Welcome to the SeeMeCNC Community

Something to note on connector pin current capacity: That's often given with only ONE pin carrying the current. If you have a bunch together carrying max current the connector may melt. It's best to spread out the power/ground pins so you have at least one or two signals between them. Makes layout a pain unless you're using a multi-layer board, then it's not a problem.

jdurand wrote:Something to note on connector pin current capacity: That's often given with only ONE pin carrying the current. If you have a bunch together carrying max current the connector may melt. It's best to spread out the power/ground pins so you have at least one or two signals between them. Makes layout a pain unless you're using a multi-layer board, then it's not a problem.

These are speced that I can use the 8 specific power pins and 8 grounds and get the 1.1A. But I do have to use the 'right pins', which are going to be the ground and power pins on an actual PCIE slot to get that rating.
I will be using a 4 layer board - 2 layer boards are sucky at EMI and routing when you have a 144 pin part on the board.

Those PCIE latches take up a lot of space. I may have to stagger my connectors if I am going to use them.

Just checked the spec, worst case they are 40mOhm contact resistance. that will create 0.05 W heat for each power and ground pin with 1.1A on them. I think we are safe from melting unless we exceed the 1.1A by a lot.

Water cooling capability

Just thought I'd toss out the power note, I've seen a lot of melted connectors.

I use more 2-layer boards than I used to when appropriate, modern ARM Cortex processors can be screaming along and barely sipping power which makes layout a LOT easier. Still, the latest Apple accessory has a 4-layer board since it's got the processor, analog, and a motor driver on it and it's only running at 2.8V.

Haven't done really high pin-count parts for a while (hundreds to a thousand on BGA parts), always a pain. One board wound up being 14 layers with micro vias on both sides and a couple pair of buried vias inside. 50 Watts on a single U6 sized board, we hoped the cooling fans would never fail.

I forgot to add, that big board didn't have room for heatsinks. Just barely enough room (one slot) to fit the board in.

You could just put a power connector on the shield, like they do with video cards. Everyone running an ATX power supply has loads of yellow +12V wires. Some of them even have power connectors specific to the video cards. You could use the female "VGA power connector" on the shield, and include the male connector for people to crimp if they don't already have that coming off the power supply.

626Pilot wrote:You could just put a power connector on the shield, like they do with video cards. Everyone running an ATX power supply has loads of yellow +12V wires. Some of them even have power connectors specific to the video cards. You could use the female "VGA power connector" on the shield, and include the male connector for people to crimp if they don't already have that coming off the power supply.

Yes, the plan is that all drivers can use either Bus power from the main board, or if that isn't the right voltage or enough current, power can be fed directly to the board.

And while most of the driver boards will not be provided with external power input connectors, things like a bed heater or enclosure heater would have the connector, and probably the PC video card spec connector would work, although maybe a screw/pluggable type would be better.

Scratch what I said about 64K being enough. The Smoothieboard has 64K. I got confused by something, maybe it has two banks of 32 or similar. I would aim for 128-256K. Eventually people are going to want nicer LCDs with color and icons, and Smoothieware is already close to tapped out at 64K. wolfmanjm (lead dev I think) says there's only a few K of ram available even now. Make it as futureproof as you can!

Also consider airflow. Oldish apple desktops had nice plexiglass parts inside to optimize the airflow. Made it much quieter. (it looked like a breakfast tray).

A 3d printed box that had fans would be ideal for this.

teoman wrote:Also consider airflow.

Renha wrote:Water cooling

626Pilot wrote:Scratch what I said about 64K being enough. The Smoothieboard has 64K. I got confused by something, maybe it has two banks of 32 or similar. I would aim for 128-256K. Eventually people are going to want nicer LCDs with color and icons, and Smoothieware is already close to tapped out at 64K. Wolfmanjm (lead dev I think) says there's only a few K of ram available even now. Make it as futureproof as you can!

The freescale cortex m4 has 128k ram. And floating point hardware.

Renha wrote:

teoman wrote:Also consider airflow.

Renha wrote:Water cooling

With the drivers all vertical they get much nicer air than all the existing controllers

Water cooling is I think overkill and too expensive

Basically I would like a PMDX-126 with a PMDX-134 but specifically built for the 3d printing world. I personally like modular just in-case you burn something, it's easier to replace and honestly the size doesn't matter to me either.

And yes, I'm a big fan of geckodrive myself.

http://www.pmdx.com/pmdx-126
http://www.pmdx.com/PMDX-134

KAS wrote:Basically I would like a PMDX-126 with a PMDX-134 but specifically built for the 3d printing world. I personally like modular just in-case you burn something, it's easier to replace and honestly the size doesn't matter to me either.

And yes, I'm a big fan of geckodrive myself.

http://www.pmdx.com/pmdx-126
http://www.pmdx.com/PMDX-134

And PMDX is pretty awesome, I have one of there PMDX-132 boards. When it quit, they repaired it, and reasonably so.

How's this project coming along? With rumors of a new version of Smoothie, Duet, and RAMBO all in the works, I'm curious how this will compare.