What is wrong with my system?

[Mariss Freimanis]

Tom,

Please contact me off-line. I found out what happened to your drive.

Mariss

[tjskcnc]

"I apologize for the babble. It's two grizzled old EE's talking in shorthand. We just started using the forum to try and pin the problem down in the shortest time. Normally this sort of exchange is off line. If it's causing any problems we can take it to private e-mails."

Actually, this has all been very educational. It's what this whole forum is about - education and information. Most of what I know about CNC has come from this forum, so keep posting. It's how us newbies become 'grizzled old EE's' and can start helping others like you and Mariss arre doing.

Thanks for all the help!

[rowbare]

And you are just going to leave us hanging? No fair!

Tom,

Please contact me off-line. I found out what happened to your drive.

Mariss

[Torchhead]

It's a long technical diagnosis with scope traces and timing issues. Mariss will need to distill it down somewhat. The important thing is he stayed with it until the testing pointed up the problem. It was the proverbial needle in the haystack. It's not a design flaw and does not occur but in perhaps 3 out of a hundred 250 series drives and then only under certain circumstances. I know they now test for that specific parameter and he is working with the component supplier to verify a fix for the small number involved. Now that he has been able to duplicate the problem and knows how to test for it than I am sure he will be in a position to fix any drive that exhibits the problem.

97.0% of position problems with a G540 are not from drive problems. Usually it's bad tuning in the software or bad signals from the PC. If you have one axis that misbehaves (especially on very slow RPM of the motor) and the others run fine, then you may have one of the drives in question. If you swap that drive with another and the problem follows the drive then that is another bit of smoke from the gun.


TOM Caudle
www.CandCNC.com

[Larken]

Im a bit confused

What has Tom Candles drive got to do with tjskcnc who started this thread ?

[ger21]

Im a bit confused

What has Tom Candles drive got to do with tjskcnc who started this thread ?

The OP has a G540. Tom worked with Mariss at Gecko to find a problem that has plagued the OP and a few others with G540's. Tom doesn't make drives, he sells Geckos.

[Mariss Freimanis]

The "Sherline Mode" mystery may be solved.

It involves strange behavior of some Xilinx CPLDs (about 3&#37 that form the core of the G250 drives used in the G540. The problem traces to Xilinix at this time and I have been in contact with their engineers regarding what we are observing. I have developed a work-around but I'm uncomfortable implementing it until Xilinx can identify the cause, consequences and cure for this phenomena.

In essence, a small percentage of CPLDs ignore the step pulse input signal for 6 to 10 microseconds. "Sherline Mode" cures this because it stretches the step pulse to a 15 microsecond duration. Compounding the strange behavior is this phenomena only manifests itself at low step pulse rates and disappears at higher step rates.

Most all CPLDs respond instantly to the step input which is why most people don't have to use Sherline Mode in Mach3. For them, a 2 microsecond pulse works just fine.

I'm awaiting a response from Xilinx on what they find. The problem seems to be CPLD hardware related (yield or design) and has nothing to do with the code programmed to the CPLD. The work-around is board level hardware related (resistor value change) and seems to eliminate the problem. As soon as I hear back from Xilinx, I'll post.

It does clear up why Sherline Mode works though the reason is arcane.

Mariss

[foam27]

So, the bottom line is, there are defective components in some of the drives you have been selling.

Is that correct?

[Mariss Freimanis]

Yes foam27, that is correct. We see to it 3 customers out of 100 get these defective drives.

Why 3 out of 100? Well, I opened a cheap can of mixed nuts and counted the pecans and peanuts. It worked out to 3 pecans for every 97 peanuts and I figured if it's good enough for Planters Peanuts then it's good enough for Geckodrive.:-)

Kidding aside, let's say you are me. What would you do?

Option 1: Stonewall and say there is no problem. If you are the 3 out of 100, it's your fault because you aren't using the drive right. You twiddle your thumbs and do absolutely nothing to find out if there really is a problem. You don't care because you know customers are suckers.

Option 2: Stonewall and say there is no problem. Continue to say if you are the 3 out of 100, it's your fault because you aren't using the drive right. Track down the problem eventually and quietly correct it. Figure time and the flood of corrected drives will drown out those who complain with a flood of happy new customers. You congratulate yourself on being so clever.

Option 3: You hear there are problems. You cannot replicate them in your lab and so you think it must be something else. Still it bothers you so you put out the word you need to have a bad-boy drive back from a customer. You even put out a bounty for it (post #10). Tom Caudle supplies one.

Tom also supplied specific, usable information under what input conditions it malfunctioned. You replicate the conditions and voila, it malfunctions. As a designer, you use the best components and you integrate them into the design without stretching any of the component ratings. Xilinx is such a vendor as is International Rectifier, Fairchild and National Semiconductor. It comes as a surprise some parts (3&#37 don't do what they are supposed to.

What do you do?

1) Say there is no problem. It's not my fault, it's Xilinx fault.
2) Quietly correct the problem while saying there is no problem. Fixed it, no one will know.
3) Come out and say "there is a problem, here is a work-around and it takes care of it".

The internet is a hard place. Knowing the stuff that will come down on your head if you admit to a problem makes you ache to choose (1) or (2). I have always been dumb so I always pick door number (3).

Long and short:

1) If the G540 misbehaves, set Mach3 to Sherline Mode. It completely eliminates the problem.

2) If that is unsatisfactory, return the G540 for a hardware modification. It will completely eliminate the problem without having to use Sherline Mode.

Mariss

[lowpull]

No,,NO,,,

Chose option 4. Claim it is an undocumented xilinx feature.


Mariss seems to be one of the most open vendors out there, and I find it refreshing.

[SUPER-CNC]

Mariss,

How do we test to see if we have a defective Gecko?

Are there only certain serial numbers that are effected?

Thanks for keeping us posted,

SUPER-CNC

[tjskcnc]

So basically, now that you have found the problem, if we are satisfied with the results we get using the Sherline mode (or just stepping up the pulse rate to 15) then we can just continue using the drive as-is and there is no problem from this that will accumulate and eventually destroy the drive. Or, if we wish to, send you the drive and it can be repaired to correct this situation.

Will there be any charge to fix the drive and what would be the approximate turn-around time?

Also, couldn't we just order a G250 and replace the drive on the offending axis?

[LeeWay]

I hate to throw another curve into this.
I now seem to have a 201 that does not like anything but Sherline mode.

This was a spare drive that I had used on several machines in the past. I finally pulled it from my lathe controller and relegated it to a spare.
A while back, I started having some step missing issues on my Z axis. During testing, I swapped out the 203 with the 201. I ultimately ordered a larger stepper and that solved the issue.
Then I changed BOB's to one with a quasi charge pump. My x axis started missing steps. After lots of swapping out of spare parts and reordering the drives to see if trouble followed, I clicked on the sherline mode. The problem went away. As luck would have it, the 201 was in the X position at the time of this discovery.

I then tried to swap back and the problem persisted. Severe missed steps on X.
I tried older versions of Mach 3 as well as the newest lock down.
No change. I stick the original 203 back in and have no trouble so far.
When changing the drives, I am aware that the 201 needs 5VDC on common and the 203 needs 0VDC.

I had also tried different PP cables as well as computers. Just really odd.
I need to order another 203 as a spare, but would you like to see this 201 personally, Mariss?

The reason I can't use the sherline mode on this machine is because it doesn't play well with my spindle. It won't let me adjust my speed controller correctly, otherwise I would have been okay with it.

[foam27]

Mariss,

Now matter what you do it will be the right thing because you are a stand up guy that runs a stand up company.

I have bought some drives from you in the past and had no problems with them.

I was thinking about going for the G540, but.....

I have been patiently waiting for you to come out with some drives that take 120VAC in and offer a 160V bus, since drives that offer this blow me away, low amps, high voltage, and no external power supply to purchase and run. I think you know what drives I am talking about, as there are quite a few, that do this, and have done this since 1997.

Still got the Gecko keychain

[Mariss Freimanis]

SUPER-CNC,

The G250s and G540 don't have serial numbers. The Sherline Mode thing first cropped up shortly after the introduction G540 which means this has been a problem from the very beginning with 3% of the Xilinx XC2C32A CPLDs. This rules out a bad batch problem with Xilinx or a process problem with us. Many batches have been purchased and our process has changed considerably (low-temperature reflow for leaded solder in the beginning, high-temperature reflow for unleaded solder now).

How to test the G540:

1) Set step pulse 'active high' and step pulse width '2 microseconds' in Mach3.
2) Jog the axis slowly at less than 30 RPM on the motor.
3) If the motor moves, it's OK.
4) If it doesn't move, set Mach3 to 'Sherline Mode'.

Mariss

[Mariss Freimanis]

tjskcnc,

Simplified, the problem is the drive ignores step pulses for 8 millionths of a second below some speed. If the sent step pulse is 2 millionths of a second wide, the affected drive will not "see" it. Sherline Mode makes the pulse 15 millionths of a second wide. The affected drive "wakes up" after 8 millionths of a second and "sees" there is a step pulse.

There is no cumulative effect or deterioration with time. The phenomena is restricted to the step input pin only. Proof is affected drives are still running after a year and a half.

Mariss

[Mariss Freimanis]

foam27,

A line operated drive (115 AC) is potentially a lethal drive. Do something wrong with a G201 and you get a puff of smoke. Do something wrong with a line operated drive and you can become metaphysically challenged (dead).

Mariss

[foam27]

lol, I think I'll skip this one...

[CJL5585]

lol, I think I'll skip this one...

Foam27,

Many foreign and some american companies make potentially deadly products which operate directly from 120 VAC.

The only protection the consumer has is a voltage isolation stage, transformer, or power supply with proper fusing or circuit breakers. Some of these products are housed in plastic or non-conductive enclosures, and many do not have a ground plug which plugs into the 120 VAC receptacle.

Open the non-conductive case and touch any active component, and one might find himself face to face with his creator, if he is in any way, shape, or form in contact with anything that is at earth ground potential.

The bottom Line: Be wary when using any product with a two conductor power cord, or any 120 VAC powered product in a non-conductive (plastic usually) enclosure.

[Torchhead]

Voltages as low as 60VDC (or lower depending on skin resistance) with a few mils of current flow across the human chest area is potentially lethal. So if you place yourself across a voltage that is high enough to penetrate skin resistance (which varies wildly) regragless of where it is, you can be just as dead as the guy that grabs the AC line cord. Electricity is to be respected. The two methods of protecting humans is to :

A: Make it so the human cannot contact the hot side of the circuit (insulate)
B: Make sure the case is at earth ground so someone touching the (external) case cannot get current flow to a source of earth ground.
.
Every appliance in your house is "potentially deadly". Lamps, toasters, blenders, electric hotwater heaters, etc.

Open them up and poke around and nature will punish the stupidity.

Open the case on anything and touch the hot side of the circuit and have yourself grounded and you will be "alerted" to your mistake. You can put in interlocks that turn off power when you open the case but they are easily defeated.

Truth is, if you want to live with modern appliances, modern transportation, and not live in the woods (where the primary danger is being eaten by wild animals) you have to accept a certain level of risk.

So if you grab the 90VDC wires to a servo and get that across your heart you will be just as dead as if it was 160VDC of the main Bus.

The primary danger with higher voltages (besides causing higher currents to flow across a given conductor (you, if you are the conductor) is that it tends to lock muscles (overrides tha electrical impulses the brain uses to move them) and keep the victim locked to the source.

To a lot of CNC DIY guys, electricity is some invisible magic force that should be avoided at all costs. Feeding that fear with statements about potentially deadly products is counter productive.

Your odds are 1 in about 10,000 you will die from any form of electrocution. Compare that to more common causes like auto accidents, medical mistakes, planes, trains, falls and alcohol poisoning and it highlights what the true risks are. Death by electricity (including lightening strikes) is in the BOTTOM 10 causes of death in the US.

So your chances of death are 1 in 84 from a Motor Vehicle Accident; 1 in 218 from a simple fall, and 1 in 1008 from drowning. You would be better advised to stay away from water (even bath water), ladders and vehicles, than from electrical appliances.

If someone works hard enough they can maim/kill/hurt themselves on a glass marble.

TOM Caudle
www.CandCNC.com