Design Opinions Appreciated!

Fella's..

As a part of my "project" for the final year of college, I'm designing a constant-current power supply which will supply a small current through the windings of a motor whilst it's not in use, this is to maintain an elevated temperature within the windings to prevent condensation forming.

The design will use the 440VAC 60Hz power that's supplied to perform the heating, whilst a separate 220VAC auxiliary voltage will be used to power the control circuit.

The first step I've taken is to design the "High-Voltage" side, which is rectified, filtered and then controlled by an IGBT - through a current sensing resistor to the motor windings.

Here is the schematic as it stands now:
http://i177.photobucket.com/albums/w219/br...zps21e76fb3.jpg

The control side will feature a PIC which will control the IGBT. It will have a "soft-start" to begin with, as soon as power is available - and then will regulate the current to a pre-set level. The current passes through a 0.005 Ohm resistor which will act as a current sense, the voltage drop across this resistor will be measured by an op-amp in differential configuration (which should subtract the Vout from Vin, leaving the voltage drop).

My initial thoughts are that the PIC will initially have the IGBT in the "off" state, then will turn on the IGBT until the current that's sensed matches the set-point - the IGBT will then be turned off. This cycle will repeat for as long as the motor remains off, and should provide a constant current that's sufficient to keep the windings up to temperature.

Any thoughts on my idea would be greatly appreciated, as much of this has been a constant learning process so far!

Cheers!

Ed: pic link

This post has been edited by Sch3mat1c on February 08, 2013 03:35 am

Interesting. Have you done any testing yet? I'm wondering if it's going to create enough heat without trying to rotate the shaft.

You might consider adding hysteresis (such as a window comparator) to prevent constant on/off cycling when the desired temp is reached, or... you might consider using the voltage variance to control the IGBT somewhat directly, and turn it more/less on rather than just on and off.

A temp sensor (thermistor?) attached directly to the motor housing might be simpler btw.

Just a few thoughts.

The motors that it'll be applicable to are three phase, so if the heating is kept to one winding it shouldn't (hopefully!) try to rotate the shaft.

Temperature control was a consideration, the problem is the motors are often quite a distance from the starter panels, and so it would be impractical to try and add extra cables. I did consider using the motors own supply cables as some sort of communications line - but I'm limited with time unfortunately, as the whole project has to be finished mid-April! :-(

With regards to testing, being away from home to attend College means that I will need to buy all the components before I can begin testing, let alone actually constructing the thing

Since you're feeding the motor with DC, there will be no tendency to rotate. In fact, applying DC to an AC motor has a braking effect. So, no worries there.

I question the need for filtering the DC after the rectifier. Seems unnecessary to me. You could use your differential amp and shunt resistor to measure instantaneous current, and then average that value to control the IGBT.

Well duh. I had non-local images turned off and missed the schematic.

One thought for regulating the current,maybe a low-pass filtered PWM signal? It might help with the possible hysteresis problem..maybe.
Instead of on/off/on/off, you could ramp the PWM value up/down to maintain a set temp. Err,I guess you don't have a temp sensor/feedback though.. So maybe just pick a suitable PWM value,and run with it.

Dunno,just a couple sleep deprived Vicodin induced thoughts.
At least that tooth isn't hurting anymore.

Thoughts:

1. Remember, DON'T F**K AROUND WITH 480 (or 460, or 440 -- and 240 isn't all that friendly for that matter). I am around it on a daily basis, and am always careful to keep safety procedures in mind.
http://www.youtube.com/watch?v=-iClXrd50Z8
Don't fear it, just respect it. Give it plenty of room whenever possible.

2. Is your circuit connected to the motor continuously, even while the drive is operating? If so, you will have to design it to tolerate 480VAC of any phase (whichever it happens to be hooked to). If this is impractical, then you will have to include some sort of disconnect feature, AND some means of verifying that it is, in fact, disconnected -- contactors weld shut, switches are forgotten. It would be irresponsible to allow such a simple mistake to happen.

3. What kind of current sense and IGBT drive are you expecting? Only the gate connection is shown, but of course, with the emitter bouncing around at 480V something or other, driving that from a PIC in grounded equipment obviously isn't going to accomplish much! If the PIC is "flying" with the IGBT (and current sense), then you must ensure that its power source is adequately isolated (at least 3.75kV 1 min hi-pot, "functional" insulation class or better) and that, if any control signals come in, they must be similarly isolated.

4. If you were intending on sensing current with a differential amplifier somewhere around ground, you'll need very large series resistances -- at least 10Mohm, I don't happen to know how much resistance counts as "galvanic isolation" unfortunately. These resistances must be matched extremely closely, in order to resolve that teensy milivolt signal out of the 480 common mode voltage. Even the op-amp must be very high quality to achieve this. A better alternative is a hall-effect sensor, which works very nicely in the DC-100kHz bandwidth (more than enough for a buck converter like this*), usually in the 5-100A range (they are available for lower currents, but the sensor has a many-turn winding on it, less friendly to use).

5. *Note: you've made a buck converter, working against the motor winding inductance. FWIW, pretty much any frequency from 1kHz to 100kHz will be just fine, of course you'll want to measure the inductance to verify how much peak current to expect. You need a free-wheeling diode, not a series diode, to complete the circuit -- without, the IGBT will work perfectly for the first cycle, but as soon as it turns off...

As for temperature control -- can't be ruled out just yet. If your current measurement is accurate, and you throw in a DC voltage measurement as well, you can calculate the load resistance. Copper has a strong and predictable tempco, so as long as you have the opportunity to calibrate against the winding's "cold" resistance, you can estimate the operating temperature, even put it into a PID to control by.

Final note -- don't worry about the EMI filter, at least right away. You'll most likely need it (to meet CE requirements, anyway), but evaluating what's required is another year or two of school in itself. You can always throw in a commercial unit out of good faith (the small ones are under $100, even 480V 3ph models), just make sure you reserve space in the cabinet to do so. I do note, however, that you set the inductances to 6.8H (unclear if they are coupled), which could actually be enough to handle some AC voltage on the output (6.8H is 2.56kohms at 60Hz). It's not going to be a small inductor, though!

Tim

yep ... tim is correct ... don't give 440 a chance to bite you , it'll take out a big chunk if it does .
you might think about a "safety lock out system" for the control panel .
in the real world , the power lockout would be done with a physical lock on the 3 phase power disconnect ...
what i am referencing is a method to ensure that the motor cannot be powered up at/from the control panel "accidentally" .
this would be for "minor maintenance" (and such) of the machine the motor is in .
maybe it'll garner you some brownie points in the final evaluation of your project .


i seen an electrician at a mill get hammered with 440 ...
it was his own fault :
he did not lock out the power disconnect at the distribution/control panel ... he merely "pulled the switch" .
he pulled the power to the wrong circuit .
he did not check to make sure the circuit was dead (this would have drawn attention to the above issue) .
what saved his life was the fact that he fell away from the live circuit after contacting it .

he had started work on the equipment during lunch break (normally no one around that area then) .
an employee was going back through that area to get something he had forgotten and found him laid out cold .
the guy seen the open panel and put two and two together .