Viper servo drive overview

[sbalder]

Sometimes I think I am getting to understand something and then realize how little I have learned. After reading the last several posts here's my somewhat lengthy question-

If you have a servo with a 300-count encoder and a 3000 RPM limit then (without a step pulse multiplier) it would take 1200 steps per revolution or 60khz to achieve maximum speed.

If you have the same servo with a 3000-count encoder (and again no step multiplier) then it would take 12,000 steps per revolution and 600khz to achieve max speed. If your servo drive or pulse source (computer) was unable to handle this speed and you applied a 10-times step pulse multiplier you would be back to 60khz and 1200 steps per revolution. Right???

In the above situation, is there still an advantage in using the higher count encoder? If the drive can handle the higher encoder count, won't it react to a smaller error? So if the drive begins to react to anything greater than error count 1, with the 300 line encoder it will take .3 degrees movement to generate one error count, but only .03 degrees with the 3000 count encoder.

Is this correct, or does a 300 count encoder equal a 3000 count encoder/10x step multiplier in resolution AND drive response?

As for resolution my Bridgeport has original steppers half-stepping for a maximum resolution of .0005". The ways and screws are good and I am able to route pc boards, engrave text, and machine bearing blocks in aluminum. The screws are not nearly as good in the Milltronics mill I am working on, but I will have a servo resolution of .000083" using 300 count encoders. If I had the 1800 line encoders mentioned on my servos as they are geared the resolution would be .000013". I don't think anything that makes the return trip through my shop door will ever demand that kind of precision!

[Xerxes]

sbalder, you're correct in calculations.

Higher resolution encoder has some advantage over low resolution one. Imagine encoder with very high or "almost infinite" resolution. It would equal analog encoder which eliminates servo dithering completedly.

[JerryFlyGuy]

If I had the 1800 line encoders mentioned on my servos as they are geared the resolution would be .000013". I don't think anything that makes the return trip through my shop door will ever demand that kind of precision!

Well, we each have a different system. W/ a 1800 line encoder my mill would have a resolution of 0.00008726" it's more than I need, but I can still hit thousands at that resolution, I was shooting for 0.000100 per step but had to take the closest number under that, and 1800 line encoder was what came out of it. The thing is, if my G100 can output a high frequency pulse train, then I might as well take advantage of it, like was mentioned.. it will give me a smoother motor and it's not stressing the system in ANY way.. Do I ever plan to machine something to a 0.00008" tolerance.. not on your life.. I'm just making sure that every component can run at the same level in the system. Thus I expect to have a well balanced system. Running a G100 w/ a 300 line encoder makes as much sense as.. well... um... a V8 engine mounted on a BMX bike frame..

[routerman]

Step Multiplier = Lower resolution encoder.
Why bother with a step multiplier when you can just use a lower res encoder?

You should have a higher encoder count than the steps you give it. This gives the servo a window of error correction.
A high-Res encoder with a step multiplier to lower the step frequency , will improve accuracy and smoothness.

[JerryFlyGuy]

You should have a higher encoder count than the steps you give it. This gives the servo a window of error correction.

Most servo systems [that I know of] already have this built in, Gecko's have 128 count's set right into them, which if exceeded will prompt a fault. Rutex are setable to whatever you want them to be, messing w/ the encoder counts and step multipliers isn't the place for this [as far as I can tell]

A high-Res encoder with a step multiplier to lower the step frequency , will improve accuracy and smoothness.

I disagree, my understanding is the higher the step freq [to a point] the smoother the servo operation [granted the drive can accept this higher step count]. It still makes zero sense to me, to buy a high count encoder and then handi-cap it by using a step multiplier.

I see it as driving a 5 speed car in 3rd gear, you have more potential but choose to use gearing 'to your advantage', it makes no sense.. Plus, it will REDUCE accuracy, not improve it.. there is no way that a lower count encoder [which is what a step multiplier is In-effect] is going to have an increased accuracy over a higher count encoder. Thats like saying... ' For Highway travel a first gear is better than fifth'..

[routerman]

Most servo systems [that I know of] already have this built in, Gecko's have 128 count's set right into them, which if exceeded will prompt a fault. Rutex are setable to whatever you want them to be, messing w/ the encoder counts and step multipliers isn't the place for this [as far as I can tell]

I'm not talking the error trip range, (which BTW is programmable in the viper 1-4000 counts).

Your example using a car is a good one, but not the way you say.
A step multiplier is like having the engine turn 20 times for the wheels to turn 1 time. Its like gearing.

[JerryFlyGuy]

I'm not talking the error trip range, (which BTW is programmable in the viper 1-4000 counts).

Ok, then your going to have to explain to me what you mean by "window of error correction"

Your example using a car is a good one, but not the way you say.
A step multiplier is like having the engine turn 20 times for the wheels to turn 1 time.

True, except that in this situation it's backward, the tires and the transmission can turn at the required speed [the motor and the encoder] but the engine can't drive them at this speed, so you need to gear the engine [drive] so that it turns once [steps once and then is mulitiplied] for 20 times on the tires [ In-effect ].

So then tell me how does the step multiplier know how to interpolate between steps? Ie: The computer can only output 1 step per inch because thats the limit the drive imposes on the system. [ I'm REALLY over exagerating-- I know.. ] but we have an 1800 line encoder which will have a resolution of 0.0001" per line in Quad.

How does the computer tell the servo to move 0.0002" ?

Its like gearing.

Sure it is, the problem being is, your control has less steps per inch, your resolution suffer's because of it. I still see it as a bandaide fix for the problem.

[Mike Nash]

If I gave you a set of digital calipers that only read out 0-6" as 0,1,2,3,4,5 or 6 and asked you to set them for 3", how close could you get? Would you be at 3" when you they first changed from 2 to 3 going up, or would 4 to 3 going down be okay? Now if they read out to 0.001" then I expect you to set them to 3.000" even though I said only 3". Now figure that servo that's given 0.0005" steps but can resolve 10 times better than that on the feedback. It's not going to be hunting back and forth around that setpoint by 0.0005" increments as a minimum, but hopefully by smaller increments. Just because your indexer can't put out the steps fast enough to use really fine increments doesn't mean it's not useful to be able to resolve them.

Just my take on this.

[JerryFlyGuy]

Mike I can kind of see where your going w/ this, basically your saying [from my analogy above] that if the drive is limiting my steps so that they are 1 step per inch yet the servo has an encoder which gives a resolution of 0.0001" that it will move in 1" incriments and be accurate to less than 0.0001"?

[Mike Nash]

That's the way I see it. The servo's job is get to and hold commanded position. It should do that to the resolution of the feedback regardless of the set position resolution. While I remember all the blah, blah, blah from school about number of places of accuracy being no better than the least accurate value, I don't buy into it in all cases. Kinda like taxes. It ticks me off when I have to pay 49 cents more for the year than I should have to because they want to round to the nearest dollar. Of all the years I have checked, it averages in their favor.

Mike I can kind of see where your going w/ this, basically your saying [from my analogy above] that if the drive is limiting my steps so that they are 1 step per inch yet the servo has an encoder which gives a resolution of 0.0001" that it will move in 1" incriments and be accurate to less than 0.0001"?

[Larken]

Jerry, i think you have the impression that the encoder resolution is going to be your machine resolution. That would only be if servo error was ZERO all the time, but a servo works differently from a stepper.

Depending on how fast you are cutting and the load on the servo, (also the PID gain parameters) the drive tries to keep the error as low as it can, but there is usually always some error while the machine is moving.

But a good rule of thumb is have 4-10 times encoder resolution for required machine accuracy.

[JerryFlyGuy]

Nope, I think I stated someplace [up above] that I had no intention of hitting my 0.00008" encoder resolution. I know I never will. Due to thermal expansion my machine grows 0.00288"/degree [change in temp] There is no way that it's going to hit some small number like 0.00008". Much like a stepper using microstepping won't be able to count on it's 2000 [10x microstep] steps per rev being 0.18deg each. The point being is that I would like to have accuracy down into the 1 or 2 thousands range. I would like to have the smoothest system I can get, and I'd like my drive to be up to the task of which the controller can put out. In the end I want a very well balanced system. There is no point in my buying a expensive yet VERY capable controler if the rest of my system isn't up to the task of working with it. That is like my V8 in a ?BMX bike frame analogy.. it would make no sense..The suggested bandaide fix of a step multiplier is a solution to a problem that I don't have, given I pick the correct drive for the job.

I think it's great that your producing these drives! It's innovations like yourself that have got affordable CNC where it is today! I think that they would be a great product to use in conjunction with a PP, however they don't seem up to the task of coping with my system and it's pulse rates, esp if I go ahead and double my reduction over what I've currently been using.

Don't take my comments as a kick against what your producing, originally I just wanted to confirm that they would work w/ what I'm using. This whole conversation evolved from there with others trying to offer solutions to make it work.