High voltage & high power servo drive design

Since our upcoming servo drive VSD-E will be 12-80 VDC drive and there is a clear market for high voltage drive that can be operated directly from mains voltages (115/230VAC), I'm creating this poll to gather user opinions about preferred servo drive constructions.

I have 3 different drive types in mind that would allow such operation:

1. VSD-E without on-board power stage. It would just have connector for external power stage.
2. Small sized control card with pin headers. This could be plugged on "mother board" containing the power stage.
3. Monolithic design. Everything on one board and possibly a metal enclosure for it.

Options 1-2 allows controlling virtually any power and option 3 would be the easiest to use.

Attached picture shows VSD-E drive if one isn't already familiar with it. Option 1 would look much like, but 1/2 of this size.

The board looks great, except it looks like it's missing a rather large heat-sink. Adding a heatsink looks difficult.

I LOVE the voltage range, it's perfect. Most serious servo motors want 80-90VDC for their rated performance specs. I'm so sick and tired of drives and amps which seem aimed at the combat robot/wheelchair style builder who is limited to 24VDC.

I note you call it power supply, though you may actually mean servo drive, which uses a high voltage/high current power supply to power a high-torque motor or gearmotor combination.

FYI, my motors are the odd duck ServoDisc motors, so my power supply is customized for them. My development is still underway, the first phase will be limited to about 31VDC, and later expanded to approx 80VDC.

Since I've purchased a lower voltage drive for 3 channels, I've been thinking about using it's output to modulate an external circuit running at 80VDC (MOSFET switching of ground). I worry about presenting enough of a load to the amp for it to remain happy.

Comments?

Hi,
I vote for a control card with separate powerstage. That way you can have 2-3 different powerstages available so the user can choose which one he/she wants when ordering. 80V/20A, 200V/10A, 200V/25A or something like that.

That way it's also possible for the end-user to design his own powerstage. However, that may end up as a huge support issue for you...

Next choice would be a VSD-E with a voltage rating of 200V. I guess that would be number (3).

/Henrik.

Billhill, the drive in picture is without heatsink because powerstageless pure control card wouldn't need any. You need to add inductor in series with disc servo to be able to use it with a normal servo drive.

Henrik, I'm planning to try some day equipping VSD-E with 200V MOSFETs. Drive rating could be around 160-180VDC max. However current rating would be only about half of 80VDC verson.

Personally I'm feeling that option 2 would be my choice. It has the best flexibility and could be used to build type 3 drives too. Attached picture shows sketch of option 2.

Quote:

---

Henrik, I'm planning to try some day equipping VSD-E with 200V MOSFETs. Drive rating could be around 160-180VDC max. However current rating would be only about half of 80VDC verson.

---

But then it's about the same specs as a VSD-A, right?

It would be weaker than VSD-A, maybe 5-8A cont and 15A peak max. There aren't very good 200V MOSFETs available but I was thinking to try IRF640N. Because demand is for higher currents, I'm wondering if it's worth effort doing 160V version of VSD-E at all.

Well, I can only speak for myself but I'd like to see a 160V/25A VSD

Regarding the availabillity of 200V MOSFETs you probably know that better than me but I think you need to move to larger MOSFETS than what is available in TO220 package. As you might know the HP-UHU uses IRFP264N in TO247 package, rating is 250V, 44A.

The features and performance (true torque loop, feedforward, high res encoder etc etc) of the VSD coupled to a powerstage rated at 160-200V, 20-25A continous would be a sweet thing IMO.

Again, this is just my personal thoughts, you certainly have a better overall view of the market than I do but it's not uncommon that people ask for 200V/25A drives on the Mach3 list for example. With Rutex currently in a somewhat unknown state and the Pixies S/D->analog converter being discontinued you really don't have many options when it comes to driving larger servos.

/Henrik.

The problem is not finding a large enough MOSFET but finding a good enough. High voltage mosfets have nasty properties compared to <100V ones, so they need snubbers etc to work nicely. It needs much more design effort than just equipping existing PCB with alternative parts. Perhaps some day (not anytime soon) I start such project for high power VSD-E.

However, controller-only card allows doing that any time. It allows making drive that works straight from 110-230VAC which is a good advantage. So when (inexpensive) controller card is available, I hope to see third parties offering reasonably priced drives built around it.

Quote:

---

I do but it's not uncommon that people ask for 200V/25A drives on the Mach3 list for example. With Rutex currently in a somewhat unknown state and the Pixies S/D->analog converter being discontinued you really don't have many options when it comes to driving larger servos.

---

Larkens Viper 200 drives will drive motors up to 170 volts and 20 amps. This is a DC servo drive, not an ac servo drive though.
http://www.larkencnc.com/viper/index.shtml


Here is a big mill retrofitted with them
http://www.cnczone.com/forums/showthread.php?t=52696

Hi,
Yes, I know about the Vipers. I looked at them too when shopping for drives but didn't go that route due to a couple of reasons.

Anyway, if they are running that big Hurco they might be 'ready'. If none of my current options works out (god forbid...) I might try them out.

I have nearly finished the "stamp" design (see post #4) and a reference schematics design for drive around it. It will take time to build prototypes and write firmware for it.

If product gets out, I would publish reference schematics of example drives. I have been thinking to design also a reference 400V/40A DC servo drive around it. It could end the lack of high power DC drives once and for all :)

Ingegration of stamp (VSD-Core) could look like this. Only few external components required.

Quote:

---

If product gets out, I would publish reference schematics of example drives. I have been thinking to design also a reference 400V/40A DC servo drive around it. It could end the lack of high power DC drives once and for all

---

Now we're talking :)

Seriously though, I really like my VSD-A's but unfortunatly my biggest concern seems to be true - my motors are just too big, or so it seems :-/

I'm still running bench tests but I just can't seem to get from 0-2000rpm in less than 0.5s and that's with the motor unloaded and the following error limit set to 180° (7200 counts in my case) which is way to much. With the accel set to ~0.5s the motor faults during accel as soon as I put just a bit of load on it. I need more power, Scotty...sorry I mean Tero.... ;-)

I do have my HP UHU's too but those can't handle the encoders I have on the motors so....aaargh ;-)

You're doing great work Tero, keep it up!

/Henrik.

Henrik, you could do the acceleration test in GDtool. In scope, select also commanded & actual torque signals so you could see if the drive is actually outputting maximum current to motor. Let me see the saved plot file if you do this test.

Do you know motor torque constant & motor inertia? Theoretical maximum acceleration rate can be also calculated. DC motors typically have high inertia so acceleration is slower compared to AC servos.

Hi Tero,
I've plotted commanded and achieved torque and the drive does output maximum current. I don't have any plots saved but I'll do that and send them to you next time I'm working on it.

Here's some data on the motor:

Code:

---

Continous Current:                19A
Max. peak current:                150A
Permissible peak voltage:        170V
Voltage constant:                0.47Vs/rad
Torque Constant:                0.47Nm/A
Rotor moment of inertia:        0.0075kgm²
Mechanical time constant:        8.1ms
Weight:                                24kg

---

The motor is large, 140mm in diameter and ~400mm long. I was prepared for somewhat limited acceleration from the beginning but was kind of hoping it would be 'enough' with the 15A peak. Anyway, if you're able to calculate max theroetical acceleration I'll do the plots, send them your way and we'll go from there.

Thanks and sorry, I didn't mean to hijack this thread....

/Henrik.

If I did calculate correctly, it should be able to accelerate from 0 to 2000 rpm in about 0.22 seconds at 15A (without friction & losses).
(0.47Nm/A * 15A) / 0.0075kgm² results 940 rad/s² acceleration and 2000 rpm equals 209 rad/s.

http://hyperphysics.phy-astr.gsu.edu/hbase/mi.html

It also can produce 4.7Nm continuous torque at 10A which is plenty for most machines.

Lets continue discussion in this thread instead:
http://cnczone.com/forums/showthread.php?t=38455

I invested in 3 FC15030 30A filters from Advanced Motion Controls which can handle 400V and adds 300 microhenries.

I've built a PWM motor tester using a couple of medium sized lead acid batteries, and run the pot up and down. The motors are strong - gripping the shaft harder and harder by hand doesn't seem to slow them down much at all.

I did have a frown during testing - The smaller normal DC brushed motors could be slowed and well controlled at slower speeds. The ServoMotors didn't ever come to a complete stop, so I figured my low power waveform was still plenty of energy for them.

I appreciated your comments.

I took a look at your thread, the products are awesome and priced well.

I wish there was an interactive DC servo motor calculator where I could put it what I know, and get out accelerations, and running speeds.

Last question - anyone know where to find the amount of energy needed for the cut (for steel, aluminum, brass, polyurethane). I figure the axis move is proceeding along a a minimal power budget until it encounters the object to be milled. Then the amps go up, depending in size of cutter, depth of cut, speed of motion, etc. Are there any guestimates for torque needed for certain cutting depths, assuming standardized cutter sizes?

Thanks everyone for your posts. I'm reading everything I can find here.

Aloha
Bill

Bill,

FC15030 looks suitable for this application. However, it is only 0.3mH, so I would recommend it for under 60Vdc (or 100V as absolute maximum).

I think you may have mistakenly posted in wrong thread. Lets keep this thread on-topic :-)

Tero

I like the idea of a "stamp" type controller. One suggestion, I would think of a different name before bringing it to market, I think "stamp" may be trademarked.

Also realize that by making a "kit" type controller will open you to a flood of technical support inquiries. "I bought your control card and made a crappy bridge circuit, why won't it work?" Things like that I would think will be fairly common. You may want to specify what type of support is available for this type of product. Ie. You will help with control/interface problems, you won't help diagnose power stage problems. You can get into a situation where people talk bad about your product because they can't make it work with their amateur power stage. Tread carefully.

Keep up the great work
Matt

Matt,
Thanks for sharing your thoughts!

I have been thinking to name it something else than stamp, probably VSD-Core. It is targetted mainly for OEMs which may need different ratings that our standard products offer but can be used in single setups too. You are right that troubles may arise when inexperienced hobbyist tries to use it. However, we'll see how it goes and do something about strategy if situation start getting bad.

I just reviewed this thread quickly...

I am glad to see that hi-voltage parts are being developed, too :) Can't wait to see what will come out!
M preference would be in the same order as you quoted them, the monolithic being the "worst case"

A question: why do you want to use MOSFET-s ? at higher voltages they are sensitive, as well as high currents are problematic. I'd suggest IGBT-s they are superior, when it is about voltages of ~300V and currents of tens of Amps
(a mains-powered 220V AC version would have ca. 300V bus voltage)

Quote:

---

Originally Posted by hunserv

A question: why do you want to use MOSFET-s ? at higher voltages they are sensitive, as well as high currents are problematic. I'd suggest IGBT-s they are superior, when it is about voltages of ~300V and currents of tens of Amps
(a mains-powered 220V AC version would have ca. 300V bus voltage)

---

MOSFETs are superior in about 0-150V range and IGBT 250-xxxxV. In about 150-250V range there is no clear winner so it depends on application.

VSD-A has IGBT stage and VSD-E MOSFETs. I could consider IGBT for higher voltage VSD-E, too.

Count me in if you can produce a drive that will actually supply 200V at 20A continuous, and at least 50A intermittent.
A dumping circuit would be nice, so it won't fry, when the inevitable crash happens.
Keep up the good work.

Quote:

---

Originally Posted by Tig

Count me in if you can produce a drive that will actually supply 200V at 20A continuous, and at least 50A intermittent.
A dumping circuit would be nice, so it won't fry, when the inevitable crash happens.
Keep up the good work.

---

Dumping circuit is not necessary to protect drive in hard stop, no matter how much energy is stored in motion. Also regenerative voltage won't hurt it since drive disables before voltage goes too high.

Attach an lead screw to the moon and use servo motor with VSD drive to crash moon on earth near light speed. No damage to VSD will happen due to continuous current control (unless drive gets crushed by the moon).

Hi, I'd have some questions.
-I sadly see on yor site, that VSD-A is no longer produced. does it mean that some new is coming out soon?
-Within this planned modular design, are you planning a powerstage which supports regenerating to supply?
-Will the new design have the same PID algorithm as previous ones?
-Is there any reference where we could see which exact PID scheme it uses? (hopefully not the "textbook" basic one, but some more advanced, as differential term acting on feedback signal only, etc...)
-Are you planning some more features, e.g. making real-time motor-load outputs (I asked it earlier)

Thanks,

Aks

Quote:

---

Originally Posted by hunserv

Hi, I'd have some questions.
-I sadly see on yor site, that VSD-A is no longer produced. does it mean that some new is coming out soon?
-Within this planned modular design, are you planning a powerstage which supports regenerating to supply?
-Will the new design have the same PID algorithm as previous ones?
-Is there any reference where we could see which exact PID scheme it uses? (hopefully not the "textbook" basic one, but some more advanced, as differential term acting on feedback signal only, etc...)
-Are you planning some more features, e.g. making real-time motor-load outputs (I asked it earlier)

---

-Yes, we should get VSD-E final protos on week 32 and the first production run soon after that.

-Regenerative braking "shunt regulator" can be built without help of drive logic. However, one output pin could be assigned for regen brake control if needed.

-Same PID but also a PIV algorithm. I try to find time to document both well.

-Hardware doesn't have analog outputs. Digital output pins could be used to monitor conditions (follow error warning for example). SPI bus can be used for monitoring, too.

A phase controlled SCR bridge can be made for peanuts and handle enough to blow the wiring off the wall without failing.
Ultra simple drive, no big capacitors needed. Just a few snubbers.

Do you need full 4 quadrant control, or is reverse with a switch/contactor OK?

Any comments?

Quote:

---

Originally Posted by Xerxes

...Attach an lead screw to the moon and use servo motor with VSD drive to crash moon on earth near light speed. No damage to VSD will happen due to continuous current control (unless drive gets crushed by the moon).

---

So, at this point someone is suppose to jump in with all the numbers required to accel the moon to near lightspeed before it hits the earth, and how much energy that would require given the mass of the moon, and then congratulate Xerxes on building a drive that can output that much power. ;)

neilw, SCR drive is Ok for turning pulleys in spindle and other raw power applications, but not for positioning servos. No enough control bandwidth.

--

The moon acceleration might require a few VSD's connected in parallel :)

Xerxes..

Oops... too late here. I missed the thread (of things) :tired::o Go fast IGBT's driven by IR parts. in a H-Bridge.

SPI bus monitoring is OK, too imho, and if someone wants analog can hook a microcontroller on it :)

And some more questions...
-When do you think you will have time to describe these algorithms?

-What development time do you predict for the modular controllers?

I am really keen on it, I can't say enough that I was impressed by the performace/cost ratio of the VSD-A :)

As I see most guys here are beyond the hobby-category, (low power) but there are still few freaks like me, who need higher outputs, which are closer to the industrial grades...

Quote:

---

Originally Posted by hunserv

SPI bus monitoring is OK, too imho, and if someone wants analog can hook a microcontroller on it :)

And some more questions...
-When do you think you will have time to describe these algorithms?

-What development time do you predict for the modular controllers?

I am really keen on it, I can't say enough that I was impressed by the performace/cost ratio of the VSD-A :)

As I see most guys here are beyond the hobby-category, (low power) but there are still few freaks like me, who need higher outputs, which are closer to the industrial grades...

---

I'm doing my best to get detailed block diagrams in VSD-E manual which should be ready soon.

About VSD-Core I don't have any schedule yet. There are other things to be finished first. Product release date could be within a year.

xerxes did you get a chance to read my PM?