Looking for Testers for my new "Step2Linear" (Step/Dir to analog ) module

[Larken]

2- As soon as I release the emerg from Mach3, the motor runs away. I have switch encoder A and B as suggested in the manual without effect. It's almost like noise out of the step to linear (more or less
24mV) or the breakout board maybe. <b>If I turn on the scope feature, the encoder count goes up, </b>the error count goes up to and the steps stay at 0.

The problem you are having is you ARE getting pulses from the encoder ( as you said the encoder count on the screen changes) The random encoder pulses are making the step2lin produce a output.
Try this test. Disconnect the encoder and fire it up. With no steop and encoder pulses the motor shouldn't move . Then lightly jog mach3 and you should be able to make the motor go left and right.

Now, to test your linear scale , disconnect the analog output from the step2lin to the drive. With the encoder connected and reading it with viper tune, use a small voltage (a 1.5volt battery) to make the motor move the axis left and right. You should see the encoder count increase and decrease as you make the axis move left and right.
'
It could be noise being picked up by your circuit producing the encoder pulses. Keep the encoder wires away from power wires. Also what PS are you using to 5volt power the Encoder circuit ? It would be best to use a isolated wall adapter.
BTW, did you add pullup resistors to the outputs of the 393 comparator (i forgot to add) ? Having a scope would make it alot easier to get the scale working.

Larry

[mike_Kilroy]

It is great that you are making a small low cost position controller for folks to use with the likes of Mach3 for simple CNC use! Until your unit, the cost was much higher for a position controller from the likes of Galil, Delta Tau, who offer complete MOTION controllers, not just a single axis position controller. Then too it is much lower cost than the position controller products that some others like Machmotion.com & others use; for instance the Apallo or Interpreter 1000 position controllers, even with their 3-4-5-6 axis in one packages. One reason for their much higher cost of course is their update times are typically in the low single digit milliseconds or faster rather than step2linear's fastest 500msec loop update time. But your design is what it takes to reduce the cost of the position controller by a factor of 10 or so as you have done. Folks just need to take this into account on the performance they will get; it will not perform like a position controller with usec update times. That said, Good job!

You call this a converter but should maybe instead call it what it is: a position controller.

verb (used with object): to change (something) into a different form or properties; transmute; transform. Your step2linear is not a converter, it is a single axis positioner. It accepts a position command, compares it to an encoder feedback, then sends a position error analog output which it appears you call a command for speed. I cannot tell if you expect this to be a command for speed or rather a command for current. You should identify which is a better match for your positioner. I suspect it would work best with an analog velocity loop drive as you do not include a velocity loop in your unit.

I am curious why you feed motor current back to this unit for?

[Larken]

Yes, the name Step2linear is a bit lame. I will add positioner into the description, but it doesnt't have motion commands built in , although is could.

The position loop speed is programmable and the Slowest loop time on it is 500mS (500 loop /sec) and the Fastest is 250 uS (4000 loops /sec). If there was a need, i the code can run up to about 10,000 loops/sec. ( In the viper drive code the current sampling loop is running with the pwm at 20khz )

The current input is for limiting the output Torque and for tuning and display in vipertune. I don't know too much about current mode drives, thats why i want testers to try it with different drives and see how it responds. But the analog output is designed primarily for velocity mode where the 0-10 volt controls the speed of the motor., like the PWM does in my viper drives.

Larry K

[mike_Kilroy]

Yes, the name Step2linear is a bit lame. I will add positioner into the description, but it doesnt't have motion commands built in , although is could.

The current input is for limiting the output Torque and for tuning and display in vipertune. I don't know too much about current mode drives, thats why i want testers to try it with different drives and see how it responds. But the analog output is designed primarily for velocity mode where the 0-10 volt controls the speed of the motor., like the PWM does in my viper drives.

Larry K

Agreed, you should call/add 'position control loop' or some such to name to identify that it is not a converter that just changes one set of units to another. The output is totally modified from the input the input so no converter action there. And right, it is not a motion controller so that name is not right. The day you start adding motion commands to it, you begin competing against Galil, Delta Tau, etc, and THEY have 150 engineers and 40 years head start on you - I doubt you really want to go there.....
A typical Motion controller has those other functions too.

Ok so you let the external drive be both current loop and then velocity loop around that. I do not believe you will be able to control the current in a position loop unit - too far removed. Why no one else does that. Current control is to be done in the current loop. Even if you have all your customers dial your loop update time to 250usec (4khz), that will not be fast enough to try to change your velocity error analog output signal to limit current effectively in a 1-2khz current loop. In order to effect current control in your position loop, you would need an update rate effectively 10x faster than the current loop bandwidth as a m inimum, then I would serious have to think about how you get THAT answer filtered down through the 50hz bandwidth velocity loop to the current loop. Just too far away and too slow. You might be better off putting your effort into the logistics control of the loop as you are finding - like how to zero the PE before re-enabling, etc. There are like 75,121 similar things to consider to get reallyi good control over your position loop - way too many in my opinion to take non existent time out to try to do the current loops job. Let it limit the current itself I'd say.

And why are you limiting your market to non-spindles? I assume you know the firmware code or have a really good hot-shot who does? So just turn off the encoder counts - don't add them into the summing bucket - walla, you have a CONVERTER rather than position controller; it is a simple freq->analog 0-10vdc CONVERTER; with the dir input it is a simple freq->analog +/-10vdc CONVERTER. I have papermill applications I supply $ 300.00 units that do this to simply get a master line speed signal for my servos for #/inch tension control. Your price would be 4 times less for the same function...... market idea? Stop telling those folks on here you cannot run a spindle motor with mach 3 with this!

[G59]

Stop telling those folks on here you cannot run a spindle motor with mach 3 with this!

I agree 100%.
Larken opened the door to a new way of controlling the spindle, and I tried to tell him, but I think he wasn't too interested. Even after I told him it could be done and reliably, as I currently have a 5 horse dc motor that is controlled by Mach3, in the same fashion.

[Larken]

Ok so you let the external drive be both current loop and then velocity loop around that. I do not believe you will be able to control the current in a position loop unit - too far removed. Current control is to be done in the current loop.

No, i would only use the current output from the drive for a Load control, or overload monitor. If the current level stays above a set level for more than a second or 2 then it would cause a fault. Yes, it would be too slow for current mode control.

You might be better off putting your effort into the logistics control of the loop as you are finding - like how to zero the PE before re-enabling, etc.

Yes I do clear the encoder count before re-enabling the drive and it does a soft start on the output signal too. ( i got the idea reading from a Granite devices post )

The day you start adding motion commands to it, you begin competing against Galil, Delta Tau, etc, and THEY have 150 engineers and 40 years head start on you - I doubt you really want to go there.....

lol, actually, almost all the products here on cnczone are created by 1 person. Gecko, mach3, granite devices, Kflop, (my stuff), etc... I think in this economy , its the big guys that are worried.

And why are you limiting your market to non-spindles? I So just turn off the encoder counts - don't add them into the summing bucket - walla, you have a CONVERTER rather than position controller;

People want the closed loop so they can do rigid tapping , but there is a problem running a servo controller on a spindle. When running at high rpm under load the spindle will load down, since the powersupply usually is close to max and doesn't have the head room to fully correct the load. This 'Slip' will accumulate error until the drive hits its max error limit and trips. Any spindle will load down and loose some rpm under load at high speed. I guess you could disable the max error trip setting.

[ech2o2]

Hi,
it turns out that my linear Heidenhain are producing a small voltage at the diodes: it swings from +360 to -360mV. As per our discussion, i modified your sketch with a simple voltage divider at pin 1 and 5 and the outputs 2 and 6 are now swinging from 1.3V to almost 2V. I used the pot to set the comparator to approximately 1.7V and it started to count the pulses no problem. I get 500pulses per revolution of the screw. I get a very steady pulse and get back to zero when reversing. That is when i turn the motor by hand, as soon as i apply power it counts 20 000 per revolutions! Obviously it is very sensitive to electrical noises. The converter is built on a breadboard, it is powered by the steptolinear which is powered by a separate power supply. The encoder wire is 10ft long, and runs along the motor for about a foot (original setup). My breadboard is on the cabinet containing the transfo and drives about 4 feet from the motor. I made sure that the encoder wire is shielded to ground, however I am using about 8inches of unshield twisted wire to get to the breadboard and about 6inches of unshielded twisted wires from the breadboard to the step2linear.
I have tried to use 200ohms resistors at the end of the encoder terminals, no changes. I have tried a few capacitors between pins 1 and 2 and 5 and 6. If the capacitor is too large, it misses some pulses by hand but still gets out of control when the motors runs. Any other ideas to try to curb that noise?
Thanks
Eric

[mike_Kilroy]

Replace lm393 with proper line receiver chip instead; it can be adjusted to give you your gain too.

[Larken]

Did you try putting the 1meg feedback resistors back on ? Without that the comparator will have jitter at the crossing point.

You may need a lowpass filter between the photodiodes and opamp

[Al_The_Man]

Larry, what is the nature/configuration of the step/dir inputs?
Can the P.P. drive these direct, using with the 5v supplied from the PC?
Al.

[Larken]

Larry, what is the nature/configuration of the step/dir inputs?
Can the P.P. drive these direct, using with the 5v supplied from the PC?
Al.

Yes, you can power it from the parallel port (a bob is recommended , but direct connect is OK for testing).

IF the PC has a 3.3 volt mother board and PP output, you have to reduce the (5volt from the USB) down to 3.3 volt , or the opto won't turn off when the level is high.

Larry

[ech2o2]

Hello, I gave up for now on converting my linear encoders. I got it to the point where it seems to give me square waves but since I don't have a oscilloscope it was hard to see if the circuit was really efficient. I got the motor to gently oscillate back and forth but only on occasions. It was hard to get steady results and I don't know why. The translation of the linear Heidenhain to TTL will be a project on its own at a later date.

I have decided to use a cheap, single ended rotary encoder from automationtechnologies.com mounted on the motor. I used the resistors as per the diagram in the instruction manual, I plugged it in the step2linear and the motor ran away, as you recommended, I reversed the channels and VOILA! Super stable.! After playing around with the vipertune parameters, i found that the best setup is: loop time to 8, as my motor is fairly big, kp=50, kd=100, ki=1, step=1. From creeping speed to 80in/min it is very stable. I have not tried any faster but I will eventually. Once I used the rotary encoder, the step2linear was actually very simple to use, interface with and setup. I need now to set up the other 2 axis but I am very confident that it should work.

On a different note, as I am new to Mach3, it took me a while to figure out why the midi and the step jog did not work. the continuous jog was the only thing I could use to make this machine move until I figured out that you must hit the stop button on Mach3 for Midi and step jog to become active!

I'll put some pictures and videos once I am a little further along. Thanks for all the support.

Eric

[Larken]

Great . glad you got it working.
Are you using the AMT encoder ? I see now they make a differential version the AMT113Q.

Larry

[ech2o2]

First aluminium chips last night! And it went very well.
I am using the cheaper version encoder AMT 102 or 103 I can't remember (single ended anyway). The motors vibrate a little bit when on hold. I have tried to adjust the loop time parameter, proportional and derivative gain but it does not seem to have any effect. Only that if i increase proportional gain to much it really vibrates! So for now kp=50, kd=100 and loop time 8 seems to be the best. Everything else is at 0 or 1. the 3 motors are really smooth when running.
Thanks again.

Eric

[bhurts]

If you move to us digital or similar glass disk encoders I'd bet you could get much better performance. The software I tune with allows me to see exactly what the encoders are doing. The amt encoders are repeatable but there performance is sub par. They have a lot of dither while idle that causes the motor to hunt constantly. They are also the only encoders I have tested that show small spikes due to being non linear.

Ben

[ech2o2]

Thanks Ben,
I understand. Larry from Viperservo is also saying that us digital is a better product. Since I am a new at cnc conversion and trying a fairly new product from Viperservos to convert PWM to +/-10VDC for my older drives I wanted to keep the investment to a minimum. I am now convinced that the step2linear will work for me and I would not hesitate to invest in a better encoder. I'll simply keep these cheaper encoders for another application or as spares.
Eric

[Larken]

Eric, I added a deadband setting to reduce dither. The 'u' command can be entered with the terminal.
What is does is when the motor is not turning (idling) it allows the drive not to correct a small error (1-5 counts)
Then the motor is turning its not in effect.

EG: so if you enter
u3
and the drive will not correct if the error is less than 3 counts.
the parameter will be saved with the burn command.
you can also see the setting when displaying parameter set 1 or 2, as 'dz' (deadzone)

Btw, I did notice that there is more dither with the step2lin than on my Viper drives which hold very still, but i still with do more development on the firmware soon. I want to get a couple more amps and motor combos to try (ebay) over the winter.

Larry

[ech2o2]

Hi.Hope everyone is well. I have explored cad/cam software to try my new set up. I have found that I have some imprecision due to a bit of backlash in the screws and I am getting a bit of jerking is some of the movements not necessarily during a change of direction. I used vipertune to follow the error between steps and encoder and came to realize that my step2linear was far from being properly set up. By increasing some of the parameters I have reduced the error to close to 0 for most of the movements. The machine seems to run much smoother and this jerkyness I had seems to be gone. It is still a work in progress but I am getting pretty good results with the following:

Loop =7
proportional=200
derivative=550
integral=3
feedforward=450

That being said, as long as my usb cable is on the step2linear, i can change the parameters, update them and read them back and burn them...or what looks like i can burn them to the drive. The motors do behave differently so it looks like i do interface with the step2linear. Without turning the power off,if I take the usb cable off and back on, relog with viper tune, a push of the read drive button seems to read the original default parameters? Yet, the step2linear seems to perform as well as with the updated parameters...I don't understand. Also, although i reset the step2linear with the low voltage input, the trip error will not go away? Instead of reading back my input, it will read ERROR?
Thanks
Eric

[Larken]

Try saving by using the 's' command directly in the terminal. You should see a line of ++++ when its saving.
you put s in the box then hit send.

Also, Sometimes you have to hit the read a few times for it to get in sync.

I will check out that reconnection problem, and try to have a updated vipertune by the weekend .

Larry

[spectrumsas]

I'm getting some feedback from customers drives and these MTS automation drives have a + - 10 volt output on the current monitor.

Input/Output Interface: Analog Signals
Velocity Command Input -- Differential input 0 to ±10 Vdc (15 Vdc Max)
Auxiliary Input --- Differential input 0 to ±10 Vdc (15 Vdc Max)
Velocity Output --- Select 1.25, 2.07, 3.71, or 4.54 volts per 1000 rpm
Current Output ---- ±10 volts = ± Peak Current

The AMC drive i have is a is a 25A8 , it has 2 current output lines.

CURRENT MONITOR
Current Monitor. Analog output signal proportional to the actual current output. Scaling is 4.4 A/V. Measure relative to signal ground.

CURR REF OUT
Measures the command signal to the internal current-loop. This pin has a maximum output of ±7.25 V when the drive outputs maximum peak current.
Measure relative to signal ground

Al, are you familiar with the CURR REF OUT and what it does ?
Larry

Dear Sir
Please send me full details with connection diagrams to connect my hurco electrocraft max 400 analogue drives to run with mach3.
Sending photo of drive.