EE Help Needed: Motherboard tray docking circuit

[Captain Slug]

I am working on a project that has one particular requirement that is beyond my expertise.

These pictures explain what I need fairly well
http://www.captainslug.com/modding/mk6_d_1.jpg
http://www.captainslug.com/modding/mk6_d_4.jpg

Essentially I will be needing two PCBs (2"x12" in size) that will act as a pass-through connector allowing all of the cables from the various parts of the machine to connect in a central location and then interface with the motherboard tray through a second PCB. The two will connect together (using typical sized pins used for motherboard headers) when the tray is inserted.
If there ends up being enough space left over an intergrated fan controller would also be extremely useful. Since the circuit is also extending the power/reset switches and LCD indication from the motherboard a front panel PCB may also be required (8" x 1" in size).

I am more than willing to pay for this to be designed and made and I would appreciate any thought, comments, questions, etc.

A theoretical requirement that I am not quite sure how to implement would be a locking mechanism that when the machine is powered-on will not allow the PCBs to be disconnected. I would love to hear any ideas on how to accomplish this.

http://www.captainslug.com/modding/mk6_d_5.jpg

The upper section contains all of the power connectors and except for a few power traces to feed the light power output and fan controller it's an independent unit and can have it's own type of interconnect capable of handling the required current load. Whatever interconnect is used the pin requirement is only around 50 for this layout.

The middle section will just be a series of precisely positioned threaded mounting holes for attaching the flush-mount cable extensions (on the motherboard side) and the flush-mount rounded IDE and SATA cables (on the case side). This section will require custom cables, but no actual circuitry.

The lower section serves 3 functions.
1. Front-Panel extensions (Power/Reset, LEDs, USB2.0, Front Audio, 9-pin Serial header)
2. Light control (2 switches that toggle the 4-pin connectors)
3. Fan controller (on the Front-Panel PCB itself)

The front panel PCB will contain all the switches, ports, and fan controlling electronics. It can connect to the Plug-Bus with one or two cables. It's size limit is 8x2 inches.
The serial header extension is meant as a pass-through for an LCD unit. This could either be a header on the plug-bus, or an actual DB9 connector.

[noxqzs]

Seems like an interesting project. Regarding the 2 12x2 boards adjacent to the mobo. Why not make one PCB instead. It would be a much neater design, and the extra real-estate would help with the ATX plug traces. E-mail me a detailed layout of exactly what you want done. I use PADS and WinBoard for schematics, so a fan controller is also possible.

[four4875]

If it were me tackling this project, i'd use an edgeboard connector system for at least the power. it seems to work great in the rackmount case's PSU i have, although the load might be a little more with your needs. i can grab some pics if needed.

Then I'd put matching sockets on each side and use jumpers, to go from one ATX plug ot the one on the mobo. then same with all the switches, SATA, IDE, USB, LEDs and switches. then you could use IDE cable (s) to connect the front panel to the case mounted board. I'd put the socket for the edge connector on the case side.

i'm guesing this is for some sweet mobo tray type thing?

and i noticed you changed your av.. is it like the first time in 3 years?

[Captain Slug]

Why not make one PCB instead.
I need something to act as an interface to allow the removal of the motherboard tray without having to unplug any cables. I'm not just trying to centralize cable connections, but rather make the entire motherboard tray a modular unit. If you need to increase the size of either PCB or can think of a better way to do this by all means let me know.

The 3D conceptual image is how I want the layout of the connectors setup, and the upper and lower halves of the plug-bus are almost entirely independent. The only traces connecting the two would be the power connections for the fans and lights on the lower half.
I do however need some expert advice on what interconnects to use for connecting the two PCB halves. I was thinking that a mixture of barrel connectors and D-subs would work fine depending on what the voltage requirements for a certain connection will be. Edge connectors were also discussed, but could become more difficult to implement.

If you need me to I'll try to work up a more detail schematic, but if you understand what is I'm working towards I would seriously appreciate your input or any practical design work you can come up with.

If it were me tackling this project, i'd use an edgeboard connector system for at least the power. it seems to work great in the rackmount case's PSU i have, although the load might be a little more with your needs. i can grab some pics if needed.
And edge connector would work for the majority of the connections, I'm a little wary of using them when the requirement for the 12V connections are around 5A though. I was thinking that most of the lower voltage connections could be consolidated into a 50P D-sub connector.

Then I'd put matching sockets on each side and use jumpers, to go from one ATX plug ot the one on the mobo. then same with all the switches...
Yes, the plan is for the motherboard tray half of the plug-bus to connect to the motherboard with short interface cables. The data cables however will just be cable extensions surface-mounted onto the plug-bus. That was decided early on as the best way to reduce potential resistance of the device.

The end goal is to have a modular motherboard tray which can be removed without having to detach any cables.

and i noticed you changed your av.. is it like the first time in 3 years?
I change it once a year, but it's been bomberman in one form or another for the past 3 years.

[speed bump]

Why not use two 24pin atx connectors to interafce the power together instead of the indvidual connectors when going in from PSU. that would make docking alittle simpler.

Next for your locking device you could try an elctromagnet.

Finally for a good connector to put it all together i'll look at the Mouser catalog tonite

[noxqzs]

Ok. just so I understand right. You want a MOBO tray, that is removable. On this tray you want one PCB interface board, with connectors to the motherboard. In addition, you want a master bus of some sorts to interface with the PCB mounted in the case. The PCB within the case has all the interfaces for components, front panel, fan controller, etc...


I have a few ideas how to handle the power. Do you want to integrate data into the PCB, and which type, parallel or serial?

[Captain Slug]

Ok. just so I understand right. You want a MOBO tray, that is removable. On this tray you want one PCB interface board, with connectors to the motherboard. In addition, you want a master bus of some sorts to interface with the PCB mounted in the case. The PCB within the case has all the interfaces for components, front panel, fan controller, etc...
Exactly
I have a few ideas how to handle the power.
Do tell.
Do you want to integrate data into the PCB, and which type, parallel or serial?
After alot of discussion with some other individuals it was mentioned several times that the added resistance of trying to integrate the data cables (especially SATA) would lead to data loss, so I revised to design so that the PSB halves DO NOT bridge those connections. This saves PCB space, and significantly reduces the pin requirements.

The data cables will simply connect to the motherboard via an extension cable. Then both cable connector halves will screw-mount onto the PCB (or an enclosure over the PCB) to facilitate the connection and seperation of those cables with the rest of the motherboard tray.

[kayson]

That's a great idea. I wish I could help you, but my knowledge in this area isn't too great. But I thought I'd let you know that this is the best idea I have seen in a long time.

[four4875]

without IDE and the likes, it gets MUCH easier. i think edge connectors are the way to go, nice and easy to keep lined up and nice contact. they can be nice and wide to accomodate higher loads and all, it seems to work fine for the rackmount PSUs that are slide out like. i might be able to start workin on a PCB layout or similar, but basically dont hold your breath, i have ADD and will lose motivation and get distracted VERY quickly, so it might get started for like a couple hours then ill be outside burning the neighbors cat (joking on the cat burning)

[noxqzs]

If you didn't mind going in the 3rd dimension, and stacked PCBs, that would solve alot of issues. I just did a small test with a 30 inch ATA extension, and drives still worked. A small bridge like you want to implement shouldn't cause a problem for this, especially for serial which has much higher tolerances.

And while you are on the subject of connectors. Stick with the norm, and get the standard MOLEX ones used on the MOBO and components. It will make the design much easier to digest.

[Captain Slug]

If you didn't mind going in the 3rd dimension, and stacked PCBs, that would solve alot of issues.
Stacked PCBs will be fine and understandably would make routing traces to spec much easier. If there are any other space saving measure you can think of just post them.
And while you are on the subject of connectors. Stick with the norm, and get the standard MOLEX ones used on the MOBO and components. It will make the design much easier to digest.
If you're talking about molex connectors for the interconnect, those might work but how much force would be required to insert/remove 50 ATX pins at the same time?
I just did a small test with a 30 inch ATA extension, and drives still worked. A small bridge like you want to implement shouldn't cause a problem for this, especially for serial which has much higher tolerances.
If you can get the Data cables added without any problems that would be spectacular. From what I've been research Serial ATA has some issues and limitations concerning resistance affecting its signals.

And if you can, please make the mounting holes for the PCBs at the standard ATX spacings (http://www.formfactors.org/developer%5Cspecs%5Catx2_2.pdf) so that this can be mounted on the eATX stand-off holes.


I'm more than willing to compensate you for your time, parts, etc if you can help me make a finished usable part. And once we're both sure it works maybe I could even make a duplicate case (that this PSB will be a central part of) for you at cost.

Edit: I would also like to see a locking mechanism if possible. It's not essential but it's a feature I would like to eventually see as a fail-safe. It would also make the device more fool-proof. :beer:

I DON'T REALLY HAVE A TIME-TABLE FOR THIS but I'm hoping to have something that works sometime before or around December so I can start work on the project that this part is designed around.

[four4875]

for a locking mechanism... so something to use a double action selenoid, so when power is applied it snaps it shut, when released it reverses. only problem i see is getting it to shut off after latching. maybe it would be easier for a mechanical latch, which has a switch to break the pin14 connection from the psu when not latched, so if unlatched it shuts off? a selenoid of sorts seems easiest, but getting it to open when power is turned off.... unless you can find one that will handle constant power while the comp is running withough heating up and melting..... pic coming in a few.

it works as... the selenoid has power, so it keeps pushing the latch shut. let off power, spring pulls it open, and it can slide apart. not too complicated.

[four4875]

a question.... how many connections will you need for front panel switches, LEDs, sound, USB... all that, so we can start gettin ideas for connectors to use. im thinking like the kind used in removable hard drive trays might work, and the plug parallel printers have (not the D sub, the other one)

is there anything keeping the boards from mounting with their larg faces parallel to each other? they could then mount to the board with angle steel or whatever. then it could expand the possibilities for connector arangment.

I'm thinking towards a layout that anyone could put together, not having to have some fancy special lab put the multiple layers of PCB together and all, just a gut with components and a soldering iron (After the board is made that is) this sounds like a really fun project.

[Captain Slug]

it works as... the selenoid has power, so it keeps pushing the latch shut.
Most spring return solenoids, especially if their current load is limited, should work fine as a pin-through-tab lock rather than a latch mechanism.

a question.... how many connections will you need for front panel switches, LEDs, sound, USB... all that, so we can start gettin ideas for connectors to use. im thinking like the kind used in removable hard drive trays might work, and the plug parallel printers have (not the D sub, the other one)
Front Panel output
2 x USB 2.0 = 8 pins
Audio FPC (One Headphone, one Mic) = 6 pins
Reset = 2 pins
Power = 2 pins
LEDs x2 = 4 pins
Light toggle switch = 2 pins

Total = 24

Optional additions (if space allows)
Another Light toggle switch = 2 pins
1 x Firewire = 8 pins

Sub Total = 34

The fan controller would be independent from the rest of the PCB and would receive power from one of the case-side plug-bus molex connectors, then output power to the lower 3-pin connectors on the plug-bus. Doing it this way keeps the front-panel cable and plug a low voltage path with more connector options.
is there anything keeping the boards from mounting with their larg faces parallel to each other? they could then mount to the board with angle steel or whatever. then it could expand the possibilities for connector arangment.
If there's an easier way for this to be made that still allows for durability and a general consolidation of wires I'm all for it. Another potential solution would be to have all the molex connectors installed as 90-degree plugs off one edge of the board, then have the two boards connect together with a simple pin header array.

I'm thinking towards a layout that anyone could put together, not having to have some fancy special lab put the multiple layers of PCB together and all, just a gut with components and a soldering iron (After the board is made that is) this sounds like a really fun project.
I was just thinking PCBs would save space and be easier to mass-produce. I'm sure something like this could be made with ALOT of wire and solder, but I wouldn't trust that kind of solution with anything other than power (which is what I was leaning towards). The remaining issue them would be trying to keep the enclosure small, but still functional.

[four4875]

i was getting at that single layer PCBs are much easier to etch and drill and all, most of the members here could build one if they set their minds to it. but multi layered are scary, so i'd stay away form them lol.

I was thinkin of some of the connectors like on old vid cards.. how they had RAM addon boards... if you know what im talkin about... like larger versionas of those or something. also might consider the plugs used in laptop docking stations, with the pins to line them up and all.

[four4875]

captain, i have the PSU out of the rackmount, ill try to get pics tonight hopefully, before i elave for columbus. if not ill have pics for ya on sunday, hpefully they'll help a little.

[noxqzs]

i was getting at that single layer PCBs are much easier to etch and drill and all, most of the members here could build one if they set their minds to it. but multi layered are scary, so i'd stay away form them lol.


I think you misunderstood what I meant. I was thinking of single-layer pcb's that are mounted on top of each other, with spacers in between.

I managed to locate 80-pin connectors you could use for data. They look something like the ones I used in my samples. They make 90*female ones, so putting boards side by side is no problem. They make larger versions for higher capacity. The ones in the pics would be good for data or < 3A. With all the leverage you would have with the mobo tray, insertion force would be negligeble. Work permitting, I can spend some time on it this weekend.

http://home.comcast.net/~noxqzs/pics/opto01.jpg
http://home.comcast.net/~noxqzs/pics/opto02.jpg

[kayson]

Any updates on this? I'm curious to see how it turns out...

[Captain Slug]

Any updates on this?
Not that I am aware of.

[Captain Slug]

After my first try I've basically started the PCB design over from scratch because of my previous inability to take into consideration the peak current load requirements of certain traces. In deciding upon different connectors and consolidating some portions of the PCB for better trace connections I've managed to turn this into a fairly simple single-sided PCB.
Here's the new schematic: http://www.captainslug.com/modding/mk6_plug-bus_th_pcb2.gif

The two halves will connect through 1 DB15 and 2 DB37 90-deg connectors.

The other half is going to take more time to design and will need to be double-sided.

The data cables are no longer included because I can simply use an array of these mounted on-top of the fairly blank portion of the PCB.
http://www.coolgear.com/images/221005A.jpg

I haven't made any head-way on finding a workable latching mechanism but hopefully I can discuss this further with the local electrical engineer I finally found.

[four4875]

damn thats nice. you've made it much farther than i would have lol.

i can see this turned into a compact rack system for a folding borg lol. each mobo tray and everything slides out for diagnostic / repair / replacement... then you have all the folding rigs runnign at once and have one big psu incorporated... that would be fun to do lol.... but i would lack motivation and ADD would kick in haha.

[Captain Slug]

The case-half of the PCB is designed.
http://www.captainslug.com/modding/mk6_plug-bus_ch_pcb2.gif

The case-half will be only female connectors while the tray half will be only male. This made the traces easier to do since Male and Female connectors have reverse pinouts allowing me to simply mirror most of the elements.

The last part to design is the front panel interface PCB.

[Captain Slug]

The front-panel PCB is done.
http://www.captainslug.com/modding/mk6_plug-bus_fp_pcb2.gif
I've inspected all three designs completely to make sure traces are going where they need to. Now I just need professional help making it myself.

[four4875]

i think theres mention of a company that does custom PCBs in the audio DIY section, but i remember there being a broup buy involved to meet quantities, so the price for just a single run might be a little steep.

[Captain Slug]

I have a friend who will be helping me with the soldering (because I don't have the most steady hands). And his Dad used to make custom PCBs for sensor-arrays so I have him to go to for advice on the process.
I just need to confer with him concerning my design and then make a rather long shopping list.

[four4875]

the part of the pcb making that would scare me would be the drilling,and keeping spacing right. its nothin to etch a board, ask solidxsnake, hes done it. the scary part is keeping spacing correct for the holes, and with so many pins on some of the connectors... I guess that with a decent drill press and taking alot of time to do the layout work... oh how i hate layout work, keeping everything accurate and all. and the layout i do is on slightly larger scale stuff, but tolerances must be kept regardless.

soldering isnt that bad on something like that, the hands steady out when you can rest the iron against the parts you're soldering to. hell you even have to apply some pressure to it.

when drilling, i reccomend you use a carbide bit caus the fibreglass might do a number on a steel bit that small. and if the bit starts dulling you might get some holes that arent the cleanest.

something to consider is copper thickness on the board. i this application, i would think that as thick as you can would be desired as the higher current capabilities, although it might take a little longer to etch.

[Captain Slug]

If I can't do the drilling myself I can go get the smaller holes drilled with a CNC mill table.
And yes my friend's Dad mentioned that I may want to go with a heavier weight. They're quite hard to find, especially since I need a 12x12" double-sided sheet.

Edit: Whoa! I found 12x24" 3 oz. surplus mil-spec copper-clad board for $7 each. Would be a pain to etch but it's super cheap and would be a lovely amount of over-kill.
http://www.surplussales.com/RF/RFMicrowaveCir.html

[Captain Slug]

After some concerns with using D-Sub connectors I broke down and decided the use high-amp (4.5AMP max!) capacity .030" diam. Mill-Max Pin Header, which look like this.
http://www.mill-max.com/images/products/fullsize/79D.gif
One row of 100 pins on each PCB makes the traces much less complicated and allows me to use wider traces for almost everything. In other words EVERY is connected with an overkill amount of trace so this should be a very durable device. These PIN headers also have a longer service life. I also added optional board underlighting LED sockets to opposing corners of all the boards for an aesthetic touch.

All of this meant redoing all three boards, but in the end I'm much more satisfied with the results.
This PCB will attach to the motherboard tray
http://www.captainslug.com/modding/mk6_plug-bus_th_pcb3.gif
And it will dock into this PCB which will be hard-mounted in the case
http://www.captainslug.com/modding/mk6_plug-bus_ch_pcb3.gif
And all of the front panel connections will be forwarded to this PCB
http://www.captainslug.com/modding/mk6_plug-bus_fp_pcb3.gif

The case-half and front-panel PCBs will interface with eachother through a rounded floppy cable instead of a D-Sub. This will not only reduce weight and improve cable management, but it will also make the end unit much cheaper. The revisions I made reduced the end-cost by atleast $20.

And my last question is: If anyone knows of a good source, I need some 5V or 12V push solenoids with a low current draw and 100% duty cycle. Force application is not an issue here I simply need one or two solenoids to lock the tray in place when the machine is turned on.

[Captain Slug]

I've begun ordering parts for this project and after discovering no viable sources for double-sided 3oz copper-clad board I alleviated the problem by making the case-half PCB single-sided like the rest of the boards. This allowed me to buy more affordable 3oz single-sided copper-clad board from eBay.
I now need to decide on an etching method.

[four4875]

i have a mill-max mousepad at home, and a catalog and some VHS tape they sent somewheres... lol.

never have used their products, but somehow got on their marketing list.

Again i congratulate you on the progress you've made, much more than i would have done haha.

I know solidxsnake found somewhere how you can use a transparency printed with a laser printer, then it can transfer to the board as an etch resist, but i dunno the process exactly and dont have a link.

[Captain Slug]

You print the design onto glossy paper using a laser printer then use an iron to burn the toner onto the PCB. This seems like the easiest method and I have a friend who owns a print shop that will be able to help be print the large format sheets I need.

[four4875]

looks like you're already ahead of things lol. best of luck with it!

[zebkoolindc]

At my school they have a full mill that takes in files from Eagle PCB software and spits out the board with traces and wholes.

It mills away the copper that you don't want and then drills the holes. If you have a schematic I might be able to have the macine do all the work.

[four4875]

basically any cnc mill would do, just have to program it or use some CAM software to do the programmin for you, then set the PRZ correctly and it should all be well, but if you set the prz too low or too high on the Z axis it might not take off al the copper or take off too far into the PCB, so you'd have to be careful there.

[Captain Slug]

At my school they have a full mill that takes in files from Eagle PCB software and spits out the board with traces and wholes.

It mills away the copper that you don't want and then drills the holes. If you have a schematic I might be able to have the macine do all the work.
:drool:
Can I send you the copper-clad board to use as well? I have a bunch of 12x4-inch single-sided sheets of 3oz weight on order that I could cut to size.

[Captain Slug]

I've begun the slow process or recreating the circuit boards in Eagle so that I can actually export them in other file formats.

[Skeith]

i have a mill-max mousepad at home, and a catalog and some VHS tape they sent somewheres... lol.

never have used their products, but somehow got on their marketing list.

Again i congratulate you on the progress you've made, much more than i would have done haha.

I know solidxsnake found somewhere how you can use a transparency printed with a laser printer, then it can transfer to the board as an etch resist, but i dunno the process exactly and dont have a link.

Yeah I use that method all the time. The ink from the lazer printer is heat transferable so after the copper board is clean you place a paper on it and heat it up for a min with a clothes iron. once its heated the paper comes off and you lay the transparancy down (ink side down you can tell because the side with the ink on it looks matted while the other side is shiny) the ink should kinda stick. once the transparancy is down DONT MOVE IT! or you rist messing up the board and have to start all over again. put a piece of paper over the transparancy and heat it with the iron till its good an hot, but not so much that the plastic melts. then peel of the transparacy. The ink should now be transfered to the copper. Little pieces that get messed up aka not enough ink, can be touched up with a permanent marker. Then put it in the etchant. Most commonly Feric Chloride. And etch away.

At my school they have a full mill that takes in files from Eagle PCB software and spits out the board with traces and wholes.

It mills away the copper that you don't want and then drills the holes. If you have a schematic I might be able to have the macine do all the work.
Yeah my school has a fancy $300k CNC machine that will do that too but its so much easier and faster to do it by ironing on. Unless your adept at using the Milling software wich in my case I am not,... just been too lazy to learn it.

[Captain Slug]

The Eagle .brd files are done and available for review by all capable persons. The front-panel PCB is the only one that will have the traces on it's top side. Which is due to the USB headers having SMT contacts. I couldn't find any through-hole ones.

http://www.captainslug.com/modding/plug-bus_eagle.zip

The advantage I can see at this point in using the mill would be not having to drill several hundred tiny holes. :eek:

Now I'm just waiting for my 3oz weight copper-clad boards to come in the mail...