[Xerxes]

Hi,

This is the poll I discussed here.
And whole thread can be seen here.

In short this is a poll to ask whether you would like to get AC (and DC) servo drive industrially assembled and tested or as DIY kit. I estimate that a batch of 100 units could be produced at end price of ~$150 or hopefully even less.

About sinusoidal AC servo drive project:
http://www.students.tut.fi/~kontkant...drive.php.html
The drive is designed using Sanyo P5 servos and especially optimized for CNC use.

Please express your thoughts! :-)

[HillBilly]

Xerxes,

Will the Industrial made version be able to handle all the P5 motors up to the 1000 watt?

Darek

[Ken_Shea]

servo drive industrially assembled and tested or as DIY kit

Xerxes,
Why not offer both?
Personally I would opt for an assembled and tested drives, others of course would prefer to save as many dollars as possible, or enjoy the assembly.

Ken

[ger21]

I voted for build, but if the price is right (~ $125-150), I might be tempted to buy. But, I'm not sure when I'll be ready to buy, and I'd be temted to buy a kit sooner and have them in hand, rather than wait a year or two (when I'd actually need them) for something that may no longer be available in the future. Does that make any sense?

[vroemm]

and I'd be temted to buy a kit sooner and have them in hand, rather than wait a year or two (when I'd actually need them) for something that may no longer be available in the future. Does that make any sense?

Yes, it makes sense to me.

But there is a other side.
Maybe in a few years there wil be something even better and/or cheaper.

Difficult choice.

[CNCAddict]

Xerxes, have you considered adding feedforward gains on your controller? Feedforward is VERY important on pulse/direction servo drives to keep the actual vs. commanded position errors low. Only using a PID loop will result in large errors during acceleration and possibly at high speeds. The velocity feedforward may not be needed if the Integral term is high, but increasing "I" is never a good idea. I'm not sure if you have any "industrial" servo drives to compare with, but I have Teknic unit here if you want to borrow it.

[JFettig]

I voted 175 if its going to be reliable, if its going to be exceptionally reliable I would pay 200 I bet.

Jon

[Halfnutz]

I wanted to try out industrial servo's so I bought three Omron step and direction drives and matching motors for 300 bucks off eBay. One 100 watt and 2 200 watt units. Small but suficient for comparison to the other drive system I will evaluate, Geckoe's and DC brushed motors.

If the boards end up at over 150.00 it would make more sense to me to buy used industrial drive and motors off eBay.

The Omrons are awesome. Smooth, lightning fast and completely silent. I have yet to put them under load, besides my fingers, but it has been really nice working with the software and jogging them around. I'm sold on these AC systems. They have all the features talked about, and I expect they will perform flawlessly. The major concern for me was support, but Omron's has been great. I am going to try and snag a Kollmorgan Servostar CD soon, it is another step & direction drive that is available often, and Dahnaher is the support company.

I will definately be watching this project and hope to try one out as well. One thing I dont understand is the hype over the Sanyo P5 motors. What makes them so special?

[Xerxes]

I listed some pros and cons of DIY and industrial assembly:

Industrially assembled:
+Works for sure, every drive personally tested by me
+Components are cheaper in quantity purchases. If volume is large enough it will make industrial and DIY price difference insignificant.
+If the drive is success, there will be more
-Starting costs

DIY
+It’s nice to build something
+Cheaper in small volume
-More defects. Most problems are probably caused by small assembly mistakes that may be hard to find. The final drive may have chips with 0.5 mm lead pitch which can be shorted very easily and can be difficult to repair. Most hobbyers don’t have tools for desoldering TQFP packaged microcontroller without breaking it.

Now something I haven’t mentioned here before:
I and one of my friends are seriously planning to start developing drives to compete existing commercial solutions in price AND quality. Our goal would be to make industrial quality drive that second to nothing in the market. That is the reason why I have been so careful about source code. I’m personally very open source minded. I have contributed to several OS projects and have released some my own too but in this case I’m careful because of possible business. I’m most probably releasing the source if I find it riskless or if there will be no business and drive remains only as hobby project.

[Xerxes]

HillBilly:
One possible mosfet is IRF640N which has been rated for 200 volts and 18 amps current. To be realistic it could be stressed no more than 8-10 amps continuously. If you calculate 200V*8..10A, it theoretically would give 1.6..2 kW. In practice it would be less because motor voltage is less than 200V.

CNCAddict:
Sure, there will be tunable feedforward paths (first and second order). And integrator will have anti-windup logic to make nicer recovery from output saturation. There will be also PI tunable current regulation feedback that is not available even in many commercial drives. It would be interesting to get hands on some commercial drive for comparison. :-)

[vroemm]

I listed some pros and cons of DIY and industrial assembly:

Industrially assembled:
+Components are cheaper in quantity purchases. If volume is large enough it will make industrial and DIY price difference insignificant.

I was thinking, you could also sell a complete DIY kit, all parts included and ready build ones.
But then you have to calculate many will have problems building their own, and you will have to support them, which costs a lot.
Serious problem.

I and one of my friends are seriously planning to start developing drives to compete existing commercial solutions in price AND quality. Our goal would be to make industrial quality drive that second to nothing in the market.

Go for it, if you think its the way to go :-)

I would propably not buy one, DIY is cheaper, more a challenge, more fun.

And there is a small change someone will design something like it, and put it on the internet for free, open source and all.
This might hurt your bussines, and investments.

That is the reason why I have been so careful about source code.

I understand. :-)

[Xerxes]

But then you have to calculate many will have problems building their own, and you will have to support them, which costs a lot.
Serious problem.

True. But support is critical in industrial drive, too.

And there is a small change someone will design something like it, and put it on the internet for free, open source and all.
This might hurt your bussines, and investments.

I'm not worried about that. I think drive market is still big enough even when there are DIY builders.

I would happily share also my source code for DIYers but there is a problem called industrial piracy. See some posts from Mariss Freimanis, he doesn't seem to be very happy about industrially mass produced pirate G201's. :|

[vroemm]

I would happily share also my source code for DIYers but there is a problem called industrial piracy.

I understand.
With the UHU servo controller there was the same discussion.

[martzis]

Hi Xerxes

Nice to hear about your project. It seems VERY promising!

I listed some pros and cons of DIY and industrial assembly:

DIY
+It’s nice to build something
+Cheaper in small volume
-More defects. Most problems are probably caused by small assembly mistakes that may be hard to find. The final drive may have chips with 0.5 mm lead pitch which can be shorted very easily and can be difficult to repair. Most hobbyers don’t have tools for desoldering TQFP packaged microcontroller without breaking it.

Have you considered that you could make a semi assembled DIY version which would contain a pcb with SMD componts soldered into it only? I have seen few this kind of DIY kits before.

I do not know how much this kind of solution would affect to the price, but I have understand that though hole components are harder to solder automatically than SMD components.

Martzis

[Xerxes]

Have you considered that you could make a semi assembled DIY version which would contain a pcb with SMD componts soldered into it only?

Not a bad idea :-)

I will find out how much that would save cost when I make contact with some contract manufacturers. I called yesterday one manufacturer but they didn't give any price quote yet.

Meanwhile I updated my project page and collected a listing of features that have been planned for the drive:
http://www.students.tut.fi/~kontkant...drive.php.html

[Halfnutz]

Xerxes, Your board layout looks very nice, as good as anything I've seen available anywhere. Packing all the features discussed into that real estate is amazing. Congratulations on a great job. When do you anticipate being able to sell boards, or preferably kits? Is anything going to be available in the neer future? Do you have a schedual of target dates for release stages?

[walter]

I voted 175 if its going to be reliable, if its going to be exceptionally reliable I would pay 200 I bet.

Jon

I feel the same way.. I`d pay more for reliability.

[Xerxes]

My goal is no less than rock solid reliability. I'm checking that every component operats at safe operating area. Every code line is written carefully thinking and in case of bugs there is firmware update possibility. I think it would be good to provide some boards for testing for someone else before final release in addition to my own tests.

About deadline, I don't have one, it available when it's done. I don't want to sell unfinished drive but I hope it wouldn't take too long. Hopefully it gets ready for production at end of this summer.

Its really hard to predict timeline since I'm still considering quite large modifications for current drive. I was planning to try replacing Allegro hall effect current sensors with op-amp based circut that measures current differentially, amplifies it differentially and finally it gets AD converted differentially. Hopefully that would be the end of noise (and it's cheaper, too). But it's possible that I stick with Allegro sensors to speed up design process. Motor hiss isn't really that bad, i'm just perfectionist.

[vroemm]

I think it would be good to provide some boards for testing for someone else before final release in addition to my own tests.

Maybe people can make different motors avaible for you for testing.

[Xerxes]

vroemm,
I will test as wide range of motors as I can. I already have 20W, 40W (low voltage), 300W (100V), 920W (230V) AC servos to test. I would like to test it with 1kW sanyo, too.

I got answer from a local electronics manufacturer and they estemated that dropping TH (through hole) components would save costs about 10 euros/board. But one should remember that assembling component assortments in small plastic bags can be quite time consuming too. A more serious drawback would be that drives couldn't be tested at all.