Abene vhf-3 tnc retrofit questions

Hello,

I have an abene VHF-3 with a tnc150 control. The controller isn't working properly (and doesn't have drip feeding) and I would like to retrofit it with linuxcnc with the MESA 5i25 and 7i77 cards. The Heidenhain glass scale on the X-axis is broken as well.

I have searched for info on this forum and so far I know I'll have to buy 2 heidenhain EXE converters to convert the heidenhain scales' 11 µApp signal to a TTL signal. Would it be possible to replace the broken scale with a (much cheaper) sino glass scale (5µm resolution)? I was thinking this would be easier and cheaper because of the price of the scale and also because I won't need an EXE signal convertor.

I've read that the best way to use the glass scales with linuxcnc is to use the servo motor encoder for the P and D parts of the servo loop and the glass scale for the I part. However, my machine has tachos instead of encoders. These are built in the motors and I don't think they are easy to replace. I'm also worried that the servo system becomes unstable if I use an encoder on the motor for position feedback (there is a normal gear with some backlash between the servo and the ball screw)

Is it possible to use the glass scale for position feedback and the tacho for velocity feedback (as is now the case)?

I would also like to keep the current servo drivers. These take a 0-10V signal from the controller so I assume this would be no problem?

All help is appreciated. (I apologize for any bad english)

It should be possible to do a direct retrofit on this using the existing analog drives (if the drives and motors are in good shape)

If your motors have tachometers it means that you already have a velocity mode drive system (The 'D' term is handled by the drives/tachometers)
There is no reason that you should not be able to make this work as well
or better than the original control, using only the velocity mode drives and the linear scales

As you may have guessed, backlash will be an issue so should be
minimized as much as possible or you will have to program the PID parameters with a large dead zone to prevent 'hunting'

Ok, this makes sense. The tacho signal does indeed go to the drive, while the glass scale signal goes to the controller.

If the driver controls the speed, does this mean the +-10V signal that goes from from the controller to the drive only "chooses" the speed (I mean, for example 5V= 1.5mm/m and 10V=3m/min)?

In practice, how would this exactly work with linuxcnc?

andypugh wrote:

You use a PID controller. This looks at how far the axis is from the programmed position, and commands a velocity proportional to the error in the direction to make the error smaller.
Think of the commanded position being the car in front, and yourself as the axis, in a following car. You operate the throttle pedal to maintain the correct speed, which makes you the PID controller in that system.

I understand the principle of a PID controller. What I don't understand is how the combination of a servo driver controlling the velocity, and the controller controlling the position works.

andypugh wrote:

Jerre122 wrote:

I have an abene VHF-3 with a tnc150 control.

www.lathes.co.uk/abene/ ?

Interesting machine, if so, and presumably not originally CNC.

It is an interesting machine, and compared to my previous elliot milling machine very high quality. My machine is original cnc, just like Deckel made cnc versions of the FP-series, Abene made cnc versions of the VHF-3.

Its very common in control loops to have a distributed control (like your velocity mode servos) That is, there are loops within loops.

Usually the high bandwidth portion of the loop is 'nearer' the controlled parameter.

For example in your servo drive there is yet another loop that you do not normally have any access to, this is the current (torque) control loop. This is a high bandwidth PI loop (several KHz) that forces the motor current to match the requested value regardless of power supply voltage, motor resistance, motor inductance and motor back-EMF variations.

The local velocity loop that compares the tachometer output to the commanded velocity controls the setpoint to this inner current control loop, just as LinuxCNCs position control loop controls the setpoint to the drives velocity loop.

Does the Mesa 7i77 accept RS-422 encoder signals? I need a replacement for the x-axis scale but all glass scale's I can buy here (except expensive mitutoyo or heidenhain scales) have RS-422 output. The problem is the other heidenhain scales are fine and I was thinking of using them with EXE converters. These take the 11µApp signal from the scales and converted it into a TTL signal.
I thought that finding a new scale with direct TTL output wouldn't be a problem but apparently it is.

I know I can set the inputs to differential, but the RS 422 from the glass scales have 6 outputs, while the mesa cards have 3 inputs for every encoder (unless I am reading the manual wrong). I might be asking dumb questions but electronics are not my strong point.

I'd like to be sure if everything will work before I start buying things.

EDIT: I read the manual wrong, they have 8 inputs.

The 7I77 has 6 encoder inputs, each input can be TTL or differential (RS-422)
All 6 encoder connections have 8 wires (6 for differential A,B,Z and 2 for power and ground)
TTL wiring only uses 5 wires per connector (A,B,Z, power, ground), differential uses all 8

Has anyone ever ordered from www.retrofit-plus.at/ ? After a lot of trouble trying to buy directly from mesa (mastercard issues) I ordered from them 2 weeks ago. I asked them for a tracking number almost a week ago but they haven't answered my mail yet. I have a feeling the cards haven't even been send yet.