[Mariss Freimanis]

I just realized this concludes the G250/G251/G540 project from a hot-tub inspired idea back in January to a full-scale production ready design in July. I has taken 6 months to the day; January 23 (first post) today. Eerie coincidence.:-) I was driving home from work this evening and by habit I started thinking about what needed doing next on this project when I suddenly realized the "to do" list was empty. Nothing was left; it's all done.

I have learned a lot of things in these 6 months.

I learned optimizing a design for minimum cost takes more effort and cleverness than I originally estimated by a long shot. What I arrogantly underestimated and allowed weeks for instead took months. If poetry is saying something in few words that a novelist says in a thousand, then I learned a poet sweats over a single word every bit as much as a novelist sweats over a paragraph.

I learned going down just one scale in component size (0603 vs. 0805) requires double the quality of printed circuit boards and double the control over the production process used to assemble them. Good things came from this; learning what was required transfered back to improving our 0805 component process. A lot of reading, learning and experiments (in a word, research) went into understanding. Understanding trumps cut-and-try every day of the week.

Another thing I learned was how to integrate SMT production techniques with traditional through-hole methods. I believed the two were incompatible separates and never gave a thought about combining them. SMT parts were for pick-and-place machines, through hole parts were for people wielding soldering irons or wave soldering. The economic imperative to eliminate manual labor tasks required this portion of the process to be investigated. The result was through hole reflow where everything gets soldered at the same time, SMT and through hole parts, in a single pass through the conveyor oven. It works so well and saves so much that all of our other drives will be tailored for this process.

Finally, I learned not to include business considerations when sharing the genesis of a project from idea to production. I was as open with that as I was with the technical stuff and frankly in retrospect, it didn't belong. The reaction frankly took a lot of joy, excitement and happiness out of what I love about what I do. Everyone eats sausages, not everyone should go to a sausage factory to see them made. Omitting the biz considerations would have saved people a trip to the sausage factory. Everyone would have been better off, myself included.

Sorry for the ramble.

Mariss

[tauntdesigns]

I'd like to thank you for the thread,

I loved the whole thread - minus a few posts.

Human being, Being human, Ramble On!
Thanks again,
Jack

[yngndrw]

I'd like to thank you for the thread,

I loved the whole thread - minus a few posts.

Human being, Being human, Ramble On!
Thanks again,
Jack

I have to agree, I for one have learnt a lot about a range of subjects from this thread.

I'd like to thank you Mariss, as well as the whole Gecko team for showing us "A 6 months in the life of the Gecko Drive team".

[Mariss Freimanis]

yngndrw,

The BAT54S is a dual series connected Schottky diode. I design for surface-mount so I tend to pick surface-mount components. The BAT54S offers me the convenience of having two diodes in a single, small SOT-23 package. You can always bust-out combined parts such as the BAT54S into their discrete through-hole equivalents. Schottky diodes are an overkill in a 12VDC circuit, their 0.2V forward drop in a 5VDC circuit is almost a requirement when compared to 0.6V Si diodes. The circuit is somewhat universal in that it will work with 3.3V logic if need be. I'm lazy; If at all possible I try to design circuits I know will work with 12V logic as well as 3.3V logic.

Mariss

[Scott McCray]

Mariss,

Thank you for sharing this amazing journey with us...please do it again soon!

I understand the need to leave the business/financial aspect of things (the sausage factory) out of the mix in the future, but this has been a rare insight into a high quality company's product development cycle.

I feel privileged and honored that you shared it with us.

Scott

[joeybagadonuts]

Ok now that we all heard about the Kiełbasa,
WHERE IS THE BEEF?

JoeyB

[Mariss Freimanis]

Ok now that we all heard about the Kiełbasa,
WHERE IS THE BEEF?

JoeyB

Honestly JoeyB, you are one of the people that took some of the joy out of this. Not criticizing you but just want you to know it nonetheless. Please have a good day.

Mariss

[joeybagadonuts]

No Hard feelings,

Still going to buy a dozen G251's to go with my coffee.

JoeyB

[yngndrw]

yngndrw,

The BAT54S is a dual series connected Schottky diode. I design for surface-mount so I tend to pick surface-mount components. The BAT54S offers me the convenience of having two diodes in a single, small SOT-23 package. You can always bust-out combined parts such as the BAT54S into their discrete through-hole equivalents. Schottky diodes are an overkill in a 12VDC circuit, their 0.2V forward drop in a 5VDC circuit is almost a requirement when compared to 0.6V Si diodes. The circuit is somewhat universal in that it will work with 3.3V logic if need be. I'm lazy; If at all possible I try to design circuits I know will work with 12V logic as well as 3.3V logic.

Mariss

Ah I see, thanks for clearing that up.

[wolfdagon]

Mariss, I have also enjoyed this thread, aside from the obvious. I have learned much and hope that you will do similar threads on future products. Thanks for the time that you put into all of this.

Kenith

[phomann]

If anyone can fill me in on how Mach3 handles VFD drives ( if PWM, what frequency, which pin preferred, etc) and maybe the next revision of the G540 can have an optoisolated VFD output as well.

Mariss

Mach generates the PWM as follows.

The resolution of the PWM mark/space is dependant on the kernel speed for Mach3 and the PWM period the use sets. For instance, if you use a kernel speed of 45Kz and you want a PWM frequency of 50Kz then you get;

45000/50 = 900 steps from 0 to 100% of the max output voltage. To apply this for a spindle, if you max speed of say 4000rpm then you end up with a speed resolution of 4000/900 = 4.44 rpm.

If you use a kernel speed of 25K, then the speed resolution is 8rpm

I'd design it to work with a PWM frequency of 50Hz. If you go much higher, a minority of people will complain that the resolution is not adequate.

Mach can also now have a kernel speed of 60kHz and I think 100KHz but I wouldn't design your converter to rely on it. I'd suggest the 25 - 45 KHz is what the majority of uses use.

Cheers,

Peter.

[CROSSHATCH]

Uh I'm just so excited for the 540 release! I need my machine up and running!

Good thread Mariss with much insight!

-Jason

[yngndrw]

Mach generates the PWM as follows.

The resolution of the PWM mark/space is dependant on the kernel speed for Mach3 and the PWM period the use sets. For instance, if you use a kernel speed of 45Kz and you want a PWM frequency of 50Kz then you get;

45000/50 = 900 steps from 0 to 100% of the max output voltage. To apply this for a spindle, if you max speed of say 4000rpm then you end up with a speed resolution of 4000/900 = 4.44 rpm.

If you use a kernel speed of 25K, then the speed resolution is 8rpm

I'd design it to work with a PWM frequency of 50Hz. If you go much higher, a minority of people will complain that the resolution is not adequate.

Mach can also now have a kernel speed of 60kHz and I think 100KHz but I wouldn't design your converter to rely on it. I'd suggest the 25 - 45 KHz is what the majority of uses use.

Cheers,

Peter.

Question, why would you want such a high PWM frequency ? By using a lower frequency you will have more accuracy per unit of kernel speed. 50KHz seems a bit high.

[kawazuki]

Just a quickie - 540 release sometime in August? Will you announce here and on your website or just here initally? You'll get my order anyhow, just want to know were to keep looking:cheers:
Hope the postage to the UK ain't gonna hurt too much

[phomann]

Question, why would you want such a high PWM frequency ? By using a lower frequency you will have more accuracy per unit of kernel speed. 50KHz seems a bit high.

A higher PWM frequency provides for a better response to changes in the PWM value. This can be important in getting the Mach3 closed loop spindle control to work.

Also a higher frequency means that smaller capacitors (cheaper) can be used in the filter circuit.

Cheers,

Peter.

[yngndrw]

A higher PWM frequency provides for a better response to changes in the PWM value. This can be important in getting the Mach3 closed loop spindle control to work.

Also a higher frequency means that smaller capacitors (cheaper) can be used in the filter circuit.

Cheers,

Peter.

Ah I see, cheers for that.

[Torchhead]

First kudos to the Gecko design team. Their behind the scenes assistance and support have been exceptional! Thanks again for Mariss and his love of engineering (which I share) and for his willingness to share knowledge. While I tend to be more introverted and do my designs behind closed doors, I always watch with interest his narrative of "Engineering Challenges". It is, for us Engineers, the series of challenges and technical puzzles that becomes the drug of addiction.

For our G250 based design we decided to take a slightly different direction. We approached it as a system design rather than a component. By that I mean that we wanted to offer a total solution for builders that addressed the finer issues of constructing a full featured machine.

We started with our own design for 1 to 5 Gecko 250 drives. We took our expertise with the larger ESP Power Controllers (they use G203 and G320), and the modular block concept, and came up with a Motherboard/Blade concept. With input (and encouragement) from Mariss and Marcus along the way we came up with the EZPlugG250-5. That then became a main component of the BladeRunner series that integrates a unique new single port I/O card (UBOB), logic power supplies, diagnostic LED's, and finished enclosure. Our various packages offer different configurations and levels of features. The EZPlugG250-5 Motherboard has it's own 40Mhz processor that monitors and measures all critical systems parameters including each Axis current draw (load) and the Motor DC volts. We have a temp probe (reads out in degs on the MACH screen) that sits between the X and Y blades and auto shutdown on over temp. All blades have opto isolation for Step & Dir.

The BladeRunner is part of our modular series and can be expanded up to 5 full axis along with other feature modules to be offered in the future. First plug-in module will be a low cost precision isolated Spindle Speed (0-10VDC analog out). It's due in September.

We think the Gecko G540 is a good fit for the roll-ur-own controller crowd. It has most of the base features you need to do a 4 axis CNC build. For those less electronically inclined we offer a more plug-n-run approach with MACH profiles, custom screens and complete systems.

We do plan on making the EZplugG250-5 sub-system available as a separate product line this Fall. The BladeRunner is due to start shipping Aug 15th.

For more info and photos:

http://www.CandCNC.com
http://www.CandCNC.com/Bladerunner.htm
http://www.CandCNC.com/EZPlug250-5.htm



TOM CAUDLE
www.CandCNC.com

[Mariss Freimanis]

OK, the PWM frequency is 50Hz. Is my understanding correct that VFD drives allow for remote potentiometers and typically supply 10VDC at a few mA? If so, I can cobble-up a 3-pole Butterworth low-pass filter. A cutoff frequency of 5Hz gives 60db attenuation at the 50Hz PWM carrier frequency. That should be plenty.

By the way. I use a terrific free application called Filter Pro from Texas Instruments for active filter design. Go to www.ti.com and tap in "Filter Pro" in the search box. That will take you to the download page.

Mariss

[Mariss Freimanis]

Tom,

Very nice work!

Mariss

[Torchhead]

OK, the PWM frequency is 50Hz. Is my understanding correct that VFD drives allow for remote potentiometers and typically supply 10VDC at a few mA? If so, I can cobble-up a 3-pole Butterworth low-pass filter. A cutoff frequency of 5Hz gives 60db attenuation at the 50Hz PWM carrier frequency. That should be plenty.

By the way. I use a terrific free application called Filter Pro from Texas Instruments for active filter design. Go to www.ti.com and tap in "Filter Pro" in the search box. That will take you to the download page.

Mariss

Some (not all) VFD's need the 0-10VDC to be isolated from the PC logic side. If nothing else but for noise, it's a good idea. The challenge then becomes getting across the isolation barrier in a linear format. Easy to do with PWM, optos and the filter but linearity and DC offset can be an issue.