Reprap Electronics Devolpment

I agree, Reprap is such an interesting project because it has almost everything I can think of under one umbrella: electronics, real-time programming, mechanics and traditional software. So naturally the electronics should be considered just as eligible for tweaking as any other part of the project.

In terms of design philosophy, I - personally - like the philosophy behind Unix, where several small modules are connected to perform some task, having each module do only one small thing, but doing it well. That way, things may be improved evolutionary rather than revolutionary.

Like any other open source project benefiting from the efforts of other projects will often make things move along a lot faster, if not reusing the whole, then at least reusing the basic innovations. Currently a lot of small and not so small stores are starting to make pre-assembled Arduino and Arduino clones available, and all sorts of small projects are cranking out new and better ways of using that platform. IMO it would be great if Reprap would/could benefit from these kinds of projects.

Now in terms of how much intelligence to add to the Reprap main board? The blueprints for the designs printed by the Reprapper needs to come from somewhere, right now I see two sources for those blueprints, either a 3D scanner built into the Reprap, like a 3D photo copying machine, and CAD drawings from an external source.

Right now I can see only a local computer as the source of those blueprints, though further out in the future it may be possible to create a machine, which retrieves the design for a full meal, food, plate, utensils and everything from a site much like Thingieverse,

Short term, I think the blueprints from a local computer and/or an inbuilt 3D scanner are the only realistic sources. IMO Dell, Leonovo et.al. are - currently - in a better position to crank out high performance computing devices, but have no clue about making Reprappers, so again IMO it makes more sense to focus the creative effort on areas where the Reprap community is likely to make a difference.

Quote

I wish we would get others in this discussion like Nophead, Forrest, Vic..

I stopped discussing electronic architecture a long time ago. There is no consensus so I simply do my own thing and stopped wasting time trying to convince anybody else.

I will control my Darwin and forthcoming Mendel with an ARM based mother board with four microstepping drives for axis control, Ethernet, RS485 and an SD Card. I will probably also have a display, some switches, USB A & B, perhaps CAN, but they are not essential as I can have a web interface to set up the machine and then let it run from SD.

I have already published an extruder controller with RS485. The axis controller plus networked head controller topology gives a cost effective solution with very simple wiring but does not limit new head developments.

I will probably include enough on the motherboard to control one stepper extruder using the 4th drive. Adding the extruder controller on RS485 would free up the I/O on the mother board to drive a heated bed and a fourth axis. That gives a cheap single board basic solution and two board solution for a machine with the current capabilities of HydraRaptor.

I can't see the point of backplanes, etc. It just adds cost. I will just bring out any spare pins to a simple connector.

Edited 3 time(s). Last edit at 02/01/2010 04:59PM by nophead.

---

[www.hydraraptor.blogspot.com]

nophead Wrote:

> I will control my Darwin and forthcoming Mendel
> with an ARM based mother board with four
> microstepping drives for axis control, Ethernet,
> RS485 and an SD Card. I will probably also have a
> display, some switches, USB A & B, perhaps CAN,
> but they are not essential as I can have a web
> interface to set up the machine and then let it
> run from SD.

Would you happen to have a link to your architecture? I'd like to take a peek.

> I can't see the point of backplanes, etc. It just
> adds cost. I will just bring out any spare pins to
> a simple connector.

The biggest advantage I can see would be to swap out the stepper drivers with something else; you've made an implied assumption that the 3axis machine will be stepper based. In this regard, the Gen1 electronics had better modularity.

RS485 has appeal too, but that adds cost if your daughterboards were simple logic boards, instead of slave/microprocessor controlled.

Having a 'standard' allows sourcing electronic components from multiple suppliers -- where there is reasonable expectation that the components will interoperate without too much effort. Case in point, my RepolaRap design will use steppers + 3 line each encoders (QUAD1+QUAD2+ABSPOS) for X+Y; I'll be able to use the existing Z. To get this working, I will more than likely need to solder and adjust the MakerBot motherboard. I would much prefer a plugable device that I could expand, and then buy a new motherboard from Makerbot if my daughterboard experiments were successful. If they failed, I'd much rather be able to plug the motherboard back into my future Mental printer and have everything happy and printing (risk of problems if I have to solder the motherboard to make adjustments.)

What you propose seems more appealing if RepRap modules, boards, electronics, firmware, I/O, etc. remained static and unlikely to change. I don't see that as reality yet.

Basically, I believe the drive for pin headers, card slots, backplane, etc. is to support a more 'hacker' mentality. Remember the user port on the C64? I made cards that plugged into that, and it was awesome. I did *NOT* have to take the computer apart and make soldering changes to the internal board (well, to be honest here, I sort of did that when one stopped working. )

> Would you happen to have a link to your
> architecture? I'd like to take a peek.

No I haven't documented anything yet but I will start designing it this week.

>
> > I can't see the point of backplanes, etc. It
> just
> > adds cost. I will just bring out any spare pins
> to
> > a simple connector.
>
> The biggest advantage I can see would be to swap
> out the stepper drivers with something else;
> you've made an implied assumption that the 3axis
> machine will be stepper based. In this regard,
> the Gen1 electronics had better modularity.

Yes I have assumed steppers because they are simplest solution. Now we have switched to NEMA17 for $15 they are a reasonable price for a precision device and give good performance. A proper 3 phase BLDC servo with encoder is going to cost more and be a lot more complicated. Home rolled servos made from DC motors and encoders will not have the performance or lifetime that I desire.

>
> RS485 has appeal too, but that adds cost if your
> daughterboards were simple logic boards, instead
> of slave/microprocessor controlled.

It doesn't add much because the connectors and wiring get cheaper.

>
> Having a 'standard' allows sourcing electronic
> components from multiple suppliers -- where there
> is reasonable expectation that the components will
> interoperate without too much effort. Case in
> point, my RepolaRap design will use steppers + 3
> line each encoders (QUAD1+QUAD2+ABSPOS) for X+Y;
> I'll be able to use the existing Z. To get this
> working, I will more than likely need to solder
> and adjust the MakerBot motherboard. I would much
> prefer a plugable device that I could expand, and
> then buy a new motherboard from Makerbot if my
> daughterboard experiments were successful.

Assuming there are enough pins left on my board you should be able to connect you encoders simply by making up a cable. I have qaud and index on two of my axes on HydraRaptor but I only use them for sanity checks as the steppers never skip.

If
> they failed, I'd much rather be able to plug the
> motherboard back into my future Mental printer and
> have everything happy and printing (risk of
> problems if I have to solder the motherboard to
> make adjustments.)
>
> What you propose seems more appealing if RepRap
> modules, boards, electronics, firmware, I/O, etc.
> remained static and unlikely to change. I don't
> see that as reality yet.

There will be no need to use any of the RepRap electronics as this provides a superset of functionality on the main board and could support the existing extruder controller.

>
> Basically, I believe the drive for pin headers,
> card slots, backplane, etc. is to support a more
> 'hacker' mentality. Remember the user port on the
> C64? I made cards that plugged into that, and it
> was awesome. I did *NOT* have to take the
> computer apart and make soldering changes to the
> internal board (well, to be honest here, I sort of
> did that when one stopped working. )

Yes but modern expandability is done with serial connections, nothing has bus or parallel expansion these days. Also there is very small percentage of hackers among the RepRap community. Most people just want a cheap 3d printer with electronics that are cheap and pre-assembled.

As I said before, there is no consensus so I won't waste any more time here. I will just do it and publish like everything else I do. People can take it or leave it and it doesn't matter to me which.

Edited 1 time(s). Last edit at 02/01/2010 06:20PM by nophead.

---

[www.hydraraptor.blogspot.com]

i wonder if there ever will be a consensus on parts for this project, that is the one good thing about makerbot is that they can have one for there systems, and i think if it works well for them it will be adopted to a point here,

---

[mike-mack.blogspot.com]

The Makerbot electronics were designed by Zach so it didn't need a consensus, he just went ahead and did it and then formed Makerbot.

---

[www.hydraraptor.blogspot.com]

Consensus's are possible where the one you create does not exclude those who would otherwise be interested in supporting it.

At the point your consensus view prescribes one technological route to the exclusion of others your consensus will be no more.

Insisting on one processor over another, or language over another are particular cases in point.

The trick is to specify the interfaces in such a way that they do not prescribe the technology that resides behind them.

So long as it is compatible at the interface and can therefore do the job that is required does an end user realy care.

The only folk who do care are those that actively create the functionality behind the interface.

The backplane arangment is an enabler. It specifies the interface but leaves it up to the card implementers as to how they do the sub tasks that make up the system.

If as an individual you have the where withall to do all of a system (Software, Fimrware and Hardware) or even purchase it in. Then consensus is not neccesary for you to proceed, and you should do so.

Alternatively if you have or wish to develop skills in one particular area and still want to make a valid contribution a degree of consesus is neccesary. You have to feel confident that the part you can contribut to will work with a part that someone else contributes. If you contribute a part that will not work with anyone elses so as to acheive the given objective you have arguably wasted your time. I can't think of many individuals that would care to commit time, effort and cash to such an uncertain outcome.

A sensible degree of consensus then is an enabler towards greater participation.

Sensible in this case is clearly a route that does not prevent other rotes of experimentation.

---

Necessity hopefully becomes the absentee parent of successfully invented children.

Ladies and Gentlemen, for your consideration.

Please find attached a first stab (literally that, a first stab for discussion) at some form of backplane pinout guide, the word GUIDE is stressed very heavily. It is not written in stone and being a passive backplane any buss line can be used for any purpose.

You as individual experimenters are more than likely to tweak as needed for your chosen area of interestt/contribution.

The row count currently stands at 38 rows and is a double column layout giving 76 individual connections. Due to the way pin strips are bought in I would like to sugest we work towards reducing the pin count to at least 32 rows by 2.

Bearing in mind that most micro-controlers etc have special function pins (PWM,ADC,Serial, etc) that can be used as Digital IO, but Digital IO can not always be used for whatever special function you like. I would like to suggest that the Digital IO pins are perhaps the best ones to whittle down first. But this is of course up for discussion along with everything else.

The pins are grouped from top to bottom thus:-

Power
Reset
Datacomms (Bidirectional)
PWM (Output)
Digital IO (Bidirectional)
Analogue to Digital Conversion (Input)
Power

The power connections are doubled up and mirrored top to bottom for the following reasons.

1. To reduce likely-hood of the plug-in boards glitching due to poor power connections.

2. The mirroring top to bottom is specifically to aid in the use of plug-in boards with the minimum number of layers. Having optional power tracks top and bottom of a layout greatly aids this. It helps greatly with prototyping card layout also.

3. Increase the current density available to the plug in cards as they may be driving one or more Axes directly.

Cheers

aka47

PS the digital pins reserved to bring the row count down to 32 could potentially be earmarked for LCD interface and a minimalist (4key) keypad as justification, again this is up for discussion.

---

Necessity hopefully becomes the absentee parent of successfully invented children.

So if you are to make a backplane will it be for the Megarduino?

And maybe we should think about a Sanguino in the Arduino header pin layouts?

Take the following comments as BS (brainstorming), rather than needs/wants:

1. Put a group pins likely to be used together on a single side; RX/TX together, RS485 pins together on same side. This would support using single sided cards more easily.

2. As Bruce implied, I think it would be useful to show profiles for lower IO pin CPUs; AVR based as well as non-AVR, (I.E, similar to what you've already done... this block will be ADC. This block straight digital. This block PWM. Etc.)

3. Maybe consider breaking out smaller 'functional' slots along these same lines. Idea -- the 10 pin sockets for stepper boards on makerbot electronics could provide something like 4 general I/O (first two RX/TX capable if possible), 2 analog, 2 PWM, GND, +5V. This could easily be adapted to drive steppers by sending 12V power using a segregated cable for high power needs (to reduce interference.) Or, allow a completely functional card to be plugged in that used only the first 12 lines on side A to have this kind of minimal configuration; this might argue for a non-bus architecture, where there were a set of shared pins on one side + half the first side, and a set of dedicated pins per io slot on the other half of the first side (for simple small devices.)

Ok on controlers that expose less IO than the buss is pinned out for.

This is by design.

I have puposely put more IO on the Backplane than you need to run a machine. This is so that all controlers will be able to work with it.

The pins are grouped functionaly and organised low order to high order from bottom to top.

Where a controler (arm, diecimilla, sanguino, Xcore, what ever you like) exposes less pins of a functional group than the back plane provides for, these will be connected to the backplane from low to high.

The backplane lines and sockets etc are there whether you connect anythign to them or not. You will only drive the backplane lines that you have enough IO for.

The IO you drive should still though be conected to the back plane using their functional group's. This gives IO cards or axis cards etc the greatest chance of being pin compatible with whatever controler floats your boat.

So if you use a controler that has say only six ADC lines thes will use the bus lines from the ADC grouping, in order, starting from the lowest number upwards.

On 1. pin arangment, it needs to be considered in the light of the notes above. Consider also the groupings are aranged functionaly to help reduce some of the cross-talk. This is why the PWN and Datacoms (fast and noisy) are grouped away from the ADC lines (analogue and suceptible to noise). The digital IO arguably slower and less noisy, or can be aranged such is in the middle to space these conflicting groups. Re single siding boards this is doable however you want to arange the io lines and makes no difference. I can also guarantee that a Pin arangment that works better for board layout for one person will be considered counter productive for another. It is better to acheive a functional consensus. And choose layout along the lines of practical electrical needs. On the serial pins I have purposely put them on opposite sides because if you are making up cables or using reduced pins on the plugin card the pins come in strips that are more peasily cut into pairs across wise. Trying to cut double row pinstrips longditudinaly where there are no scorings is rather dificult and fails more often than not.

On 2. see opening comments on other devices. If this is problematic to visualise we can relabel the pins in a diferent way. The function, rational and arangment is probably better maintined though for all the reasons above. I did expect folk to get hung up about the fact I had chosen one controlers notation to model the design on, In reality it could have been any one at all or none at all. It has been done to illustrate the method only. Hence the emphasis on GUIDE.

On 3. Your plugin cards should have on them only the pins that they need to acheive their function. So your pins on the plugin card would be only those you needed to connect. Virtualy all plugin cards will have the full complement of power and reset pins. WHat else youinstall pins for depends on what you want to connecto to from the back plane. What each developer chooses to use those pins for will be down to what they want to do. I am attemtping to stear clear of what you use the line for and concentrate on it's electrical function. I feel that if we start to lean too much on what the end purpose of the line is used for we will loose any chance of acheiving consensus. (stepper folk will fight with servo folk, heated bed folk will fight cold bed folks, DC motor extruder folk will fight stepper extruder folk and so on ad infinitum) Primarily because each of us wants to use it in very different ways. A Servo Motor Axis plugin card for example will have a very different pin usage than a Stepper Motor Axis plugin card.

Overall valid and useful points to raise. I hope some of the explanations as to what has influenced the design decisions are useful.

aka47

Edited 1 time(s). Last edit at 02/02/2010 05:52PM by aka47.

---

Necessity hopefully becomes the absentee parent of successfully invented children.

Ok guys

If there's not much else to add here's where we will go.

I will put together Kicad designs for a 5 slot backplane (single sided PC and a prototyping card (single sided PC.

Note to self, need to ask sebastien to set up repository access for me.

I will not be selling these. It is a DIY or send the designs to your preferred Board House thing. If someone else wants to sell them fine. Be my guest , just accredit the project and source.

I will be using the pin out in the text file. but will be relabeling the pins according to functional grouping, such that folk don't get hung up on the idea it is only for controler X.

I will be reducing the row count down to 32, by taking away some of the Digital IO lines.

The backplane will be a passive non terminated design. (very simple fine for IO) If you want to terminate anything (eg RS485) put it on your plug-in cards or on whatever your plug-in card eventually leads to.

If You as a designer want to add higher current and or higher voltage feeds onto plug in cards make it part of your design. Avoid putting it on the back plane(personally I would go with something like Hard Drive power connectors or screw terminals on the non back plane end of your board)

The important card/back plane dimensions will be :-

30mm spacing between cards (connectors spaced on 30mm centres), this should be enough for reasonably sized heat sinks and daughter board's (ie an Arduino-U-Like on a controller adapter card)

Card width of approx 10cm (100mm) (space needed for connector, plus a little for tracks to get around the end rows of pins, Plus clearance on the sides of the board for sliding into rails/slots if you rep rap a card frame/case to suit)

Card Length of whatever you want.

All in all I will not start until tomorrow morning. Just in case there is any last minute feedback. Or in case the silence is actually disinterest as opposed to no objections. (if no one wants it I won't wast your time with it, much less mine)

aka47

PS the dimensions are close enough to the Eurocard standard that you should be able to use Eurocard prototype cards, cases, rails frames and accesories with a little bit of a tweak, if you want to.

---

Necessity hopefully becomes the absentee parent of successfully invented children.

Thats annoying the forum software takes PCB followed by a closing bracket ) as PC

---

Necessity hopefully becomes the absentee parent of successfully invented children.

aka47 Wrote:
-------------------------------------------------------
> Thats annoying the forum software takes PCB
> followed by a closing bracket ) as PC



And USB ) becomes US. That threw me for a while too. US? US? Wait.. what? Lol.

What do you guys want to call this version of the electronics interconnect.

I was thinking something along the lines of CEEF or something pronounced like Keith.

AS opposed to jim bob or some such.

Common, Experimental, Elecronics, Framework.

Its not very smooth, clever or insulting though. Anyone think of an acronym that is all three ???

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

The brackets come out fine in the RSS feed so the previous 2 posts about them make no sense until I look at the actual forum. Something I normally only do to post.

---

[www.hydraraptor.blogspot.com]

aka47 Wrote:
-------------------------------------------------------
> What do you guys want to call this version of the
> electronics interconnect.
>
> I was thinking something along the lines of CEEF
> or something pronounced like Keith.
>
> AS opposed to jim bob or some such.
>
> Common, Experimental, Elecronics, Framework.
>
> Its not very smooth, clever or insulting though.
> Anyone think of an acronym that is all three ???

Give it a working name until you decide it is a stable release, then release it with a better final name.

How about "BEEF" -- Back-plane Experimental Electronics Framework; and then you can tell people where to find the beef..

Lol I like that one.

I can't think of an acronym but how about "Prime Porky Worky". It could be insulting enough, but not clever or smooth.

Beef is sort of topical given how much we all argue over stuff... What's the BEEF....

ROFL

Just had word back from Sebastien, there is going to be a slight delay with repository setup cos he is mega busy with some other changes, bout a week.

I will start on the designs anyway in the morning and post up a zip for you guys to have a rummage through if you want to.

Cheers

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

So the plan is to make it piggie back onto an Arduino board.

I think we might want to look into making a Sanguino board that has the Arduino foot print. And also be compatable with the Megadruino.

Bruce

>"So the plan is to make it piggie back..."

I found the name I like - >Piggy Back!

---

B^2 : [replibot.blogspot.com]

~~ We Are The Factory ~~

Bruce re the controllers. You have one of several routes, but principally two.

1. Custom Controler, You make up a completely new board having all of the functionality you want but with dimensions and plug strip that are compatible with the backplane. Plug it in and go. (some might choose to just rework the open source schematics for the board they fancy into the BEEF format)

2. Adapter for Existing Controler, You take an existing off the shelf controller, Diecimilla, Sanguino, Mega.... whatever-u-like, and you either make up a custom single sided board. wait for someone else to do it. Or use the prototyping board I will be designing to patch the shield connections to the back plane connections on the plug strip. Plug it in then go.


When I have completed the basic boards, Backplane, Prototype & Cable Break-out. I could then take a look at ATX PSU board with Power-Good driven reset and dummy load, followed by Arduino Mega adapter board. Unless someone has beaten me to it. Followed by dual axis boards etc. Again unless someone has beaten me to it.

until there are dedicated boards experimenters use the prototype board to roll their own. Maybe the cable breakout board to connect off to bits and pieces that have beenscavanged or bought in, from say the RRRF etc.

The choice is yours.

I expect perhaps in time folk wanting to experiment with other PSU's will make them up to plug in, or Other Micro Boards etc.

The aim is if you want to experiment with for example alternative power supply designs you can just work on that bit. Or axis drivers, or ................

If for example you wanted to replace one or more of your stepper drivers with the pollolu micro-stepper board, you would use the prototyping board to rearrange the connections as you needed and add any glue circuitry you needed and then plug it in.

The objective is to facilitate experimentation with isolated parts of the electronics without having to redesign the whole lot. (like you would have to on a single board design)

You as the designer then can focus on just the bit that you want most, whilst being reasonably sure that it will talk to the other bits that someone else has done

Also to make it easier to take the part you are experimenting on, out and put it back without a lot of connector swapping or undoing screw terminals. Just unplug/slide-out the board you want to work on , work on it then slide it back in to test.

Right now there is some work to do to get it all to that point. I don't mind starting this ball rolling even though it is currently off my planned track. In time when enough folk have done the bit that they wanted the range of plug-in parts will be quite something. Everyone will eventually have the choice of sharing someone else's contribution. (at least electronically)

Instead of being all generation 1 electronics or generation 2 or generation x electronics like now. You will have your own choice of mix and match capability eg Generation 3 Arduino Mega controller board, with Generation 1 Car Battery PSU board, Version aka47 micro stepper board for XYZ axes but generation 10 Nophead board for extruder control. etc etc etc.

Hope this helps/explains.

cheers

aka47

Edited 1 time(s). Last edit at 02/04/2010 08:20PM by aka47.

---

Necessity hopefully becomes the absentee parent of successfully invented children.

"I found the name I like - >Piggy Back!"

Struth, that puts me in mind of deliverance.......

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

I can still hear dueling banjo's

Ok heres the draft schematic for the backplane, board next.

I will make all the development files available as soon as we have somewhere to put them.

Cheers

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

The backplane is a very interesting idea.

I looked through the pinouts but wasn't sure where I would connect a few items.

For example if I have an analog output. I see analog input but no analog output.

Another question is about alternative voltages. Lets say I want to run something, steppers for example, at 15 volts. It would be great to be able to supply that voltage to the pack-plane instead of having to supply it to each device needing it.

-- Chris

hmm i dont know about running each stepper at a different voltage, i think that would make it allot harder. but i would like to see a single power plug on the backplain so that it would be less clutter on the device overall. i am also thinking it might be a good idea to make some sort of power breakout board to feed the different boards the arduino and the backplain and so on,

---

[mike-mack.blogspot.com]

dissidence Wrote:
-------------------------------------------------------
> hmm i dont know about running each stepper at a
> different voltage, i think that would make it
> allot harder. but i would like to see a single
> power plug on the backplain so that it would be
> less clutter on the device overall. i am also
> thinking it might be a good idea to make some sort
> of power breakout board to feed the different
> boards the arduino and the backplain and so on,

With the kind of current the steppers operate at, you're looking at a switch-mode DCDC converter, probably a step-down buck.

If there are benefits to running the steppers at a reduced voltage, then it is feasable, but will add cost. The cost add on depends on the features required.

Secondly, the design of said power supply is critical and will require some analysis. This is where the challenge would lie.

The benefit of this type of circuit is the removal of the need for high current heat sinks over a linear arrangement and better energy efficiency.

@Krafter

The current microcontrolers have no analogue output. If you want analogue output you would construct it on a plugin card and drive the plugin card from the microcontroller plugin card of your choice along the backplane.

You can use serial to analogue devices, (pick one most serial types can be done across the backplane.) or you would use an R2R Ladder and drive it using digital IO lines as IO, Or you would use PWM with a low pass filter to get a DC voltage that is a function of the PWN period.

There are a range of ways you can go, which one is up to you.

@ Dissidence...

I am not sure that I am following your statement re Steppers and voltage, The current steppers are run at 12V which you would pick up from the Back plane. (no change)

They would all run at the same voltage, drive is a function of the stepper driver card you plugged into the backplane and how you drive it from the microcontroller across the backplane.

Hope this helps.....

cheers

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

i was thinking different voltages for one stepper and the same for others, i was thinking that would make it more complex to make the backplain thats more what i was thinking

---

[mike-mack.blogspot.com]

heh um the current chopper turns the stepper winding into the primary inductor of a buck converter, thereby automagically controlling "stepper voltage" as a side effect of current control. Only place that this seems to be missing is on the extruder board where you have to manually set a PWM value because there's no current chopper.

---

-----------------------------------------------
Wooden Mendel
Teacup Firmware

I guess I miss-understood what you were getting at. I thought the idea was a generic backplane so that I can plug in any kind of board I want. Some with stepper drives, some with CPU's, etc, etc.

The voltage thing I mentioned is simply that lets say I want to drive my steppers at 15 volts. The driver chips support up to 35 volts. There is no way for me to feed 15 volts, or any other voltage, into the backplane. All I was suggesting is that it might be nice to have a few pins set aside for alternative voltages.

As far as analog output, yeah, the 644p doesn't have it but you can fake it with the PWM. There is a writeup on the Atmel site about it.

Plus, again I must have misunderstood, I didn't think the goal was to just take the current design and put it into a backplane. I thought you were trying to establish a standard backplane. Kind of like the old S100 bus, so we could plug in whatever we come up with.

-- Chris