New Machine Build- G540 problems and support

[chuck1024]

A while back I picked a prolight 1000 CNC mill. I was advised to upgrade to Mach3 and gecko drivers. I did some research and decided on a Gecko G540, AnTek 48V power supply and Keling 23H276-28-4B steppers. I pretty much gutted the original control box and installed the G540 and PS.

The Prolight has a 1HP DC motor and the speed controller is mounted in the back of the mill. All I needed to do was connect the solid state relay to one of the G540 outputs and it should work just dandy.

I managed to get Mach3 running. I did some simple G code moves but discovered that my G3 commands were not really make a circle. After a while I figured out that the G540B.XML file had the outputs all active high. Changing them to active low seemed to fix the problem. Later on I would discover that the G540B.XML file also is not setup correctly for PWM spindle control either.

After getting the steppers to move, I tried to get G540 output 1 to control my solid state relay. It would work ok for a short while. But after setting in the off state for a while, the motor would start to cycle on and off rapidly and would eventually just stay on. I monitered the G540 output pin with a scope and it would go high and would start to drift down. At a certain level the relay starts to cycle and after it drifts down some more the relay is on solid. I attemped to discuss this with Marcuss in email but his comment was my PS was dying. After a few attemts at working this issue in email, I called about mid December. I had a brief conversation with Marcus and he told me he would do some testing and call the next day. In the mean time I got the scope out and monitored the parallel port connections and sent the info to Marcus via email. After waiting for a couple days, I called back and was told he was not available but I could talk to tech support. I attempted to explain the problem to tech support but he really was not interested in a technical discussion on 12-23. He said there was a recall on the G540 and I should return it for an exchange. I made a quick trip to the UPS store and sent a followup email via their web page requestion quick turn around so I was not out of business for the entire holiday break. I got a phone call telling me there was a dip switch inside the G540 that might help. Too late, I already shipped it. BTW, there is no mention of the switch in the manual. Turns out the switch disables the charge pump which I doubt is related to the relays outputs. Anyway the person on the phone said they would ship a replacement drive asap. Maybe that day (12-23) or monday. Well both days came and went and I still don't have a drive. My old drive arrived at Gecko on 1-2 but was not delivered because they were closed.. How nice. Too bad the tech didn't think about the fact they were closed when he told me to return the drive. EVen though the outputs were not working the stepper drivers were working and I still could have used it for the entire holiday break.

While trying to resolve the output issue, I experimented with PWM speed control. I can make mach3 output a PWM signal and see it on the parallel port with a scope. However the analog signal out of the G540 is always 2.7 volts into an open circuit or into the Prolight's Miniark speed controller. I have no clue how to debug this because the manual has no info on output impedance or any clue how this circuit works. Email requests for assistance have gone unanswered.

So maybe other people have good luck with Gecko, mine has not been so good. I am still waiting for Gecko to test my old drive.

I am beginning to think that I should not have used the G540. The stepper drives seem to work ok, but nearly all aspects of the integrated breakout board do not work. BTW, I am an electrical and computer science engineer, so I have a pretty good idea how to hook things up.

chuck in naperville IL

[David Campbell]

Chuck,

GeckoDrives have a yahoo groups (http://finance.groups.yahoo.com/group/geckodrive/). Apparently the circuit details for the breakout board is available from the files area (you need to be a registered member to gain access).

I am not suprised by the christmas shutdown. You should try Australia as just about every business is closed over the Christmas / New Year period (three public holidays within 9 days, peak of summer, we're outa here mate!)

With the G3 command issue "not making a circle", if you excuted a pair of G3 commands to make a full 360' did the mill bit return to the original starting position? If you system was working I would suggest chucking up a felt tip marker and "milling" some paper to see where the errors are occuring:
- Is it in only one of the axis?
- What is the acceleration set in Mach3?

The acceleration has a big impact on the torque load that the stepper experiences. On my mill the contributing factors for torque in descending order are: shaft rotational intertia, mill bit cutting forces, mill bed friction / inertia.

Sorry I can't help you any more...

David Campbell

[chuck1024]

I have the Velocity set to 50IPM and acceleration set to 3. 50 IPM should be very consiverative settings for this machine. The prolight is speced for 50 IPM with the original 180oz steppers. The new motors are 270 oz so they should do better. By the way I could not find a motor with 2.5mH inductance. The 3.6mH was the lowest I could find and keep the current under 3.5A.

I was drawing a circle with a pencel in the spindle with a G3 command. I was primarly loosing steps in the Y axis when it changed directions. It would actually draw a straight line for quite a ways near Y=1 and X=0. I could see the pulses on the Y step pin on the parallel port but the motor was not moving. Also it would only loose steps when the feed was low (like 4) but would do much better when the feed was 10 or 20.

The problem seemed to go away when I changed the step and direction pins to active low. Someday I will have a drive and hopeful be able to test it again.

If it matters the ball screws are 5 turns per inch and the steppers are connected to the ball screws with 2:1 cog belts (size 80XL).

Right now I am happy with the drives, but I need to figure out how to make the relay outputs work.

chuck

[chuck1024]

I managed to find a schematic for the G540 over on yahoo. It is an old version and does not show the PWM spindle analog output.

Parallel port puts out nice +5 volt signals when the G540 is not connected.
When I connect the G540 I see this on the parallel port.

Pin 16 (charge pump) square wave with period of 82 uS. Peak voltage varies between 2.0 and 2.5V (Ie the square wave is not really square).

Pin 2 (X axis step) is a 3.6 uS 2.8 volt uS pulse.

Pin 14 (VFD PWM). I can vary the duty cycle by changing the S word in the M03 command. It is a nice pulse with high voltage of 3.6 volts. VFD output is always 2.775 volts reguardless of duty cycle. It does not matter if the output is open or connected to Miniark speed controller.

pin 17 (output 1) is loaded down to 1.4 volts (from 5V) when the G540 is connected. 1k resistor put on G540 main connector position 5(output1) for testing. Output starts high 48V and slowly drifts down. I watched it go down to about 30 volts with the scope. I tried the same test with 25V instead of 48 and it operates about the same.

The fundamental question is why is the G540 loading the parallel port down to 1.4Volts??

[Richards]

The parallel port must be capable of sourcing the ILQ74 opto-isolators used in the G540. The G540 schematic shows that a 200 ohm resistor is being used. At 5V, that opto-isolator is going to pull about 18mA ((5V - 1.3V) / 200).

Some parallel ports can't supply that kind of current. I've always added a PCI parallel port to my computers, rather than depending on the built-in parallel port on the motherboard.

[Crevice Reamer]

Originally Posted by Richards The parallel port must be capable of sourcing the ILQ74 opto-isolators used in the G540. The G540 schematic shows that a 200 ohm resistor is being used. At 5V, that opto-isolator is going to pull about 18mA ((5V - 1.3V) / 200). Some parallel ports can't supply that kind of current. I've always added a PCI parallel port to my computers, rather than depending on the built-in parallel port on the motherboard.

This is an EXCELLENT suggestion.

CR.

[LeeWay]

I'm not as up on the 540 as I would like. It seems to me though that a standard BOB will operate a relay without issues. The 540 should be able to do that.
I use the C10's from CNC4PC on all mine. For the router, I just use a simple solid state relay.
On the mill, however, I have to use a DC-06 from Homann Designs.
It actually does the controlling of the little relay on the DC-06 as well as controls the DC motor controller.
You adjust the settings in Mach 3 to get the exact speeds you want on the motor tuning page under spindle and then on the spindle setup itself.
I had a lot of trouble with a different relay board. This one turned out to be perfect however. So good in fact, that I will redo my new lathe controller to be identical to the mill.

The DC-06 is also a tiny little thing and would easily fit in tight quarters.
Just over 2 1/8th" by 1 3/4".

http://homanndesigns.com/store/index...5a0a9573079f2d

[chuck1024]

Before blaming the parallel port, we need to explain why all the step and direction outputs can source the required current and old this output connected to the ILQ74 is having problems.

I tried to install a NetMos parallel port. Windows put it at some odd address like AC00 and split the address into two parts. I could not make mach talk to it.

Another question. The solid state relays will energize on 2ma. I suspect the parallel port can drive them directly and skip all the output port logic.

Another idea I had was to install a c26 buffer from CNC4pc. It would not load the parallel port and could drive the optos in the G540. It can be powered off a USB port.

[Mariss Freimanis]

The parallel port must be set to 'EPP' in BIOS or the outputs won't work. A 'standard' parallel port has the auxiliary outputs 'open collector' while EPP makes them 'active pull-up'.

Mariss

[chuck1024]

I shipped by G540 back to Gecko on 12-23. I received a replacement on Jan 13. Paperwork said they upgraded the motherboard to Rev 4. No indication if they found a problem with the returned unit or not.

I installed replacement G540 and attemped to test using existing Mach config. G540 would not come out of fault. I cycled power a few times and I got a green light.

Y and Z axis would move ok. X axis would only move in one direction. Now the fault light is back. I discovered that touching the back of the G540 would remove the fault light with or without the computer connected. (The fault light should always be on without the charge pump connected)

Things were flakey enough that I decided to open the G540 and have a look. The attached pictures shows four parts near the DB25 connector are not on their mounting pads and are bunched up and soldered to each other and to the traces on top of the circuit board shorting things out.

I really wish I had looked at the first board because I would bet it had a similar problem. I got two bad ones in a row so I must have really bad luck or there is something else going on. What really bugs me is that I got a board with problems that even I could visually detect and obviously was not tested.

So what shall I do now. Send the G540 back and wait another 3 weeks for a replacement and hope that it might actually work? Maybe I should consider a PMDX breakout board?

[Mariss Freimanis]

chuck1024,

This is the kind of stuff that makes me pull my hair out! I'm embarrassed and I apologize for the problems this has caused you. Here's what we will do for you:

1) We will FedEx you overnight a new motherboard.
2) We will refund you $300 for the cost of the G540 for your aggravation.

Things will get a little ugly here later this morning. The motherboard is clearly from the rework bin, not from the tested and inspected bin. The DB-25 connector is hand-soldered by our assemblers. The assemblers should have caught this error and marked the board. The test technician caught the error and should have marked the board with an orange rework sticker. I assume you didn't see a sticker on the board. Finally, the board should have gone in the rework bin.

The process broke down for this board. I've called a meeting with everyone involved to find out how and why it broke down. This must never happen again.

Mariss

[chuck1024]

I have the prolight 1000 up and running pretty well with one exception.

Some background info. I modernized a Prolight 1000 with G540, Keling 280Oz steppers and a AnTek 48 volt power supply. The Prolight has ballscrews and thompson linear rails on all axis. IMO, it is a wonderful machine.

I experimented with Mach3 and EMC2. Right now I am running EMC2 because it supports things like Loops and braches in the G code that Mach3 does not. The only issue I had with EMC2 was getting the chargepump to run. The EMC2 configue ties the chargepump to the Estop and I could not make the charge pump work consistantly in that mode. It works fine when this link is removed but I still have the G540 optioned to ignore the chargepump because I have to pry the case open to change it.

The orignal prolight hardward used L298 chips. I actually made it run with the old hardware and software on a 386. I believe the original hardware was only capable of 1/2 steps and it would play a different note for every feed rate. It acutally sounded kind of cool. The G540 drives seem pretty monotonic compared to the old drives. Another interesting sidebar is that the old motors have very inconsistant feel. Some turn harder than others and the detents feel very different. My conclusion is that advice to replace old round steppers is very good advice!

I cut some slots with simple MDI commands and the distance between slots was right on. I cut some circles with G3 commands spiraling down to about 1/4 inch deep. A light finish pass leaves a beauty finish and the dimensions are good and the holes seem to be round. Sometimes I can see a dwell mark when the Y asix changes direction but I don't expect a G3 to do as well as a boring head. I used the a Python lettering wizard interface to EMC2 to write names into brass for my kid with a 0.050 endmill. Looks awesome.

I am controlling the spindle with a solid state relay connected to the G540 output 1 which is a relay driver. Works well.

I have PWM signal into the G540 with a freq of 50Hz. I can vary the duty cycle with S word observed by attaching a scope to pin 14 on the parallel port. I apply 10 volts from an isolated supply to the VFD +10 input pin on the G540 and the neg side is connected to the VFD Grd pin. The VFD output pin is always 0 volts into a 22k ohm resistor. I also tried a 1k resistor thinking it might be load dependent but no change. On my first motherboard, the VFD output was always 2.7 volts. I give up on this feature and I am running manual spidle speed control. I will probably find another way to do this unless I can find a schematic for the G540 that contains the PWM circuity.
chuck