UCB Construction and Testing - catch 22?

Yes, I'm finally building a UCB :-)

I tried very hard to follow the instructions at [reprap.org] the UniversalControllerBoard_1_2 Wiki page. All went well until I found the stepper testing stuff did not work, lots of sending error, retrying messages.

I went to use poke (the instructions don't suggest poke for some reason?). This works... once! After that the UCB (with an X-axis PIC in it) NAKs everything. Power cycling the UCB solves this -- for one command!

OK, having read the firmware source code recently trying to help Eric out, I knew exactly what this was -- the PIC is sitting around waiting for a SYNC edge pulse. OK, just jumper P3, I said to myself.

Wait a minute! There *is* no P3 at this stage of construction!

It seems that, to test a UCB as the wiki page suggests, it is necessary to add R10 and R11 and P3, and then jumper P3. After that, the tests all work as expected.

Questions arising:

Q1: Why aren't we using poke to test the UCB?

Q2: Why do the test instructions "not notice" the lack of the sync line -- am I really the first person to test them?? Did whoever wrote them fail to test them? Or at the time of testing did the sync line not exist?

Q3: What is the best way to amend the wiki page to overcome this -- adding R10 and R11 and P3 is fine for a UCB destined to become a stepper controller, but is undesirable for a UCB destined to become an extruder controller.

The good news is... have a working UCB :-)

Jonathan

P.S. Having a schematic for the board easily visible on the Wiki page describing its construction and testing would make this sort of troubleshooting a lot easier. Is there some reason we are not displaying the schematic along with all the pretty pic of partially constructed boards?

jmarsden Wrote:
-------------------------------------------------------
> Yes, I'm finally building a UCB :-)

yay!

>
> Q1: Why aren't we using poke to test the UCB?

i wasnt familiar with it when i went over the instructions. it would be a good idea to add this, or add a firmware that is a generic universal tester firmware.

> Q2: Why do the test instructions "not notice" the
> lack of the sync line -- am I really the first
> person to test them?? Did whoever wrote them fail
> to test them? Or at the time of testing did the
> sync line not exist?

you may be the first person to follow them exactly step by step. i did the testing stuff, but it was usually on a finished board. you just found some documentation bugs =)

> Q3: What is the best way to amend the wiki page to
> overcome this -- adding R10 and R11 and P3 is fine
> for a UCB destined to become a stepper controller,
> but is undesirable for a UCB destined to become an
> extruder controller.

i would say for the time being, post instructions on using poke. it would be nice to have a UCB tester firmware and a corresponding test area in the host software that will test the various things it can do. this will be much easier once we have ICSP on the board so you can program the chip to X axis, extruder, etc for real once you complete it.

> The good news is... have a working UCB :-)
>
> Jonathan
>
> P.S. Having a schematic for the board easily
> visible on the Wiki page describing its
> construction and testing would make this sort of
> troubleshooting a lot easier. Is there some
> reason we are not displaying the schematic along
> with all the pretty pic of partially constructed
> boards?

thats a good idea. its not too hard to add (and used to be there), it was just a question of time and priority.

I think Jonathan may very well be at least one of the first to follow everything step by step - when I made my boards, I had to go between the 1.1 and 1.2 instructions because the 1.2 stuff was in the middle of construction and not complete. I think if you review the 1.1 instructions, there was a good bit more testing info in there (including some that were erroneous - tests for the previous board/firmware) that should be added to the 1.2 instructions.

OK. Even poke will only work *once* without R10/R11/P3 and a P3 jumper in place. Once per power-cycle of the UCB, that is.

Actually I think a high value pulldown resistor on the Port A1 pin of the PIC might solve this, since the firmware only enables sync if it sees SYNCA high. So that input pin must be floating high when unconnected.

If someone more expert (Simon?) could confirm that, it would be good to add say a 100K pulldown resistor there, or else tweak the firmware to not enable sync unless it can first verify that there really is another PIC out there hooked to the SYNC line?

Meanwhile, I think the instructions probably need to have the user do the one poke to verify comms to the UCB, and then say "in order to test further, you need to start making this board a Stepper Controller Board, add R10, R11, P3 and a jumper across P3" before they can do any more poking or run the stepper exerciser? Does that make sense? Also, if you add those 2 resistors and a 2 pin header, test further, and then make the board into an Extruder controller, would you have to desolder the resistors, or would they just go unused?

When my son gets off the one Windows PC in the house that has a serial port, I'll hook up my Reprap electronics bundle over there, and see what I can do about a poke or poke-equivalent test program.

Now I need someone to discover a great RepRap compatible stepper motor available cheap here in the USA, so I can hook it up and watch it spin! Or I could get a really cheap one just for playing with, I suppose, even if it is too puny or too imprecise for actual RepRap use.

Jonathan

i've had success with both a Jameco stepper, as well as an AutomationDirect stepper. i'll probably add them to the BOM later today, even though they are 200 step motors rather than 400 step motors.

here are the links:

[web1.automationdirect.com]

jameco #: 237577

Johnathan, I have a testing motor for you. Brand new surplus I bought a while ago. Vexta Px243-03BA 12VDC .I even have the manual for it. It can be wired up for bipolar for testing purposes. I will give it to you for free, just pay for the shipping. What is your address? This isn't a NEMA 23, but it is great for testing purposes

Johnathon, give me an address or P.O. box or something and I'll send you a free 12VDC 6 lead stepper for testing. It is a new VEXTA stepper and I have the manual sheet with it(In Japanese and English). I put it in a box. You will need a 4 hole shell, I'll put crimp on female pins for you, Hell I'll crimp them on before I ship it! I can't find anymore 4 pin shells around my house. I'll put some heat shrink around the crimped pins and you can slide it down as you put the female pins on the 4 male pins. It has 6 leads, (4 coil ends and two center taps) I'll write a color sequence for pin 1 to 4. As long as the sequence is right, you can go 4 to 1 or 1 to 4, it will just go reverse.

I crimped female connector pins and I put heat shrink you can slide down to protect the pins from shorting, if you can't find a 4 pin shell. I also marked the sequence on the instruction sheet and I verified the operation of the motor just now. It spins great!

yes! i love the camaraderie and helpful spirit of reprappers. ... and i'm not even getting anything free =)

I found a 7 hole shell and put the 4 female pins in it. Boy is it ever keyed, you can only plug this in one way, that will work! Let me know where to send it

Englewood,

Thanks. My mailing address is in a PM to you here on the forums.

Now I'd better add the L298N and the 4pin connector and diodes to my UCB, so I have a completed Stepper Controller Board ready for the motor. Since it is 12V, I won't need any series resistors, which is good :-)

I might be able to do some sort of RepRap presentation / electronics demo at my local LUG (Linux User Group) once I have something that moves. It's a very small LUG, and I know they'd be happy to get a presentation out of me!

Jonathan

God don't buy a L298N! Go to www.st.com and sample several L298HN. Just make up a company name, tell them it for a new project and they will send you up to three free, I then got my other two by sampling again They are bent horizontal. solder in the 15 pins, then bend it vertical and attach the heat sink. I see if I have a spare heat sink lying around and I'll throw it in the box, I'll puts some heat sink compound on it and a 4-40 nut and screw. I also threw in a thermistor that use 1.99 microfarad cap( I used 2 X 1 microfarad in parallel). I measured the resistance and used a digital thermometer. I then heated up my oven. I have an oven from the 50's , but I installed a Watlow digital controller, J type thermocouple and a Solid state relay for heater control, my pies come out PERFECT! I ran the temp up until it steadied and confirmed the temp with the digital thermometer. They agreed to within one degree F. I then measured my new resistance and calculated beta. I'll send you the data later and you can double check. I also threw in 9 diodes. Mendelson's is moving out, so they are trying to all the electronic junk they have, it will take forever!

> God don't buy a L298N!

My sentiments exactly! Nobody uses darlingtons anymore. They switch too slow (2uS!) and have a very bad voltage drop (3-5V!!!). And that means you have to muck about with big heatsinks.

Use a nice NMOS driver instead, Allegro has some very good ones for instance. And since you don't need a heatsink anymore, they are cheaper too.

--Blerik

Thanks for all the help! Actually, I already have one L298N for this board (I ordered a set of parts for one PowerComms board and one UCB/Stepper Controller from Mouser a while back, both as a test of the BOM and so I would have some electronics to test with).

If there is an already tested way to replace the L298N on an RRRF-supplied UCB with a cheaper/heatsinkless equivalent, and the firmware and host code need no changes to accommodate this substitution -- great! Please provide specifics on what part to use instead. If the modifications you are describing make my electronics "non-standard" in terms of being able to test the "official" Reprap project host software and firmware, then I'll need to avoid them in order to be able to test "the real thing" rather than a modified variant of it.

Most people seem to want to build a Reprap to make stuff. Which is cool, but not my main current motivation! I want enough of a RepRap to be able to test our build instructions, and test the software and firmware I build, at least for now! So staying "standard" as far as all software/firmware interfaces are concerned is more important to me than it may be for some others. Even if it means using "old fashioned" L298N drivers and big heatsinks.

My current plan is to wait on more electronics purchases until the new BOM Generator has settled down a bit and has some useful cheap US mail order suppliers in it for electronics, then pick one and buy enough parts to stuff at least a couple more Stepper Controller boards. Maybe the 3 opto endstop boards too. An Extruder is not going to be much use until I have some sort of mechanical robot to move it around with, so I'm likely to delay constructing an extruder control board for some time, unless I can find a source for RP-ed parts at a local university or similar.

Jonathan

If the part needs to fit on the v1.2 board, the previously mentioned L298HN might be your best bet. The SLA706xM series are about the same footprint, but are driven completely different so that is not an option for you even though it is a very nice part.

If you also cannot find an L298HN, search octopart.com for a dual full bridge driver, you should be able to find some alternatives...

--Blerik

Can you recommend a supplier for the Allegro parts? I've read good things about their chips and I've been meaning to tack one onto my next order, but I haven't run across anyone that carries them.

try [www.octopart.com] and search for the chip you're looking for

www.icmaster.com I have an old 3 volume IC mater catalog. MOS bridges have less voltage drop across the source drain but they need a voltage driver and they would still need the eight free wheeling diodes to protect the transistors from reverse voltage, (VL=-Ldi/dt)

Yes, using just mosfets is more trouble than using the L298N's. But that would be silly, as there are lots of integrated solutions out there. Both STMicro and Allegro (and lots of others) have fully integrated MOS drivers with current chopping, smart decaying (so no diodes anymore), simple step/dir interfacing and lots more features. And the parts cost only a couple of dollars each. For instance, look at the Allegro A3967. Up to 30V and 0.75A per coil, no heatsink needed (even though it isn't mosfet based), and about $3. Sparkfun sells these fully assembled as the EasyDriver v3 for $15. They are so simple to use, you could control them directly from the parallel port of your pc using emc2. If the RepRap host software could print gcode that would be everything you need, but alas...

--Blerik

(Yes, I am currently looking at outputting gcode from the RepRap host software but that could take a little while... lots of other stuff to do as well)

cool, it sounds like we have lots of room for improvement on our motor circuits. i'm not sure how long until we are ready to do a v2.0 of the boards, but we'll definitely want to solicit community input on the best type of circuit to use.

If you're thinking of changing the motor circuits around, you may also consider the Sherline stepper motors (http://www.sherlinedirect.com/merchant.cfm?pid=125&step=4). They are 1.8deg/step, but they have dual shafts and are not horribly pricey at $75 each, but they need tweaking of the motor circuit to run efficiently.

Just throwing it out there in case it's a good idea.