Viper 200 Initial Impressions and Experiences

[sbalder]

My Viper 200 Servo Drivers arrived the other day and I have had the chance to work with them over the past 2 evenings. I thought I would post some of my experiences and I encourage others to do the same.

These will be for a Milltronics Partner mill that I am restoring. I am using the original DC brushed servos, made by SEM, rated 4000RPM at 140Volts and 11 Amps. One motor had a bad encoder, so I have replaced all encoders with 300-line units from US Digital. I have worked with several stepper systems before, but this is my first experience with servos and I am learning as I go.

I started by making a power supply and the necessary cables. Larken included all of the mating connectors for the board to make it easier (although I do wish they had included enough jumpers for setting all parameters on the board- c'mon they are like a penny a piece!) Salvaging transformers, caps, and a rectifier from the Milltronics hulk I built a power supply that runs on 120V and outputs a clean 86Vdc. This should give me about 2500 RPM or 250IPM as the mill is currently geared- well more than enough speed.

The instructions are adequate, but not exceptional. I'd suggest a few changes:
+Adding to the "connections" photo explanations of all the lights
+Better details on configuring the pc terminal program (including a screenshot)
+A true step-by-step set of setup instructions (or even an online video!)
+Clearer instructions on how to wire encoders with only one wire per channel

My testing: I hooked everything up, testing the low voltage control stuff first. The pc interface using a serial terminal works well and provides nice clear information. The default values in the drive were set way too high, causing an oscillation (I=6). I hooked a function generator up along with some switches to provide manual step and direction for testing. Upon first startup the motor shook violently, and I reversed motor connections per the manual to correct the condition.

I fumbled blindly through my settings, not knowing what is "normal". I wound up with P=700, I=1, and D=1400. This gave a strong motor hold at stop, a crisp stop from 500RPM, and no unwanted oscillation or humming. I've looked at current/response curves on different websites, but was unable to get my oscilloscope to pick up a clean trace. I'd like a slightly more scientific way to test my settings other than "that looks about right". A test point that I could hook my scope to, like in the Gecko drives, would be nice.

All in all, a nice product so far. I'll post more as I get more experience.

[Larken]

Wow, it didn't take long for you to get it running. I just mailed them last friday.
Have you got the mill moving with mach 3 yet?

Don't forget to connect the fault output back the the Estop, so it stops your machine if you overload it.

[sbalder]

Like I said, they were easy to setup so it really didn't take long to get them running on the bench. The mill itself is still in pieces completing the restoration so maybe another week or so. As for software I am sticking with TurboCNC for a couple of reasons.

Do my PID settings seem reasonable or are they too high? Also, is there a way to test or tune these setting that is better than "feel"? I tried to hook up a scope to measure current where it would do a single sweep when the direction is changed, but mine requires the external trigger to be on the same ground path so I couldn't make it work.

[Larken]

Do my PID settings seem reasonable or are they too high? Also, is there a way to test or tune these setting that is better than "feel"?

No, they seem about right. Did you change the L (looptime) parameter. This is the one that adjusts to the motors responce time.

I hope to have a software tuning program that displays error vs time soon.

What diameter are those motors ? size 42 or bigger ? I was looking to buy one of those mills for my shop years ago, but bought a Leadwell V40 instead.

[sbalder]

These servos are Nema 42, I believe. I moved the loop timing up to 5 and found 0 to be the best "I" factor.

Being a novice to servos I didn't think about using the servo error reading for tuning. I wrote a simple VBA application in an Excel workbook that reads the servo error 1000 times. While it is running I switch on the pulse generator (to start the motor spinning) and flip the direction a few times. Then it is simply a matter of plotting the direction change. Based on this I increased my Derivative 100 points and got what I hope is a nice pattern:



The servo error hovers around 70 while running. For fun and education I put in some deliberately bad parameters just to see what it would look like. Here's a P of 100 and a D of 200:





The curves look about the same, but the error hovers around 500!
Then I set a moderate P (400) and a very low D (100). Here you can the servo underdamped as it oscillates back and forth during a reverse:








Anyone who wants this spreadsheet please send me a PM.

[Larken]

So, your reading the error over the serial port? How many times a second? Does it do the HEx to decimal conversion ?
Are you getting a good comunication with low error rate. I program mostly in Delphi for the PC, but i don't have a good serial interface routine.

I am going to add a FeedForward setting soon to reduce running error. Without a Feed forward, a step/servo will produce a error proportional to speed.

[sbalder]

I am reading the error through the serial port. I moved the application to Visual Basic 6 so I could add time measurements and got an average of 47 milliseconds per reading with a standard deviation of 6 milliseconds. This is about 5 readings per second slower than just in the VBA Excel worksheet, so I am inclined to keep it the worksheet where I can get a solid 25 readings per second. This is not stunning resolution, but I think it is adequate for tuning purposes.

The program reads the returned text string, and parses it into the Excel workbook. Then I use the Hex2DEC worksheet function to convert the output into decimal. I'm most comfortable with graphing in Minitab, but obviously once this data is in Excel you can use Excel's graphing functions to make the same kind of charts.

As for the serial interface, I'm kind of a newbie at that kind of programming. I came across an ActiveX freeware control that is similar to Microsoft's non-free comm control called NetComm. Anyone using my application would have to have to download and install this control here:

http://home.comcast.net/~hardandsoftware/NETCommOCX.htm


I put the Excel workbook application here:

http://www.geocities.com/eba1972/ViperServoAnalyzer.xls

[DennisCNC]

How is this project going?

[sbalder]

I have had to put more work into the basic machine than I thought I would, taking nearly everything down to the castings for cleaning and refinishing. If this thing was any filthier the Discovery Channel would be featuring me on Dirty Jobs!

I'll have the basic machine assembled by the end of the weekend, all repainted and with servos, sensors, and lubrication installed. Then I can finish fabricating and installing the electronics!

[sbalder]

Wow how my life is passing me by so fast! I can't believe it was 6-29 when I last posted on this. I got diverted on several other projects and the work on the Milltronics mill took more effort than I thought. Anyway last night I finally finished the mill and got everything working! The Viper servo drives have performed well in initial tests, but the Y-axis was "jumpy" and weak at low jog speed only. Hooking the serial cable up I was able to adjust the parameters (mostly lowering the derivative) and got all 3 axes running.

This weekend I'll finish fine tuning and cut some test parts. I'll post a few pictures of the final setup.

[sbalder]

I got to work this weekend and tied up all the little loose ends on the mill project, such as calibrating each axis and cleaning up the horrible mess I have made in my shop! Anyway here are a few pictures of the completed mill/electronics setup.

[routerman]

Nice job, looks clean. What program is that running ?