[Mariss Freimanis]

Same thing, pdf this time. Thumbnailing a fine .gif seems to turn it into a totally crap .jpg which is unreadable.

Mariss

[bobchiloquin]

Mariss,

Looking at your schematic, I have a couple of questions. How does the +15v to+50v get to the +12v regulator circuit, and pins 5-8 on the G250 sockets? On the G250 pcb, is there a ground connection between pins 1-4 & pin 30? If not how does pin 30 on the G250 sockets connect to ground? Where does the voltage for pin 26 on the G250 connectors come from?

Bob Hayes

[phomann]

Mariss,

Looking at your schematic, I have a couple of questions. How does the +15v to+50v get to the +12v regulator circuit, and pins 5-8 on the G250 sockets? On the G250 pcb, is there a ground connection between pins 1-4 & pin 30? If not how does pin 30 on the G250 sockets connect to ground? Where does the voltage for pin 26 on the G250 connectors come from?

Bob Hayes

Bob,

From the 50V terminal go through the 10A fuse, then take a left turn, turn right at the 10mOhm current sense resistor. Go through the resistor, then do a left hand turn at the T junction. Fly over two intersections and do a right hand turn at the 4 way junction. Continue on until you get to a T junction. Turn right and follow it around until you arrive at the drain of the IRF540N FET.

It is a bit difficult to see where crossing lines intersect (small dot) and where they pass over (small break in the line).

Cheers,

Peter.

[Mariss Freimanis]

1) I should have shown the 15 to 50V bus using a thicker line. It's there, just trace back the line that enters the G250 socket nearest the letter "k" in "socket".

2) Pin 30 on the G250 connects internally to pins 1 thru 4. The difference is pin 30 is a "quiet" ground. It only makes a Kelvin connection to the "noisy" pins 1 thru 4 ground and it is reserved for use with the HCPL-2531 optos. This insures there are no noise issues with step and direction.

3) Each G250 has a 1.8V, 3.3V and 12V regulator onboard. They generate these voltages straight from the +15 to 50VDC bus to power their internal circuitry. The 12VDC is brought out only to back-bias the HCPL-2531 photodiodes.

Mariss

[bobchiloquin]

Peter,

After I increased the display to 200%, I could see the connection dots.

Mariss,

Thanks for the explaination. When I was working for a living, all the schematics I looked at were drawn so two wires crossing and making a connection were drawn as two "T" connections. That way no matter what size the drawing is, it's easy to see where a connection is. Makes troubleshooting 5v, 105A power supplies a whole lot easier.

Bob Hayes

[Mariss Freimanis]

Many conventions in that regards. The one you mentioned, having "U" shaped crossings, A space like I use and even nothing at all. What always marks a connection is the "dot" between lines. Finally, nothing works like reading a schematic. You see what makes sense and what doesn't.

Mariss

[dmauch]

It would be nice if you have room on your motherboard to add a pulse generator using something like a 555 tmer and a few other parts and a 3 pin header so that the ownmer could use the G540 pulse generator as a self test and more importantly as a power feed for those not into cnc but want 3 axis power feed. I visualize a USER provided selector switch, a pot to adjust the speed and a switch to change directions.
Dan Mauch

[eman5oh]

You could just make a box with the 555 pot an d switches that has a db 25 cord on it and connect it to the G540, kind of like a pendant.

It would be nice if you have room on your motherboard to add a pulse generator using something like a 555 tmer and a few other parts and a 3 pin header so that the ownmer could use the G540 pulse generator as a self test and more importantly as a power feed for those not into cnc but want 3 axis power feed. I visualize a USER provided selector switch, a pot to adjust the speed and a switch to change directions.
Dan Mauch

[dmauch]

I already thought of that but figured Mariss would be clever enought to find about .5 sq in and add it in. Would probably cost less than .50 in parts and it would reduce customer service by being able to generate a pulse plus I am enconomical on my expended labor and thin it would be better on the motherboard.. Sure there will still need to be the external switch and pot.
Dan Mauch

You could just make a box with the 555 pot an d switches that has a db 25 cord on it and connect it to the G540, kind of like a pendant.

[Mariss Freimanis]

The G250 is alive! I got it running for the first time at 8:15AM this morning so I've had 2 hours to run some gentle tests considering it's the only working example there is.

Maximum Tested Speed: >250,000 microsteps per second (7,500 RPM).
Maximum Tested Motor Current: 2.97 Amps per phase.
Tested Voltage Range: 13.5VDC to 55VDC
Midband Resonance Compensation: Tested and it works.
Adjust Trimpot: Scaled correctly for high L, low VDC to low L, high VDC range.


Comments:

What has surprises me is how cool the MOSFETs stay. They are tiny TO-251 (I-PACK) devices 0.25" by 0.25" in size. The above tests were run without any heatsinking of the MOSFETs at all; just the bare devices sitting on the board. At 3A and 55VDC, I estimated a case temperature of <50C using my temperature calibrated fingertip. It's probably dissipating less than 500mW to have that. Astounding, given what I expected so I may raise the max phase current to 3.5A.

Otherwise, it performs just like a G201 set at 3A and run with 50VDC. Totally silent at low speeds and no perceptible motor vibration. Please see the attached motor phase current for how nice the current looks.:-)


Mariss

[pminmo]

Excellent, so what do you attribute the motor quietness?

[Mariss Freimanis]

Like the G201 from which it borrows heavily, the G250 is not a chopper but rather a synchronous PWM drive. That means there is a single 20kHz clock that is the timebase for both motor winding current circuits. Because of this common clock, the circuit is fully synchronous and cannot produce any audible beat frequencies.

A chopper (L297, Allegro, et all) has two free-running 20 - 25kHz oscillators. These oscillators phase-lock and break phase-lock hundreds to few thousand times a second. This "make, break" phenomena produces beat frequencies that fall smack-dab into the audible range producing the grunting, hissing, squealing and whistling everyone is so fond of.:-)

A synchronous PWM switching topology is more complex than a chopper which needs minimal gating logic only a one-shot (mono-stable multivibrator) per winding driver to function. What comes with that simplicity are some irritations.

Mariss

[klmjr]

Did you need any mods to the board to make it work? Or was it plug and go at the first revision? Fewer mods means we get new toys sooner I would hope.
klmjr

[Mariss Freimanis]

klmjr,

For a clean-sheet design prototype finally tortured into running, the G250 looks remarkably good after the experience; better than most prototypes actually. A few Exacto knife scars and jumper wires but nothing that won't easily heal.:-)

Here's how the process of starting up the results a new design works after the board is populated:

1) Fire up the board very gradually and check that all the voltage regulators work properly. Log supply current and regulated voltages against an ever increasing input supply voltage. Hope nothing gets hot, keep your hand near the lab supply output banana plug cable. Jerk the plug out even if you just imagine something is not right. You lose no points because no one saw you. Remind yourself it took hours under a microscope to populate the drive; you don't want to repeat that ordeal.

2) JTAG the onboard CPLD with the first cut of the configuration program. Tickle the inputs, watch what the outputs do. If the waveforms are wrong, edit the source code in Verilog, generate a new configuration file, JTAG the CPLD again. Repeat many times, assure yourself the many iterations converge to a solution. Pray that each change doesn't increase the macrocell, P-terms and registers used count.

3) Connect the smallest motor you have, a 1A, 4V NEMA-17. Set the lab supply to the minimum voltage the drive is expected to run at. Set the lab supply current limit to the minimum current you expect for that motor. Tell yourself 13.5VDC and 200mA is not a lot of Watts. The parts on the board are mighty small though.

4) Part (3) involves visiting part (2) a few more times. It also involves sitting down with the schematic to see how come you have to visit part (2) again. Same prayers apply.

5) Cut traces, solder-on jumper wires, change component values and reprogram the CPLD until the drive shows life. Every change you make gets documented on a "ding" sheet. This sheet is diary of all the mistakes and changes made to get the drive to run. It will be invaluable in bringing the pcb layout, schematic and source code into synchronization with the eventually running prototype.

6) It's alive! Celebrate for 5 minutes, then start testing the drive very gradually to it's design limits. Supply voltage first, then rated current followed by both simultaneously works for me. Document everything along the way; if it blows, documentation will be the only forensic info you have. Steps (2) through (5) may get revisited again but usually not as often as before.

7) It's time for pictures. It works and you have learned all the vital information and documented it. Now is a good time to update the schematic and pcb layout; the source code got taken care of on every update. Comment the changes in the code and why they were made.

8) The prototype drive is now expendable. If it blows, nothing vital gets lost. Time to test it beyond its ratings to see where it ultimately fails. Doing this is important because it gives hard proof of the safety margins of the maximum ratings. Document everything because it may be a one-shot deal now. This is done in as non-destructive way as possible but damage is almost certain to occur. Repair the drive until it becomes too chewed up to endure another repair. Throw it away afterwards or save it if you liked how it held up.


I'm nearly finished with (7) and (8) starts next Monday. The idea for the G250 originated Jan 23 while soaking in a hot-tub and here it is Mar 6, about 6-weeks later. Fastest idea to working drive design in my history.

What's coming up: Step (7) has taken all week. Step (8) will take most or all of next depending on the G250's will to live.:-) Next Friday the pcb gerber files get let for fabrication of the first 2,000 drives. The turn time is 4-weeks for boards. Allowing 1 week for slipping them into the SMT production line here, figure on late April for availability in quantity. There will be only the single prototype until then and I'm keeping it. Then we'll see what happens.

Mariss

[.xXACEXx.]

mariss...
g250 ......in the event it does work flawlessly....do you ever set back and say.."it was so simple i dont know why i never thought of it before" it still amazes me that the g250 will handle all that voltage and amps...for something so small...i keep trying to pull one of those g250's off of my screen ....gj i say! oh to get to push one of those to its breaking point...always wanted to be a superbike tester...."here go out and run the living heck outta this gsxr 1000 ,and come back after you have destroyed it,and let us know what it needs done to it..." that,.. like pushing any thing to its limits,would be a kewl job...

[Monte]

wow. I'm amazed by the stuff put out by Geckodrives. I haven't been playing with my mill for over a year, I come back to try again to get it running and check out Gecko and see more than one new driver. Cool.

This thread is titled "Cheap drives", the first post had this list in it:

1) 3A per phase, 50VDC max. Discrete all n-channel MOSFETs.
2) 10 microsteps per step, CPLD design.
3) No midband resonance compensation, no morphing.
4) No trimpot crossover adjust.
5) Audibly silent design (not a chopper, same PWM as a G203V).
6) No plate, no can, no nice connectors. Just a board and header.
7) No opto-isolator.
8) No heatsinking needed. Everything is surface-mount.
9) No protection circuitry.
10) Target price: $29 single quantity.

Now on page 25 or so there is this list:

1) I have added trimpot "Smoothness Adjust" access holes, a "Power" LED and a "Fault" LED to the 4-axis unit. The miscellaneous I/O connector is back to being the same removable terminal block we use on our other drives. It is now located in what would be the interior of the control box if the unit is panel mounted as expected. Pictures later today or tomorrow. As you can see, I too haven't met my "personal goals" this week.:-)

2) The Cheap Drive is now called a G250, the 4-axis unit is called a G540. Something was needed for the pcb silk-screened part numbers so they are it.

3) We sent the G250 and G540 board files in for fabrication Friday afternoon and we should have the boards a week from this coming Monday. We had a sit-down meeting Friday evening where we ran the numbers; the G250 should come in at $30 to $36, the G540 at $250 to $300. This is as close as I can nail it down until we churn out a thousand or so and then know the real numbers.

4) G250:
0 to 3A, 15VDC to 50VDC rated
10-microstep resolution
mid-band resonance compensated
smoothness adjust trimpot
dual full-bridge heatsinkable all n-channel TO-251AA power MOSFETs
2 row, 15 position 0.1" (2.54mm) pin header connector
1.3" by 1.6" by 0.5" size (33mm by 41mm by 13mm)
4-layer PCB, solder-masked, silk-screened

5)G540
contains 4 heatsinked G250s
short-circuit, motor disconnect, over/under voltage, reversed polarity, thermal protect.
4 DB9 motor cable connectors
DB25 parallel port connector ported for Mach3
12 position 5mm removable terminal block misc I/O and power
Power and Fault indicator LEDs
anodized aluminum enclosure
internal socketed fuse
2.4" by 5.7" panel mount cut-out needed, 1" deep (61mm by 145mm by 26mm)
1.5" by 6" 4 mounting screw pattern, up to #6 screw size (38mm by 152mm)
every parallel port I/O pin is optoisolated
every output is 100mA 48VDC rated
every input is filtered, pulled up to 5VDC and takes a SPST switch to GND to operate

It looks like things might have changed a little, for the good(?). I'm always a little awe struck by the stuff put out by people with knowledge of electronics.

Thanks Mariss and Co., this is cool.

Monte

[Mariss Freimanis]

monte,

Thank you. I am one very lucky and grateful guy. Lucky because I found a niche in life that truly fits me like shoe; I cannot imagine doing anything else that would give such satisfaction and pleasure. Grateful because I have had the opportunity to perfect my skills to do something well in life. I am also grateful I have the means and the good people that work with me who can put these designs into production and turn out many, many copies with care and attention to workmanship and provide the support products like these require. But especially I'm thankful for my loving wife, best friend and partner in life who always believed in me back when there was little material to show for all that faith.

When it comes to the G250, go with the latest specifications, not the earlier ones. The product has evolved during the course of this thread.

A lot of new stuff is lined up behind the G250 / G540. Once the G250 goes to pcb fabrication, I have some bare G380 prototype boards needing to be populated, tested and manufactured. This should begin as early as the third week in this month. Expect a similar rate of progress then on as with the G250 now.

After that immediately is the G213V, a G203V with an integral 4-quadrant unfoolable pulse multiplier. The pcb layout is finished; it just needs to pass design check before being let for prototype boards. It is where the G250 was 3 weeks ago.

Finally, that will clear the decks for the rest of the year for what I believe will be the most significant design of all, a true PID step motor servodrive. This will trump everything Geckodrive has done since we started.

After that, who knows. I may want to veg-out in Hawaii for 6 months in 2009. Recharge my batteries, take up oil-painting, come up with something new or not; I have no clue except to say once these new drives (G250, G540, G380, G213) and the step servo is finished there is nothing warming up in the bullpen. After this year my tank is on "E" until it gets refilled. Either way it will be fun.

Mariss

[Mariss Freimanis]

I ran more tests this evening. The prototype G250 avalanches at 68.6VDC (destructive breakdown voltage), the rating is 50VDC so there's plenty of margin. The MOSFETs start getting hot to the touch at 4A and not hot at all at 3A so it looks like 3.5A is a good max rating with a very modest heasink plate. The MOSFETs are 16A continuous rated with an infinite heatsink so there is a lot of margin there as well. Finally, I spun-up a NEMA-23 3A motor to 10,000 RPM at 48VDC (333,000 microsteps per second). There should be no problem with max speed either. Mid-band resonance compensation is very effective, suppressing resonance with excellent damping.

This is turning out to be a very good little drive; no bad habits at all so far. I'm beginning to like it a lot.:-)

Mariss

[klmjr]

Mariss,
We are starting to like them a lot also and we don't even have one to play with yet. I have a old t-tech base i want to mount a couple of the G250's right into the base, about 1/2 inch thick aluminum. That should do fine for heat sinking. And they are so SMALL wont be in the way of any of the moving parts. Great job. Ive been lurking since you started the geckodrive group on yahoo, have always liked your posts. And your drives.
klmjr

[pfeist]

I'm getting re-excited about a long term project of mine; CNC my Enco Mill-Drill...

I have these motors; "Step motor 680 oz in 2A 3.4. ohms" at home..

I'm looking at this power supply;
"KL-350-48 48V/7.3A 110V/230V"
From http://www.kelinginc.net/SwitchingPowerSupply.html


I'm bordering on an electrical idiot... (I can USUALLY get the right wires connected to each other, as long as I have a diagram and ask a few stupid questions...). So here's my dumb question...


My question is not "how do I build this stuff into a controller/driver for my Mill-Drill"... but rather;
"In my long parts-collection phase, would I be on the right track if I bought that power supply and a Gecko G540 later on when they're avaialble?".

Questions about the details will be months down the line... just want to make sure I'm collecting the right parts in my "dollar at a time" project planning.


Paul F.