Realistic Gecko 320 following error

[eldata]

I have one of these drives and know that the maximum position error after a move or disturbance is + or - 1 step. However, I have no idea of what to expect in terms of following error while moving/cutting. What range of following error can be expected from these drives while accelerating , cutting and decelarating?

[HillBilly]

There is alot to consider to answer that.

How many pulses per inch?
What load is being moved?
What speed?
Servo wattage?

Are the first to come to mind.

Darek

[eldata]

>There is alot to consider to answer that.

Are there any step compatible servo drives on the market that give an indication of following error (FE)? Seems as though FE is only used to fault the drive when it reaches a given threshold. Isn't it odd that servo drives should know the FE at all times yet not share this information with the user?

IMO, FE is probably the most important thing to know while cutting, otherwise, you have to use some low cutting feedrate and hope for the best as with open loop stepper setups. Mach3 seems to allow encoder connections to the computer, can it be used to measure following error while stepping or do you need two computers for that?

[eldata]

>There is alot to consider to answer that.

Just took a look at the 320 manual and all indications seem to suggest that measuring the voltage at the position error test point should give both the position error (no step input) and the following error (step input). A peak hold DMM or storage oscilloscope would come in handy here.

[CJL5585]

Originally Posted by eldata >There is alot to consider to answer that. Just took a look at the 320 manual and all indications seem to suggest that measuring the voltage at the position error test point should give both the position error (no step input) and the following error (step input). A peak hold DMM or storage oscilloscope would come in handy here.

+ or - 128 (Count) of Leading or Following Error before drive faults. This is what I have read. I believe it is all based on the max pulse frequency from the encoder which if I remember correctly is about 20,000 hertz.

My Gecko 320's have never given any faults during operation, and do not oscillate when stopped, at least enough that I can see or feel any oscillation.

A 500 cpr encoder is the max recommended, although some people attempt to install larger cpr encoders. (Internally the Gecko's use the leading edges and trailing edges of the pulses to get a resolution of 2000 lines on a 500 cpr encoder.)

I use 250 cpr encoders which the Gecko multiplies by 4 to give a resolution of 1 part in 1000.

Hope this helps.

There is a thread on this site relating to this subject, but I do not have a link to it at this time, and cannot do a search while making this post.

.

[kreutz]

+ or - 128 (Count) of Leading or Following Error before drive faults. This is what I have read. I believe it is all based on the max pulse frequency from the encoder which if I remember correctly is about 20,000 hertz.

Actually what limits the maximum error in any direction is the method Mariss used to calculate the difference between actual and desired position. He uses two 8 bit counters one for input step counting (step + direction(up/down)= desired position counter), the other for encoder pulse counting(actual position), them he uses an 8 bit adder to subtract one count from the other, its output (error) can go only from +127 to - 128 (without overflow), so that is the error limit. There is absolutely nothing wrong with that.

[eldata]

>There is absolutely nothing wrong with that

Most step compatible server drives I've seen take a similar approach except the MCU based ones of today typically use 16 or 32 bit counters and allow for greater overflow thresholds. My grouse with just about all of them is the open loop stepper mentality that influenced their design.

These drives all know the following error at any point in time during step input, otherwise, they wouldn't be able to fault when thresholds are reached. If the user got an indication of this error then one would not have to choose some overly conservative feedrate while cutting as is often the case with open loop steppers. In other words these designs are not leveraging, and passing on to the end user, all the benefits of a closed loop setup.

[ger21]

Originally Posted by eldata If the user got an indication of this error then one would not have to choose some overly conservative feedrate while cutting as is often the case with open loop steppers. In other words these designs are not leveraging, and passing on to the end user, all the benefits of a closed loop setup.

You get what you pay for.

There is aboard available for use with Mach3 that monitors actual position and will pause the machine when a user defined position deviation has occurred.
http://rogersmachine.net/encoderinterface.html

But you'll still need to use an "overly conservative" feedrate to keep from exceeding the predefined error tolerance. But at least you'll know when you exceed it.

[eldata]

>But you'll still need to use an "overly conservative" feedrate to keep from
>exceeding the predefined error tolerance.

Now we're getting somewhere....Cutting to the chase.....Just how good is the 320's integrator during step input? By good I mean how fast can it charge/discharge and track? If it is as slow as it is with no step input then the 320 should have poor following error characteristics even with no cutting load.

[ger21]

When I said predefined error tolerance, I meant the tolerance specified for the encoder board I mentioned, not for the drive. Just in case there was some confusion there.

[HillBilly]

Following error will be present in any closed loop servo system. Take any commercial CNC control and display following error. Run some linear moves with different feedrates ( do a rapid move also.) and watch the value go up the faster you move. The more power your system has the better it will deal with this.

Darek

[eldata]

>Following error will be present in any closed loop servo system.

They are not all created equally. Integrator strategies from high-end servo systems are trickling down to step compatible drives.

>Run some linear moves with different feedrates ( do a rapid move also.) and
>watch the value go up the faster you move.

I'm doing that with a drive that displays the FE and it is below 6 steps while loaded and during acceleration and deceleration provided it isn't loaded to the point of faulting. The 8 bit PWM drive does boast a relatively high maximum stable proportional gain of 16 and an ultra fast (.0007 sec response time) tracking integrator.

>The more power your system has the better it will deal with this

True for a drive with a "pedestrian" integrator depending on gain too much. Torque requirements do not influence the FE of the above mentioned drive in any way whatsoever due to integrator characteristics.