trying GB.

[Davor]

This scheme is OK. For EF it is only necessary to represent a difference between the target sample and GB pulse, and it is irrelevant whether you calculate the difference, or you actually fire switches in opposite polarity of target and GB pulses.
Since you are experimenting, let me draw your attention to a somewhat better integrator, a so-called 2C integrator. Unlike the classic inverting integrator that you use, this one is insensitive to the switching glitches:

http://images.books24x7.com/bookimag.../fig450_02.jpg

The inverting integrator relies on op-amp speed to deal with step voltages. A garden variety op-amp with FET inputs is not that fast, so the 2C integrator is a much better choice, and glitch-free with PI inputs.

[green]

(Integrator out is a lot more stable since I added EF. Got in a hurry to take data before, should have known better)

Charted a 4 inch o0o coil with the diff amplifier and EF. No GB. It's at least 10 times more stable, mostly due to adding EF compensation I think. The 12 inch measurement was varying a couple counts, charted what I wrote down.