Just Checking In and Catching Up #44 - Delhi India

[quadtech]

Anil, Congrats! Great work, and done really quick too.
So #33 and #44 are in India. Perhaps someone here
will get #55 too :-)

Do post some technical details of your homebrew
driver board and BOB.

[Rad Racer]

Nicely done Anil, congratulations on #44.

[aniljangra]

Thank you Wayne.

Originally Posted by quadtech

Anil, Congrats! Great work, and done really quick too.
So #33 and #44 are in India. Perhaps someone here
will get #55 too :-)

Do post some technical details of your homebrew driver board and BOB.

Thanks, Pieces keep on fitting in as you move along, its a nice design just get started and in no time you will be moving fast

I am doing a controller board for managing all inputs from proxy and switches and redoing the BoB to add an intellegent charge pump (uC based so that it never missfires on noise) and including a PWM to 1-10V for spindle control from mach on same chip.

I'll update as soon as I test them and send the design/code to anyone interested.

[riesvantwisk]

Originally Posted by aniljangra

Thank you Wayne.
intellegent charge pump (uC based so that it never missfires on noise)

Anil,

I was thinking the same, however would not a simple bandpass filter be enough? Then EMC2 or MAch3 just needs to put out a specific frequency.
That said, I am not sure how the current implementation is of the CP on the PMDX122 is, does the flip frequency matter?

I did notice that EMC2 puts out a very low (2-4hz??) frequency for the CP, I am not sure what will happen if that is increased.

Ries

[jhiggins7]

Anil,

Great build and a wonderful job documenting the build. Congratulations on Serial #44, the first Serial # of 2010 just as you had hoped!

The Updated Builder's Log is here.


Please let me know the "cutting" dimensions of your MechMate for the Builder's Log. Also, please review your entry and provide any changes you would like.

[aniljangra]

Thank you, I was just waiting for you

For log cutting dimensions are 2500x1220, Entry look fine, if you can please update the comment section as "Belt reduction per Chopper design, DIY Drives & BoB"

BoB yet to be completed as PCB is not yet ready but right now I'm using a DIY handwired BoB.

Thanks again

[aniljangra]

Originally Posted by riesvantwisk

Anil,

I was thinking the same, however would not a simple bandpass filter be enough? Then EMC2 or MAch3 just needs to put out a specific frequency.
That said, I am not sure how the current implementation is of the CP on the PMDX122 is, does the flip frequency matter?

I did notice that EMC2 puts out a very low (2-4hz??) frequency for the CP, I am not sure what will happen if that is increased.

Ries

Ries, A small 8 pin uC is cheaper than few passive components moreover think of this as a band pass digital filter with added hysteresis. Hysteresis can be smaller for ON and a bit longer for OFF so that a glitch in CP pulse train don't send a stop signal.

For PWM to 0-10v conversion, of course a simple passive network will do and that's exactly I am thinking of doing.

I'll do some testing on both and post the results.

[jhiggins7]

Updates completed, here.

Again, great job, Anil.

[jhiggins7]

Anil,

Like "quadtech",

Originally Posted by quadtech

...Do post some technical details of your homebrew
driver board and BOB.

I'm also interested in additional details of your DIY drivers and BoB. I'm familiar with the AVRSTMD.COM site.

You mentioned Gerbers. Would you be willing to share your Gerbers?

[aniljangra]

John,

Don't know if I should post the files here so I am PMing you the Egale files for the Drives, Change as you like and generate Gerber or anything you want . Fell free to ask if I missed anything. Download free version of Eagle, the PCB is within the limits of free version.

I'll post my version of BoB with CP and PWM as well soon. Also there is a controller board that rides on gantry for taking inputs from proxy and buttons in progress, will be done and shared soon, thanks for the interest.

As the real work is done on weekends so my progress is slow

[aniljangra]

Thought of testing for backlash by milling the contours of pcb tracks.
pcb-test.jpg
Roughness is due to me not using proper bit but a broken drill bit as cutter and high feed rate of 100 ipm, whole thing milled in 34 seconds flat . See for yourself the capability of MM , perfect usable PCB can be produced with proper bit and feed.

May be if I can get a bit I'll do the BoB and cntroller board for MM on MM itself

Note : I still need to put the spoil board and surface it.

[aniljangra]

Can anyone using VFD with 0-10v speed control unit confirm that a input with about 500mv ripple will work on VFD ?

I tested a simple RC network today and its producing DC 0-10V with a ripple of 500mV from PWM signal at 1500Hz.

How far can we increase the PWM freq from Mach ?

[bradm]

Anil, I'm running off a PWM signal and a simple RC network; never bothered to measure the ripple. The first shot used a very small capacitor, and it showed up as a random small speed variance at the spindle (I had the VFD display the spindle freq). I upped the capacitance until the variance went away.

Assuming that your VFD linearly allocates the 0-10v across 0.0-400.0hz ( mapping to 24000 rpm), it's 10v divided into 400 (or 4000) levels, so 25mv (or 2.5mv) per level, with each level representing 60rpm (or 6 rpm). 500mv ripple could represent up to 1200rpm or 20hz variance, assuming that the VFDs sampling of the control signal was worst case timing with no hysteresis. Real world variance would likely be less; mine was about 3hz before I fixed it.

I suspect you'll want to increase your capacitance a bit, but testing will show you what you need to do. The down side of too much capacitance would be a VFD that was slow to respond to control signal changes.

[aniljangra]

Thanks Brad, I guess there is bit of hysteresis or averaging or both built into VFDs so don't want to increase cap too much slowing the response time.

That's exactly the reason I asked for how much high Mach can reliably put out PWM freq (assuming 25Khz kernel freq).

Whats your PWM base freq set at ?

[bradm]

I'm using EMC2, not Mach3, and the setting I'm using is 50hz (I just triple checked this, I'm using the pwmgen module for EMC2). Practically speaking, the physical limits of spindle accel and decel are far greater than even lots of capacitance here, as long as you don't use a spare power supply filter cap . 1500hz should be more than enough.

[aniljangra]

Thanks Brad for confirming the freq, you are right about the response time in practical world I'll increase the cap value to get the ripple down. Bad news is my DSO died suddenly for no reason , I suspect the grinding dust from rail grinding got in and shorted something inside.

[Gerald D]

I can't remember what a DSO is?

[aniljangra]

Sorry Gerald for the confusion, "Digital Storage Oscilloscope" is not a part of MM, thats the reason you don't remember

[quadtech]

Hope the scope issue turns out to be something
simple like a fuse.

Can you get the LMD18245T locally?

[aniljangra]

Quad, I got LMD18245 from farnell (http://in.farnell.com), you need 2 of them for each drive. Micro-controller is also available on same site.

[aniljangra]

Kobus asked if its possible to do isolation milling on MM, I did trail and here is the proof that its surely possible.
pcb test.jpg

Note that I am using broken drill for milling, so the isolations are wide and the actual track/pad width looks small. But with actual bit the output is a usable PCB as all tracks are intact

By the way this is the actual PCB for MM to combine all inputs.

[Kobus_Joubert]

And with Vectric V-Carve pro I can do the PECK Drilling of the holes at the same time.

[aniljangra]

Kobus, Eagle generates a NC file for drilling as well. I didn't bother to complete the milling and do drilling as its just a test, for you

[jhiggins7]

Anil,,

Your PCB looks good...even with the limitations you talked about. Thanks for showing us more of the capabilities of the MechMate.

I know you may be busy with getting your spindle working and that you have a "day job" to attend to, so this discussion may need to be delayed...BUT, I'm very interested in your experience with the Open Circuit AVRSTMD drivers.

First, could you confirm the specifications of the stepper motors you are using? I believe you are using the 3.5 amp, 4.5 volt, 450 oz-inch HOLD, Superior Electrics for the X and Y and a 300 oz-inch for the Z. Is this correct? What current setting are you using on the AVRSTMD? Also, what voltage are you running the stepper motors at?

Next, I'm wondering about mid-band resonance. I don't believe the AVRSTMD driver design has mid-band resonance compensation. On the other-hand, ALL of the Geckodrive stepper motor drives have mid-band resonance compensation. So, do you notice any problems with your MechMate that would be the result of mid-band resonance issues? I'm no expert, but I believe the effects would be strange noises and even stalls or lost steps.

Finally, I'm wondering about heat. The AVRSTMD uses some significant heat sinks. It appears that your drivers are mounted on metal stand-offs directly to the back of the enclosure. Do you notice any heat build-up inside your enclosure? Did you vent your enclosure? Do you use any circulation fans? I don't see any vents or fans in the pictures of your enclosure.

[aniljangra]

Thanks and first of all let me make it clear that I wanted to avoid Geckos because of cost (my whole set of drives cost me about the cost of one 203 ) and believed that if motors are chosen and connected correctly 3Amp limit of drivers chips (LMD18245) is enough for MM.

I am running chips at 2.5 Amps with single coil on x and y and series connected coils on z (because z motor is 4.6 amps).

You won't believe but only Y motor is 450oz-in and all other motors are approx 300 oz-in and I am comfortable at running any axis rapid at 500ipm without loosing steps (of course you can't cut at 500ipm anyway).

I have not noticed any mid-band problem till now, I guess motors accelerate fast enough through the region. May be I need to run slow enough to notice the oscillations. (whats the accel you are using ?)

I've multiple taps on transformer so that I can get 45, 50, 60 volts, I tested drives on 45 and 50 volts.

Though I use large heat sinks on my drives (because I already had many of those laying with me), I think smaller will do, after-all its a DMOS device unlike 298 which generates lot of heat.

I don't have a fan but I have ventilation holes right above the drives, not visible in pics because those are covered by keyboard sitting above them

Computer CPU is generating more heat than these drives , Its winter season so I am not installing the fan right now, may when its hot I'll add one.

I guess you already know, PICstep is another similar option available for drives using LMD18245.

[jhiggins7]

Anil,

Thanks for taking the time to respond to my questions. I appreciate the thoroughness of your response.

I'm planning to build a second MechMate. So, I'm wondering about using the AVRSTMD drives. I think I can build them for $20-30 each.

Keep up the great work! And please continue to share your results with us.

Two more questions. Have you "sensed" the temperature of the drives during operation of the MechMate (like touch them or nearly touch them or measure them with one of these)?
Same question for the motors?

[aniljangra]

I don't have device for temp measurement, but I can touch them.

Motors are only slightly warm.

I can say that its worth to give it a try as you can build 4-5 of them for the cost of one Gecko and in case you feel you need to upgrade, you can use them for Y,Z,A etc and just but Gecko for X.

I'll do a temp measurement circuit as I get time and do the measurements on drives, inside box and motors.

Also again, do look at picstep as well.

[jhiggins7]

Thanks again, Anil. I'll be interested in your observations once you get your spindle working and start using your MechMate for long cutting sessions.

For temperature monitoring, I purchased a temperature probe like this from Ebay and mounted it in the front panel of my enclosure. I attached the probe end to the driver heat sink.

I'm going to build at least one AVRSTMD. and check it out.

I've programmed both PIC and AVR. I find AVR simpler to program, so I'll go with the AVRSTMD for now.

I also notice that the Geckodrive G250 (which is the module used in the G540) is only $61 each and it can handle 3.5 amps and 50 volts. It also has the mid-range resonance compensation that I mentioned in an earlier post.

Keep up the great work Anil. Your build is an inspiration.

[aniljangra]

Thanks you John and good luck for you next MM.
Installed the spoil board and tried to set gantry at 90 deg to X, not yet done but will do next week as its I think its off more than that can be fixed by shimming the gantry wheels. Need to adjust the Y rails first on gantry tubes a bit before shimming the gantry wheels.

For now before surfacing the spoilboard, wanted to spoil it a bit so here is the cartoon file again

spoil.jpg

[aniljangra]

My first attempt at 3D on MM ..
first-3d.jpg

Sorry for the bad pictures I always post, my handycam don't do better than this on stills , Next time may be I'll try video.