PowerComms power out.

While mucking with crimp on connectors and that whole tedious mess, I pondered how we could improve the power connection part of the board. One of the things I'm a big supporter of is the use (and re-use) of connectors that are easy to find. For example, we use a molex connector for the power in on the PowerComms board. It was a nice hack, and now there is no danger of reversing the polarity on that board. I'd like to take it a bit further.

We are already cannibalizing a normal PC power supply for the machine, and there are TONS of useless connectors hanging off it anyway. The disk drive power ones are nice, but it doesn't really make sense to use them for power output. However, there is one connector that has nice, big wires, and also a ton of connectors! Its the one that you are supposed to plug into the motherboard. Simply cut it off the power supply, plug the connector to our powercomms board, and then route the power to where it needs to go. If the wires aren't long enough, you can simply splice longer wires on (so much easier than crimp-on connectors) I propose we change the connector on the PowerComms board from a the 12 pin connector to the ATX style mobo connector.

I've tried looking it up in mouser, but i dont really know where to start. Anyone willing to dig in and see what they can find?

Thoughts? Suggestions? Problems? I'd probably make this change for the PowerComms v1.3 board which is a month or so down the road, but I'd love to make it even easier to use.

I already put the connectors on my latest powercomms board, but mine were for getting the power in, not out. I used the two-row 20/24 pin connector that most recent motherboards use. I'm not sure if that's what you were thinking of, or if you meant the two AT-style 1x6 connectors?

If you'll permit a gratuitous photo:
[img340.imageshack.us]

I was actually thinking in the opposite direction, prevent the user from having to modify the power supply at all. It's also kind of cool because the host software can turn the reprap power on and off on and off under software control, but admittedly I did that just because I could :-)

Another way to go would be just stick a 4-pin hard drive power connector on each board. I don't have a good feel for how big darwin is, but most ATX supplies I have lying around have at least five disk connectors, and Y adaptors are pretty easy to find if you need more, the boards just need to be within a few feet of each other.

Anyway, the connectors you want are:

39-29-9202 molex 20 pin ATX motherboard connector (mouser, $2.57, $1.93 for 10+)
39-29-9242 molex 24 pin ATX motherboard connector for 20+4 (mouser, $2.18, $1.82 for 10+)

The 24-pin will accept 20-pin connectors (the latch won't engage, but there's plenty of friction to hold it in). All the power supplies I've seen so far that support the 24-pin connector are in a 20+4 arrangement, but I don't know if it will always be that way. I've got an eagle library for the 24-pin one that you might be able to convert to kicad to save some work.

wow, thats a cool board you got there.

i like the idea of using molex connectors for each board, that makes alot of sense. if we did that, then the powercomms board would really become the comms board, and we'd just let the power supply take care of the rest. i dont think thats the route we'll take though, as we'd like to keep it all running straight off 12v, so people could power it from say car batteries, solar, etc. who knows.

nice board, btw. do you plan on writing up any info on it? i know we'd like to switch to USB for v2.0, although with the usb -> serial converter we're basically already there (and have backwards compatibility w/ serial.)

i guess i'll try and mockup a v1.3 board in this style and see how it feels. it would be nice to have to make as few cables as possible. i'd also like to switch to using ethernet cables for tx/rx as well. the price is basically the same, and there are no cables to make!!!

We aren't necessarily cannibalizing a PSU -- we're *using* one. We can even use the one inside our Reprap host PC (or another nearby PC if you have multiple PCs!) if it has a spare power output wire -- just route it out the back of the PC! This is what I currently have set up. It may not work when I have 3 beefy steppers to power, but for testing it is highly convenient!

I'm not sure about using motherboard AT/ATX connectors, precisely because it makes the PowerComms board suited for only one or the other style of PSU. The current molex connector works well, because both PSU styles have them.

Putting the same 4pin Molex connector on each UCB for power in is fine with me. Then people can either cannibalize a PSU for wires from the output pins of the PowerComms board to the UCBs, or they can go direct from a PSU if they wish. No loss of flexibility there, and the common use of those connectors *is* to supply DC power, so we're using them as intended.

If we try using Ethernet cables, *please* carefully consider the effects when people accidentally plug the Reprap board into a PC NIC or into a router or switch -- both the Reprap hardware *and* the NIC/router/switch need to be able to survive this abuse! It *will* happen -- using home-made wires and the two-pin headers is good in this regard, because it is unlikely there is much else around that those wires will easily connect to :-)

I'd suggest that the 4pin molexes for power input is probably a good move, despite their size. I'm much less sure about using RJ-45 "Ethernet" wiring for non-Ethernet signals; I think you'll risk ending up with people destroying network hardware or reprap hardware that way.

Jonathan

I think Ethernet cables have 4 twisted pairs and normally only two pairs are used. So you could use one of the "spare" pairs. They may get used for POE so you could get big problems if you plugged it into a network rather than a PC though.

I can't say I like the idea, it seems a but over the top to use 8 way connectors for a simple two way circuit. Surely its cheaper to use two pin headers and housings and some twisted wire than it is to buy a pair of Ethernet connectors and a patch cable. It also makes the wiring neater as you can tailor the length exactly to the machine and the cables is less bulky.

I am not overly fond of the disk connectors either. They are big, hard to get on and off and often don't make a good connection.

---

[www.hydraraptor.blogspot.com]

ZachHoeken Wrote:
-------------------------------------------------------
> i dont think thats the
> route we'll take though, as we'd like to keep it
> all running straight off 12v, so people could
> power it from say car batteries, solar, etc. who
> knows.

Makes sense.

I don't have any great ideas for the power connectors. I'm using the 1.1 stepper boards with identical connectors for power and data, and I've come close to mixing them up a few times. It looks like the 1.2 version would save me from that by using screw terminals for power, but I'm not thrilled with that because it's a pain to disconnect one board when they're all tied into the same harness on the other end. If you switch to hard disk connectors, I'm sure I'll find something to complain about there too =)

I'm already tired of making molex connectors and I'm only half done. RJ-11/12/45 sounds tempting, but you might not have a very good selection of short cables in your local stores. Making your own is an option, but for beginners I think crimping your own cable has a much lower success rate than soldering molex connectors.

> nice board, btw. do you plan on writing up any
> info on it? i know we'd like to switch to USB for
> v2.0, although with the usb -> serial converter
> we're basically already there (and have backwards
> compatibility w/ serial.)

Thanks. I can put the boards and hex files up if anyone wants them. Sticking with requiring a USB->serial converter is probably more practical, I just don't find them very appealing (and I don't have one :-). The USB power/comm board costs about $3 more than the serial version, and you need a PIC programmer to make it so it's chicken & egg all over again. It starts to be more attractive if the project moved towards having an intelligent controller that would run on the PIC like TylerM had posted about. It could also be combined with the IoBox since the PIC has at least a dozen unused pins.

I was thinking that the transition to USB from serial was (in part) to enable the use of power-in over the USB port. Then we can do away with the power supply altogether and have eg. RJ45 power over ethernet connections from the power/comms board.

Did I pick something up wrong, or did I just have a good idea at some point?

Then of course you get further away from having the standalone machine that you just put a memory stick in and hit print. But I think you also have a less complex machine for the newbie to try and build.

If you really want the standalone machine you could design an additional box thats just the power supply in a case with a couple of usb ports, buttons etc. and a little circuit board they plug into with the right firmware on board. Hook them together and you're good to go.

As for the idea of using car batteries etc. A quick search on google brings up [www.proporta.com] as the first entry. This little gadget plugs into the cigarette lighter port in a car to charge/power any USB device.

It'd be really nice if that would work, but I think we're always going to be stuck needing an external power supply. The RepRap machine needs a fair amount of power, and a USB port just won't give you enough. I think the USB spec says you can get a maximum of 500mA at 5V, so that's 2.5 watts. Our three stepper motors are specced at around 850mA at 12V each, so together they would be trying to suck 30W. The extruder motor and heater are going to want some power too, I'm not sure how much they need off the top of my head.

The USB port could easily power all of the logic in a RepRap, it just wasn't designed to power big loads like motors and heaters.

Yeah, we definitely need the power supply for heater / motors / etc. I think USB is mainly just for easy interface. Also, we may need the faster speed at some point in the future.