What do you want on your next control board?

User avatar
Tonkabot
Printmaster!
Posts: 251
Joined: Wed Mar 05, 2014 1:28 pm
Location: Minnesota

What do you want on your next control board?

Post by Tonkabot »

Ok, I am working on my own improved version of a smoothieboard that runs smoothieware, and can handle many more steppers, heaters, servos, or whatever.

The idea is this:
1) use smoothieware [because that is much easier than starting over with the software]
2) Use a ARM M4 instead of the ARM M3 the smoothie has - this M4 has floating point hardware and more RAM and ROM
3) Have many sockets for expansion - like 12 or so. The sockets are PCI-E which makes them incredibly inexpensive, and expansion card that has a stepper driver on it can be very inexpensive, as it does not need any 0.100 pins added to the board to attach.
4) it would have ethernet, USB, SD card (microSD), on the board (although putting ethernet onto an expansion card might be desireable).



With this you could easily run a delta with a 4 extruder head, or a 10-axis robot arm, or whatever you can imagine.

The mainboard would be targeted to a much lower price than the smoothie, and then you add as many driver boards as you need for your setup.

I would appreciate any constructive suggestions - anything you think I must have, or what things should be optional.
JFettig
Printmaster!
Posts: 824
Joined: Tue Nov 18, 2014 4:39 pm
Location: Minnesota

Re: What do you want on your next control board?

Post by JFettig »

Awesome - I'm local if you want to chat about this - I am a machine design engineer and would love to have some input.

Stuff I'd like to see: off board motor drives - or at least easy to use step/dir pins - yes I know most boards have headers for these. OR use drivers that can handle higher current - like 3-5 amps or more for Nema 23 motors.
Separate power for the steppers - 24v-60v?
More digital/analog inputs and outputs.
Higher power outputs for things like high current heated beds(24v rostock bed would be nice not to have to use a SSR), or just require things like solid state relays for these outputs and make them low power

A board to look at that I found really interesting is the Alligator board http://reprap.org/wiki/Alligator_Board" onclick="window.open(this.href);return false;

If I was to design a board, it would be simple - it would be the processor and a ton of IO so the end user can configure it how they want. At the same time, many builders want a board they can just plug in and go.
Polygonhell
ULTIMATE 3D JEDI
Posts: 2430
Joined: Mon Mar 26, 2012 1:44 pm
Location: Redmond WA

Re: What do you want on your next control board?

Post by Polygonhell »

I was looking at doing this at some point.

Basically coming up with a modula design with off board stepper controllers, and Hotend drivers connected via some sort of simple connector. Ideally things would be interchangeable, i.e. I can plug a Hotend driver or a motor controller into the connector, and ideally the firmware would self configure, though you have to start looking at the number of available pins to ensure it's practical.

The issue becomes supplying power to the separate components, you really don't want to run the 12V lines ajacent too the stepper control signals, so getting a usable physical layout becomes challenging.

I would go with Etherrnet as the primary comms protocol, rather than USB, USB is designed to send large packets of data to a device not the 1 line handshakes comms that all the host software uses. Unfortunately you have to then add support to the host software.

The other issue when you start moving things like stepper drivers off board is packaging and the associated costs, in the end you probably end up costing significantly more than existing solutions, it might be better initially tagprgetting the CNC crowd, they are significantly less price sensitive than the DIY printer guys.

As to processors, if you go M4 I'd strongly recommend the STM32 series.
User avatar
jdurand
Printmaster!
Posts: 397
Joined: Mon Dec 01, 2014 12:41 am
Location: Silicon Valley, California
Contact:

Re: What do you want on your next control board?

Post by jdurand »

One note on the M4 and floating point. During some contracting work for NXP I found that double floats run SLOWER if you enable the floating point hardware. I asked ARM directly and they said it's because the hardware is single precision and to do a double float it goes through MORE trouble than it would if it didn't use the hardware at all. (!!!) I said that was a bug, they said it was a feature. Actually, they just said that's the way it was.
Standing on the edge of reality... (me)
Quando omni flunkus moritati (Red Green)
Let no man belong to another that can belong to himself. (Paracelsus)
All things are poison and nothing is without poison; only the dose makes a thing not a poison. (Ibid.)
User avatar
jdurand
Printmaster!
Posts: 397
Joined: Mon Dec 01, 2014 12:41 am
Location: Silicon Valley, California
Contact:

Re: What do you want on your next control board?

Post by jdurand »

BTW, anyone want an old M4 development board for the LPC4300? I have one with a spare J-Link debugging pod that can go cheaply.
Standing on the edge of reality... (me)
Quando omni flunkus moritati (Red Green)
Let no man belong to another that can belong to himself. (Paracelsus)
All things are poison and nothing is without poison; only the dose makes a thing not a poison. (Ibid.)
User avatar
teoman
ULTIMATE 3D JEDI
Posts: 1783
Joined: Sat May 24, 2014 5:43 pm

Re: What do you want on your next control board?

Post by teoman »

What i would like to see is something like the repli cape but with more features.

The intel edison has 70 pins at its disposal, builtin wifi and ethernet and usb. There are a bunch of i2c pins so you can ann extra stuff if need be.
When on mobile I am brief and may be perceived as an arsl.
User avatar
Tonkabot
Printmaster!
Posts: 251
Joined: Wed Mar 05, 2014 1:28 pm
Location: Minnesota

Re: What do you want on your next control board?

Post by Tonkabot »

JFettig wrote:Awesome - I'm local if you want to chat about this - I am a machine design engineer and would love to have some input.

Stuff I'd like to see: off board motor drives - or at least easy to use step/dir pins - yes I know most boards have headers for these. OR use drivers that can handle higher current - like 3-5 amps or more for Nema 23 motors.
Separate power for the steppers - 24v-60v?
More digital/analog inputs and outputs.
Higher power outputs for things like high current heated beds(24v rostock bed would be nice not to have to use a SSR), or just require things like solid state relays for these outputs and make them low power

A board to look at that I found really interesting is the Alligator board http://reprap.org/wiki/Alligator_Board" onclick="window.open(this.href);return false;

If I was to design a board, it would be simple - it would be the processor and a ton of IO so the end user can configure it how they want. At the same time, many builders want a board they can just plug in and go.

Since my board has no drivers on it , all drivers slot into the expansion connectors. And I plan on having various drivers (8 microstep, 16, 32) integrated solutions (DRV88xx, Allegro chips) or even discrete (think microstep controller with 8 big fets, giving say the ability to drive 10A steppers).

Obviously there would be a bed heater board (just a 5 mOhm FET that can handle a 500W+ bed heater), and nozzle heater boards (probably more than 1 nozzle per board).
IO expansion boards, ...

The big power boards would not have to be supplied from the motherboard.

A fan would be able to blow across the expansion boards for added cooling.

The 'rats nest' of wiring in my printer is something that I will try to minimize by design, or at least organize by design.


Polygonhell wrote:I was looking at doing this at some point.

Basically coming up with a modula design with off board stepper controllers, and Hotend drivers connected via some sort of simple connector. Ideally things would be interchangeable, i.e. I can plug a Hotend driver or a motor controller into the connector, and ideally the firmware would self configure, though you have to start looking at the number of available pins to ensure it's practical.

The issue becomes supplying power to the separate components, you really don't want to run the 12V lines ajacent too the stepper control signals, so getting a usable physical layout becomes challenging.

I would go with Etherrnet as the primary comms protocol, rather than USB, USB is designed to send large packets of data to a device not the 1 line handshakes comms that all the host software uses. Unfortunately you have to then add support to the host software.

The other issue when you start moving things like stepper drivers off board is packaging and the associated costs, in the end you probably end up costing significantly more than existing solutions, it might be better initially tagprgetting the CNC crowd, they are significantly less price sensitive than the DIY printer guys.

As to processors, if you go M4 I'd strongly recommend the STM32 series.
Right, I have several good idea's on distributing power nicely, and I also will make the addressing of the boards logical (each socket provides a unique address to whatever board is placed in it, and possibly every board has an ID so an autoconfigure could work.

I thought the smoothieware now supports the ethernet, at least for a web interface. Do you mean it should have like a TCP socket interface [which I am guessing isn't in smoothieware?]

I have worked on lots of different processors over the years and Like the Freescale stuff. Always thought the STM stuff was overpriced and had less support. However, I will do more homework to see if the STM32 would do a better job for this than the MK64FN1M0VLQ12 that is currently the leading candidate.



I'll look up a repli cape and see what that is. and the edison.

There was a kickstarter that failed that at least used my idea of PCIE sockets and driver boards, but it also just used an ATMega.
(Which for 8bit stuff I prefer the Atmels, but I don't think they are right for this controller)
User avatar
Tonkabot
Printmaster!
Posts: 251
Joined: Wed Mar 05, 2014 1:28 pm
Location: Minnesota

Re: What do you want on your next control board?

Post by Tonkabot »

Polygonhell wrote: [snip]
As to processors, if you go M4 I'd strongly recommend the STM32 series.
Ah, I think you mean the new Cortex-M7 of which the STM32 F7 seems to be the first on the market. Freescale has a license also, so they are probably coming.
These are a lot faster. I'll look into them, they might make sense if they are really available yet. [Hmm, I think I'll call up my Freescale people and see what they can tell me]
User avatar
Tonkabot
Printmaster!
Posts: 251
Joined: Wed Mar 05, 2014 1:28 pm
Location: Minnesota

Re: What do you want on your next control board?

Post by Tonkabot »

teoman wrote:What i would like to see is something like the repli cape but with more features.

The intel edison has 70 pins at its disposal, builtin wifi and ethernet and usb. There are a bunch of i2c pins so you can ann extra stuff if need be.

I checked out the replicape and as a beaglebone cape it may sound good on paper, but one look at the picture tells me they haven't paid enough attention to heat.
They have many stepper drivers packed into a tiny area, which might work okay for steppers running maybe 1/2 amp or less, but there is no chance your getting 2 amps into a stepper from that board for very long unless there is a leaf blower to keep it cool. Also I don't see very large caps, another requirement for a well designed driver (large in value is usually large in size)

Now the edison is very interesting as a raspberry pi competitor. right now if I want cheap wifi the way to go is a Pi + cheap USB wifi dongle + cheap SD card. But the edison has the wifi builtin, and bluetooth, and 4GB flash and 1GB RAM - which unless I need a display (HDMI) is a better deal.

But for my board I don't think having wired ethernet is that big a problem. Anyone who is running everything in their home off WiFi should really take a look at how much better it will work if they run cables to everything that is not 'portable'. My blu-ray player and TV can do WiFi, but the viewing experience suffers compared to 1000T cables. And yes, that is even comparing to my WiFi router which is about the best/fastest one can buy [it was for a few minutes there anyways].
User avatar
teoman
ULTIMATE 3D JEDI
Posts: 1783
Joined: Sat May 24, 2014 5:43 pm

Re: What do you want on your next control board?

Post by teoman »

I have most of the small pc boards.

Let me know if you need anything.


My list is:

Edison with arduino board and brakeout board
Beagle bone black
Pcduino
Banana pro
Raspberry pi
Arduino Yun
All of the other arduinos except the lilly.


If the mossfets used have a low enough resistance the heat is not such a problem. I have 2 in parallel running a total of 24A and they dont even get warm. Their internal resistance was rated at 30mOhm if i recall correctly.

I saw some pcbs that were made out of aluminium. I want to use them but i just cannot find a project to justify using them.

Replicape:
They claim the high power mossfets are rated for 30A and the motor drivers can do 1.75 A rms continuously.
http://wiki.thing-printer.com/index.php?title=Replicape" onclick="window.open(this.href);return false;
When on mobile I am brief and may be perceived as an arsl.
Eric
Printmaster!
Posts: 726
Joined: Sat Aug 18, 2012 4:09 am
Location: Chula Vista, CA

Re: What do you want on your next control board?

Post by Eric »

They just launched the Raspberry Pi 2, at the same $35 price. The major upgrades are a quad-core ARMv7 CPU and 1G RAM.
User avatar
Tonkabot
Printmaster!
Posts: 251
Joined: Wed Mar 05, 2014 1:28 pm
Location: Minnesota

Re: What do you want on your next control board?

Post by Tonkabot »

teoman wrote:I have most of the small pc boards.

Let me know if you need anything.


My list is:

Edison with arduino board and brakeout board
Beagle bone black
Pcduino
Banana pro
Raspberry pi
Arduino Yun
All of the other arduinos except the lilly.


If the mossfets used have a low enough resistance the heat is not such a problem. I have 2 in parallel running a total of 24A and they dont even get warm. Their internal resistance was rated at 30mOhm if i recall correctly.

I saw some pcbs that were made out of aluminium. I want to use them but i just cannot find a project to justify using them.

Replicape:
They claim the high power mossfets are rated for 30A and the motor drivers can do 1.75 A rms continuously.
http://wiki.thing-printer.com/index.php?title=Replicape" onclick="window.open(this.href);return false;
If you are running 24A through two parallel 30mOhm fets they are disapating 8.64 Watts of heat. (12*12*.03*2) [12=1/2 of 24, for each parallel FET)

Yeah, I just looked again at the replicape specs. the bed heater FET will dissapate 3.7W at 32 A (voltage doesn't matter, use I*I*R for a FET) (they are rated at 3.6 mOhm)
The DRV8825's will be making 3W of heat at their RMS 1.75A number. Each.
I'll eat their board if it doesn't burn up - at 22W total for 6 steppers and a 32A bed heater. Maybe with a big loud fan it will stay cool. Maybe.

There have been PCBs made out of ceramics for multi-chip modules forever. Think military. Berg has some kind of copper clad PCBs. IIRC.
Those become giant heat sinks.
User avatar
626Pilot
ULTIMATE 3D JEDI
Posts: 1720
Joined: Tue May 14, 2013 12:52 pm

Re: What do you want on your next control board?

Post by 626Pilot »

I have owned both variants of Smoothie-compatible boards already, so here is my feedback.

Azteeg X5
Love: Small footprint, 32x microstepping, all power connectors in one location, USB/PS jumper, reset + bootloader + "play" button onboard make it a more or less complete solution (even without an LCD)
Hate: One blew a power FET and the other has a bad CPU. Each one lasted for about 2-3 months before dying. Not a lot of connections either - only three max endstops and one min (for a Z-probe), and two or MAYBE three thermistors. Also, they ran out of stock of the MakerBot connector version and I had to use some breakaway pin connectors with their screw terminals. That's... no. That's like using duck tape to fix your car.

Smoothieboard 5X
Love: So far it hasn't blown up, and it's not an Azteeg of questionable longevity. Has 3 min and 3 max endstop connectors (which is what I really want, for FSR-triangulated probing) and enough connectors to hook up two extruders. Capable of running the board off USB with the power electronics on or off, so I can switch it off at the end of the day without losing the Repetier connection
Hate: Much larger, 16x steppers, some of the connectors are inexplicably upside down, no USB/PS jumper (you have to power it off USB or buy your own voltage regulator and solder it in), you have to connect power in three physically distinct places, it's short one power connector (so you can't run it off 5V from the PSU - you need USB power, or to solder in the regulator). They don't give you reset or bootloader buttons. Site says, "Breaking pins out doesn't cost much, right…" Very funny! Except I would have paid the extra $2 to have those buttons there. You need them. If the board bricks during a firmware update, you can fix it by holding bootloader and tapping reset, assuming you have them. Without them, you have to get out a wiring diagram and hack together some ugly nonsense with wires. For the 95% of 3D printer owners who aren't comfortable enough with electronic circuits to risk frying their board, this is no good!

Stuff I would do if I was rolling my own:
  • Use better components than the others, and don't be afraid to over-engineer it a little. Maybe it costs $40 extra, but what is $40? Not even a single hour of my time. I would pay that without blinking, if I knew I was getting something that won't break and make me wait days for a replacement part. Someone who bellyaches "it's too expensive" can buy an Azteeg. Someone whose printer has to earn its keep, who wants to run four extruders and all that, is going to recognize the value of buying the right board, the first time, rather than blowing money needlessly on lesser solutions that then have to be replaced with what should've been bought on day one.
  • Use really good power FETs, capable of running a serious heated bead like the Onyx, for years, without blowing. I don't want to have to mess around with an SSR or any other dumb stuff. In fact, take the money I'd have to spend on an SSR and farm that into getting better FETs. I don't want the hassle of extra components, figuring out a mount, losing space in the bay, wondering if the ebay vendor I got it from is crap and it's going to catch fire, etc.
  • 32x steppers would be nice.
  • Any fuses should either be blade fuses I can replace at any auto shop, or the self-resetting variety seen on the RAMBo. I didn't like when my RAMBo blew a tiny weird fuse, and I had to wait days for a replacement, because (even in Los Angeles) no one local had them in stock. No one that I could find online, anyway.
  • Micro USB. The square-shaped "USB B" or whatever it's called is a waste of space, and micro USB cords are already very common among people who have cell phones. (Not mini USB - the connectors wear out quicker, and fewer people have them these days.)
  • 40-pin expansion header like the Raspberry Pi. You can then sandwich a shield on top, or use a cheap 40-pin IDE connector to put it somewhere else.
  • Two six-pin headers for connecting to a RRD LCD, or if not that, some decent graphic LCD with an encoder, reset and stop buttons, and a honker.
  • Left (hingeward) edge: Power connectors. ALL OF THEM. No spaghetti with three different connectors in three totally different places. Putting them by the hinge means they get the least stress possible when the door is opened.
  • Right edge: microSD, Ethernet (if included), USB, and reset + bootloader buttons.
  • Top/bottom edges: Connectors for steppers, thermistors, hot ends, etc. All connectors should be VERTICAL. Horizontal connectors waste space and make the wires turn 90 degrees to go towards whatever they hook up to, and the pins get pulled on when the door is opened. Gross.
  • Speaking of connectors, use standard MakerBot. I hate having to re-splice connectors every time I swap boards.
Tonkabot wrote:1) use smoothieware [because that is much easier than starting over with the software]
My favorite!
2) Use a ARM M4 instead of the ARM M3 the smoothie has - this M4 has floating point hardware and more RAM and ROM
Good. I feel the decision to use a processor with 32K RAM, when it would be... what... one or two dollars to get 64K or more... was not a good choice. In Smoothie IRC, they are always griping about developers using too much of the paltry 32K RAM, and even now they're down to a few K free. I have been doing a lot with the Teensy 3.1 - a hair under $20, AND it has 64K RAM. Why not a chip with 128-256K? People want localization and fancy menus with dozens and dozens of nested entries. Saving a few bucks on a controller that's already well over $100, so the developers can spend hours wringing their hands over memory issues, makes no sense to me.
3) Have many sockets for expansion - like 12 or so. The sockets are PCI-E
You mean the six-pin power connectors, right? I don't know if I'd have a use for 12. 8 is already enough to run a four-color printer with a tool changer. It might not seem like a big deal, but PCB milling costs will have to be managed, and they ding you for every square centimeter!
teoman wrote:The intel edison has 70 pins at its disposal, builtin wifi and ethernet and usb. There are a bunch of i2c pins so you can ann extra stuff if need be.
That's probably why Smoothie is being ported to it :o
User avatar
Renha
Printmaster!
Posts: 282
Joined: Fri Apr 11, 2014 2:18 am
Location: Balashikha, Russian Federation
Contact:

Re: What do you want on your next control board?

Post by Renha »

Hey, look to Quadstepper board from sparkfun, it only needs 6 wires to drive 4 steppers when you use them one-by-one (as for extruders), you could make compatible port for it. I use it now on my RMAX2, and it is wonderful to have stepper controller placed where steppers are placed (in top plate) :)
3D printing is magic!

Sorry my engrish :-/ you could PM me in Russian.
User avatar
Tonkabot
Printmaster!
Posts: 251
Joined: Wed Mar 05, 2014 1:28 pm
Location: Minnesota

Re: What do you want on your next control board?

Post by Tonkabot »

Good stuff 626pilot, I will keep your ideas in mind. One thing you don't understand about my board is that the idea is a 'motherboard' with about 12 expansion slots.
Yes that means there is going to be some 'height' to my board (But I am pretty sure the board size can be rambo size or smaller)

The concept is to use the 36 pin PCI-express connectors you find on computer motherboards, because they are the cheapest card edge connector I can find.
So the motherboard has 12 of these sockets, the ARM M4, and ethernet/usb/ a few more things common to all boards (I'll include a reset and boot switch for you)

The cards that fit into the sockets have the stepper motor drivers or bed heaters or a number of other options (one has the 6 end stops)
These cards can be made very inexpensive, because they have just the stepper driver, pcb area as heat sink, and output connector. In cases where maybe a 'high power' stepper driver or bed heater board with discrete FETs on might require more power than can be supplied through the motherboard, it can have it's own power input and output [also allowing a different voltage than the rest]

You see some boards that use the pololu stepper modules, my expansion boards should be as cheap because they have no header pins, although they do need your maker stepper socket [ which because of my form factor is going to have to be a right angle connector, sorry]

I think that I need to make sure my board with expansion boards populated fits in my rostock. Clearly because of the height of the board it might have trouble in some cases.

However, my expansion boards may only need to be 1.5 to 2 inches high. so maybe 1.5 x 2.5 inches? I have also been thinking that some kind of board lock will be required so the expansion boards cannot wiggle or fall out.


Renha, I also have been kicking around the idea of remote stepper drivers which would help reduce EMI noise [electrical noise], like the reason the endstop wires cannot run parallel to the long stepper wires. Interestingly sparkfun says that you will not be able to run all 4 steppers at 2 amps because of heat - and they have only 4 steppers in an area much bigger than the Replicape has 6.

My design could easily handle remote steppers with out changes to the motherboard, just an expansion board that has proper signalling [differential pairs are nice in a noisy environment] up to the remote end.

Automobiles used to have huge wiring harnesses with like 50 wires going to the driver door for all the crap (power mirror, power mirror switches, power window, switches, ...) Now they use a network and have like 6 wires doing the same job. I see 3d printers right now as having a mass of cabling, and reducing that will add to reliability and looks and assembly ease.
User avatar
Nylocke
ULTIMATE 3D JEDI
Posts: 1421
Joined: Sun Jun 23, 2013 1:43 pm
Location: Iowa

Re: What do you want on your next control board?

Post by Nylocke »

My 2¢, keep it all in one. I've only ever seen a multi board solution once, it was on an old Mendel, and it was the biggest waste of space I've ever seen. Modular is cool and all but my X3 Pro is quite nice. Separate power inputs for motors, heaters, and the bed is nice too. X32 drivers direct to the board, or plugin pololu type works for me. I've had good experiences with either.

Personally, I'd love it if it were to fit the X3 Pro foot print so I could just drop it in with no fuss. I think a fair amount of your sales would come from X3 Pro users wanting something better and faster, so I think it could be advantageous from a sales standpoint.
User avatar
Tonkabot
Printmaster!
Posts: 251
Joined: Wed Mar 05, 2014 1:28 pm
Location: Minnesota

Re: What do you want on your next control board?

Post by Tonkabot »

As I see it the problem with putting it all on one board is the heat sink area needed for all the many drivers, plus very few people need every driver populated. The drivers are relatively expensive, so having say 7 or 8 on the board is going to put a price premium on the board for those that don't need more than 4 or 5. To do a reliable board the drivers need space, and that means lots of surface area. both surface area and paying for just what you need for drivers work very well with my idea.

You say your okay with the pololu type stepper modules, well, my plan isn't all that different except the modules are at 90 degrees to the board instead of hovering over it. The pololu modules do not solve the surface area issues at all, in fact they get worse because you got all those pins running power up to the module, then back down to the connector, and the pololu module doesn't have much surface area for heat disapation.

The X3 pro is 6.3x3 inches. Rambo is like 4x5. That footprint is doable, but I'll have to see what works out.

The area above most boards is filled with wires and air space, I am thinking that if my expansion boards are 2 inches up off the mainboard I shouldn't cause too much fitment troubles.
Mac The Knife
ULTIMATE 3D JEDI
Posts: 1409
Joined: Sun May 11, 2014 6:18 pm

Re: What do you want on your next control board?

Post by Mac The Knife »

I would like to be able to install stepper or servo drives of my choice, (Geckodrives). Accidents happen, and being able to replace a single drive without sending the whole board in is a big plus. Plus Geckodrives are pretty tough.
R-Max V2
Eris
Folger Tech FT-5 R2
User avatar
Tonkabot
Printmaster!
Posts: 251
Joined: Wed Mar 05, 2014 1:28 pm
Location: Minnesota

Re: What do you want on your next control board?

Post by Tonkabot »

Mac The Knife wrote:I would like to be able to install stepper or servo drives of my choice, (Geckodrives). Accidents happen, and being able to replace a single drive without sending the whole board in is a big plus. Plus Geckodrives are pretty tough.
Right, and my system allows for multiple different stepper boards DRV8825, Axxxx, or discrete fets, or a simple breakout board that lets you route the step,direction, and enable lines out to an external controller. In addition to a few standard driver types, I'll also have a 'proto' board that just has a .100 grid, and also I could make a board that has a DB25 that runs out to a Gekko G540 board.

My system allows you to replace just one driver board in the event of a failure, something difficult if they are soldered to the board. If all the power components are on modules, you'll never have to replace the motherboard because of a failure.
Polygonhell
ULTIMATE 3D JEDI
Posts: 2430
Joined: Mon Mar 26, 2012 1:44 pm
Location: Redmond WA

Re: What do you want on your next control board?

Post by Polygonhell »

The thing you want to avoid to ensure no main board failures, is to not use the Analog pins on the CPU, they are all sensitive to any sore of over voltage, so any short in wiring can take them out, and commonly there is really only one ADC in the silicon that is multiplexed across the inputs, so by frying one you can fry them all.
The common use-age for them is the thermistor, providing an external thermocouple board is one alternative.

The reason I didn't end up following through with my work (other than lack of time) was just what it would end up costing, I think you can make the cost fly in the DIY CNC space which is a lot less cost conscious, but I don't know how much scope there is for a higher end controller in the 3D printer space which seems incredibly cost sensitive.
User avatar
Tonkabot
Printmaster!
Posts: 251
Joined: Wed Mar 05, 2014 1:28 pm
Location: Minnesota

Re: What do you want on your next control board?

Post by Tonkabot »

Polygonhell wrote:The thing you want to avoid to ensure no main board failures, is to not use the Analog pins on the CPU, they are all sensitive to any sore of over voltage, so any short in wiring can take them out, and commonly there is really only one ADC in the silicon that is multiplexed across the inputs, so by frying one you can fry them all.
The common use-age for them is the thermistor, providing an external thermocouple board is one alternative.

The reason I didn't end up following through with my work (other than lack of time) was just what it would end up costing, I think you can make the cost fly in the DIY CNC space which is a lot less cost conscious, but I don't know how much scope there is for a higher end controller in the 3D printer space which seems incredibly cost sensitive.

Right, so I need to diode clamp the analog inputs or otherwise protect them. That will go on the to do list. Having a module that does serveral thermocouples (or other analog that needs an amplifier [like a strain gauge]) does make sense, it can do the protection on the analog lines. Although with only 36 pins to work with I might not be able to run analog lines to all my IO sockets. All my sockets will have SPI ports, and unique chip select per socket, so I could read an analog chip that way.

I was looking at the smoothie board price of $160 or so for the x5. I think I can beat that, but my model will be different - mainboard might be $75 and a stepper driver board $10, or something. Don't etch those in stone, I have a lot of work to do. But I see the smoothie boards selling 500ish per year, so I think there is room in the market.

Also as you can see my avatar picture I am also into robotics, and this board would work well for driving them, also. Although my Tryclops robot you see actually has 3 real servos, brushed DC motors with 20:1 gearbox and quadrature encoders on each wheel. To control tryclops with my new board I would also need a 'smart' expansion board that could do real servos and quadrature decode, like it would have a DSP and 3-phase driver on it. [obviously with a 3-phase driver you can also drive a brushed motor,too.] That board will be bigger and much more expensive than a stepper board.
Eric
Printmaster!
Posts: 726
Joined: Sat Aug 18, 2012 4:09 am
Location: Chula Vista, CA

Re: What do you want on your next control board?

Post by Eric »

I'm not convinced that a pcb edge connector is appropriate for what must be considered a high-vibration environment, especially if it's free-standing in the socket. To make it work, you'd need a card cage with a screw-down or spring tension to prevent constant micro-motions of the contact surfaces.

From a hackers point of view, I like simple pin headers because I can do what I want with them, without having to buy a breakout board or pci-e cable adapter. But I know I'm not your typical audience, either.

If you do use a PCI-E connector, make sure the boards are dimensioned in a way that makes it impossible to insert in a PC motherboard. Otherwise, someone is bound to destroy both the board AND their PC out of ignorance.
User avatar
626Pilot
ULTIMATE 3D JEDI
Posts: 1720
Joined: Tue May 14, 2013 12:52 pm

Re: What do you want on your next control board?

Post by 626Pilot »

PCI-E card edge connectors have built-in spring detents. You have to push a little lever sideways to unseat the card. I wouldn't rely on it, though. Some standoffs that let you fasten the shield to the motherboard would be ideal. I agree with the comment that it should be made impossible to seat a regular PCI-E card. Perhaps you could physically block some of the slot (glue in a custom plastic insert), and make your shield's edge connector work with it. For high current, you can just use multiple pins.

I would like to see back EMF protection on the steppers, if possible. There is always some warning about moving the arms when the printer is powered down, in case it fries the board.

If this thing came with an option for a parallel (is that DB25?) port, I could see you selling a lot of these to CNC people.
Eric
Printmaster!
Posts: 726
Joined: Sat Aug 18, 2012 4:09 am
Location: Chula Vista, CA

Re: What do you want on your next control board?

Post by Eric »

626Pilot wrote:PCI-E card edge connectors have built-in spring detents. You have to push a little lever sideways to unseat the card.
That's only true of PCI-E x16 connectors (162 contacts), normally used for a video card. The tab helps secure the back end of the longish card connector, while the front end is secured by the screw-down bracket. I've never seen any of the shorter versions (x8, x4, x1) with a similar lock tab.

In this case, Tonkabot is talking about using a PCI-E x1 connector (36 contacts). Many PC motherboards have at least one x1 connector. You can find x1 cards made for things like network cards, modems, sound, USB ports, sata 3 ports, raid controllers, ssd drives, low-end video cards, and a few other things. In my own case, I used it to add USB 3.0 ports to a older motherboard that didn't come with USB 3.0, which I then used for an external book drive (it worked on USB 2, but is way faster on 3).
User avatar
Tonkabot
Printmaster!
Posts: 251
Joined: Wed Mar 05, 2014 1:28 pm
Location: Minnesota

Re: What do you want on your next control board?

Post by Tonkabot »

Eric wrote:
626Pilot wrote:PCI-E card edge connectors have built-in spring detents. You have to push a little lever sideways to unseat the card.
That's only true of PCI-E x16 connectors (162 contacts), normally used for a video card. The tab helps secure the back end of the longish card connector, while the front end is secured by the screw-down bracket. I've never seen any of the shorter versions (x8, x4, x1) with a similar lock tab.

In this case, Tonkabot is talking about using a PCI-E x1 connector (36 contacts). Many PC motherboards have at least one x1 connector. You can find x1 cards made for things like network cards, modems, sound, USB ports, sata 3 ports, raid controllers, ssd drives, low-end video cards, and a few other things. In my own case, I used it to add USB 3.0 ports to a older motherboard that didn't come with USB 3.0, which I then used for an external book drive (it worked on USB 2, but is way faster on 3).

Correct, I am thinking the PCI-E x1 36 pins should be enough, but I will have to use like 4 for gnd and 4 for Vbb (12-36v or whatever). They are 1.1A per pin. Bed heater power will not come through the PCIE connector.

I was thinking that they need to be firmly secured. I have thought of a couple ways to do that so far, but I think the leading idea is to have a hole in the board near the connector (maybe on each end) that snap into a thing on the mainboard. You'll have seen similar things on video cards and stuff in the bigger slots. I suppose another option is to use a SIMM or DIMM memory card socket, but a PCIE socket is like $0.15 and a DIMM socket is going to be way more [from memory somthing like several dollars] [but I should check, maybe there are a billion SIMM sockets in a warehouse somewhere] The SIMM and DIMM would be much bigger than I need, The laptop SO-DIMM might be doable, but I still bet it is over a dollar. The point behind using a SO-DIMM instead of the PCI-E is that there is a positive latch to hold a DIMM in, I would not have to add something.

I see the retention mechanism is also about $0.15 in qty. [you can get them seperate from the PCIE x16 sockets, some of which have them integrated] doubles the PCIE connector price, but probably worth it.
Post Reply

Return to “Smoothieboard and variants”