Universal Controller v1.3.0

Now that the PowerComms v1.3 board is essentially finished, lets revisit the Universal Controller v1.3 board. Here is the wiki page on its progress so far:

[reprap.org]

Note that all the schematic stuff is marked as done, and none of the layout stuff is. Are there any changes we should make to the schematic before it is laid out? I'm pretty sure that ICSP was the biggest thing we were going to add to this board, so please nothing major, just glaring errors, or simple but major improvements.

Is someone able to do these as a 'green wire' mod to a v1.2.1 board? i'd be willing to ship you a free PCB to do it on. =)

i've attached the most recent version of the board as a .zip file... if you have any changes, feel free to use those files and repost here.

oops, forgot to attach the file.

I lost track of exactly which ICSP mod was agreed on, but I remember it had at least four or five components. This looks like an earlier version, Adrian or nophead should know the fix.

Is there anything stopping us from replacing the TIP110 transistors?

They will drop up to 2.5V (and dissipate 5Watts if run at their full current rating). This generates excess heat and limits the power to the heater.

IN the same package you can get a
[www.mouser.com]

This mosfet that would only drop 68mV and dissipate 136mW. In addition you could potentially drive 35! amps through this part.

A more generic question: Is the electronics design trying to stay away from surface mount parts at this stage? Are there component guidelines?

Edited 1 time(s). Last edit at 09/06/2007 04:16PM by d0ubled.

d0ubled Wrote:
-------------------------------------------------------
> Is there anything stopping us from replacing the
> TIP110 transistors?

i'm not sure... but i dont have the technical prowess to weight in on that.

>
> This mosfet that would only drop 68mV and
> dissipate 136mW. In addition you could
> potentially drive 35! amps through this part.
>

awesome! those seem very superior to the TIP110's, and at a decent price too still. does anyone have any objections to this change?

also, does this change require re-wiring at all? or are they completely drop-in replacements?

> A more generic question: Is the electronics
> design trying to stay away from surface mount
> parts at this stage? Are there component
> guidelines?

there are no hard rules, but we're generally trying to stay away from SMD components. from the way things look, people are going to be assembling their own RepRap boards for a while, so we'd like to make that process as easy as possible. here are the guidelines i've been using:

1. easy
2. reliable
3. cheap

(in essentially that order)

d0ubled Wrote:
> A more generic question: Is the electronics
> design trying to stay away from surface mount
> parts at this stage? Are there component
> guidelines?

Please let's stay away from surface mount stuff for a while. So far there doesn't seem to be any compelling reason to switch to surface mount components (except, I guess, the magnetic quadrature encoders for Mendel if they work out). We've already seen reports of a few soldering errors, things could only get worse if we switched to surface mount.

As long as the components are widely available and fairly robust, I'd think they'd be ok. There have been some discussions of switching to less antiquated stepper driver chips, more powerful microcontrollers, fancier connectors... I don't think anything is set in stone, but there's a desire not to make too many unnecessary changes in the 1.x boards.

4A x (1 extruder/3 stepper) ~= 16 amps

It is good to have some breathing/growth room so I would aim for 50% greater current capacity than our best estimate (ouch 32 amps is quite a bit (10 gauge) hopefully the estimate is off by a bit).

[www.geocities.com]

The 4 pin mate-n-lok connectors used on the PC power supplies I believe are rated at a decent current. Aren't they already used in the design? Incidentally why are we not using 5V from modified PC power supplies (or are modified PC supplies not the intended voltage source?)?

I would be nice to replace the Darlingtons with FETs and get rid of the heatsinks. However, those FETs only spec the RDSon for 10V VGS. That would require another transistor for a driver. I prefer to use logic drive FETS with ESD, thermal, over current and over voltage protection built in. They are more expensive but give a single component solution to driving inductive and resistive loads.

I think moving from Darlingtons, which are pretty robust, to unprotected FETs, which are easily destroyed, would be a bad idea for this project.

To use the 5V rail you would have to bus it round to the other boards. Lots of scope for people destroying things with a bad connection, e.g. a missing ground would put -7V across the PIC. Noise and ground bounce due to the motor currents would get onto the PIC 5V rail.

Much better to provide a locally regulatd clean supply, and at the current involved the cheapest linear regulator suffices, so it's not expensive.

---

[www.hydraraptor.blogspot.com]

d0ubled Wrote:
-------------------------------------------------------
> 4A x (1 extruder/3 stepper) ~= 16 amps

your math is right, but your logic may be wrong. that would be the full/max requirements going *into* the powercomms board, but going out of the powercomms to each individual board, the wires only need to carry the 4A.

> The 4 pin mate-n-lok connectors used on the PC
> power supplies I believe are rated at a decent
> current. Aren't they already used in the design?
> Incidentally why are we not using 5V from modified
> PC power supplies (or are modified PC supplies not
> the intended voltage source?)?

i think that this bears some consideration. i've brought it up before, but people werent as receptive as before. here's the deal: we're already using a PC power supply to power the powercomms board, and we're only using one of the many power connectors it provides.

why not simply bypass the powercomms board and power each board directly off the PC power supply? the main reason i remember was that we're aiming to eventually have reprap be powered off car batteries. seeing as how reprap is going to be powered off cheap, $20 computer PSU's for quite a while, i think this might be a good change to make. it uses our power supplies more efficiently, adds an extra layer of safety with keyed connectors, and reduces the need for making more cables.

of course if we're going to go that route, then the power portion of powercomms would essentially be obsolete, so perhaps we're better off delaying this change for a v2.0 or later. it would offer quite a simplification to the design though, which is nice.

True only 16 amps at the input to the powercomms. But what is the molex connector rated to? Looks like 10 amps. We should add a second one if we believe 16A steady state is expected or predicted.

Well the simple step would be to use the same (male) molex connectors for all 12V board inputs. 12V outputs from the powercomms board would be (female) molexes. Then you can use of the shelf PC internal power extension cables for interconnects. (or build your own). They (connectors) are cheap and used in PCs all over. Also you can be flexible and either use the Powercomms card as power distribution or use individual PC power connectors directly from the supply.

Off the shelf internal pc power cable.
[www.cyberguys.com]

The Fet was a quick example of what is possible vs Darlingtons. If the current drive capabilies of the darlingtons is more than enough for the heater (and has been tested or is being used) then maybe no change until we have need to optimize the design. If more current drive would provide some nice flexibility during integration then we should move to a Fet.

Here is a better example. 16Amp rating. Threshold voltage 1-2.5V. 60V drain source breakdown. With a 0.1uF cap on the output should be safe against ESD. Make sure we leave the gate drive resistor. That leaves a short circuit as the danger to this part, if there is a nice integrated solution as nophead mentioned that would be worthwhile for sheer robustness.
[www.mouser.com]

The parts I have used on my machine are BTS 134D Infineon HITFETs [www.infineon.com]
mainly because I have a bag of samples I saved for 10 years. They don't need gate resistors or back emf diodes. I have not found a source for these though.

Edited 1 time(s). Last edit at 09/06/2007 02:56PM by nophead.

---

[www.hydraraptor.blogspot.com]

[www.mouser.com]

or

[www.mouser.com]

look to be some potentially nice parts. Though the short circuit protect does not kick in until 12 and 10 amps respectively.

Also I am using mouser to estimate price and availability. Since you are in the UK I am assuming you would check the availability of any potential parts on your side of the pond.

[edit] all of this assumes that it is worthwhile finding a part that has a higher current capability (or gets rid of expensive heat sinks) than the Darlingtons. By the time we get to working darwin prototypes the requirements may have taken us a different direction. Does anyone have any numbers on the current draw of the existing heaters? Would development and testing benefit from a higher power transistor?

Edited 1 time(s). Last edit at 09/06/2007 03:58PM by d0ubled.

umm... i'm pretty sure we're only drawing 1-2 amps for the heater.

Yes the latest heater spec is 6R which would be 2A when driven by a MOSFET, a bit less with a Darlington.

On the subject of current, with the standard motors and firmware I think we are only about 0.8A per motor as only one phase is on at a time. I only mentioned 4A as that is the maximum the chip can handle if people want to use other motors. Not all PC power supplies would be able to power such a machine, the one I have in front of me is 13A max on the 12V rail.

A nice find on the ST FETs. The second one has a highish RDSon so will get a bit warm but still probably OK without a heatsink.

I can't find them in RS, Farnell or Digikey though but we do buy stuff from Mouser over here.

---

[www.hydraraptor.blogspot.com]

Since 2A is the "Absolute Maximum Rating" for a TIP110 I would suggest using the aforementioned VNP14NV04 as a replacement. Running a component at or near its maximum is not good for reliability or potential growth in power requirements.

Otherwise I would use two TIP110 in parallel.

link:
[www.mouser.com]

Yes looks like a very nice device, a bit more cost than TIP110 but you save a heatsink and it is indestructible.

It should be a drop in replacement, you can either leave the resistor in series with the gate or replace it with a link.

BTW, rather than two TIP110, the wiki recommends a TIP120 for 4A.

---

[www.hydraraptor.blogspot.com]

While I have no doubt that using a FET instead of the TIP110 would improve the board's performance (getting rid of a heat sink) I find myself wondering why we would want to fix something that isn't really broken at this point?

There are a lot more pressing issues with getting Darwin working that would, imo, have a much higher priority than making parts swaps on the controller boards that don't fix something that isn't working.

There will be plenty of changes like this swap that will be well-advised for the next generation of controller boards. I know with Tommelise, I am going to swap out the 18F4610 for one that lets me do direct USB comms with the PC. That will get rid of my whole comms card and will greatly simplify my board design. That's not something that I intend to do any time soon, though. I always try to remember the old maxim, "if it ain't broke, don't fix it. "

No it isn't broken, indeed 20 years ago that would have been the way I would have done it. The only real advantage is that the FET solution is slightly easier to construct. It just seems a bit odd from my point of view to using such out of date electronics technology on a project which is otherwise pioneering. Or is it just that FDM is new to me and is infact old hat to an expert in fabrication?

---

[www.hydraraptor.blogspot.com]

nophead Wrote:
-------------------------------------------------------
> It just seems a bit
> odd from my point of view to using such out of
> date electronics technology on a project which is
> otherwise pioneering.

It's not so odd when you stop to think that the skill sets that more than a few of us are trying to apply to this project are probably a bit past their "best before" date.

Had you been around the project when the present board set got designed about a year ago the stepper controller boards would probably look considerably different than they do now. I doubt, for example, that Darwin would be using belts had we had the benefit of your insights on controlling steppers back then.

Darwin, like any design, is the net result of a lot of compromises and limitations that were dictated by both the nature of the original vision of the project and the skills that the core team members had at their disposal at the time it was getting designed. While the vision may have been flawed and the skills less than totally up-to-date that doesn't invalidate the design. There is a lot to be learned from the Darwin design experience that will make the several next-generation machines a whole lot better.

At present, we're nowhere near the end of that design experience, though. If Darwin can be made to work at all, it will be a tremendous advance in the state-of-the-art.

Even if it can't, the experience of having tried will stand us in good stead when we have another run at the problem.

Edited 1 time(s). Last edit at 09/07/2007 11:20AM by Forrest Higgs.

Well should I continue to suggest design improvements in the areas I am knowledgeable about or just keep for my own machine? I can certainly progress that faster if I stay out of these forums.

---

[www.hydraraptor.blogspot.com]

I'd personally prefer it if you kept on with your forum postings. I've learned a hell of a lot just reading your postings. I sure as hell wish you'd been around here a year ago. Life would have been a LOT simpler, for me at least.

One thing to keep in mind is that there are a lot of people who post and read here who aren't just trying to build Kosher Darwin machines. Darwin is A focus of the reprap project, not THE total focus. It isn't even close to being the only thing going on here, as I'm sure you are aware. The forums serve as a place where people exchange experience and tips on how to get things done. Your postings in that regard are extremely valuable.

It may be best to keep in mind that Darwin is kind of under the gun right now. If the core team doesn't go ahead and get it finished it stands a good chance of becoming irrelevant if some other design gets to working first. That's a real possibility. It's easy to make a working machine better. It has to be working first, though. So far, however, Darwin isn't working, though given Adrian's obvious genius I have no doubt that it eventually will.

Edited 1 time(s). Last edit at 09/07/2007 11:51AM by Forrest Higgs.

If the heater driver works then so be it, there is value in not changing it now. If it is just barely meeting our needs then at the very least we should put a TIP120 in the same footprint and leave some room for experimentation with different heater coils as well as improving the reliability of the component by running it at less than it's full (or less than 83%) current capacity. Perhaps at the same time we can have some alternates listed for those who might want to experiment. The VNP14NV04 is pin compatible with both TIP parts.

Still the TIP series darlingtons don't have short circuit protection. But that can be a problem for another day.

nophead Wrote:
> Well should I continue to suggest design
> improvements in the areas I am knowledgeable about
> or just keep for my own machine?

Please, keep 'em coming. I've learned a lot by reading your posts here and on Forrest's Blog.

yes, nophead... please keep posting. you've been very valuable, and your suggestions are great. seriously man, the v.1.3.0 board would not be able to go off without you.

i really hope that forrests gruff, but well intentioned posting hasn't scared you off. we need you man!!!

since the MOSFET suggestion is a drop-in replacement, we can make a note on the wiki, do some testing, and once the dust has cleared on the new version of the board, perhaps change the BOM to use the new transistor. its not actually a mod to the board, so that makes things much simpler. its a good suggestion, and my guess is we'll adopt it in due time.

I have a very minor request:

If the screw terminal (power) was just a mm or two farther away from the mounting hole, it'd be a lot easier to fit a screw in there (at least, the oversized screws that I happen to have on hand). As it is, I had to grind the screw head down.