Universal Controller v1.3.0

Hey all,

This post is mostly similar to the PowerComms v1.3 post, but I'll detail the changes that should be made to this board for v1.3

* diagnostic LED's for Rx/Tx
* switch to RJ-45 plugs for Rx/Tx
* anything else?

like on powercomms, if someone could either do the LED's in kicad, or even draw me up a schematic, i think i can handle it myself pretty easily. i just dont want to mess it up.

(1) Diag LEDs for min and max too, please.

(2) Check that silkscreen component names (especially for connectors and LEDs) can be seen when the board is populated.

(3) See my comments about RJ-45s in the PowerComms 1.3.0 thead.

Jonathan

Can we simplify a bit to remove some of the optional components? I'd think R7 and R8 have outlived their usefulness, anyone who was depending on their existence for an alternate min/max sensor has probably made their boards already.

If we're going blinky, we could (optionally) build the stepper tester in. Two 3mm bi-color LEDs and two resistors don't take a lot of space.

I'd like ICSP stuff. One diode, and possibly a jumper or switch to disconnect the L298 if that ends up being my problem. I'll try to debug that this weekend.

I remember seeing some weird stuff when converting to Eagle. I think the pin numbering on the transistors was wrong, so pin 3 went with the square pad, but I could be remembering wrong.

I'm not entirely sold on using RJ-45 (but that's ok, I'm making my own boards anyway :-) It just seems like an awfully big and inflexible connector to use when we're only really using one conductor. Pre-made cables might be worth the pain, though...

[edit] Someone (englewood?) had mentioned that the PIC's capacitor could be moved a bit closer to the power pin, not sure if that applies to the current rev of the board or not.

Edited 1 time(s). Last edit at 08/02/2007 03:35PM by emf.

I think some people have reported the connector holes being a bit tight.

---

[www.hydraraptor.blogspot.com]

jmarsden Wrote:
-------------------------------------------------------
> (1) Diag LEDs for min and max too, please.

good idea, lets add that to v1.3. like i said, i can lay out a PCB for production, but i cant do electronics design... if someone makes a drawing or actually does it in kicad, then we can get it done =)

lights on means a limit has been reached?

> (2) Check that silkscreen component names
> (especially for connectors and LEDs) can be seen
> when the board is populated.

most definitely.

emf Wrote:
-------------------------------------------------------
> Can we simplify a bit to remove some of the
> optional components? I'd think R7 and R8 have
> outlived their usefulness, anyone who was
> depending on their existence for an alternate
> min/max sensor has probably made their boards
> already.

i agree. if someone absolutely has to use non-standard ones, they can probably go the extra length to solder in their own resistor, and cut the traces. lets get rid of those meaningless resistors.

> If we're going blinky, we could (optionally) build
> the stepper tester in. Two 3mm bi-color LEDs and
> two resistors don't take a lot of space.

i'm a fan of that. as long as the LED diagnostic stuff is optional, then theres no reason not to include it on the PCB. at the levels we're doing board buys, the extra square inch or two wont effect price much.

> I'd like ICSP stuff. One diode, and possibly a
> jumper or switch to disconnect the L298 if that
> ends up being my problem. I'll try to debug that
> this weekend.

that would be fantastic, if you can get that stuff working in the next week, i can see it getting into v1.3. otherwise, i think it should absolutely be part of v1.4.

> * switch to RJ-45 plugs for Rx/Tx

Why? Much more complicated, and no advantage that I can see. Remember that _none_ of the signals go to and from the same place; they go in a ring (i.e. just one wire between each board and the next). And, of course, it only takes ten seconds to solder the end of a single wire to make a fully insulated pin socket for an 2.54mm pin using a shell socket and a bit of heat shrink.

We must get 4 screw holes on a rectangle in the corners, not just 3.

Can we have all the LEDs down one front edge so they can be easily seen, especially when the boards are being used as extruder controllers?

Building in the stepper tester: if you make the external stepper tester and you have X, Y, and Z controllers you can test them all without buying 12 LEDs and 6 resistors...

---

best wishes

Adrian

[reprap.org]
[reprapltd.com]

Adrian Bowyer Wrote:
-------------------------------------------------------
> > * switch to RJ-45 plugs for Rx/Tx
>
> Why? Much more complicated, and no advantage that
> I can see. Remember that _none_ of the signals go
> to and from the same place; they go in a ring
> (i.e. just one wire between each board and the
> next). And, of course, it only takes ten seconds
> to solder the end of a single wire to make a fully
> insulated pin socket for an 2.54mm pin using a
> shell socket and a bit of heat shrink.

I'm going to have to respectfully disagree with you Adrian. The advantages seem obvious to me, and maybe if I repeat them enough, you'll agree too Also, you may be misunderstanding what I'm proposing. We will still be using the token ring, everything of that nature is staying the same. the ONLY thing that is changing is the connection method BETWEEN boards. Using premade cables saves alot of time, alot of frustration, and solves lots of problems. If you have to spend 2 hours making all the cables for your reprap machine, thats 2 less hours you have to do something useful. For a hobbyist like me, thats basically an entire day wasted. Also, you only have to mess up one connector to make debugging the thing a nightmare (read the other posts in this forum if you dont believe me)

Let me break it down by advantage:

Cost: hand-made (by about $0.50)
Time: RJ11
Ease of use / simplicity: RJ11
Debugging: RJ11
Reliability: RJ11

Factoring in time to the cost, RJ11 wins there as well.

The .100 header pins are GREAT for prototyping, but we're beyond the prototyping stage, and we need to start think about making this easier for your average user, not just pros like you. (thats a compliment :^)

> We must get 4 screw holes on a rectangle in the
> corners, not just 3.

already fixed in v1.2.1

> Can we have all the LEDs down one front edge so
> they can be easily seen, especially when the
> boards are being used as extruder controllers?

sounds reasonable. i'm probably going to use smaller traces, as nobody is making these by hand. that will make laying out the board MUCH easier.

> Building in the stepper tester: if you make the
> external stepper tester and you have X, Y, and Z
> controllers you can test them all without buying
> 12 LEDs and 6 resistors...

yeah, i agree... it might make sense to add a spot for a male header on the stepper tester so you could have both plugged in at the same time. eg UCB -> stepper tester -> stepper motor

2 hours? We are talking about 5 one pin cables so that is 12 minutes per connection. I can see why it would take more than a minute per pin which is 10 minutes in total. Double it for somebody with no experience.

---

[www.hydraraptor.blogspot.com]

Zach,

Are you seriously suggesting replacing *all* wiring in Darwin with pre-made RJ-11 or RJ-45 wiring??? Power wiring too? Stepper motor connections?? Extruder connections?? Endstop wiring?? If not, saving the 2 hours you state as a result of your proposal to change just SNAP serial wiring seems *wildly* pessimistic.

For the stage most of those reporting the issue we are trying to address have reached (one PowerComm and one UCB only, Test #3), the total wiring needed is:

(a) An RS232 cable. Your proposal leaves this unchanged.
(b) A Powercomm T to UCB R cable, single wire, two crimps.
(c) A UCB T to Powercomm R cable, single wire, two crimps.
(d) 12V DC power and ground from PowerComms to UCB. Two wires, one crimp each, one bare end each. Extra time required to check you didn't wire them backwards. Your proposal leaves this unchanged (I think and hope!).
(e) A 4pin Molex from a PSU to the Powercomms board. Usually the lead is already on your PSU, so no construction needed. Your proposal leaves this unchanged, too.
(f) A sync loopback jumper. Single short wire, two crimps. Your proposal leaves this unchanged.
(g) Two jumpers across max and min sensor connectors. Your proposal leaves this unchanged. Hard to know if these qualify as "wiring", but included for completeness.

So, the wannabee RepRap constructor still requires sufficient crimping skills for the remaining non-premade connections. Whatever learning curve the constructor needs to gain such skills has not been eliminated or shortened in any way by your proposal. Therefore, we're looking at going from creating 5 pieces of wire with a total of 8 crimp pins, to "only" 3 pieces of wire and 4 crimp pins, plus two premade cables.

Saving what? Your proposal saves exactly two cuts of a length of wire off the spool, and the stripping and crimping of 4 wire ends, to reach Test #3. How does this take anyone two hours???

I was very slow and very careful and double-checked stuff a lot (like reading schematics and then following board traces to check which pin of the connectors was T or R and which was G, even, since my UCB had no silk screen and I knew one connector was reversed!), and I *know* the wiring didn't take me anything like that long to get to this (one PowerComm and one UCB ) stage.

Two hours to stuff the two boards, including making sure components were inserted right way around and in the right holes, and soldering, and visually checking each joint looks decent, sure. Maybe even a bit more than two hours, if you are being really really careful, looking up some components online to double-check which lead of a LED is the anode (Yes, I admit it, I did that. I really am not any kind of electronics expert!), etc. as you go. But your proposal doesn't simplify component picking and placement or soldering, so *that* time is not saved.

My estimate: at most, this proposed change will save a beginner 10 minutes out of the 2 or 3 hours of actual construction time it could take a beginner to reach Test #3 on their first UCB.

Jonathan

"Power wiring too?"

No you need thicker wire for the power.

---

[www.hydraraptor.blogspot.com]

I think the answer is no all along. I just wanted confirmation that Zach's proposal is restricted to changing the SNAP Tx and Rx wires only.

Theoretically (STRANGE THOUGHT EXPERIMENT ONLY!), you could maybe use all 8 wires in a Cat 5 cable and get enough power that way (shudder!). Two RJ-45s, so 8 wires for +12V and 8 for ground? :-) I've never measured the current-carrying capacity of Cat-5 cables! We need what, 2 amps per stepper, so would each wire in a Cat5 cable handle 250mA? I think it might be possible, if you were weird enough to try it! This would probably make a strange but effective WMD (Wire of Motherboard Destruction!) if you plugged such a "power port" into the RJ-45 NIC on your PC motherboard by mistake!

Jonathan

> the ONLY thing that is changing is the connection method BETWEEN boards

But that's just one (1) wire each! Are you seriously suggesting a fancy RJ11 ID connector for a single wire?

I agree with Jonathan...

---

best wishes

Adrian

[reprap.org]
[reprapltd.com]

ZachHoeken Wrote:
> > I'd like ICSP stuff.
>
> that would be fantastic, if you can get that stuff
> working in the next week, i can see it getting
> into v1.3. otherwise, i think it should
> absolutely be part of v1.4.

Running behind schedule, as usual... I spent quite a few hours debugging my setup and it turns out that adding ICSP the obvious way works fine, no switches needed. I had to increase my delays a bit to get my JDM programmer to work with ICSP properly. I also had to kick myself for not checking the wiring on my ICD2's ICSP connector -- it's wired in the opposite polarity from normal for no good reason. And I'm sure there were a few other stupid things I did along the way that made it seem harder than it actually was.

Anyway, I'll email the Eagle files I used tomorrow for reference (or PDFs if you'd rather). It just amounts to adding a diode to VPP and re-arranging the expansion header a bit to fit the usual ICSP configuration.

AWESOME!

i'd love to get this into v1.3. this is a very useful and excellent addition to the board.

i'd prefer PDF as I'm running OSX. thanks very much for your contribution!

What news of 1.3? I'm torn between ordering the last 1.2.1 board in stock from the RRRF, and waiting for 1.3. The biggest advantage of 1.3 that I see is the ICSP stuff (and it sounds like there's a good chance that'll make it in?), but pre-made cables are a big plus too. So Zach, and/or anyone else involved: about how long are we looking at waiting for the next version? If it's just one or two weeks I'll wait it out, but if we're looking at one or two months I'd like to get it while I can. Thanks.

Kyle

Kyle:

Its probably going to be at least a month until we have them for sale on the store. Here's our current status:

* We have to finish the schematics (a week)
* We have to lay out the PCB (a week)
* We have to order a test-batch (10 days)
* We have to test that batch to make sure the design is correct (???)
* We then order a batch for the RRRF (10 days)

assuming everything goes flawlessly, we're still looking at a decent wait.

i have a new order of 1.2.1 boards for the store coming soon, and those will be in supply until they run out or we get 1.3.0 boards.

I've just made up 4 Universal boards in a one-man assembly line (up to the power test point) and the only comment I have to make is to take back a comment I made in the past about the size of the holes for the 2 pin and 3 pin header. I had previously said they were too small (actually its literally because its square pegs, round holes). Now that I'm used to the amount of force needed I find it a lot easier as I can put all the headers in at once and they will quite happily sit there so its very easy to flip the board and solder them all in at once.

So, my vote is now to keep the size of those holes as is but put notes in the correct places about needing a bit of force to get them in.

PCB holes have a tolerance so you need a bit of clearance otherwise you may get a batch of boards that are very hard to press the headers in.

There is a technique called pinning, where you don't have a header body, you just ram square pins into round holes, but that requires big a machine. Hole tolerance becomes a big issue then because if they are too big the fall out and if they are too small they bend or crack the PCB.

---

[www.hydraraptor.blogspot.com]

cool. i think i'll be working on the boards today, so i'll *slightly* increase the size. but not by very much... i also like the tight fit. it makes soldering easier.