Mesa 7i92 Installation, considering old ATX psu for I/O voltage source...

Hello all, and greetings from the USA. This week I finally got a working plan sorted out for a 4Axis desktop mill build after spending alot of time combing thru several data sources accross the web (primarily linuxcnc, much thanks for the years of good contributions from the folks here, this site has been very helpful). So, I will try to get to the point of this post and list the general harware information I have already got on hand then state what my current questions pertain to, and also I will note some additional build information to provide a better idea of the direction I intend to head with this project.

Hardware on hand:
Pc: Older Dell Vostro with intel i3cpu (quad-core), 4GB ram (generic matched set), native graphics, onboard ethernet port reserved for the Mesa OS: Fresh working install of Linuxcnc uspace(Debian Bookworm)
I/O: Mesa 7i92, 7i84, 7i85S (haven't connected any of these yet, but I do have cables)
PSU for I/O? Couple of different older ATX style computer PSU's, atypical 3.3v, 5v, and 12v outputs.

Hardware currently on the buylist:
Mill: Taig(Peatol) 5019 CNC ready with 5tpi ballscrews and extended bed
4thAxis: Sherline Rotary Table
Steppers: Presumably going to use the stock Taig(Lin Eng.?)DSLS 200oz-in steppers with encoders, probably around 4mH rating
Stepper Controllers: Either going to run Gecko G540 or G214s depending on what exact steppers I am able to source
Stepper PSU: Meanwell 1600w 110v/220v AC input, 48v-65v DC output. (again, the exact model depends on the stepper selection and Geck req.)

Alright, I believe that should cover most of the basic harware involved. A few items to note about the above listed hardware; As per the suggestions I noticed on the wiki and the forum I did go thru my bios bootmenu and turned off just about all the intel nonsense that was optional, I havent quite gotten deep enough into my config files or hal yet to really tune other settings as I only just got my Mesa cards a couple days ago. Currently I believe the cpu is still running on all 4 CORES but when I ran the base jitter tests they didn't look too bad ~26,000 max under moderate task loading if I really need to shut down to 1 core I will but wanted to ask for a little more clarity on why before doing so.
Mesa notes; I am hoping to utilize a G540 so I went for the 7i92f to have a more commonly implemented configuration there considering they were out of the 7i76 series when I looked. To support the encoders and additional I/O I picked up a 7i84 and 7i85S. I realize these two may require some hal work to get everything to play nice, but in light of availability this looked like a reasonable starting point.
Stepper notes; I am currently trying to get information from a couple of US retailers of TAIG equipment to see what options the have available without buying their control hardware/software as a package deal...(most of the taig controllers seem to perform rather poorly and are mach3/4 dependent). Hopefully, I can get a set of their 200oz/in DSLS steppers with encoders to be able to run a closed loop system, but I am not holding my breathe with current availabillity issues being what they are in that market here. I may have to outsource a different supplier for a 260oz/in low impedence dual shaft motor and figure out where in the blazes I can get an encoder from to tie it back into linuxcnc...

Main question at this time: I have a couple ATX psu's on hand I would like to repurpose IF they would be a good power source to drive various I/O thru the Mesa Cards, but a few hours of searching for clear information on the matter, I didn't have much luck. Biggest questions regarding that point being; would that power supply be a clean/stable power source that would work well with the mesa cards? This would be my first time repurposing one of these so is there any sense wires or psu regulation behaviour I should expect? And finally, just to get the mesa cards configured, would the ethernet line to the 7i92 be adequate to supply the other 2 cards until I start connection I/O lines, or would something like a split usb cable be a good choice from an extra motherboard port until I sort out my stepper motors and order a Meanwell for their drivers...

Sorry for the long-winded post, I seem to be compelled to be too thorough at times...

Sorry for the bad indentation/wraparound text on the first post, didn't realize the text input window was formatted at a wider width than the main forum. I tried to go back into that post and edit to adjust the spacing so it would be easier to follow, but other formatting started getting a little buggy so I cancelled the edit before it got worse... I will note the page width difference and try to keep things a little cleaner with that in mind next time.

Thanks for the input on the ATX supply, I will start setting that up tomorrow when I go back out to my desktop.

I was aware the G540 is fused at something like 6A internally, guess I fudged the amp calc and accidentally
multiplied each axis by a transposed 5A figure when I was doing rough estimates, (720w just for 3Axis, lol)...

Main reason for the higher voltage on the psu is to be able to turn up my feedrates substantially to
make use of the ballscrews. I do really want to find some sub 2mH steppers (i did find some nice
german steppers under 1mH, but special order only...) but I was having a very hard time finding any
in stock over here with documentation and a dual shaft setup for encoders. So, it was looking
like I may need to run something like a common 4mH which would require alot of Volts to get
any appreciable rapids or torque, and with that higher impedance the G214 would come into play.
Do you think it is worth waiting on a 6week lead time to get a 1mH stepper? I really want to have
this thing running by about that time, and some lead times are 4months from now, hard to justify.

Are those german drivers similar to the ones kollmorgen markets? I had happened to notice
something about a programmer for theirs too but they didn't have alot of Amperage overhead
until you moved into the bigger industrial models and those looked REALLY expensive...
In any case, I will look into those drivers, 28euros on a programmer doesn't sound that bad if the
drivers don't add up to more than around 500euros...

Ok, so I just had a look at the mechapro site and it looks like the LAM drivers are completely different from
the ones I had noticed on the kellmorgen pdf. Probably a good thing tho, the smaller kellmorgen drivers were
more similar to the gecko and their larger ones were probably very pricey.


The LAM DS1076 does look like a higher quality unit compared to a gecko G214, now I have just got to get
my hands on some steppers to know what my supply voltage needs are going to be there...


Just to clarify for anybody reading from this point on, the 36V and 48V power supply's referenced in the last
few posts are referring to what I am going to power my STEPPER DRIVERS with, the ATX psu I primarily
started this thread to discuss is for 3.3V, 5V and possibly also 12V for other various I/O applications thru
the Mesa boards.

Very long day, finally got a chance to slow down and check in here to post updates, then forum ate my post 3/4
of the way thru. Just one of those days i guess...  So here goes again.

Good information on the LAM, and yes I want to go closed loop with encoders but need to figure out a valid
encoder that will work on the dual shaft steppers I end up with. Honestly tho, the stepper drives with all the active
amperage/efficiency management are rather interesting when you think about it. I have read alot of posts of guys
mentioning midrange performance boosts using the better gecko options too, so it sounds like either make is up
to the task.

Honestly I am leaning towards the geckos due to them being based out of the states and there appears to
be more documentation/support for Taig mills with the gecko drivers on them over here. But, on a different
build, I would really like to order a LAM drive to see how they handle as well.

I certainly would like to have found an available Japanese or German made stepper here in the States, but after
a second session of mucking thru the catalogs of the major regional suppliers here, I was left with only the
questionable and recently acquired keling name, so I took THAT risk... Didn't like that decision, but I am trying
to keep to a rather tight timeframe with this machine build and it was taking way too much time finding another
option. I intend to upgrade away from this set of ordered steppers sometime later, they were cheap enough to
justify the compromise for now.

Keling nema23 dual shaft hybrid stepper specs:
p/n: KL23H276-30-8B
282oz/in rated holding torque
1/4" / 6.35mm shaft with a flat    (idk if the flat will slip in the Taig couplers, but we will find out/make it work...)
4.2A
0.46ohm
~2.2mH (this is what they were claiming at least)
8wire bipolar parallel, no connector, just leads with a grommet... Back case doesnt look encoder friendly either...

So ya, not my first choice of motor, but I can probably work around these things to get this machine operational.

np on all the info, I will utilize the data I can gather and rationalize my final choices with consideration to available resources.
 

ATX psu progress...

Sat down and split up / tested all the wires on this old unit today, overall it went rather well. After I got through
the sorting and color code identification for this particular example, I regrouped the wiring to better suit my
current application. (various I/O supply for my Mesa cards and auxiliary electronics)

The longest available wiring on this unit had been for the various molex connectors that ran to dvd drives and
the like. I cut the chain of molex connectors off the end of this group of wires and then also did the same for
the adjacent wire group that had ran to the hard-drives in the original application. While either of these groups
were not terribly long, they should be adequate to reach a couple of common terminal blocks that I can branch
out from in my wiring enclosure, and from there supply the Mesa cards. While I could have tried to implement several
direct wiring paths by utilizing more leads from the main wiring group that had the 24pin motherboard connector,
i believe the terminal block approach (using the longer wiring groups) will ultimately be more organized and allow
easier service in the future.

Once I had cut the groups of connectors off, reordering and bundling similar wire groups was just a function of time.
Any extra wiring I didn't plan to use I moved out of the way and just capped off any loose ends in those groups.
On the 24pin connector there was a -12v (Blue) , two -5v (White), and also a small (Brown)
wire that had shared a pin with a larger 3.3v (Orange) wire (I just left that pair connected for now as I wasn't too sure
of the Brown wires function...) these were the outliers that I didn't have much interest in. Most of the rest of the
main 24pin connector was an assortment of Red or Orange positive wires, and Black common wires. However this
left just...
 

... THE 2 MOST IMPORTANT wires on the 24pin connector: The Gray and the Green, these two get connected together with a switch to power up the power supply.

Remember, the color coding i have listed is just what I happened to find on this unit, others will vary. Of several other similar power supplies i have, the GRAY and GREEN are the same however.


General wire color coding and Voltage on this old dell ATX psu:
Gray ---> Green (these two turn the unit on)
Black: Common
Orange: 3.3V
Red: 5V
Yellow: 12V  (Not in the main 24pin connector on this unit, used in some of the other connectors however)

Blue: -12V (low Amp rating)
White: -5V (low Amp rating)
Purple: 5v (low Amp rating, this one is special, it functions different than others when on standby I believe)

Brown: ? idk on this one, left it with the Orange 3.3V it was paired with originally.
 

A meanwell MDR-60-24 power supply would be much more compact than the PC one. You may need to buy a similar 5 volt supply
Maybe a MDR-20-5

If you want a high quality stepper motor look at moon's industries. They are way above the average stepper online standard. They have stock or an office in the US.

I have paired these with LAM drivers. The Lam boost function is driven by a custom component that monitors the velocity of the drive. if it is constant (eg not accellerating), it drops the current to a defined value say 30-50% of the actual. This allows the drive to cool when its not working. Some drives we profiled only used 12% of rated power once the motor was at speed.