Gerber 408 retrofit

Considering buying a $4000 Gerber 408 in my area. Your conversion to Linux is making the idea very tempting! How much do you charge for your kit?

If you are interested in buying a retrofit kit from me, please private message me.

--Forest

Interested, don't know how to send you a PM on here.

I am preparing for a Gerber 408 retrofit. The machine was given to me (can't beat FREE). It does work with original software.

I will probably go ahead and replace original servo motors with steppers. Anyone with experience on this have a recommendation for size of motor. Looks like NEMA23/24  would fit, but I'm worried they won't have enough torque. Would love to hear your experience.

Gerber408 wrote:

I am preparing for a Gerber 408 retrofit. The machine was given to me (can't beat FREE). It does work with original software.

I will probably go ahead and replace original servo motors with steppers. Anyone with experience on this have a recommendation for size of motor. Looks like NEMA23/24  would fit, but I'm worried they won't have enough torque. Would love to hear your experience.
 

Replacing the brushed DC servos with steppers will reduce the maximum speed of the machine and lower the accuracy/reliability. The original motors have an unusual frame size, neither NEMA (23, 34, etc.) nor metric servo (60mm, 80mm, etc.) standards. You'll not only have to fabricate adapter plates, but also replace some other parts such as the timing belt pulleys since they were meant for the odd-sized 11mm shaft that Gerber put on their motors.

I have yet to see a Gerber a bad servo motor. I have looked into getting the encoders replaced should they go bad (haven't seen that), and I have replaced the brushes on the motors (preventative maintenance). The "Achilles heel" of Gerber Sabre is that their Y axis leadscrews (or at least their plastic leadnuts) wear out. I am looking into a ballscrew replacement solution, I know of another person that had success with this.

Regarding controlling the original motors, if you want to retrofit with LinuxCNC you can keep the amplifiers and use the encoder feedback to control them in "torque mode" from LinuxCNC. This is what I recommend as it gives you the best performance, and you don't have to replace anything. As I have posted, I am actively working on a drop-in motherboard for people such as yourself.

However, if you want to control the servos from simplistic CNC controls that only deal with open loop step/direction, then motor power voltage permitting you can use a Gecko G320X:

www.geckodrive.com/product/g320x-digital-servo-drive/

The downside is that you will have to tune the PID settings using a screwdriver with the Gecko drive's potentiometers, and can only tell what is going on using the LEDs and audible oscillation. Very different from being able to use halscope!

You'll also want to convert the encoder's differential signal to single ended as close to the Gecko drive as possible so that you maintain noise immunity, using something like Mesa's ENCY:

store.mesanet.com/index.php?route=produc...oduct&product_id=304

Will you also provide the LinuxCNC config that goes along with the board when you get it done. A "plug and play" solution.

SkyMoCo wrote:

Will you also provide the LinuxCNC config that goes along with the board when you get it done. A "plug and play" solution.
 

Yes, not only will I provide the LinuxCNC configuration, but I will provide a completely pre-configured LinuxCNC compatible PC with the kit. I use HP Compaq Elite 8200/8300 models, as they have been proven to have good real-time performance.

I also may be interested in this solution. How can I contact you? I can't seem to find a DM function on this board.

Gerber408 wrote:

I also may be interested in this solution. How can I contact you? I can't seem to find a DM function on this board.
 

I cannot find any way to direct message either... Maybe they removed it due to spam problems? At any rate, you can contact me by emailing my forum username at gmail.

tommylight wrote:

fdarling wrote:

The downside is that you will have to tune the PID settings using a screwdriver with the Gecko drive's potentiometers, and can only tell what is going on using the LEDs and audible oscillation. Very different from being able to use halscope!

You can use halscope, but it does require having differential output encoders and a Mesa board with encoder inputs, and very good knowledge of wiring/powering stuff as Gecko has its own power for encoders and so does Mesa, making sure those are at the same 0/zero/gnd/potential is absolutely important, and using the +5V only from Gecko to power encoders.
So not easy/comfortable/possible always.
 

You are correct that you can still use HALscope with the Gecko drives, however if you already have the hardware that can deal with interfacing the encoders to LinuxCNC, then why aren't you just using that setup permanently? It's much easier to tune the PID from the "calibration" dialog within LinuxCNC, rather than use the potentiometers on the Gecko drive.

Mesa cards support single ended encoders via jumpers, but the Gecko drive only supports single ended not differential. You can simply use half the differential signals to effectively look single ended, however you lose the noise immunity if they are run a long distance (through the drag chain from the motors). This is why I am suggesting the use of the ENCY encoder repeater, so that it takes in the differential signal and cleans it up into a new differential output, where you can opt to use only half the outputs as single ended. You can send the repeated signal to the Gecko drive, and possibly the secondary repeated signal to a Mesa card if present for tuning purposes. Again, I would instead forgo the Gecko drive and just use a Mesa card (or my kit)!