[MechanoMan]

Well, still not able to get the mill I got from DeepGroove1 to work all that well. Prone to stalls. Dampers do nothing. At best, this thing "limps", and tends to stall unpredictably. The only sure safe speed seems to be like a maddeningly slow 10ipm.

Being an electrical engineer, I had to dig into what this setup actually is. I found this out.

Steppers: 3.3v 2.8A "280 oz" 1.8 deg NEMA23 double-stack. No info on what make or model it is and DeepGroove1 will not say. The motors provided are unlabeled, the 280oz came from his eBay description and 3.3v 2.8A comes from a pic on his website.

Power supply: 24v 4.5A. I know for sure there is no current problem, putting it on a scope there is no drop in voltage that indicates a shortage of current and IIRC I set them all to jog at once. I'm not surprised here since the way these drivers work is a buck configuration so average input current is much less than the average output current.

Driver: Said to be "Allegro". Someone said they thought it was a Xylotex inside but I don't think so. It says "DeepGroove1" on the board's silkscreen and I doubt he had Xylotex make a special mfg run here. So it's a locally grown board. There are heatsinks glued to the chips so no telling what Allegro chip it is (they make many) and DG1 won't say. There are no external transistors. This IS a microstepper, it's 32,000 steps/in so with a 20tpi leadscrew that's 8 microsteps per full motor step. The fans don't seem to be necessary, Allegro wouldn't have rated a chip below above what it could drive without heatsinking and it doesn't seem like 1.2A would be "pushing" the ratings (there are 750mA,1.2A,and 2.5A Allegro driver chips, bet this was a 1.2A). In any case Allegros HAVE a thermal pad on bottom for cooling and it's kinda funky to try to cool them through the top through the case.

Putting a shunt resistor in series with the motor and scoping it, this thing puts out ~1.2A to a winding. MAX. The motors can be left on indefinitely and do not warm up. The current is extremely noisy. The winding's PWM freq is ~38KHz.

So I think this setup is wrong for several reasons:
1. The supply voltage is far too low. The motor's coil current will rise much faster with a higher source voltage so it can successfully step faster with a higher voltage source. Well electrically that is certainly possible, dI/dT=V/inductance so that makes sense, and some people DO seem to say this can be a limiting factor below 50v or so. Also they say that the bigger motors have higher inductance as well as higher target current and thus require more voltage. Well, these are supposed to be 280oz which is on the "big" side. Many Allegro chips have a low Maximum Voltage and that may have been the reason there.

2. The motors aren't being driven with the right current at all. Not even half of what they're supposed to have. Since all I'm going on is DG1's picture, could the motor current specification be wrong? Could that have been oddly stated, that is was 2.8A as a sum of the two winding currents? NO. No because the motors never warm up at all. Motor current is supposed to be derived as the maximum that won't overheat the motor in normal service. So it's gonna get real warm at least when run at that current. These don't.
This low current choice probably came from a limitation with the choice of Allegro chips, most of the ones with the internal power output stage have relatively low current capability.

3. The PWM freq of the driver may be suspect. It's kinda low. I don't know enough about this driver, but I have a theory that there may be an interaction here (like a resonance, well electrically this is not "resonance", but call it whatever you want) as the step freq lines up with the voltage pulses to the motor.

It's probably a swell driver for smaller steppers. I think the whole problem here is it's a terrible choice for the bigger motor on the mill it's sold with.

[jalessi]

MechanoMan,

You did not mention the size cutting tool you are running however I still have a bit of advise.

Do yourself a favor, pickup a Gecko G540 and a 48 volt power supply.

It will triple your Taigs performance.

Either that or bite the bigger bullet and ditch the stepper setup and upgrade to servo's.

What is your goal with your setup?

Jeff...

[Crevice Reamer]

I second Jeff's recommendation.

http://geckodrive.com/product.aspx?c=3&i=14469

http://kelinginc.net/SwitchingPowerSupply.html

Assuming that the motors are 8 wire type, it sounds like they might be wired Bipolar Series. You will want to wire them Bipolar Parallel with the G540.

CR.

[TOTALLYRC]

IIRC Xylotex will do a run of boards for you if you ask and will buy enough of them.

With the advent of the G540/G251/250 steppers, all other low cost driver boards have been made obsolete IMHO. Even when looking at price, the low cost Gecko's are a far better value and the customer service is second to none. Not to mention that when you figure in a bob and spindle control, the G540 is about the same cost as a Xylotex board.

I have a Taig with an actual Xylotex board and the 425oz steppers and it will run all day at 10ipm, wether jogging or cutting. Anything above 12 gives unpredictability and frustration.

I have G251's waiting in the wings to make it all better and a 50v power supply to make it fly.

Mike

[MechanoMan]

Yeah, I know, this driver needs to be replaced.

Eh, I considered the Geckos strongly when they had the sale going on, but I wasn't completely sure what the problem was and didn't want to rush into buying something else when I didn't fully understand what I already had and the sale was almost over anyways.

With the Gecko price bumped way back up, they're looking a lot less attractive.

I may try to build a driver with the Allegro A3896, which seems to be pretty awesome. I kinda wanted the learning experience anyways. I recently found the CNCZone thread where people are building around that A3896.

[MechanoMan]

Assuming that the motors are 8 wire type, it sounds like they might be wired Bipolar Series. You will want to wire them Bipolar Parallel with the G540.

CR.

It's got 8 wires out of the motor, the cable parallels them into 4.
Hey, that "2.8A" number... being that the windings are in parallel, would the convention meaning of "stepper amps" mean that this should be driven to 5.6A?? Is 5.6A even plausible for a NEMA23?

My goal is to simply run this at a conventional speed or better with reliability, like 40ipm or more. I keep looking at the calculated times for finishing work on curves on aluminum parts and they're just too long at 10ipm, even though 10ipm may be faster than it should be pushed when initially hogging through aluminum. Also, running engravers on acrylic, etc etc. Speed is useful.

Like I say, I already did a lot of work with dampeners, actually some pretty advanced stuff. I don't think midband resonance has much if anything to do with the problem, just too low a voltage and current on the driver.

Say, there ARE a lot of DIP switches on the driver. Unlabeled. I wonder if those are set up to adjust the current and DG1 just sent it with the wrong positions, but the driver is capable of sending more current? Well I'm not gonna be stupid and just start flipping switches. If DG1 won't give me any info- which seems the case- I might try Xylotex on the off chance that they DID make this board.

EDIT: bummer, I finally got to pg 40+ of the A3986 driver projects and see where they discovered the blanking time problem.

[TOC Admin]

I bought my first Taig CNC mill from him Several years ago and did get good results at 10ipm but had to run my Table kinda Loose on the gib Adjustments.

This time around I went with the Gecko 540 Setup and it Really is a Huge difference, Plus I can run my table nice and Snug on the gibs.....

Its too Bad really that guys are buying these with high Expectations and soon after Having Problems with little to no Support.....

[Crevice Reamer]

It's got 8 wires out of the motor, the cable parallels them into 4.
Hey, that "2.8A" number... being that the windings are in parallel, would the convention meaning of "stepper amps" mean that this should be driven to 5.6A?? Is 5.6A even plausible for a NEMA23?

NO, 2.8A is the max amps for your motor--and THAT would be wired in Bipolar Parallel. Series takes less amps.

This diagram is NOT your motor, but is representative of 8 wire motors and shows the different wiring configurations:

http://kelinginc.net/KL23H286-20-08B.pdf

On the off chance that your board IS a Zylotex, maybe the switches will match up:

http://www.xylotex.com/3axBoard.htm

CR.

[Jeff-Birt]

The dip-switches are most likely to set the stepping mode (Full, 1/2, etc). Do you see any pots on the board, should be one for each axis. As suggested by CR go to the Xylotex website and compare the physical appearance of the board.

I have used a Xylotex drive on a project last year (custom lead forming machine) and found it worked well but have not tried to do a head-to-head comparison to anything else. Perhaps I should pull the Xylotex drive box off the lead machine and compare it to the MicroMill 2000 driver. Would anyone else be interested in the results?

I am a bit curious what other 'think' a good fast jog speed would be on a Taig. Taig rates the MM 2000 at 30 IPM and the DSLS 3000 at 100 IPM! I would think that 30IPM should be perfectly obtainable with a Xylotex driver.

The other factor to consider in all of this is the controlling PC. If your using the LPT and your PC can't generate a stable pulse train then your stuck no matter what drives you use. I have used a SmoothStepper with MachIII on a Vista laptop loaded with all sorts of crap and had very good success as the pulse generation is done on the SmoothStepper (which produces a very smooth and accurate pulse train.)

Don't forget about other issues that can wreak havoc as well. Double check that your providing a long enough pulse width for your driver and that the step polarity is correct (well guess you can't if the MFG won't give you any docs!) Also, a bad LPT cable can really mess everything up. If you have access to an o-scope is can help immensly with testing.

Hope that helps...

[MechanoMan]

No, the DG1 board looks nothing like that Xylotex picture. There is no pot on board but like 12 DIP switches per driver, in a big row. I don't think there's a problem with the serial port anymore, I think it comes from driving the motor at only half its specified current and with a fairly low source voltage too.

EDIT: heard back from Xylotex. They do NOT make DG1's boards. He had his own made.

[lsces]

I've been playing with this myself in the UK.
Rather than bringing over the Taig/Microproto controllers given the current shipping cost and exchange rates, I've been using the Chinese motors and drivers.
A 36 volt supply and running at 3A on 2.8Amp rated motors gives me a reliable 45IPM quarter stepping on mach3. Running at 60IPM is almost reliable and I think would be stable if I could increase the voltage over the 40 volt limit of the drivers. I'm looking at a NatSemi based 55V driver for the extra power, and I think the extra 15GBP per channel is probably justifiable. Certainly 24V gives a MUCH lower performance on the same motors. Heating of the motor only becomes a problem with leaving things stalled for long periods, and possible miss alignment due to the quarter stepping is controlled by ensuring that you return to 0,0 when finished. When repositioning the origin I'll estop to ensure motor is at a stable position before running job.

[MechanoMan]

What are these "Chinese motors and drivers"? I'm not sure what you're talking about.

The higher rpm rates are prone to midband resonance if you don't have drivers than compensate for that and have not installed mechanical dampeners. Higher voltages and currents will not prevent missed steps and stalled motors.

Heating of the motor only becomes a problem with leaving things stalled for long periods

You mean "idle" or "holding", right? Stalling is where the motor's trying to turn but gets stuck and just hums.

It's a bad idea to run over the specified current. Heat generated goes up with the square of the current so you're making 14% extra heat but only getting 7% more current. And by exceeding the designed motor temp you may be reducing the motor's available torque anyways.

[lsces]

What are these "Chinese motors and drivers"? I'm not sure what you're talking about.

In the UK we see either 'American' or 'Chinese' supplies. You can tell which is which by the price. Before the current exchange rate collapse, a suitable motor and driver was 50GBP ( likely to be 70GBP when they come back into stock ). So one could convert a CNC ready Taig for around £300.

The higher rpm rates are prone to midband resonance if you don't have drivers than compensate for that and have not installed mechanical dampeners. Higher voltages and currents will not prevent missed steps and stalled motors.

Alternatively if you don't run the stepper 'flat out' your midband is up near the top end speed you are using. PRACTICAL use of these economic parts with a CNC ready Taig at reasonable speeds is giving reliable performance. I would like to be able to investigate the performance more accurately, but practical use of the machines is not resulting in the sort of questions other driver solutions are evoking

You mean "idle" or "holding", right? Stalling is where the motor's trying to turn but gets stuck and just hums.
It's a bad idea to run over the specified current. Heat generated goes up with the square of the current so you're making 14% extra heat but only getting 7% more current. And by exceeding the designed motor temp you may be reducing the motor's available torque anyways.

Yep - I meant idle. The Chinese set-up motors are cooler after a session than the Microproto set-up, but these are different motors so a direct comparison is not appropriate. Actually the driver drops back on power when not running which is probably the real reason for the temperature difference.

( I'm probably going to drop of this forum again, not being able to post via email *IS* the headache that I remember. I could have answered this in a quarter of the time via the normal email method )

[MechanoMan]

Actually part of my motivation here is just to have this where Google can find it. To serve as a warning to others. Because I'm too lazy to set up my own website just to warn people that DeepGroove's hardware is inadequate for even casual hobbyist use.

[japroach]

Driver: Said to be "Allegro". Someone said they thought it was a Xylotex inside but I don't think so. It says "DeepGroove1" on the board's silkscreen and I doubt he had Xylotex make a special mfg run here. So it's a locally grown board. There are heatsinks glued to the chips so no telling what Allegro chip it is (they make many) and DG1 won't say. There are no external transistors. This IS a microstepper, it's 32,000 steps/in so with a 20tpi leadscrew that's 8 microsteps per full motor step. The fans don't seem to be necessary, Allegro wouldn't have rated a chip below above what it could drive without heatsinking and it doesn't seem like 1.2A would be "pushing" the ratings (there are 750mA,1.2A,and 2.5A Allegro driver chips, bet this was a 1.2A). In any case Allegros HAVE a thermal pad on bottom for cooling and it's kinda funky to try to cool them through the top through the case.

To be fair the xylotex board also uses heatsinks. Sometimes having a thermal connection to the PCB isn't always enough, a finned heatsink will provide much more dissipation than a few square cm of 2oz copper (but the copper is "free", and has a better path to the IC, so you might as well use it).

Are you sure the heatsinks are glued on, or is thermal tape used? If you want to try, you can lightly heat one with a hair dryer and you may be able to carefully twist it off.

Or if you could supply an image of the board that may help to identify the IC and if a current limit can be changed.

[MechanoMan]

From the fellow here who sold his DeepGroove1 on eBay:

That's not a pot amid the array of SMD components, it's a TO92 device.

Those are TO220 heatsinks glued to the driver chips.

[lsces]

Actually part of my motivation here is just to have this where Google can find it. To serve as a warning to others. Because I'm too lazy to set up my own website just to warn people that DeepGroove's hardware is inadequate for even casual hobbyist use.

Much the same reason I'm looking to put together an economic replacement controller. People are bringing these 'cheap' kits over here from the US and then finding that while they work, they do not work well. Motors supplied in the kit are usable, just need a controller with a bit more power to actually drive them. I've not seen if the DeepGrove motor is wired in series, but you do need parallel and a good current source to drive them well.
( Although with the collapse of the exchange rate, the temptation to try and save a few hundred pounds has now evaporated )

[japroach]

From the fellow here who sold his DeepGroove1 on eBay:

That's not a pot amid the array of SMD components, it's a TO92 device.

Those are TO220 heatsinks glued to the driver chips.

Thanks, thats a lot of dip switches per IC (12)... If they are all used then there is definitely control other than microstep resolution (only 2 switches required for that).

From the datasheet: Itripmax = Vref/8Rs. Rs are those two large resistors to the left and right of each driver IC.

If we could probe the Ref pin, and determine the voltage, then we can calculate the current limit. Also maybe see if it leads back to one of the switches and can be adjusted. But obviously I wouldn't switch anything without figuring out the driver IC first. If it is 24 pin I would guess A3983, if its 28 pin maybe A3977?

Their page for the controller, does state 2.5A per phase. So hopefully it is the A3977 or better.

[slyguyfry]

Hello all,
First post, I am currently a college student studying mechanical engineering. I have been monitoring cnc zone for about a half year now had not had to post anything till now because of the great build up of knowledge stored on this sight and the use of the search button. However I would really have liked to have found this post a couple of days ago. Currently I have a DG1 currently on its way to my house. O well ya live ya learn. Ill let you gents know if I find anything different or interesting from the mill I will receive.

[TOTALLYRC]

If possible, I would cancel the order.
Buy a cnc ready tiag and add the g540.

Your credit card company may be able to help out if that is how you paid for it.

Mike