[donl517]

1992 Fadal CNC88. I was machining a small nest yesterday when I noticed the machine would move a little ways, stop, and continue on. It is not related to a single axis nor is it distance related it is a timing thing. When it would rapid it would go about 4 times as far and stop, and then continue on. I was machining in X and rapiding in Z when I noticed it. It didn't start until I had been maching for probably 25 -30 minutes so I'm guessing something is getting warm. Any thoughts?

FYI: I was doing a DNC but this was happening on an 8", single line, linear move it would stop 4 - 5 times. Of course the rapid in Z was to a tool change position, around 18", it only stopped once.

Thanks,
Don

[gar]

070728-1724 EST USA

donl517:

Since as you describe the problem as occuring within a single command, then there only an extremely small probabilty that DNC is the cause. To remove this or some interaction with that mode as the cause load a program into memory and run from memory to see if you can repeat the problem.

.

[donl517]

Hi Gar,

Yeah I'm 99.99% certain that it has nothing to do with DNC, I just through that in the description because I figured it would be the first suspect of any responses and I wanted to rule it out right away. When I get a chance I'll put the machine in a loop and watch for the problem, just to completely rule out the chance that it could be a DNC issue. Do you have any suggestions as to what it might be?

Thanks,
Don

[gar]

070728-1859 EST USA

donl517:

I have no idea.

Write the very simplest program. Like a G00 in one direction for the 8" distance you referenced, then back and repeat. Does this cause a problem? If it occurs measure the time between pauses if possible. Next change to G01 and do the same thing. Maybe use different feed rates to see if there is a time change or is it related to distance traveled.

Try this separately on each axis one axis at a time.

.

[captainasty]

If you are running cutter comp and doing a lot of small steps, the machine will stutter after about 30 ipm feed rate.

[donl517]

If you are running cutter comp and doing a lot of small steps, the machine will stutter after about 30 ipm feed rate.

Hi Cap,

Wasn't running cutter comp and making an 8" linear cut (one x move) stopped 3 or 4 times. I'm thinking it is something electronic but I don't know what would effect all 3 axis at the same time. Any thoughts?

Thanks,
Don

[captainasty]

Good morning, If all three, I would check the power and heat build up in the cabnet. I had another machine (not Fadal) that the amp cards were over heating due to fan failure.
Back to the point, check for "Deceleration or Ramp",G9 " in postition mode. That will cause the machine to stop at points. BUT yours is in the middle and after some warmup time.
Check for heat in the cabnets, aka fans working.

[donl517]

Hi Cap,

Yep I have checked the fan (my first suspect as well) it is running great moving alot of air. I ran with the door open and the air conditioning on just to make sure heat wasn't building up; and I had the same condition in about the same time frame. Somewhere in the machine there has to be a bottleneck that all the info is going through. That is where I would like to start, but what is passing info to all three servo's? Is there a CPU, for lack of a better term, and if so, does the info have to go through there before it goes to the servo's?

Thanks,
Don

[donl517]

OK I found it!!! It is communication related. I am using the NCFadal software to DNC my program and at the exact moment it gives me a "Transfer Complete" dialog box the CNC starts to do its litte stutter moves. So, Can anyone suggest what to look for?

Thanks,
Don

[donl517]

I've narrowed it down to the cable. I'm going to ring that out in the morning to see if it is wired correctly. If I unplug it after I get the "Transfer Complete" dialog the problem goes away.

Don

[gar]

070730-2108 EST USA

donl517:

Are you saying that data is being transferred via DNC (direct numerical control)(drip feed) during the time that the motion interruptions occur?

Or has all the data been transferred prior to these occurances and only because the RS232 cable is connected the hesitations occur.

If the problems occur during RS232 data transfer, then is handshaking working? If it is not, then you might be overloading the CNC CPU, but this would cause data errors at some point.

Does the Fadal use a Windows based computer? How large is the buffer for incoming RS232 data. On HAAS all of the unused program memory is available for DNC (drip feed) buffering. So on a 1 meg machine you can have nearly a 1 meg buffer.

If you have an electrical noise problem via the RS232 cable, then our I232 Isolator System would probably solve your problem. See
www.beta-a2.com .

Usually the type of stuttering problems that we solve are related to data starvation. Your description of your problem does not imply data starvation. It is most likely either external noise thru the RS232 cable or an internal problem in Fadal including some cause for lack of handshake.

Unless you have a large CNC program why are you running in DNC mode?

.

[donl517]

070730-2108 EST USA

donl517:

Are you saying that data is being transferred via DNC (direct numerical control)(drip feed) during the time that the motion interruptions occur?

Or has all the data been transferred prior to these occurances and only because the RS232 cable is connected the hesitations occur.

It has completed the transfer and that is when the problem starts. I spent 2 hours on the phone with Fadal today and we finally discovered when we unplugged the cable after the transfer completed, even though the machine is still running, the problem went away

Unless you have a large CNC program why are you running in DNC mode?

43K of memory. I'm looking into upgrades now.

Thanks,
Don

[gar]

070731-0556 EST USA

donl517:

Since the transfer has completed and the problem still exists this means the problem is not related to the CPU, timesharing, and handshaking relative to an expected data signal.

In all probability the problem is ground path noise, an unexpected signal, between the CNC and computer causing random noise to in some fashion disturb Fadal. I believe on at least some Fadal machines that a CPU, and maybe its the only CPU, is constantly monitoring the RS232 channel for new data. Bursts of noise might overload the CPU. Noise coming in will not be well behaved and may cause many parity, framing, bit overflow, and buffer overflow errors that all require CPU attention. These noise problems may take excessive CPU time causing the stutter.

Ideally if a single CPU was used in Fadal its speed would be fast enough to simultaneous handle incoming RS232 data and/or noise, and motion control without hesitations. But Fadal may not be fast enough. Better yet if multiple CPUs were used, then one CPU for motion control functions could be uninterrupted during other housekeeping operations.

My phone number is on my previously referenced web site. Call me.

.

[donl517]

Gar,

yes I'm sure it is ground path noise. I'm looking into it now. I had noticed that one of the servo's had a high pitched whine to it. I didn't think to much about it until I unplugged the RS232 cable and the whine went away. I'll dig a little deeper.

Thanks,
Don

[donl517]

I traced it back to the serial port on the computer sending the DNC. I tried a different computer and everything worked fine. I'll try a driver update, if that doesn't work I'll put XP Pro on it. (Windows 2000 now)

Thanks,
Don

[gar]

070809-1208 EST USA

donl517:

Was the computer that caused error problems a laptop?

For the good and the bad computers what kind of RS232 port do you have?

.

[donl517]

070809-1208 EST USA

donl517:

Was the computer that caused error problems a laptop?

For the good and the bad computers what kind of RS232 port do you have?

.

No, it was an IBM Netvista desktop. It has 2 built in RS232 ports. I also tried a USB to Serial adapter and had the same problem with that. I haven't tried my laptop yet to see if it has the same issue. It doesn't have a built in serial port, so I have to use USB to serial on that.

The PC that worked is my desktop I use for design work.

Don

[DareBee]

Don
In regards to your memory (or lack thereof).
I maxed out my old girl last year (A whopping 400 and some odd Kb woohoo).
Seriously though - this is a massive upgrade compared to 43Kb.
Will hold about 30,000 lines now and was about $500.
Upgrading to a -5 control was out of the question (I would retrofit before I would do that).

[donl517]

Don
In regards to your memory (or lack thereof).
I maxed out my old girl last year (A whopping 400 and some odd Kb woohoo).
Seriously though - this is a massive upgrade compared to 43Kb.
Will hold about 30,000 lines now and was about $500.
Upgrading to a -5 control was out of the question (I would retrofit before I would do that).

Yeah I talked to an independent service tech I know, he said he has a couple of the 240k (not sure if that is the right value) cards. He said he would give me one if he did the upgrade. I'm going to have him in to do some other work and I'll have that done then. All the cards in this machine were upgraded to a -5 when I bought it. They told me it was a $10,000.00 upgrade. I don't know why you would spend that kind of money and then sell the machine, but I have found several minor maintenance issues that I have been able to resolve. I wonder if any one of them led to selling the machine. I bought it from a used machinery dealer. The funny thing is the computer with the DNC communication problem was part of the CNC package. I also found I couldn't reach the full travel of Z to the table. I kept getting an axis fault with an accompanying clunk. The machine has a 28" Z and the counterbalance has a block on top of it, maybe they all do, anyway the single bolt holding it on had worked loose and it had rotated and was hitting on some of the mounting brackets. Fixed that and all is well.

Have a great day!
Don

[gar]

070809-1352 EST USA

donl517:

If you have a voltmeter check the DC voltage between pin 3 and chassis (chassis and pin 5 should be the same) on the IBM Netvista when no data is being sent. I expect about -10 V. Also measure the AC voltage. Should be near zero millivolts. Obviously this is with nothing connected to the 9 pin connector except the meter.

.