[ldoiron17]

All,

This is really my first time writing to a forum, so feel free to let me know if this post is in the wrong category!

I am a senior Electrical Engineering student and am headed into a doctorate program. My senior project was 3 axis stepper motor control using a Probotix V90 setup. The purpose of this post is to inquire for advice for a motion control board/motor driver.

I would like to extend my senior project into a usable/marketable product. With the advancements in DIY CNC machines and drive systems I have seen a problem. In order to get professional speed and accuracy one needs a motion control board in order to get accurate timing. As I see it the standard way of getting a "quality" drive setup for a CNC machine that is "cheap" is:

Computer -> motion control board -> isolation board, produce step pulse and direction -> motor drivers -> motors

To me this seems a little bit much. I understand that there are all together packages offered, but most of them are rather expensive. Having gone through the design process of taking care of all of these things on a single board it is possible to offer such a product at a cheaper price.

My question is this, would the DIY CNC hobbyist community like to see an all in one package that is both robust and cheap? And if so what features would you like to see in such a product?

Thanks!
Lance

[ger21]

There are really two different hobby communities as I see it. It sounds like your target would be the "maker" crowd, who typically build poor performing, low cost, poor performing machines.

The members here tend to prefer higher quality and higher performance components. If you want to sell an all in one motion controller/stepper drive, you'd need to be able to at least equal the performance of a Gecko G540 and have all the features of an Ethernet Smoothstepper. And it needs to be 100% bug free.

[ldoiron17]

I am very familiar with gecko drives and how rugged they are. I am particularly interested in integrating with Mach3, which I have used in the past, but with a parallel port connection. If you don't mind answering a few questions or pointing me in the correct direction for more of the software side of things, I would be much appreciative.

Is there a place where I can find a "standard" of data communication from mach 3 to something like the smoothstepper? My assumption is that it sends the actual gcode commands in some fashion, but If there is a particular protocol already in existence I would like to integrate with that. Also even though the smooth stepper has a higher pulse and direction rate then the computer, as well as a more constant data flow (not having to deal with interrupts from the OS), aren't the drive systems still limited by how fast they can perform current hysteresis?

Also I know that a lot of drivers use hardware chips that deal with the current hysteresis, while I am using ADCs to do the job. What is the kind of speed of pulse and step direction that a professional grade motor driver board would have?

Thanks!

Lance

[ger21]

Info on writing Mach3 plugins can be found here. Downloads & Updates - Newfangled Solutions

I am particularly interested in integrating with Mach3, which I have used in the past, but with a parallel port connection.

Using a parallel port with Mach3 requires no integration??

The current router I'm building will use AC servos that require 500Khz pulse rate. Leadshine Technology Co., Ltd.

[ldoiron17]

I was just saying that I had used mach3 with a parallel port previously, I plan on integrating it with either USB and/or Ethernet. Thank you for the Mach3 information as well as the leadshine link.

The drive in the link only supports one motor, is this correct?

Also just out of curiosity do you know what the electronic gear rate specification is referring to? Is this the micro-stepping ability of the drive?

Thanks
Lance

[dmalicky]

Computer -> motion control board -> isolation board, produce step pulse and direction -> motor drivers -> motors

I'm an ME, not an EE, but I had thought there are two typical paths:
1. Computer that sends step and direction pulses to Parallel Port -> BoB with isolation front-end -> motor drivers -> motors
__1a: or integrate the middle two steps
2. Computer that sends G code to USB or Ethernet Port -> motion controller board with isolation front-end -> motor drivers -> motors
__2a: or integrate the middle two steps

Integration would be nice to save wiring, but to me the main disad is if a motor driver fries, the whole unit may need to be trashed. (The G540 has modular/replaceable drives, but most/all the others do not.) So that would be my first criteria in considering an integrated unit. If the drives are modular/replaceable, I do think there's a market for a lower-cost alternative to the G540. There is quite a bit of interest in the TB6600 and THB6064 chips and they are low cost; there are Chinese options on ebay but the individual drives are buggy and the integrated units are not modular (and probably similarly buggy, but I don't know). If a bug-free pre-soldered integrated/modular 3-4 axis version came along for ~$100, I bet there would be strong interest.

[ger21]

The drive in the link only supports one motor, is this correct?

Also just out of curiosity do you know what the electronic gear rate specification is referring to? Is this the micro-stepping ability of the drive?

Yes, the drive runs one AC servo motor.

Electronic gearing effectively reduces the encoder resolution, which reduced the pulse rate requirements.

[ldoiron17]

dmalicky,

I have never actually used the G540, I have used G203V Digital Step Drives. But both drives need an external motion controller/front end isolation. It would definitely be impossible to have a robust motion controller/driver all in one for $100. If you were to have the ability to buy a robust "all-in-one" motion controller/driver that simply plugged into a wall outlet what would be a reasonable price for you? This board would connect to a standard desktop computer.

[ldoiron17]

Ger21,

I am assuming that in reducing encoder resolution will also reduce the resolution at which the motor can step?

[dmalicky]

Yes, the G540 needs external motion control, but an old PC is usually ~free, and for most Mach3 users, that PC can be dedicated and optimized to produce a pulse stream that is plenty fast and clean. I mentioned the G540 because most DIY CNCers consider it a benchmark and any new drive system will be first compared to it (also, it does have optoisolation).

Before targeting a price point or even deciding the product will be all-in-one, I'd suggest understanding the market segments in more depth, to find a consistent match between the potential product and its niche. E.g., Mach3 is great hobby software, but it's not considered 'professional grade' and so the hardware doesn't need to be super fast. If you want to target the pro-leaning market, the KFLOP is in that direction, but that was developed by a very experienced professional so would probably be tough to beat.

For a product to succeed, it's critical to find and hit the right niche, within the constraints of one's resources and market exposure. Assuming this all-in-one device were from a start-up company, the price would have to be far below existing competition (~50%), because the consumer is accepting very high risk that this complex product is immature and also that the company may fold. Since that price would likely not be profitable, I'd suggest narrowing your scope to a simpler product that consumers would accept as "likely to work well", even though the product and company are new. In short, start-up companies are more likely to succeed when they start with a product that is: relatively inexpensive, modest scope, novel, and hits a home-run in a clearly defined niche. Then the company goes for more ambitious products.

From what I know, there is a market for a lower cost alternative to the G540 ($280). I.e., an integrated BoB with modular drives of decent V and A. Price should be much less than the G540 (else, people will just save for it), and probably similar to a BoB + 4 TB6600s (~$100).

For the low budget CNCer that doesn't want a parallel port or dedicated PC, there is probably a small market for a Arduino UNO or Due compatible board (using GRBL, 'GRBL32', or similar) with modular TB6600/THB6064 drives. The competitor here is the Synthetos GShield (and soon the G2 hardware), which uses low-power soldered drives.

[ldoiron17]

Dmalicky,

I really appreciate the feedback, it is very helpful! I think I am going to continue with these things in mind and see what the production price tag for the board itself comes out to be.

I do have a few other questions for you. Almost everything I see out there requires an external power supply, is there a major reason for this and would it be desirable for the user to not have to worry about the power supply?

Also I believe you have definitely helped me understand the market a lot more, in which case I definitely would be shooting for KFLOP performance. Just taking a quick look at it I can say that it is nearly impossible to beat the speed of this setup, seeing he is using a Xilinx Spartan FPGA that will clock up to 50 MHZ as well as a TI DSP. This guy definitely isn't fooling around. However, this board does need to be paired with a Kstep and together they total $450, and also needs an external power supply.

This reason for posting this thread is mainly educational for myself, thank you for helping me out!

Lance

[dmalicky]

Great to hear it's helpful--you're welcome! A built-in power supply would be convenient for the new user, but would probably be a net minus for most CNCers. The V needs to be matched to the stepper motor's inductance, and the A to the number of motors and their amps. So if a PS were included, it would often be the wrong V, too little A, or more A than needed ($). And some people prefer unregulated, while others prefer switching for the lower cost.

You've noticed there are pretty few "all-in-one" boxes available. That could be seen as a market opportunity, but since the boxes that are available tend to be pretty expensive, my guess is that it's difficult to build and sell such a box for profit (or, buyers get more choice for less $ if they build their own box).

I know little about them, but in addition to KFLOP there are a number of other machine controllers out there. The Mach compatible ones are among this list: Plugins - Newfangled Solutions
And there are others for LinuxCNC. LinuxCNC Support Forum :: Topics in Hardware & Machines (1/1)
It seemed like Beagle Bone was going to be a promising path, but I see little activity with it lately.

On the one hand, I don't want to discourage you from pursuing the machine-controller path, since that is evidently your interest, and I'm all about doing what one loves. But, it looks like there are quite a few alternative machine-controller options and none of them have much market penetration, except perhaps Smoothstepper. Others like Gerry would know more than I, though. To make it succeed, as with any product, first steps would be to find a niche market that has unmet needs, and then understand those needs as well as the natives. Experience and connections are very helpful to find that niche.

[ldoiron17]

Dmalicky,

I definitely plan on doing some more digging on what's out there and what the market needs are. My thoughts though on the expensive all in one products are that they can be made for much less, they are just charging for the work and knowledge it takes to make a robust product, which I must admit isn't easy. But however I already am on a 7th revision of a 4 axis motion control board with a USB connection (without power supply for now) that can operate up to 80V at a verified 4A per motor. As of right now it will be completely soldered in and not modular. I haven't Looked into any of the coding of the USB drivers for Mach 3 yet, but I can only assume that is a relatively simple protocol. All in all I don't think this board will be very expensive production wise.

By "matching the voltage to the inductance of the motor" are you referring to having a high enough voltage for the current to rise with the LR time constant?

Lance

[dmalicky]

80V, 4A, and a 7th revision sounds impressive -- that's great that you have kept developing it. 80V at a reasonable price may open up two niches. First, that would allow some higher inductance motors to perform well; the lower to mid priced drives usually top out around 50V. Many builders buy high inductance motors, not realizing they are slow at 48V. Second, relative to most economical drives, that is a good amount of power and would be attractive for the larger and heavier machines that NEMA34 or some big NEMA23 motors. Typically the high power drives for those motors are pretty pricey; an economical alternative would be attractive.

Yes, as you probably know, the strongest and fastest motor performance is at 32*sqrt(L). (Page 11: http://www.geckodrive.com/gecko/imag...cs%20Guide.pdf) One of the challenges builders face is choosing stepper motors and drives that are a good V & A match for each other. E.g., since the G540 is so popular, also very popular are the 381 oz-in motors since their L and A allow both motor and drive to maximize their mutual performance. Similarly, it might be a little difficult for many users to find an economical and well-matched power supply for 80V and 4A (10.7A+ for 4 axis). So another consideration up front is to plan what 'kit' the users would put together (e.g., G540 Hybrid Stepper Motor Kits | Automation Technology Inc). That way, your product fits well with the market when it gets there, and you have the homework done for the buyers.

On soldered vs modular, the usual failure mode is that a stepper wire is accidentally disconnected when powered up, which fries most designs. Some higher-end drives (e.g., Gecko G203V) have protection from that and more; if yours could, too, then that's another advantage to your product and fully soldered-in may be fine. Or if there are known components that are destroyed in that situation, perhaps just those could use sockets.

[ldoiron17]

But what if there was a built in power supply so the user didn't have to worry about the fact that 80V isn't a popular voltage? Also what if the built in power supply had an adjustable voltage, say from 40V to 80V so that the user could tune the motor inductance to the voltage of the drive?

I have never been a fan of how motor drivers connections have been done, it normally just leads to problems. How would you feel about a one robust connector that locked in for each motor instead of a screw clamp for each wire?

I am currently driving 4.1Amp NEMA23 motors, but I haven't been able to test milling through a material yet to see how much torque the motors can output. I have the theoretical datasheet curves but I am curious to see what they can actually do.

As for the start up failures, I don't exactly see how upon start up that the drive would fry if one wire wasn't conectted. I know the motor will lock up and create very odd back EMFs. Is this due to the lack of a current limit on the drive?

Lance

[dmalicky]

Yes, that's a great idea, assuming a widely adjustable PS can be made economically (I have no idea). I know std switching supplies are adjustable about 10%... if yours could do 40-80V, that should cover basically all the users' needs. One disad is that this expands the scope of the product, adding hurdles for a new company.

Yes, the motor connections are an issue. A robust connector would be great, although I'm not sure how the motor wires would be terminated to it w/o either soldering or another screw clamp?

Oh, I see my wording was unclear on the wire disconnection problem. The drives have no problem powering up with no motor wires connected. I should have said: after they are powered up, if a motor wire is then disconnected (or connected), most drives will fry. This also speaks to the motor connector quality, as I had an ebay TB6600 that died when I jostled the (cheap) motor connector. You probably know this, but here's what Leadshine says for the cause: "It is prohibited to pull and plug connector P2 while the driver is powered ON, because there is high current flowing through motor coils (even when motor is at standstill). Pulling or plugging connector P2 with power on will cause extremely high back-EMF voltage surge, which may damage the driver." http://www.leadshine.com/UploadFile/Down/MD882m.pdf

[ger21]

The software for a motion control board is 95% of what makes a good product.

[ldoiron17]

Gerry,

I definitely understand that, but as far as that goes it only takes time to code up a good product, not money. I understand most people put a price tag on a solid software implementation, but I am pro open-source and plan on releasing the code for my motion control board.

Lance

[dmalicky]

That's great you can implement a protection path for a disconnected wire, and the rugged connectors.

On comparing to Gecko, I'm not sure how all your specs and features would compare. Geckos have 80V, 7A, Vampire/unkillable, Step morphing, Mid-band resonance damping, and of course the Gecko name, quality, user base, and warranty. If a brand new company offered a product with all of those same features, for me, the price would need to be about half of Gecko's to be attractive. With lower specs and features, I would further scale the price proportionate to how important those are to me. So, not knowing your other specs and features, it's hard for me to guess a good price. I haven't followed the all-in-one market, so I don't have a good sense of how much extra that market would pay. A guess could be obtained by comparing the price of other ~all-in-one units to the cost of one's own assembly of the same parts.

You've heard this before from me... another product to consider would be just 4 of your stepper drives (protected, modular, and/or repairable) on a board. Possibly or optionally in a box with a BoB and/or power supply. I'm imagine these would be less interesting products to you, but I think the market would be more accepting of that 'simple' product since it doesn't have the software interface risks. I.e., no matter how much development time you put into motion control software and integration with Mach3/4, there will always be unanticipated bugs that the new users will have to work through. And it's tough to reassure new buyers that they won't also be beta testers. If the stepper drives prove to be a hit, then the market will be much more ready to accept a full all-in-one solution. Personally, I just wouldn't buy a motion control board from a brand new company--too much risk. But I would definitely consider stepper drives.

[ldoiron17]

I really appreciate all the advice, and by no means am I shooting to compete with Gecko. I have used their drives and, they are beyond solid.

You will probably hear back from me in a couple of months!

Lance