New sinusoidal AC servo drive development

[Xerxes]

Hello! As many might have noticed from Surplus motors thread I have been designing my own servo drive especially for those Sanyo P5 motors. Many of Sanyo owners seems to be interested about my drive so I must ask for some opinions to get the idea what kind of drive is most wanted.

I already have a working "proof of concept" running and I'm starting to build a real prototype soon. First thing to decide is the type of the power stage. Some explanations of each choice:

1) Surface mount discrete MOSFETs: Continuous output power would probably be limited to about 500 Watts. However it has the most solid contruction and the lowest total component cost.

2) IRAMS10UP60A power module (3 phase 600V 10A max). An easy solution (speeds up design process) but the voltage rating is somewhat unoptimal for 100Vac motors. Maximum output power would be around 500-800W.

3) Discrete through-hole MOSFETs with heatsink. Over 1kW output power possible but has most undesirable construction. Many parts to cool down and heatsink must be insulated etc.

4) SMD discrete MOSFETS + possibility to connect own custom external power stage of any power. Enough power for most applications (<500W) but enables to drive higher loads too. Requires lots of time to get it well tested.

5) Controller logic only without integrated power stage. An external power stage must be used. Virtually any power possible.

A typical power stage consists a) six power transistors, b) gate drivers for transistors, c) current sensing circuity and possibly d) short circuit protection (optional).

Those 1kW servos have peak power rating of 3kW for 10% duty cycle. Because motor efficiency is not 100% it would require about 4kW power stage to squeeze everything out of motors. That kind of power stage can be complex and expensive.

I wonder what kind of CNC machine it takes to utilize that kind of power. It must be a massive industrial size machine weighting tons that rapids at 1000 IPM. How many of you are planning to build one? I don't think its very good idea to pay for 4kW drives if you really need "only" 400W. Anyway I'm open to suggestions.



Xerxes

[ger21]

I've got 300w motors, so #1 sounds best to me.

What about something like a rutex, that just accepts power from a simple power supply?

[Xerxes]

I'm planning to use 150V or 200V rated parts. However I don't recommend powering drive directly from power lines. An isolation trasformer can save lives.

[DerekZahn]

You might remember from another thread that I'm also working on a similar project. I am using discrete IGBTs because I want to be able to handle higher voltage motors also (like the Compumotor SM series frequently available on Ebay). That IRAMS power module you found looks pretty sweet though also. and I might steal that idea for a future more compact version :-)

If you think the surface mount MOSFETS can handle 500W output then that option would be great for the large number of people with the 300 and 400 watt motors, and "lowest cost" will certainly have appeal!

I'll make a thread to document my progress in a week or two. Good luck with your design! It will be good to have several open source efforts in progress.

[Pophamrt1]

Just a suggestion, so take it with a grain of salt, and I'm not sure if its even feasable per Rutex's current product offerings/development path. But, I'd rather see a joint development effort to assist Vlad/Tom in a sinusoidal or otherwise modified Rutex, if that is indeed necessary to correctly drive the current AC brushless servos such as the P5.

You're obviously skilled, willing, and interested in going to production with a product. Which is great. The only additional consideration is that Rutex has an established product platform with their current offerings, which adds stablity and the "warm fuzzies" to those like me who are interested in investing in a solid platform for retrofits to ease installation and tuning issues and that will be there in a few years with a replacement when a drive finally blows.

I deal with licensing intellectual property, trade secrets, etc., in the semiconductor industry as my profession, so am aware of the difficulties and would be willing to assist if interested.

Again, just a suggestion, and the undertaking of a sinusoidal drive is appreciated as I have 10 of the 1KW P5s--so whatever happens, my suggestion would be Rutex-style with sinusoidal amplification and headroom for at least 100V, 12amps cont., and about 40 amps peak, all for use with an independent unregulated power supply (basically enough to push the motors hard enough for retrofits to small VMCs with about 200-300ipm rapids).

[waughd]

FWIW
A joint venture sounds like a good idea to me as well.
I do not profess to be an expert in these units.
Rutex seem to have most of the front end stuff already in place,just the power/motor interface side seems lacking.
I have 6 of the 1kw units myself.
Intended use are a knee mill and a gantry router/plasma cutter.
hope this assists you.
Des

[unterhaus]

why not use the IRAMX16UP60A parts that have higher output power? They are $31 single quantity instead of $23.

[ger21]

It looks like the 300-400w guys might one one type, and the 1Kw guys might want another. How about both.

[unterhaus]

personally, the reason I like the IR devices is that there is tons of circuitry in there that would have to be replicated on a circuit board with discrete devices. To me, the prices are so low it's a no-brainer.

[HillBilly]

Will the drives be adjusted with software or pots?

Darek

[CNCAddict]

Hmm, I agree that using the module is a no brainer just to simplify the circuit design by a huge amount. It might also be possible to "piggyback" the modules to scale the power output to any level possible (I may be wrong here).

What did you plan on using for the brains of the device? There are tons of MCU/DSP chips that would work. Which device are you using?? Oh, and the setup below looks like it may work for lower power outputs. Not sure about price though.

http://www.irf.com/technical-info/re...ns/iradk10.pdf

Thanks,
David Bloomfield

[andy55]

Xerxes,

what microcontroller are you using ?

are you planning on optoisolating the microcontroller from the powerstage ?

what about current feedback, measure just one phase, or two/three for vector control ?

any chance you will publish the plans and code somewhere ?

I was thinking dsPIC30F3011 + IRAMS 10A myself. with current measurement on at least two phases for possible vector control. control of amp would be a ttl pwm torque command and ttl direction signal

in the PC I would have a mesa electronics (www.mesanet.com) 5i20 PCI card giving torque commands to the servos and reading the encoders. control software would be EMC.