Paul and Sabrina's Cheap 3 Phase Inverter (AC Controller) with Field Oriented Control

[thingstodo]

Originally Posted by e*clipse

Are any of those VFD's in working condition?
If so, would you be willing to let a couple go as-is?

Yes .. No .. Maybe. Depending on what you define as 'working'.

I will let you know if my wife gets her way and I need to get rid of a few. I don't have any single-phase, so I guess none of them would count as 'working' for you.

They are *HEAVY* and they are *OLD*, so I'm not sure they would survive the shipping, or if they are worth more than the shipping would cost.

I'm looking for better ways to control my mill & lathe speeds; something in the range of 3hp to 5hp, 240V single phase.

Since this is totally OT, PM me if it's possible.

Paul seems OK with OT traffic.

[MPaulHolmes]

Heck yes I'm OK with off topic traffic! Don't make me post another naked picture of Kim Jong Il. hold me back...
EDIT: REMOVED PICTURE OF NAKED KIM JUNG IL... they are liable to take down ecomodder if it was discovered. It's not worth the risk! haha

I could definitely use beta testers. They should be fine, as it's basically the same thing I've tested. But I passed a kidney stone doing like 10000 different variations of component positions and routes so as to make it a good 2 layer layout, with good ground and power planes.

I would use shielded 7 wire cable like this for the interface:
Buy High Quality RS232 extended distance cable to eliminate signal noise and loss of data

The PWM signal should be from something like a pwm channel in a microcontroller, or an AND gate, so that it's actively forced high or low. I've tried it in a few noisy environments like big high frequency switching transformers, and haven't had any problems. Don't use a pull down resistor to cause the pwm to change from high to low (or pull up resistor to go from low to high). Different lengths of wire don't matter much. Propagation delay for current in a wire is on the order of 1 NANO second for each 6 inches of wire. But there is going to be a variation of 10's of nanoseconds from the optocouplers between different FOD8316's.

[MPaulHolmes]

OK, I'm finally working on the PI loop tuning program. jyanof mentioned this program:
Kst - Browse Files at SourceForge.net

It graphs streaming data. So, here's my plan as of this morning. enter a proportional and integral gain constant through the serial port. Then, execute a new "run-pi-test" command through the serial port. That will load the array for Id (or Iq). Then, the array will output to the screen, which can be graphed by kst2 (link above). Repeat until your PI looop is tuned really nice!

[MPaulHolmes]

OK it's almost done! And it's going to be incorporated into the regular ACController.c file. I thought that would be too much trouble, but it's not! So people won't have to buy a programmer just to tune their pi loop. haha. No separate hex file!

[thingstodo]

Originally Posted by MPaulHolmes

I could definitely use beta testers. They should be fine, as it's basically the same thing I've tested. But I passed a kidney stone doing like 10000 different variations of component positions and routes so as to make it a good 2 layer layout, with good ground and power planes.

I would use shielded 7 wire cable like this for the interface:
Buy High Quality RS232 extended distance cable to eliminate signal noise and loss of data

That cable lists one overall shield. I'm no expert .. as I understand signal noise, you are trying to isolate the higher frequency PWM signals so that they are as close to a square wave as possible. The DC on 24V and 5V is likely generated from a switching power supply and could have similar frequencies (depending on the level of filtering).

Our instrument department is *VERY* picky about which cable is installed in our mill. They specify belden cable - I will look it up on Tuesday and post. I'm sure that there is a single shield per pair as well as an overall shield. It is likely more expensive, but having your system work reliably is *PRICELESS*

[MPaulHolmes]

Your expertise would be very much appreciated!!!

[MPaulHolmes]

I have 7 holes for the wires. Would that need to change? They are all partying in China until the end of the month so they won't get to manufacturing it until then.

[thingstodo]

Originally Posted by MPaulHolmes

I have 7 holes for the wires. Would that need to change? They are all partying in China until the end of the month so they won't get to manufacturing it until then.

The wires would still terminate on the board, so if you need 7, then 7 it remains.

The shields would normally be connected on one end, usually on a single terminal block that is only for analog or digital grounds, which is connected to the negative terminal on the isolated DC output. This can be outside the board, and the connection to the board can be via the DC negative terminal.

I'll run this by our instrument supervisor on Tuesday ... sometimes I mess up the details on the grounds. I don't think you need to modify your board layout.

[e*clipse]

Boy, where to start! So much good stuff happening.

1st - thingstodo - It sounds not worth it; shipping costs can often spoil a "good deal." Thank you for considering it. :-)

2nd regarding wires - thank you, Paul & thingstodo for any suggestions on this. I've been looking into signal wire solutions and there sure seems to be a lot more to it than I originally thought.

I guess the first thing is what type of signal it is. My understanding of the gate driver signals is they are "single ended" - a 5V signal with a ground - right? So in this case, a shielded cable would be the best?

Another problem I'm working on uses differential pair signalling. In this case, I understand a twisted pair cable is the best solution, but I'm not sure how a shield or ground surface plays into it. Thingstodo - could you ask about differential pair signals? Any advice from the instrument supervisor would be invaluable.

So far, I found some info on it regarding PC board design:
http://www.pcbcarolina.com/images/pr...diff_pairs.pdf
But I'm still not sure whether a cable made of twisted pairs would be good enough in a noisy (EV) environment. If so, I found this fairly inexpensive wire that can use IDC connectors every 270mm:
http://www.mouser.com/ds/2/1/ts0999-19101.pdf

-E*clipse

[Piotrsko]

fwiw quick and dirty: twisted pairs is for NOT TRANSMITTING EMR as the twists effectively cancel any fields generated in the wires. kinda sorta works on receiving except for really high frequencies where the waveform can exist at some even multiple of the twist length ( say a couple of ghz). Sheilding is for NOT Receiving EMR. If you want to go really nuts, twisted pairs and sheilding. Ferrite is kinda both, they act like inductor cores for low values of EMR effectively stopping the ac component (2*pi*f*L {in Henries}) with reactance. Your PWM signal is basically a dc voltage so...... ferrite doesn't bother it much below a KHZ or two as the reactance is so small at those fequencies. If you need to go really high Freq, then you need co-ax cable, but that is a weird deal all its own with skin effects andthe like.

pretty much if you want to stop receipt of EMR, then sheilding on the pairs, then the bundle sheilded with a foil or a braid and tied to different GROUNDS (not negative battery) at one end only.

Not an instrumentation super, but instrumentation Tech at Edwards AFB rocket site thruster test lab.