Re: [Emc-users] Threading Hickups
If you do use the ferrites in the outputs let us know how much heat they produce. I would guess it will be a function of the quality of the VFD's output stage, but have never tried power ferrites. The filters used in theatrical dimmers can get pretty hot, and those frequencies are low compared to a VFD. Since ferrites are more lossy at high frequencies I'm curious about it. Thanks, Javid - Original Message - From: Kirk Wallace [EMAIL PROTECTED] To: Enhanced Machine Controller (EMC) emc-users@lists.sourceforge.net Sent: Sunday, October 28, 2007 6:24 PM Subject: Re: [Emc-users] Threading Hickups Jon gets a gold star. I believe his is line filter suggestion solved my noisy spindle encoder index problem. I got a couple of PREO filters: http://www.eastek-intl.com/images/PreoSeriesER.pdf One for the spindle VFD input (30 A) and another for the coolant VFD input (20 A). I also got a few large ferrite beads, for the outputs, but I am not sure how much they help. Thanks everyone for the help. I'll get some pictures posted soon. I looks like these will be standard for all my VFD's. -- Kirk Wallace (California, USA http://www.wallacecompany.com/machine_shop/ Hardinge HNC lathe Bridgeport mill conversion pending Zubal lathe conversion pending) - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
Javid Butler wrote: If you do use the ferrites in the outputs let us know how much heat they produce. I would guess it will be a function of the quality of the VFD's output stage, but have never tried power ferrites. The filters used in theatrical dimmers can get pretty hot, and those frequencies are low compared to a VFD. Since ferrites are more lossy at high frequencies I'm curious about it. In my experience, ferrites are used on VFD outputs to filter common mode noise. For that purpose, all three VFD leads go through the ferrite together, and the ferrite doesn't see the motor current, it only sees the common mode current. Under that condition, heating is minimal. If you put ferrites around the individual motor leads, the motor current will probably saturate the ferrite and make it ineffective, as well as producing core loss that heats up the ferrite. Regards, John Kasunich - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
Kirk Wallace wrote: Jon gets a gold star. I believe his is line filter suggestion solved my noisy spindle encoder index problem. I got a couple of PREO filters: http://www.eastek-intl.com/images/PreoSeriesER.pdf One for the spindle VFD input (30 A) and another for the coolant VFD input (20 A). I also got a few large ferrite beads, for the outputs, but I am not sure how much they help. Thanks everyone for the help. I'll get some pictures posted soon. I looks like these will be standard for all my VFD's. Great, glad this is helping. (I've only been working on electrical noise problems in instrumentation systems for 30 years, I'm still learning new tricks.) Jon - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
If anyone is interested, I posted a few new pictures here: http://www.wallacecompany.com/cnc_lathe/HNC/ -- Kirk Wallace (California, USA http://www.wallacecompany.com/machine_shop/ Hardinge HNC lathe Bridgeport mill conversion pending Zubal lathe conversion pending) - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
On Sun, 2007-10-21 at 09:56 -0400, Gene Heskett wrote: On Sunday 21 October 2007, [EMAIL PROTECTED] wrote: John Kasunich wrote: Even better, if you can get some, is a braid that can be expanded enough ... snip Regarding motor cables: I am using a Belden microphone cable called Star-Quad, which has 4 conductors in a good shield that I use with my xylotex volt away from the rails if possible. ... snip Food for thought guys. From an old fart C.E.T. How about RG11 or LMR400 for the spindle motor leads. I have a 2hp motor, so the 14 gauge center conductor should be large enough. With a 0.4 inch diameter, I would need to go to a larger conduit. -- Kirk Wallace (California, USA http://www.wallacecompany.com/machine_shop/ Hardinge HNC lathe Bridgeport mill conversion pending Zubal lathe conversion pending) - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
John Kasunich wrote: Even better, if you can get some, is a braid that can be expanded enough to run the three motor leads through it, then stretched lengthwise so it snugs down around the motor wires. Again, connect one end directly to the motor frame, and the other directly to the VFD ground. The idea here is to have the return path for stray currents as close as possible to the outgoing path. There are cables produced just for this purpose with a braid and outer plastic sheath. You can also get them with 2 wires extra for temperature sensor if the motor have such. It will contain noise carried by the cables provided it is connected to a good ground at least at both ends. Additional grounding along the path improves efficiency. Ground by using clamps around the braid and against the ground rail. Pigtails act as inductors and render it less efficient at higher frequencies. And of course a spiral conductor around the leads is just a mechanical protection. It is a coil, and thus not effective at AC which is what worries us here. Any ground cable that might carry noise (current spikes) should have as large surface as possible. That means many thin strands. The grounding strap from engine to chassis of modern car is a good example. Welding cable is also fine. It should go in a straight line, never coil it up. Use mentioned straps to ground your cabinets to the machine frame and each other. Use the grounding bolts built into the cabinet for this purpose, or if there is none, make sure there is no paint or corrosion hindering a good connection when you install it yourself. Don't forget to ground the cabinet door using 1 or 2 straps. Also make sure your machine have a good connection to the grounding system of your building. If you run sensor cables or data cables to the machine cabinet, those should also be shielded and connected at both ends, and the device at the other end must be considered just another machine cabinet (grounding). This last point is less strict if using galvanic isolations, and no worry if using plastic/glass fibre. Then with the appropriate filters installed, you should be able to run your machine while your wife watches the football game. ;-) Einar - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
On Sunday 21 October 2007, [EMAIL PROTECTED] wrote: John Kasunich wrote: Even better, if you can get some, is a braid that can be expanded enough to run the three motor leads through it, then stretched lengthwise so it snugs down around the motor wires. Again, connect one end directly to the motor frame, and the other directly to the VFD ground. The idea here is to have the return path for stray currents as close as possible to the outgoing path. There are cables produced just for this purpose with a braid and outer plastic sheath. You can also get them with 2 wires extra for temperature sensor if the motor have such. It will contain noise carried by the cables provided it is connected to a good ground at least at both ends. Additional grounding along the path improves efficiency. Ground by using clamps around the braid and against the ground rail. Pigtails act as inductors and render it less efficient at higher frequencies. And of course a spiral conductor around the leads is just a mechanical protection. It is a coil, and thus not effective at AC which is what worries us here. Any ground cable that might carry noise (current spikes) should have as large surface as possible. That means many thin strands. The grounding strap from engine to chassis of modern car is a good example. Welding cable is also fine. It should go in a straight line, never coil it up. Use mentioned straps to ground your cabinets to the machine frame and each other. Use the grounding bolts built into the cabinet for this purpose, or if there is none, make sure there is no paint or corrosion hindering a good connection when you install it yourself. Don't forget to ground the cabinet door using 1 or 2 straps. Also make sure your machine have a good connection to the grounding system of your building. If you run sensor cables or data cables to the machine cabinet, those should also be shielded and connected at both ends, and the device at the other end must be considered just another machine cabinet (grounding). This last point is less strict if using galvanic isolations, and no worry if using plastic/glass fibre. Then with the appropriate filters installed, you should be able to run your machine while your wife watches the football game. ;-) Einar Regarding motor cables: I am using a Belden microphone cable called Star-Quad, which has 4 conductors in a good shield that I use with my xylotex kit, one cable per motor, with the shield well grounded at the xylotex board, and not connected to anything at the motor end. This cable is available in gauges as large as 20, but what I found was 24 gauge and its working quite well. The wire layup in this cable is 1 2 3 4 and intended to be used 2 wires in parallel for low impedance mics, where 1-4 is used as one wire, and 2-3 is used as the other so the net pickup effect of external noise is that all wires pickup as if they were perfectly centered balanced. I use 1-4 for the A coil and 2-3 for the B coil. I can run an AM radio 3 feet away without hearing it. I didn't trim the length of the motor leads, so theres about a foot of wire from the plug to the motor exposed. That wasn't one of my more brilliant moves, but it worked so I haven't fixed it. My point is that this might be a good technique to use with the encoders too, by using it exactly as Belden intended from the encoder to the diff amp. Sure, that's twice the wires, but if it works, well... Another thing to consider is that the diff amps are often not very immune to 'longitudinal' noise, noise where the spikes are alike on both inputs, but the amplitude of the spikes exceeds the power supply rails. So your 'termination' resistors might have to be adjusted in value so that you can tie one end to ground, one end to the supply rail, and use the junctions /2 voltage as a bias thereby giving equal noise margins going both directions. What you are doing there is setting the input bias of the diff amps so that equal noise margins are available for both polarities of spikes. Because of unbalance in actual R values, it might be best to set that up as one pair of r's across the supply, in which case they can be a higher value, with the junction bypassed to local ground if that local ground is quiet enough otherwise don't bypass it, and the other two term r's, say 56 ohms hooked from the junction to the inputs. This would allow the use of common 10% resistors rather than fawncy and ex$pen$ive .01%ers. These bias setting r's can be higher values, but should be low enough to absorb the noise and constrain it to within the supply rails. It really should stay at least a volt away from the rails if possible. Food for thought guys. From an old fart C.E.T. -- Cheers, Gene There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order. -Ed Howdershelt (Author) Over the years, I've developed my sense of deja vu so acutely that
Re: [Emc-users] Threading Hickups
John Kasunich wrote: Kirk Wallace wrote: Thanks for the e-mail Einar. There is a cabinet on each side of the lathe. The VFD is in the right cabinet and is grounded to the cabinet. A flex conduit with a plastic outer and a metal spiral inner carries three conductors to the motor connection box which is screwed to the motor housing. So, there should be metal surrounding the VFD to motor leads for the entire distance. There may be metal surrounding it, but that spiral has a lot of inductance and is not a good high frequency ground. If this is conduit and not cable, I'd pull the three wires out, then pull them back in with at least one ground wire added (three if you have room, one loosely twisted around each power wire). Connect the ground wire(s) directly to the motor housing, and directly to the VFDs ground terminal. The ground wires can be smaller than the power wires, especially if you have three. Even better, if you can get some, is a braid that can be expanded enough to run the three motor leads through it, then stretched lengthwise so it snugs down around the motor wires. Again, connect one end directly to the motor frame, and the other directly to the VFD ground. The idea here is to have the return path for stray currents as close as possible to the outgoing path. Regards, John Kasunich As a wild idea, would it work to use hydraulic hose as a conduit for this type of interference? If you skive back the rubber on the end before you crimp on the fitting it would give you a solid layer of wire braid and two fittings of your choice to run the wires through. Ed. - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
SNIP-- Using an oscilloscope, when I probed the power supplies with the spindle running, I got about .5 Volts of ripple that had a three stair step up and down appearance. Probing the +5 Volt differential signals I got a very short 10 Volt spike on the rising edge of each pulse, but otherwise they looked well formed. Without the differential boards, the +5 Volt encoder pulses had a more drawn out spike on the rising edge and the tops varied about a Volt above +5. The whole idea of the differential boards where to reduce the susceptibility of the system to noise, but the boards themselves seem to make the effect of the noise worse. Are your differential inputs terminated? If not, that would account for the spike... CAT5 has 100 Ohm charateristic impedance, so you need a 100 Ohm resistor across each A /A, B /B, X /X pair. Peter Wallace - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
On Sat, 20 Oct 2007, Peter C. Wallace wrote: Date: Sat, 20 Oct 2007 07:36:35 -0700 (PDT) From: Peter C. Wallace [EMAIL PROTECTED] Reply-To: Enhanced Machine Controller (EMC) emc-users@lists.sourceforge.net To: Enhanced Machine Controller (EMC) emc-users@lists.sourceforge.net Subject: Re: [Emc-users] Threading Hickups SNIP-- Using an oscilloscope, when I probed the power supplies with the spindle running, I got about .5 Volts of ripple that had a three stair step up and down appearance. Probing the +5 Volt differential signals I got a very short 10 Volt spike on the rising edge of each pulse, but otherwise they looked well formed. Without the differential boards, the +5 Volt encoder pulses had a more drawn out spike on the rising edge and the tops varied about a Volt above +5. The whole idea of the differential boards where to reduce the susceptibility of the system to noise, but the boards themselves seem to make the effect of the noise worse. Are your differential inputs terminated? If not, that would account for the spike... CAT5 has 100 Ohm charateristic impedance, so you need a 100 Ohm resistor across each A /A, B /B, X /X pair. Forgot to say that these resistors are on the _receiver_ end of the CAT5 cable... Peter Wallace - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
On Sat, 2007-10-20 at 09:56 -0700, Peter C. Wallace wrote: On Sat, 20 Oct 2007, Kirk Wallace wrote: ... snip drain. I don't know its impedance, but I suppose I could start with 200 Ohms and work my way up. The data-sheet indicates that anything lower that 90 Ohms should not be used. I'll give it a try today. If you start at 200 Ohms, you need to work your way down... That is what I meant -- start at 200 and go towards 90. Thanks for clarifying it though. Shielded twisted pair is usually less than 100 Ohms impedance.. Termination need not be perfect, the overshoot is very roughly 1/2 of the impedance mis-match (20% mismatch = 10% overshoot) Open ended cables (no termination) will have 100% overshoot (likely the cause of your 10V spike) Peter Wallace -- Kirk Wallace (California, USA http://www.wallacecompany.com/machine_shop/ Hardinge HNC lathe Bridgeport mill conversion pending Zubal lathe conversion pending) - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
Kirk Wallace wrote: The whole idea of the differential boards where to reduce the susceptibility of the system to noise, but the boards themselves seem to make the effect of the noise worse. What are you using for the differential driver chip? Is it totem-pole output or open-collector? I think I need to place ferrite beads or other type of filter on the VFD inputs and outputs and then revisit the oscilloscope. I don't have much experience with tracking down noise with an oscilloscope, so if anyone has some words of wisdom, I would appreciate hearing them. This can be very difficult, as the power cord ground on the scope adds a guaranteed ground loop that can totally confuse the measurement. One way to know you are getting a ground loop through the scope's power plug is when you connect the ground clip of the scope probe to something, with nothing connected to the probe tip, but you see a signal! Anyway, the diff drivers seem to somehow be the problem. I can't even guess whether it is susceptibility to noise on the power input, noise coupling through the encoder signals, or just the fact that the diff. driver is an amplifier. If the drivers in the encoder are pretty slow, but the drivers in the diff. driver are really fast, then they could be turning a 50 ns glitch with very little amplitude into a strong pulse. I would guess that not only does the Z signal have this, but the A and B have it, too. The digital filtering of the quadrature signals have a defense that only valid quadrature transitions are accepted, but the Z doesn't have that. A filter might be the fix, or more intimate coupling/grounding/shielding between the encoder and diff driver may be the cure. If the encoder has one of those threaded military Cannon or Amphenol connectors with the 4 screws on it, you might make a little box that attaches to the encoder body between encoder and connector and put the diff driver in that. This should help prevent any interference getting into the signals between the two. If the encoder electronics are grounded to the encoder frame, opening this ground could be really helpful. That represents a classic ground loop situation. Don't put capacitors on the VFD output terminals. A series inductance can be helpful, but is not guaranteed to help. A line filter on the input is much more helpful, as capacitors can be used there. I am using a commercial line filter made by Filter Concepts, probably a Mil-spec unit, rated at 15 A, for my Bridgeport spindle. It is likely overkill, a 6 A unit would most likely handle the 1 Hp fine. I have no filtering, and no shielding, either, on the VFD output. I do have a ground wire in the 4-cond cable, tied to the VFD frame at one end, the machine frame at the other. The line input filter solved the noise issues I was having, with the computer monitor and my computer-controlled air compressor, which was only running off the same breaker panel. That line filter came out of my junk box, no idea what it came out of. Jon - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
Kirk Wallace wrote: My encoder disk problem cropped up again too. I am using a one inch U. S. Digital disk with a .466 inch hole and .500 inch shaft. There doesn't seem to be enough clamping area to keep the disk flat, so I have run the disk up towards the receiver end of the slot, so it rubs slightly in one spot, but works better. It seems that it might be best to buy the disk/hub assembly from USD. It might be better to buy a real encoder on eBay. I've always thought that the US digital stuff was just a little TOO low-budget. Jon - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
On Sat, 2007-10-20 at 13:01 -0500, Jon Elson wrote: Kirk Wallace wrote: The whole idea of the differential boards where to reduce the susceptibility of the system to noise, but the boards themselves seem to make the effect of the noise worse. What are you using for the differential driver chip? Is it totem-pole output or open-collector? I am using the DS26C31 http://www.national.com/ds/DS/DS26C31M.pdf I think I need to place ferrite beads or other type of filter on the VFD inputs and outputs and then revisit the oscilloscope. I don't have much ...snip really helpful. That represents a classic ground loop situation. Don't put capacitors on the VFD output terminals. A series inductance can be helpful, but is not guaranteed to help. A line filter on the input is much more helpful, as capacitors can be used there. I am using a commercial line filter made by Filter Concepts, probably a Mil-spec unit, rated at 15 A, for my Bridgeport spindle. It is likely overkill, a 6 A unit would most likely handle the 1 Hp fine. I have no filtering, and no shielding, either, on the VFD output. I do have a ground wire in the 4-cond cable, tied to the VFD frame at one end, the machine frame at the other. The line input filter solved the noise issues I was having, with the computer monitor and my computer-controlled air compressor, which was only running off the same breaker panel. That line filter came out of my junk box, no idea what it came out of. Jon Is it something like this: http://www.eastek-intl.com/images/PreoSeriesER.pdf ? -- Kirk Wallace (California, USA http://www.wallacecompany.com/machine_shop/ Hardinge HNC lathe Bridgeport mill conversion pending Zubal lathe conversion pending) - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
Kirk Wallace wrote: On Sat, 2007-10-20 at 13:01 -0500, Jon Elson wrote: What are you using for the differential driver chip? Is it totem-pole output or open-collector? I am using the DS26C31 http://www.national.com/ds/DS/DS26C31M.pdf Oh, my, a blast from the dark ages! I've used a 75172 in some more recent equipment, but I really don't know if that has anything to do with the problem. Is it something like this: http://www.eastek-intl.com/images/PreoSeriesER.pdf Yes, that is certainly the sort of filter module I used on the VFD. Jon - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
Thanks for the e-mail Einar. There is a cabinet on each side of the lathe. The VFD is in the right cabinet and is grounded to the cabinet. A flex conduit with a plastic outer and a metal spiral inner carries three conductors to the motor connection box which is screwed to the motor housing. So, there should be metal surrounding the VFD to motor leads for the entire distance. The control electronics are in the left cabinet. I did some more investigation today. The current encoder system starts with a Pico Systems Universal PWM Controller and a stand alone +12 Volt power supply. The UPC encoder inputs and +12 Volt supply are connected to a home built differential receiver, then a four-pair shielded cable and the receiver. The four-pair shield is grounded by a lead to cabinet. 6 inch unpaired unshielded leads connect to the spindle encoder. A +5 Volt regulator, input capacitor and output capacitor are on both the receiver and transmitter. I first by-passed as much as I could by running a CAT-5 unshielded four pair cable between the encoder and the UPC encoder inputs and on-board +5 supply. This worked properly, some of the pulses appeared fatter than others, but they where all in the right place and there were no extra pulses. I tried a thread and it came out as close to correct as I could tell. I added the differential boards to the CAT-5 and got the previous noise problem. Finally, I went directly from the encoder to the UPC and its +5 supply using the existing shielded cable which worked properly. Using an oscilloscope, when I probed the power supplies with the spindle running, I got about .5 Volts of ripple that had a three stair step up and down appearance. Probing the +5 Volt differential signals I got a very short 10 Volt spike on the rising edge of each pulse, but otherwise they looked well formed. Without the differential boards, the +5 Volt encoder pulses had a more drawn out spike on the rising edge and the tops varied about a Volt above +5. The whole idea of the differential boards where to reduce the susceptibility of the system to noise, but the boards themselves seem to make the effect of the noise worse. I think I need to place ferrite beads or other type of filter on the VFD inputs and outputs and then revisit the oscilloscope. I don't have much experience with tracking down noise with an oscilloscope, so if anyone has some words of wisdom, I would appreciate hearing them. I will try to get some scope pictures on my website too. If anyone has a good source for VFD appropriate filters, please let me know. I melted the original filters, so I know you can't use just anything. In the mean time, leaving the differential boards out, seems to work. On Fri, 2007-10-19 at 21:01 +0200, [EMAIL PROTECTED] wrote: I may not have seen all the posts regarding this, but none that I saw mentions using shielded cable from the VFD to the motor? If not already done, change to shielded cable for the motor leads too! The braid needs to be grounded at the VFD and the motor. I found another thread where someone suggested the same thing plus using an additional three conductors, grounded on each end, to make a three twisted pair motor connection. Never ever use a pigtail to extend the braid to ground. This goes for both encoder and power cables. Clamp the braid itself to ground or use a grounding feedthrough! I have a central grounding post on the backplate of my cabinet. I have been trying to run all my grounds to this post, including leads from my cable shields. Are you suggesting tying the shields to the backplate where the the shield ends? If there is any potential difference between motor and VFD (cabinet), equalise it with a multistrand ground lead with a good cross section. Don't care if this might look like creating a ground loop through the braid. The current will go through the lead with the least resistance (your ground lead). Hence the solid cross section. Einar Sjaavik I have tried to avoid the dreaded ground loop (thats funny, pilots have a dreaded ground loop too) but the wiring has gotten so complex, that I probably wouldn't see it. -- Kirk Wallace (California, USA http://www.wallacecompany.com/machine_shop/ Hardinge HNC lathe Bridgeport mill conversion pending Zubal lathe conversion pending) - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
Kirk Wallace wrote: Thanks for the e-mail Einar. There is a cabinet on each side of the lathe. The VFD is in the right cabinet and is grounded to the cabinet. A flex conduit with a plastic outer and a metal spiral inner carries three conductors to the motor connection box which is screwed to the motor housing. So, there should be metal surrounding the VFD to motor leads for the entire distance. There may be metal surrounding it, but that spiral has a lot of inductance and is not a good high frequency ground. If this is conduit and not cable, I'd pull the three wires out, then pull them back in with at least one ground wire added (three if you have room, one loosely twisted around each power wire). Connect the ground wire(s) directly to the motor housing, and directly to the VFDs ground terminal. The ground wires can be smaller than the power wires, especially if you have three. Even better, if you can get some, is a braid that can be expanded enough to run the three motor leads through it, then stretched lengthwise so it snugs down around the motor wires. Again, connect one end directly to the motor frame, and the other directly to the VFD ground. The idea here is to have the return path for stray currents as close as possible to the outgoing path. Regards, John Kasunich - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
I may not have seen all the posts regarding this, but none that I saw mentions using shielded cable from the VFD to the motor? If not already done, change to shielded cable for the motor leads too! The braid needs to be grounded at the VFD and the motor. Never ever use a pigtail to extend the braid to ground. This goes for both encoder and power cables. Clamp the braid itself to ground or use a grounding feedthrough! If there is any potential difference between motor and VFD (cabinet), equalise it with a multistrand ground lead with a good cross section. Don't care if this might look like creating a ground loop through the braid. The current will go through the lead with the least resistance (your ground lead). Hence the solid cross section. Einar Sjaavik - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
On Thursday 18 October 2007, Kirk Wallace wrote: On Thu, 2007-10-18 at 00:17 -0400, Jim Coleman wrote: On 10/18/07, Jon Elson [EMAIL PROTECTED] wrote: Kirk Wallace wrote: Thanks again Jon and Chris. What should have been obvious turns ... snip interference sources like motor-VFD cables and any other that might have 120 or 240 VAC on them. Jon in my hitachi seiki manual, it says that twisted pair wire should be used for the encoders. not sure if this would help eliminate noise in your situation or not. I used the existing cable which has four shielded twisted pairs. I have one pair for each A, B, I and power. How about the ends of the shielding wrap at the machine? They should be well grounded at the computer interface end and get as close as possible to the encoder, but should probably be isolated from the machine frame to avoid picking up noise or enabling a ground loop, which with variable phase angle thyristor speed controls on the spindle motor, can contain a lot of hf noises. And if the encoder has a ground, can it be isolated from the machine and tied to one of the shields? I'm thinking of the usual fiber or possibly nylon washers used for transistor isolation as a method of isolating the encoder pcb mounting from the machine frame. The mica sheets for such aren't all that thick if the washer thickness might be a problem in maintaining the spacings. A possible test to see if its VFD generated noise would be to see if it largely goes away when the motor is at full speed, and gets progressively worse as the speed is lowered. It may get better when the motor is at creep speeds too depending on how much losses there are mechanically, and the VFD itself. Some are thyristor based which would act like that, and some are basicly switching regulators driving dc motors, with the switching being done at several kilohertz. Those types could get noisier as the speed goes up, or if a hand is laid on the chuck to load the motor and make it work harder. Basically, look for a correlation between spindle speed and the extra noise pulses. An oscilloscope with at least a 20 megahertz bandwidth can be handier than bottled beer for this sort of troubleshooting. also what about an inductor or capacitor at the encoder in its power line to help smooth any noise introduced there? I have a short run between my encoder and a differential driver. I am wondering about whether I should have some sort of filters on the high impedance inputs of the driver. The thing is that with 50k pulses at 3,000 I'm not sure how much of a filter I could use. This might be a good place to use a piece of that shielded cable too. -- Cheers, Gene There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order. -Ed Howdershelt (Author) If God had not given us sticky tape, it would have been necessary to invent it. - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
if there is a ground loop somewhere in your setup, it won't conduct any current unless it's ... a loop! so when you disconnect the sensor, it breaks the circuit; no current is induced in the wires by stray magnetic fields, and you dont see any voltage. if the noise were 'coming from the sensor' then you should see the noise with it hooked up in a different place, such as on a lab bench. if your differential driver has *really* sensitive inputs, i suppose you could be picking up stray electric fields. this could easily be fixed with a pull-up resistor. but this wouldn't explain the noise going away when the encoder sensor is disconnected. On Wed, 17 Oct 2007, Kirk Wallace wrote: When I was getting index pulses without an encoder disk installed, I decided to disconnect the index connections to see when the noise stopped. The noise stopped as soon as I disconnected the encoder sensor. So either the noise is coming from the sensor, or having the sensor connected promotes the noise. I am going to have to give this some thought in order to plan the next move. - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
Gene Heskett wrote: A possible test to see if its VFD generated noise would be to see if it largely goes away when the motor is at full speed, and gets progressively worse as the speed is lowered. Basically, look for a correlation between spindle speed and the extra noise pulses. A very easy way to see if the spindle drive is generating the noise is to run the spindle up to top speed and then cut the drive completely and let the spindle coast to a stop. If the signals get beautiful as soon as you turn off the VFD, you know where the noise is coming from. And if they don't, you know it's something other than the drive. Regards, John Kasunich - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
On Thu, 2007-10-18 at 09:38 -0400, John Kasunich wrote: Gene Heskett wrote: A possible test to see if its VFD generated noise would be to see if it largely goes away when the motor is at full speed, and gets progressively worse as the speed is lowered. Basically, look for a correlation between spindle speed and the extra noise pulses. A very easy way to see if the spindle drive is generating the noise is to run the spindle up to top speed and then cut the drive completely and let the spindle coast to a stop. If the signals get beautiful as soon as you turn off the VFD, you know where the noise is coming from. And if they don't, you know it's something other than the drive. Regards, John Kasunich Thanks for the tips. Here is another interesting link: http://www.eng-tips.com/viewthread.cfm?qid=196269page=5 I totally underestimated and overlooked the issues involved with VFD's. I guess they aren't a simple plug and play kind of thing. I plan on using allot more VFD's in the future and developing experience will not only help now but in the future. -- Kirk Wallace (California, USA http://www.wallacecompany.com/machine_shop/ Hardinge HNC lathe Bridgeport mill conversion pending Zubal lathe conversion pending) - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
Another possible trouble is that your motor frame is not grounded with a low impedance to earth ground. If this is the case, capacitance from motor windings to the motor frame will couple switching spikes from your VFD into your encoder via the motor shaft or encoder cover (metal covers make this effect worse) Peter Wallace - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
Kirk Wallace wrote: I used the existing cable which has four shielded twisted pairs. I have one pair for each A, B, I and power. Are the shields grounded at both ends? also what about an inductor or capacitor at the encoder in its power line to help smooth any noise introduced there? I wouldn't. Possibly adding a capacitor across +5 to ground might help. I have a short run between my encoder and a differential driver. I am wondering about whether I should have some sort of filters on the high impedance inputs of the driver. The thing is that with 50k pulses at 3,000 I'm not sure how much of a filter I could use. Is the differential driver module grounded to anything, like through mounting screws? Check if the grounds of the encoder are grounded to the machine frame. Since you have had the encoder all apart anyway, make sure it is NOT grounded to the machine. If ungrounding it is really hard, then run a huge copper braid cable from the machine to the computer/controller grounds (not a bad idea anyway). If you have a scope (not Halscope) look at the index signal all the way from the encoder itself, out of the diff driver, into the diff receiver and finally out of the diff receiver, to see where the noise is getting in. If it is coming directly out of the encoder, possibly a small cap across the power input to the encoder will help. Otherwise, maybe a better ground between the encoder and diff driver is needed so it doesn't see common-mode noise. From previous messages, I think you have 5000 counts/rev, not 50K. 3000 RPM is 50 Rev/sec, x 5K encoder pulses is 25 counts a second. The quadrature frequency would be 1/4 that, or 62.5 KHz. So, a filter with about a 8 us cutoff would be pretty good. The digital filtering I used on the A and B channels wasn't applicable to the Z channel since it doesn't have a predictable state progression, so it has less filtering, and is sampled every microsecond. But, assuming your Z pulse is at least one quadrature count wide, then an 8 us filter might do a lot of good. In your next mesaage, you said : When I was getting index pulses without an encoder disk installed, I decided to disconnect the index connections to see when the noise stopped. The noise stopped as soon as I disconnected the encoder sensor. So either the noise is coming from the sensor, or having the sensor connected promotes the noise. I am going to have to give this some thought in order to plan the next move. This may suggest problems with the encoder, but could be common-mode noise on the encoder ground or noisy power to the encoder. A simple R/C filter might be all that is needed. especially if the noise pulses are just a 100 ns wide or something, the filter should give perfect results. Jon - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
John Kasunich wrote: Gene Heskett wrote: A possible test to see if its VFD generated noise would be to see if it largely goes away when the motor is at full speed, and gets progressively worse as the speed is lowered. Basically, look for a correlation between spindle speed and the extra noise pulses. I doubt you'd see much difference. The VFD switches 340 V pulses across the motor windings no matter whether it is at 1 Hz or 400 Hz. Would it be because at high speed the motor is at greater load and the current is more in phase with the back EMF? Or, would the noise be gtting worse with reduced on-time of the transistors? I haven't had to play detective with VFD interference, so maybe it would really show such a characteristic, but I wouldn't have expected it. (Might also be a difference between the BIG drives and motors you play with and the ones I have experience with.) A very easy way to see if the spindle drive is generating the noise is to run the spindle up to top speed and then cut the drive completely and let the spindle coast to a stop. Now, this is an EXCELLENT idea! You have to set up the VFD to not do any decelerating or braking, but that is usually a setting that can be done. I did have some non-CNC interference from my mill's VFD, mostly to my computer monitor. I put a Corcom-type line filter box on the line in to the VFD, and it solved the problem. Jon Jon - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
Kirk Wallace wrote: I totally underestimated and overlooked the issues involved with VFD's. I guess they aren't a simple plug and play kind of thing. I plan on using allot more VFD's in the future and developing experience will not only help now but in the future. I'll be interested to hear what the outcome is. They do produce a LOT of noise. I had to filter the line input to my mill's VFD, it trashed the computer's video monitor. (No harm, just a lot of artifacts on the screen, looked kind of like a chain-link fence.) The line filter completely solved it. Jon - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
Jon Elson wrote: I did have some non-CNC interference from my mill's VFD, mostly to my computer monitor. I put a Corcom-type line filter box on the line in to the VFD, and it solved the problem. Jon I had the same thing with the fractional HP VFD on my drill press. It was only running a 1/2 HP three phase motor, but it put enough interference back into the AC line that you couldn't use an AM radio anywhere in the house. During baseball season that was unacceptable. A Corcom filter cleaned it right up. Filtering gets a lot more complicated when you start talking about a few HP, but for fractional HP, a 10A Corcom will do wonders for RF interference. Noise getting into encoder signals is NOT the same as AM radio interference. Although a Corcom can't hurt, I would not expect it to help much with encoder noise. That needs to be addressed by grounding, shielding, and use of differential signals. I seem to recall Kirk saying that he has a short run of single ended signals between the encoder itself and the differential driver board. That board should be as close as humanly possible to the encoder, and its ground must be connected to the encoder ground. Any stray currents flowing in the grounds between the two items will result in noise. Regards, John Kasunich - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
Thanks again Jon and Chris. What should have been obvious turns out to be the problem. Graphing the encoder index shows random pulses mixed in with the regular pulses. I really did not want this to be the problem since I had already gone through fixing another issue with this encoder. I guess wishing is not a valid method for repairing machinery. Before, I had a problem with the encoder disk position coupled with a wobble. Now it seems to be an EMI problem because I get occasional index pulses with the disk removed and the spindle turning. I haven't been able to get a reliable reproduction of the fault, so I will be busy with trouble shooting for a while. On Tue, 2007-10-16 at 21:06 -0500, Jon Elson wrote: Kirk Wallace wrote: After having some success with my tuning, I went back to making threads without much success. Randomly, but close to every third or fourth pass ... snip I assume that the spike in the spindle velocity data is due to a counter reset at the start of each pass? Another problem, that I had read about a while back, cropped up too. When I invoke the touch-off dialog, I get a Bad Number error unless the first thing I do is forward arrow and leave at least the original leading zero in the data entry box. The message just means whatever is in the box isn't valid - just correct the characters in the box. For instance, - or . are invalid all by themselves, but -.5 is OK. Jon Previously, the dialog box would complain, but as soon as I got a valid number entered it would stop complaining. Now, as soon as the error message appears, it won't go away until I close the dialog box and start over. I have learned that in order to keep from getting an error message, I must keep an editable, but always valid number in the data entry box. Even deleting the initial 0.0 will toggle the error, so I end up clicking the end of the 0.0, back spacing ( 0.0| , 0.| , 0| ) to the zero and entering the required number such as 01.032. -- Kirk Wallace (California, USA http://www.wallacecompany.com/machine_shop/ Hardinge HNC lathe Bridgeport mill conversion pending Zubal lathe conversion pending) - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
On Wed, Oct 17, 2007 at 12:35:12PM -0700, Kirk Wallace wrote: Previously, the dialog box would complain, but as soon as I got a valid number entered it would stop complaining. Now, as soon as the error message appears, it won't go away until I close the dialog box and start over. I have learned that in order to keep from getting an error message, I must keep an editable, but always valid number in the data entry box. Even deleting the initial 0.0 will toggle the error, so I end up clicking the end of the 0.0, back spacing ( 0.0| , 0.| , 0| ) to the zero and entering the required number such as 01.032. You are not the first one to say this! But, jeff and I have never seen it and are baffled by the reports. Is there any pattern you can see that would help us reproduce the problem? Chris - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
Chris Radek wrote: On Wed, Oct 17, 2007 at 12:35:12PM -0700, Kirk Wallace wrote: Previously, the dialog box would complain, but as soon as I got a valid number entered it would stop complaining. Now, as soon as the error message appears, it won't go away until I close the dialog box and start over. I have learned that in order to keep from getting an error message, I must keep an editable, but always valid number in the data entry box. Even deleting the initial 0.0 will toggle the error, so I end up clicking the end of the 0.0, back spacing ( 0.0| , 0.| , 0| ) to the zero and entering the required number such as 01.032. You are not the first one to say this! But, jeff and I have never seen it and are baffled by the reports. Is there any pattern you can see that would help us reproduce the problem? It doesn't fix the problem, but it may be easier to hit home then shift-end, then start typing. That will move to the beginning of the edit field and then select to the end of the field, so whatever you type replaces what's there. - Steve - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
Kirk Wallace wrote: Thanks again Jon and Chris. What should have been obvious turns out to be the problem. Graphing the encoder index shows random pulses mixed in with the regular pulses. I really did not want this to be the problem since I had already gone through fixing another issue with this encoder. I guess wishing is not a valid method for repairing machinery. Before, I had a problem with the encoder disk position coupled with a wobble. Now it seems to be an EMI problem because I get occasional index pulses with the disk removed and the spindle turning. I haven't been able to get a reliable reproduction of the fault, so I will be busy with trouble shooting for a while. Check the grounding carefully. The best is to not have the encoder grounded to the machine frame, but grounded through the board reading the encoder signals. Shielding of the encoder cables is almost always required. Check the lay of probable interference sources like motor-VFD cables and any other that might have 120 or 240 VAC on them. Jon - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
On 10/18/07, Jon Elson [EMAIL PROTECTED] wrote: Kirk Wallace wrote: Thanks again Jon and Chris. What should have been obvious turns out to be the problem. Graphing the encoder index shows random pulses mixed in with the regular pulses. I really did not want this to be the problem since I had already gone through fixing another issue with this encoder. I guess wishing is not a valid method for repairing machinery. Before, I had a problem with the encoder disk position coupled with a wobble. Now it seems to be an EMI problem because I get occasional index pulses with the disk removed and the spindle turning. I haven't been able to get a reliable reproduction of the fault, so I will be busy with trouble shooting for a while. Check the grounding carefully. The best is to not have the encoder grounded to the machine frame, but grounded through the board reading the encoder signals. Shielding of the encoder cables is almost always required. Check the lay of probable interference sources like motor-VFD cables and any other that might have 120 or 240 VAC on them. Jon in my hitachi seiki manual, it says that twisted pair wire should be used for the encoders. not sure if this would help eliminate noise in your situation or not. also what about an inductor or capacitor at the encoder in its power line to help smooth any noise introduced there? - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
On Thu, 2007-10-18 at 00:17 -0400, Jim Coleman wrote: On 10/18/07, Jon Elson [EMAIL PROTECTED] wrote: Kirk Wallace wrote: Thanks again Jon and Chris. What should have been obvious turns ... snip interference sources like motor-VFD cables and any other that might have 120 or 240 VAC on them. Jon in my hitachi seiki manual, it says that twisted pair wire should be used for the encoders. not sure if this would help eliminate noise in your situation or not. I used the existing cable which has four shielded twisted pairs. I have one pair for each A, B, I and power. also what about an inductor or capacitor at the encoder in its power line to help smooth any noise introduced there? I have a short run between my encoder and a differential driver. I am wondering about whether I should have some sort of filters on the high impedance inputs of the driver. The thing is that with 50k pulses at 3,000 I'm not sure how much of a filter I could use. -- Kirk Wallace (California, USA http://www.wallacecompany.com/machine_shop/ Hardinge HNC lathe Bridgeport mill conversion pending Zubal lathe conversion pending) - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
On Wed, 2007-10-17 at 23:02 -0500, Jon Elson wrote: Kirk Wallace wrote: Thanks again Jon and Chris. What should have been obvious turns out to ... snip trouble shooting for a while. Check the grounding carefully. The best is to not have the encoder grounded to the machine frame, but grounded through the board reading the encoder signals. I have the shield of encoder cable tied to a central ground on the backplate that mounts all the left side electronics. Shielding of the encoder cables is almost always required. I used the existing shielded four pair. Check the lay of probable interference sources like motor-VFD cables and any other that might have 120 or 240 VAC on them. I have a differential driver mounted next to the encoder, then the shielded cable to the receiver next to the UPC. About six inches away from the UPC are the two 100 Volt PWM amps. A little farther away are a bank of AC SSR's. I'll post some pictures soon. When I was getting index pulses without an encoder disk installed, I decided to disconnect the index connections to see when the noise stopped. The noise stopped as soon as I disconnected the encoder sensor. So either the noise is coming from the sensor, or having the sensor connected promotes the noise. I am going to have to give this some thought in order to plan the next move. Jon -- Kirk Wallace (California, USA http://www.wallacecompany.com/machine_shop/ Hardinge HNC lathe Bridgeport mill conversion pending Zubal lathe conversion pending) - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
[Emc-users] Threading Hickups
After having some success with my tuning, I went back to making threads without much success. Randomly, but close to every third or fourth pass comes out being about a half thread off. I have Halscope graphs here: http://www.wallacecompany.com/cnc_lathe/HNC/emc2/g76-bad.png http://www.wallacecompany.com/cnc_lathe/HNC/emc2/g76-good.png They aren't annotated but have the same configuration as my other graphs. For a reference, I put a bar in the collet and touched off. I invoked my G76 program and E-stopped after the first pass. I then touched-off X such that the cutter came close to the bar, but not touching on the last pass of the G76. I re-ran the program and screen copied good an bad passes determined by comparing the tool position to the reference pass I made at the beginning. I can't really see much difference between the good and bad graphs ,but there was an obvious indication with the tool and reference pass. I assume that the spike in the spindle velocity data is due to a counter reset at the start of each pass? Another problem, that I had read about a while back, cropped up too. When I invoke the touch-off dialog, I get a Bad Number error unless the first thing I do is forward arrow and leave at least the original leading zero in the data entry box. Any help with these would be appreciated. -- Kirk Wallace (California, USA http://www.wallacecompany.com/machine_shop/ Hardinge HNC lathe Bridgeport mill conversion pending Zubal lathe conversion pending) - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
Kirk Wallace wrote: After having some success with my tuning, I went back to making threads without much success. Randomly, but close to every third or fourth pass comes out being about a half thread off. I have Halscope graphs here: http://www.wallacecompany.com/cnc_lathe/HNC/emc2/g76-bad.png http://www.wallacecompany.com/cnc_lathe/HNC/emc2/g76-good.png What you want to do is put ppmc.0.encoder.02.position, (or whichever axis is the spindle encoder), ppmc.0.encoder.02.index-enable and ppmc.0.encoder.02.index on the halscope. With the spindle running, you should see regularly-spaced narrow pulses on index. If they are not regularly spaced, you are picking up noise on the encoder index channel. They aren't annotated but have the same configuration as my other graphs. For a reference, I put a bar in the collet and touched off. I invoked my G76 program and E-stopped after the first pass. I then touched-off X such that the cutter came close to the bar, but not touching on the last pass of the G76. I re-ran the program and screen copied good an bad passes determined by comparing the tool position to the reference pass I made at the beginning. I can't really see much difference between the good and bad graphs ,but there was an obvious indication with the tool and reference pass. I don't actually see much difference in the two graphs except for the trigger point. But, then the computer doesn't know the wrong place has been picked up by the encoder index signal, so the graphs shouldn't look different. But, if you display the enocder index signal too, then you will clearly see if there is an index in the wrong place. I assume that the spike in the spindle velocity data is due to a counter reset at the start of each pass? Another problem, that I had read about a while back, cropped up too. When I invoke the touch-off dialog, I get a Bad Number error unless the first thing I do is forward arrow and leave at least the original leading zero in the data entry box. The message just means whatever is in the box isn't valid - just correct the characters in the box. For instance, - or . are invalid all by themselves, but -.5 is OK. Jon - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Threading Hickups
On Tue, Oct 16, 2007 at 09:06:29PM -0500, Jon Elson wrote: With the spindle running, you should see regularly-spaced narrow pulses on index. If they are not regularly spaced, you are picking up noise on the encoder index channel. If you are getting some threading passes in the wrong place, I agree this is what's happening. - This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now http://get.splunk.com/ ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
