Thanks for the reply & its great help!

Now I know the problem is not with the OS.

A re-cap, The battery & power up issue had been solved, I had mistakenly connected the power up switch to the RESET pins... :p Now all power up nicely. One down.. more to go...

I think there are only 2 "external item", the dimm DDR2 RAM & ATA ribbon cable. Will try to find a spare in my junk pile.

It's 11pm local time, time to call it a day.

I had never encounter problems of this magnitude before this but then again, the last PC I build was in 2001... things are soo different now...

I actually enjoy this frustrating process :confused:... I think I need to see a shrink :rolleyes: ...

Bought a new RAM, found a new unused IDE cable, replaced them & still the error persist... I'll return the ATOM mobo to the supplier & see what they have to say about it.

OT, RAM price is escalating, only a few months ago, I bought a 2Gb DDR2 RAM for RM85, today its RM150...

Sent the Mobo back to the shop & thet'll get back to me tomorrow.

I decided to get on with the kitchen table project this morning.

I'd been tinkering around with the control panel for a while now, I'd decided to have the spindle & VFD on separate power line & use only a external magnetic coil contactor for it.

The control panel will house the mobo, mobo PSU, BOB, the motor drivers & the driver's power supply.

I will use a EI core 48Vac 500VA tranny; Many regard EI core tranny is "less efficient" due to its high magnetic losses but because of this, they are capable of handling high surge thanks to its "magnetic deficiency" which acts as a "buffer" for this application. Secondly, they are easily available & I can have them custom wound locally if I want, and they cost the same as off the shelve units.

The Electrical circuit will be simple, the main will come off the wall to a contactor which will tie to the E-stop buttons & the on/off button. From the contactor, I'll tap the line for the fan, the motor tranny & the mobo's PSU.

I'm torn between the following layout. Please comment.

Choice 1
Attachment 7079

Choice 2
Attachment 7080

Choice 3
Attachment 7081

Choice 4
Attachment 7082

Choice 5
Attachment 7083

I vote for choice 3 - be careful with that Mobo- might blow again with all the wires inside, prefer to house it in its own cabinet.

mine went that way and there is no warranty for a burnt out one!

I'd try to keep the motherboard as far from the drives as possible, so something like choice 2 or 3, but with the motherboard mounted low down on the back of the door.

I think that mounting items so that there is a single layer when the door is open will make any troubleshooting easier. So better to mount to the back of the door than to have components behind each other.

Hello Ken,

Nice going.

Why do want to have the mobo inside?
A mobo should work in a shielded ambience. Isn`t there danger of having
interference from the torroids or other sources?
I guess a mobo is even more sensitive than the BOB, and Bob should also parked away from any source of resonance.
Do you also plan to have a HHD inside?

Greetings Claudiu
...just a thought of Risk management.

Thanks you for the comments & sharing your experience, seems that all are concern about having things too closely packed...

Claus,
Thank you for reminding the risk management part.

There are 2 sets of reasons why I wish to have the mobo in the panel

Technically,
1) I wish to eliminate the long printer cable running across the room & risk data corruption, I may sound paranoid... :o
2) I also feel that the BOB is best install in the external PC (if one is used) as it's purpose is to beef-up the parallel port signal, hence reduce the tendency of data corruption.) Since I wish to eliminate cable (this time the E-stop, proxy switch.. etc cables.) It make sense to me to place the mobo inside the panel.

Non-technical reasons
Choosing a casing... aesthetic, size, color, location... dries up too much brain juice :mad: ... and adds on to the cost... :p

I don't use toroid, but EI core tranny for its better tolerance to withstand surge demand thanks to its poorer magnetic property, and yes, they are noisier then toroids.

Interference from other components? I really don't know. I learn from building audio amp that, air space is the best cure, if you give 25~50mm space between components, the noise will be hardly noticeable... but I really don't know how they will behave in the MM panel.

I am going to use shielded cables within the panel, I have some nice microphone wires in hand, I recon they will do a great job as signal cables.
Attachment 7089

Found some thick gauge shielded Pro-Audio cables in local pro-audio shops which I can use for the motor cable.

Going about proper wire layout practice is vital too, I won't be using cable trunking as this will force the cables to run parallel to each other (which is a bad practice). I'll do my best to avoid running power cables parallel to signal cable when possible.

Careful with ground loop is important too.

If all else fail, my plan B for noise control is the put up thin steel plates or Faraday cage around the mobo as noise shield.

Now back to more "interior design".

I actually like Choice 5, the air circulation is nice. If I bring the mobo up, then it will be as far as it can get from the tranny but some how it doesn't look right...

Choice 2 will see the mobo next to the tranny.

In Choice 3, the mobo on the cover will be in closed proximity to whatever that is in the box, about 30~40mm.

Taking Brad, Irfan #33 & Brad's comments & advises into considerations 2 new layout were conceded and they do look right to my eye.

Choice 6
Attachment 7090
The mobo will stack on top of its PSU and the BOB will be in vertical position mounted perpendicular to the mobo. The PSU fan fan blow to the right.

The supplier of the drivers said the drivers are shielded, so noise is not a problem. But need to consider cooling of the heat sink. Hence, I orientate the fins vertically for best natural convection, if the fan fails (risk management :D )

The AC fan and the tranny are at the bottom, this will be the noisy section of the panel. The drivers will partially shield the mobo from the dirty section

The free space is reserve for future A-axis driver and/or extra BOB if it ever happens.

The contactor on the cover.

Choice 7
Attachment 7091
Same concept applies.

The 2 drivers on the left are the X-axis drivers, Y in the middle & Z on the far right.

I can mount the contactor on the cover & reserve the space for future A-axis driver.

The X & Y drivers are on the fan blow path, the mobo PSU fan will blow on the Z driver then the tranny after it.

Please comment freely.

Thanks in advance.

Regards

Ken

Claus ask about the HDD,

No HDD & CD drive, I will be using a 8Gb thumb-drive for boot up & storage. in an attempt to reduce moving parts in the panel & keep it physically simple.

Cheers

Ken

Watch out for the life of that thumb-drive - I have had a couple that wore out with repeated use. (Or maybe I left them too near to a strong magnet)

Hi Gerald, I can't thank you enough for this beautiful design.

I'll keep the thumbdrive in cosy and airy location all by itself using an extention connection for mounting on the cover.

Regards

Ken

Visited the friendly computer shop & was told the mobo is fried for good, its still under warrenty but they are out of stock for the moment, will need a week for the replacement to arrive. As I'm too anxious to get on with the project, I bought a D945GCLF2 which is a dual core upgrade. & a matching PSU (this uses a 2x12 power socket & the previous one uses a 2x10 power socket.) RM290 mobo & RM35 PSU. total damage ~US$92.

Now I have the excuse to build that download station which I've been dreaming about.

Once back to me work bench, connected everything up & she boot up flawlessly.

The new test jig
Attachment 7122

Latency test 15801ns its good
Attachment 7123

The EMC2 screen shot
Attachment 7124

Now I can proceed with the control panel

My initial size is for 2440mm x 1220mm.

There had some new development progress on the A-axis indexer. I know I couldn't resist the "upgrade" in the future, so I cater for that extra in this MM build.

Attachment 7134

Have not order the steel yet which is a good thing

I do have enough racks & would think the extra steel cost justify.

I'll have the 0,0 set to the top left corner. there won't be any cross beam outside of the board area and I will have good acces loading raw board onto the work area.

The leg will be at the end of the last cross-beam which the board is mounted. Hope this will help with rigidity.

Please comment freely.

Thanks in advance

Ken

My progress is pathetically slow... :o

Apparently, my electronics & control know-how is far worst then I anticipated... :o

I've been reading the PMDX-122 manual over & over again and honestly, I'm still quite lost...

In between reading I did some drilling, cut some acrylic sheets, tapped some holes and played with jigsaw puzzle... And this is what I came up with
Attachment 7267

I experiment with double sided tape. The PSU & the DIN rail are mounted with the tape on the panel, I am please with the result. Will tape the CPU + BOB assembly & the transformers once they are completed.

I tried spiral tap bits on the acrylic plate, they are fast! Just has to be careful while retracting the bit or risk damage thread. The spiral point taps are commonly know as "normal machine tap" in our local hardware stores, about USD4~5 per piece from India.

ABB contactors, will use one eventually.

PMDX-122 will have her power from the PSU, 12V from the free disc drive plug.

If heat & ventilation is a problem, I can run one or more fans tapping power form the other free disc drive plug.

The front view... nothing to see actually, 2 push button & an indicator light all 240Vac. Dia 22 holes, USD3~4 per piece. and I just walk-in any electrical hardware store & they will spoil me with an array of choices.
Attachment 7268

The internal cramp space... I bet on good shielding of the PSU & the drivers... hope this works :)
Attachment 7269

An addition to my Mobo issue, the computer guy returned the old D945GCLF to me & said its ok. I felt defeated with the outcome...

With a stroke of luck, I learn that the RTI of the Ubuntu 8.04 isn't compatible with the BIOS installed on board, they are 2 remedies
1) Disable the LAN through BIOS setting
2) install a updated BIOS. This is available for down load from Intel site.

I disabled the LAN & installed EMC2 without a glitch. I ran the Latency test for approximately 20 minute with "torture treatment", i.e. copy huge files, played video, moving windows.... etc etc. I got Max jitter 14,952ns
Attachment 7271

Then I left it "burn-in" for 6 hours with the Latency test program alone, I got 16,313ns.
Attachment 7270

16,313ns is still a great result. For the sake of knowing, does anyone has an explaination for this?

Ken,

out of curiosity, does EMC2 not need a RTOS? (Real Time Operating System) or can it use any 'off teh shelf Linux' then there shouldn't be real problem like jitter.

Ries

I don't quite understand your question. I use the EMC2 Live-Disc which also includes the Ubuntu 8.04 LTS & other stuff.
This is my 1st Linux endeavor. From my limited understanding, EMC2 operate in real-time kernel, the problem here is the RTi version is not compatible with the one pre-load on the D945GCLF board,get an upgraded BIOS version and problem solved. No such problem with other board such as D945GCLF2. or if you run the Ubuntu 8.04.1 & 8.04.2. So, no an issue now that you know the solution...

I am also confused, please clarify for me, jitter happens when the hardware could not catch up with the software?? So logically, jitter is due to hardware? Software have to slowdown to cater the hardware?

My latency test shows 16,313ns, this gives a kernel speed of 61.3kHz w/o jitter (1/Max jitter, I hope I understand correctly) which is darn good for most application. Please correct me if I'm wrong.

Just an addition, running the OS off thumbdrive "feels" slower, but the latency test said otherwise... mind boggling...

Ken,

usually such jitter happens because a other process (kernel driver, other program) sits in the way of the active process. The active process would be the program that creates the pulses.

So Jitter happens not because the hardware is not fast enough, but because other programs take to much CPU or cannot get interrupted at that point and thus creates jitter on IO.

[Just an addition, running the OS off thumbdrive "feels" slower, but the latency test said otherwise... mind boggling...] Depends on what feels slow, and I assume the UI here, even modern Operating Systems don't understand the best way to access a thumbdrive, it's not because nothing rotates around which makes it faster.

Let me explain, Operating systems currently assume the 'disk' has a platter and things rotate and so drivers are optimized for this. thumbdrives do have very different access statistics, however they use the same 'language' so the software drivers that access a normal harddisk can also access a thumbdrives. The problem is that thumbdrives do have different access statistics and as such it can slow down a computer, usually the GUI, loading software, writing batched operations etc... (deleting from a thumbdrives works totally different then from a HD).

Anyways.. I am curious to know about EMC2, and although I am a software engineer, I do wonder how well EMC2 integrates and talks to the software I have on my mac and PC.

Ries

EMC2 requires a linux kernel with real time extensions. This kernel is available as a pre-packaged option in the Ubuntu distribution. The EMC-Live disk is an Ubuntu release with the real time kernel and the EMC2 software packages preloaded. For all intensive purposes it's simply an Ubuntu system and acts as such, although it's performance on other tasks is compromised while EMC2 is running, as it should be.

The real time aspects of the kernel can be enabled and disabled; when you start EMC2, the system switches to real time mode. However, there are certain hardware / firmware aspects of PCs that limit how perfect a real time system can be. As one example, there are PCI cards that can perform operations against main memory directly, and in so doing, cause contention with the CPU. This can cause the occasional jitter, which is why measuring it and configuring for the worst case is important.

Note that a system explicitly designed for real time operations would tend to not use hardware (network cards, fancy display cards, etc) that exhibit these issues. However, due to their low performance, these tend to be low volume specialty items, and are more expensive than their mainstream counterparts.

Ries, Brad, thanks for you explanation. really help me to visualize the scenario better.

Ries, you are right, the UI seems slower & programs upload feels slower. just noticably slow, I belive if I time them, I won't find much.
I am afraid I won't be much use with EMC2, I had just decided to act like an application engineer the to pretend like a software engineer :o

Brad, are you suggesting I strip the system down even further? I'd disabled the LAN & I intended to Disable more functions but am not sure which. Could you kindly point me the right way?

Now that my control panel actually look like one :) I need to figure out the BOB.
I am using the PMDX-122 BOB & Leadshine M860 Microstepping driver. This is what I came out with...
Attachment 7304
Please comment freely & correction is mandatory as I know I still hasn't got it right yet...

I am trying to achieve the followings
1) charge-pump feature,
2) fault feature,
3) BOB respond to the E-stop circuit,
4) Bob to turn on/off turning on the spindle c/w manual override
5) bob to turn on/off the steper motor power c/w manual override
6) Future PWM spindle speed controller

Ken,

since you have J4 free, it might be a good idea to use that to drive the stepper?

Did you check if the BOB can drive two stepper at the same time?

PS: Glad to see you have EMC running, I tried yesterday the whole evening on my laptop but the latency-test doesn't give me good results. EMC people say it's a hardware issue, but I am seriously wondering if MACH3 will also not work then.

Ries

Ries,
2 reasons to leave J4 free,
1) EMC2 doesn't have master/slave drive option as in Mech3. I learn from here that PMDX-122 can source ~35ma and my driver only sink 15ma nominal as stated in the manual, so I guess they one pot for 2 drivers should work.
2) future A-axis? maybe.... :D

I have problem with Live-CD with my notebook as well, but once I disable the power management, the WIFI, the LAN, the sound card, connect an external LCD monitor then it works just fine. Didn't give time to find which works as I'm more interested in dedicated mobo for EMC2. Have a go, should be fun.;)

Guys,
I did more manual digging and found out I made a complete mess out of the earlier layout, I came up with a hopefully "less wrong" schematic for the BOB & drivers

Attachment 7307

I really need help with this one, electronics & computers are really not my forte... so are a lot other stuff :o

Please comment freely and laugh at my mistake, I know I still has a lot to learn.

Thanks in advance.

Cheers

Ken

Ken, I haven't bothered to strip my EMC systems down, because they're fast enough to support the 500 IPM or so rapids on my MM. So I still have the network running, etc.

I could probably reduce the jitter, but it doesn't seem worth it to me. If I were going to do it, I'd remove or disable the network and sound cards, run the display with a 2D, BIOS based driver (a VGA, or frame-buffer mode that doesn't use any of the acceleration features of the GPU), and then shut down any extra background processes (daemons / services) in the system.

My MM is out in a garage shop about 30 feet behind my house. I generally develop my cut files on a computer in the house, and then squirt it down the (wired) network to the MM computer. I then walk out, set up, and cut.

Hey Ken,

could you please post a picture in a higher res. Some parts of it are not readable.
Thanks a lot
Claus

Here is the higher res.
Attachment 7322

Normally the power gets connected to the top 2 points on J8. That bottom point is marked as an output.

Do you seriously want nine limit/home switches?

Sounds like I'm over doing?? I blame it on overly safety conscious as I read of 500ipm travel...
The limit switches are USD0.90 a piece, good Japanese made Omron. When compare to proximity switch (US$15++), they are less accurate but I question the need for extreme accuracy. IMHO, for MM's application, the machine (0,0,0) position is set using zeroing technique for every single job, the sensor zero position is for parking purpose hence even a few mm off the mark won't hurt anything. BUT I can be wrong about this & I'm all ears.
Attachment 7325
I have yet to buy all but 6 mechanical micro switch for now... 9 is for minimum academic consideration, each axis have an end limit at each end of the travel & a zero(parking) position. BUT if its not necessary, the surplus can go elsewhere such as door alarm, mouse traps... etc :D

Thanks for the heads up, I'd edited the power input. Will post a fresh revision once the following are cleared.

1) will the E-Stop & limit switches scheme (J5 & J6) work correct?
2) I am not sure of JP1, JP2, JP3, JP5 & JP6.
3) Is it safer to let the PC to switch on the stepper motor power supply rather then having them power up all the time?
4) Is there anything else I missed out?

Thanks in advance.

You are overdoing the limit switches. For example, how do you detect when a cutter has gone too far in the z-direction? Suggest that you get a better understanding of Limit Switches - optical / mechanical / proximity and see that all switches easily go on one series circuit (similar to e-stop).

Thanks for the lead, I did spend a considerable effort trying to understand that thread but still cloudy. Have to consider 5V (4.2V after the PMDX), debouncing. Therre are schemes to use relays, logics or even micro processor...

This limit switch thingy is holding up progress of my build, since the MM can operate without them, I would have to leave it out until I know what to do.

Now I need to find out more about Zero-plate, apparently this is essential.

The PMDX discussion has been copied into one thread here

I have finally done it! I fried my PMDX-122 & the Mobo!!!

I tried to finish off the control panel this year if I want to get a number below #100.

I wired up the relay & E-stop circuituit + the E-stop, everthing went nicely.
Tested the Stepper motor PSU & got No load ~54Vac on the secondary & ~75Vdc at the capacitor. Loaded with a light bulb & I got ~48Vac and ~65VDC loaded.
then I wired up the tranny to the Power relay circuit & when I pressed the push buttons & the contactor made that familiar "thub!" sound when press the ON button &" tuck!" sound when pressing "OFF" & the E-Stop switch.

So far so good.

Wire up the Mobo to the BOB & BOB to the stepper driver.
It boot up nicely upon power up & suddenly the monitor went blank. I get thermal overload massage upon reboot... & I lost the keyboard & the mouse.... I check the BOB & it was fried even depriving me of a good smoke show or a tiny "bang!"...

Its almost midnight, I just pack the whole box up & stow it away in the store room until I get another BOB.

I'm jinx when it comes to electronics..... :mad:

I will order my steel tomorrow.

Ken,

this is odd...

Why do you think the BoB is fried?

THis could only happen if:
1) the input voltage on the BoB (7VAC or 12VDC) was way to high.
2) The came some voltage from the driver back into the bob

I don't have your drivers specs at hand, but assuming they are opt isolated, it mist be a power supply that was wrong.

Did you measure the voltage on the BoB's power supply?

Ries

Hi Ries,
I replaced the D945GCLF & confirmed that the mobo is partially fried.
Bob did not power up when I supply power to it.
I'm very sure I supplied 12Vdc from the harddisk power connector to the Bob and a DMM measurement confirms it.
I re-powered the Bob & I don't have any LED lit up.
The driver is opto-isolated & it even wasn't powered at the time of test...
Another abnormal is the parallel cable turn very warm.

Anyway, I ordered the steel & just received them.

The main 178x76 C channel are very straight. They have to! as I paid premium (double) for these imported Japanese channels. Judging from what I read from other build, twisted channel can be huge PITA.

I'll set up the angle grinding rig before I could start making some grind sparks soon.

Ken,

This sounds like if some ground current is running somewhere.

how did you ground the BoB and PC combination? Is the BoB and/or PC in any way connected to the metal of the control box? This is currently the only path I can imagine that could be problematic.

Did you try to connect the BoB on 12VDC without the printer cable?

Sorry to here that components are fried...

Ries

Steel Fabrication
 

Hi guys
Just an update on my progress.
I shelved the control box & ordered a chinese BOB rather then sending the PDMX122 for repair, it will be more costly in shipping in my part of the world.

Had been cutting, drilling and tapping all this while,

Here are the steel stock scattered all over the floor,
Attachment 8086

Steel stock cut to size so far,
Attachment 8085

Drilling with the guide bar,
Attachment 8088

Tapped completed
Attachment 8090

Dry fit the table for motivation.
Attachment 8087

Dry fit the Gantry
Attachment 8089

I can find straight flanged angle iron here so I will use T-bar instead. Will cut one of the short flanges with the grinder skid & work from there.
These are the 60x60x7thk T-bar. Looks like I have quite a bit of sanding to do...
Attachment 8091

Progress is slow, but its moving...

yes, it`s takung shape...:D

I'm in the process of cutting the rails.

The design say its 1.1" ~28mm, I end up with 33~34mm on the shorter flange, after I cut one of the short flange on a 60mm x 60mm x 7mm T-bar flange. Can I stick to 31mm?

I am prepared to use 30mm board rather the 25mm MDF to bridge the difference.

You can stick with 31mm. There is no issue with the MDF. The issue is whether your motor plate will swing down low enough so that the pinion gear goes under the rack. For the Americans with 1/2" square rack it is easier, but the rest of us use 16mm square rack. If you have decided that you are using big pinion diameters, you could modify the motor swing plates a bit.