PC hardware to run LinxCNC

That's an interesting question. I'll have to plug it back into the G5 to find out, but I suspect that it would. I'll try it and see.

Well. THe monitor goes to sleep when the system (OS X 10.5.8) tells it to, and wakes up when I tell it to. And it also wakes up OK when after the system puts the computer to sleep. The "no signal" message appeared on both occassions just prior to sleep mode. But the important thing is that it woke up.

It's pretty safe to say that the monitor is fine - it does what it is told to do by working hardware. You could therefore argue that (part of) the D525 hardware is not working, although that's not necessarily the case.

The size of the little 12 " screen reminded me a bit of the old Mac LC . A bit bigger than the Mac Plus or SE. That's going back about twenty years...

I've run out of ideas - not that I had many to begin with - as to what the problem might be other than frizzled hardware. I think I'm going to have to take the board back to where I got it in town (Sydney CBD) - and that's not going to be fun. They were exceedingly hard to deal with, and not just because they spoke very little English.

Make sure it is not your PSU. I was thinking of flipping the connectors for the button that power the board, but I guess it powers it on so not necessary to do that; also that it does not really matter what position you put it to on newer motherboards.
Make sure your monitor is turned on when you turn your computer on.
Also you can put the POWER LED on there if you have one and see if that lights up. You better put it correctly the first time with + being on the very corner and - right next to it. Look on the connector, the terminal with a little arrow on it is usually the +.

But go ahead and try the pico PSU on another computer and see if that is the problem.

Make sure that AC power connector you use to power the pico PSU is one that can pull of the correct amount of watts (yes there are different types).

Maybe you need to plug in your RAM and see according to what I read here:
www.tomshardware.com/forum/275064-30-turning

If that does not work, I guess you would need to go with the hard route and replace your motherboard.

1 more important thing to note:
In this picture www.flickr.com/photos/alex1n/7446804968/sizes/o/in/photostream/
I noticed your motherboard does not have mount standoffs on the right side. If that is the case, laying on a aluminum sheet like it is in the picture, that can short out connections on the bottom of the board.
Make sure your standoffs are tall enough, I noticed some terminals (those wires that are soldered like those from the capacitors) on the bottom of the board seemed to be a bit long on my board and was worried about it at first, but my stand offs were tall enough and it worked so now I'm all good.

Thanks for your input . I'll check the AC adapter. I got the second-highest-powered unit from the same people ( www.mini-box.com.au , thanks to andypugh for that) when I ordered the Pico150, and their claim is that all their AC adapters work with all their PSUs.note - actually , it Isn't 2nd highest, it's 3rd @ 80W. Highest is 192W, 2nd is 110W...

The monitor has been on first the whole time.

Regarding the RAM, I had both SO-DIMMs installed before powering up the first time (and many times thereafter!).

Well spotted with the sloping board! I'm going to wait before i turn on the thing again until I've got some snap-in nylon standoffs, before I try the thing again. One of the problems with that chassis plate is that there are no 160+/- mm spacings - and I'm still trying to work out the best way of punching holes that can be tapped 3 mm. Temporary nylon supports will get around that problem, and isolate the board from the plate. I've got to wait for a couple of couriers to turn up with what I hope are the final bits for the enclosure before I can go out shopping, so that might have to be tomorrow - if the couriers turn up today, that is...

Alternatively, I can stick some electrical tape over the top faces of the metal standoffs and (that) problem solved .

SUCCESS! Finally! I retrieved the 300W "CoolPower" ATX PSU from the old box and plugged it in (ATX and P4 connectors). It took a while to get over being left to rot but once it warmed up a bit it stabilised and started given steaady power - which seems to be what - as gera229 suggested above - the D525 wanted . I have some snaps:


1. The infamous "NO SIGNAL" message




2. Success! The D525MW just wanted a bit more oomph!




3. Added just for fun - the 12" touchscreen monitor running hooked up to the G5 PowerMac: touchscreen software not installed (at this point, anyway )



Many thanks to gera229 and andypugh for their help in trouble-shooting this problem. Now I just need to get a more powerful AC adapter - or install the 300W PSU...

SSD and bank1 DIMM next .

You do not need electrical tape on top of metal standoffs. You don't even need those paper washers that some metal stand offs come with. I mounted mines on bare metal. I mean it has nothing to short there, and the area of free space is fairly large around the holes used for the screw mounts so it will be safe.

I have a feeling that it's not that the pico PSU was supplying not enough power, but rather it's a defected unit. With the efficiency that the pico PSU gets it should have worked.
These motherboards should work fine even at 80 watts and probably only require a wattage that is less than 80 watts.

It's either the pico PSU is defected OR it requires a higher wattage AC power connector to even make the pico PSU work.

By the time I read your message I'd already taped four of them and the board was sitting on an even keel . I'm going to get in touch with the Mini-Box people here in Oz and see what they've got to say. Even though I've tracked down the ATX 24-pin pinouts I'm inclined not to probe them with the multimeter in case Mini-Box claim that I shorted something out myself (and I do in fact make a mistake and short something out and blow it up).

Here's some snaps to prove that the monitor and D525 are working:





From left to right, top to bottom: installing Ubuntu 10.04 (finally!); Downloading LinuxCNC 2.5 manually, since I seem to have grabbed the distro without EMC 2.4; The latency program running, with a few other things happening in the background.


I ramped my activity up later by installing all sorts of packages that I like to have on a Linux machine (even one that won't be used for much except the CNC, and the quite reasonable 17015 ns in the last snap eventually became 31872 at last look - that's several hours of running with glxgears (running as a backgrounded xterm process), a number of open pdfs, a bit of web downloading, quite a lot of apt-getting, and a zombie installation process (from running the ayam install via sudo) that I can't find and therefore can't attempt to nuke. Once I've finished typing here I'll reboot the D525 and try the latency sniffer again without the zombie: when the thing is being used as a mill it will be the only thing running other than essential system processes (i.e., no games ; ).

I'll read through the docs about ram-disks and the SMI stuff when I've had a good rest. At the moment the heatsink sitting on the Atom is warm, but certainly not too hot to touch. I'm keeping an eye (as it were) on it, though.

Once again, many thanks for your help, guys. I really appreciate it .

andypugh wrote:

AlexN wrote:

I'll read through the docs about ram-disks and the SMI stuff when I've had a good res).

There is no need for the SMI module on a D525, and it might actually do harm, so don't activate it.
Turning off Hyperthreading in the BIOS might help, and isolcpus is worth a bit of latency.
wiki.linuxcnc.org/cgi-bin/wiki.pl?The_Is..._Parameter_And_GRUB2

Thanks for the link. I've saved 07_rtai to disk, but won't do anything with it - I'm seeing double after all the stress of dealing with the D525 and not in a fit state to go monkeying around with something as delicate as that (relatively speaking). The latency sniffer is currently 8497 ns after about half an hour of running - starting from around 6700 ns IIRC. I'll reboot and fix the hyperthreading (which I'd originally turned off as soon as I could get at the BIOS, but turned it back on for some reason - probably because of non-booting optical discs. the latter was caused by using Toast 9 instead of disk utility to burn them. Toast, all right... I'll run the machine overnight and see what the test program ends up with - on an otherwise unloaded system.

I'm very happy to leave the SMI stuff well alone, btw .

Actually, speaking of fiddly things, something that is going to cause some more head-scratching, having done so already, is the installation of the "Short-Circuit" (ouch!) touch-screen drivers.

The supplied Ubuntu make/install script is for around Ubuntu 6 or 7(I think), and since then certain things have appear to have changed in the XFree86 file system/dependency tree, causing make to fail. I'm going to need a really clear head to deal with that one.

The "manual install" is quite complex, so that isn't necessarily quite the option it might appear on the surface - and given their age the pre-compiled drivers and so forth may not work. But that's for another day.