Continue to Site

Eng-Tips is the largest engineering community on the Internet

Intelligent Work Forums for Engineering Professionals

  • Congratulations GregLocock on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!

Home Brew Through Hole Plating tips needed

Status
Not open for further replies.

originator

Industrial
Dec 12, 2004
71
Too many vias! Over 500 vias, plus the 30+ chips on a double sided board. Fortunately I have a machine to drill it in under 15 minutes very easily. Running wires through the vias is not fun. I am looking for ideas to "presensitize" the holes, then do some home brew bath for tin (copper, etc) plating. There will be a number of revisions, so board houses would take too long. One easy way out for the vias at least is silver epoxy which conducts very well, but not sure on its shelf life.

Anybody done this before for prototypes?
 
Replies continue below

Recommended for you

Groan...

Why persist with the poisons and the drilling and the thru wires and the "where's the busted via" hunts? I get an ad a day across my desk for cheap quick prototype boards. How can your huge labor investment come out cheaper than one of these "deals"? I think you should really consider having a shop do this while you do "other" stuff to speed your project thru. This is rather like having a surgeon out in the back machining his tools for the next surgery.

If you have some simple one sided board with ten parts on it rollit yourself "might" fit.

Keith Cress
Flamin Systems, Inc.-
 
Too many vias, indeed. You need to change the rules on your router, or let it run longer, or get a better one.



Mike Halloran
Pembroke Pines, FL, USA
 
Too many vias comes from trying to fit too much stuff on a space. I am using Eagle, and have tried every possible angle and setting for getting the least amount of vias, but when the board is packed, it just adds more vias. I don't want to make a bigger board, it is 6x9 now.

LFKF is sending overnight their conductive paste that applies in minutes with a squeegie. You vacuum the excess out of the holes and stick the board in the oven for 30 minutes. It is solderable to 420F, and I think it is tin platable also, since it is a silver based ploymer. After curing, I'll bath it in a tin coating brew for 20 minutes. All in all, the real manpower minus learning curve time should be just minutes to do it. I think it is worth it to have a system to knock off a number of complicated boards in a days time.

I have quotes from many board house for $500-1200 for 2-3 day turn around. A company in China will deliver to my door in 7 days a minimum 5 pieces at $20 each on a 6x9 board. That will be for after the bigs are worked out.


 
I just stumbled across this site. Very low cost, especially if you are comparing it to home made silkscreenless sodermaskless boards.

They also provide a free layout tool who's demos are so clever I would imagine the tool works pretty well.



Keith Cress
Flamin Systems, Inc.-
 

Hey I used this advice, sent them 4 eagle board files only. They shipped within 48 hous and the stuff looks great. I simply uploaded the eagle files (.brd), entered a credit card, thats it. They converted the files free.

Making PCBs for yourself is a headache.


Why persist with the poisons and the drilling and the thru wires and the "where's the busted via" hunts? That is the truth.
 
HAHAHAHA!

Thanks a bunch for the feedback originator!

Glad it worked out.

I have actually found that up to about a week you rarely need to worry about 'lost time' because after a bunch of focus on doing the layout you usually need to stop and regroup anyway. Find those last few parts, dig out the emulators, jig up the test support stuff, make a few cables, deal with unrelated stuff that you'd put on hold for the big layout push, etc. I couldn't even guess the number of times I have selected a shorter delivery time for something, UPS red/blue, 2 day turn instead of 4 days. Then inevitably, in-fact almost without exception, the thing sits for a day or two sometimes even three because other priorities appeared. [banghead]



Keith Cress
Flamin Systems, Inc.-
 
Yes thats how it goes. I usually just get so burned out that I blow up what I am working on anyway, and spend 4 extra days trying to find the burnt traces and fatigued parts. I am getting the Parallax stamp and Demo board in today. Amazing how getting some new stuff to try out still feels like Santa Claus, even at my age. Can't wait to ditch some of this logic stuff off my boards.

I have been exploring methods to theoretically gang 500 systems on a single wire to a central unit with a software to monitor the systems, send out a serial number, get a response back that says it is there, then go to the next number. One guy tells me to use ethernet all the way, and use IP's to send and respond, then if there is a problem, the unit can send the problem over the ethernet. Parallax says I can simply query each unit with a SEROUT, wait for response, then move on. Going the ethernet route, I can already be set up for web remote control, so that may be more attractive to end users. Parallax has a third party board called Pink that plugs right into the stamp. 1-wire is a thought, but I think it has limited capabilities. Any votes for ethernet for monitoring sensors and remote control?
 
Yeah I'll vote for Ethernet since it is very fast, very well established, annnnnnnnnnnnd the real reason is that by default it is isolated. Serial is not and takes special, careful effort to make work.

Ethernet only allows about 250 units in a constellation and is a star layout when talking 10BaseT(the only game in town these days). this could cause you a real headache. Got a bunch of routers and switches? (A whole bunch!)

You could daisy chain 485 and so have only two or three points on a computer. The number of drops is limited on multi-drop serial too, but for loading reasons as opposed to address reasons.

If I were going to do this I would look very long and hard at Ad-Hock Wireless LAN system. This will get you everything you want. You get the isolation, the 250 units per router, the ability to poll them all, and no wires! This may seem a bit expensive but in fact it isn't because the routers are dirt cheap < $50 and you don't have this huge number of wires and to run everywhere and deal with. This makes the cost of ownership low.

Normally 250 units of anything on a wireless router would be a disaster but in your case it would be prefect as only a few units at a time ever need to communicate at one time.

Also in the wireless LAN mode you get to have an emergency blurted out whenever it occurs whereas in a serial system yo would 'generally' only find out once you'd polled a unit that it was unhappy.

There are a truckload of off-the-self embeddable wireless Ethernet solutions out there.

What the heck are you doing with these 500 points anyway?? All I can think of is you are using them for a pot growing operation or a Uranium diffusion system...

Keith Cress
Flamin Systems, Inc.-
 
Thanks for the ideas!

I am developing a start trek type door, technically a pocket door. hardened laminated starfire glass (or fine wood veneer, stainless etc), there 3 inner layers, two layers determine color, one is ito film, which make the glass touch sensitive. there are touch sensitive (qprox) engraved anodized aluminum panels on each side of the wall (divided into sections as "buttons"). one "button" opens and closes, one is the "lock" with led indication, others are move left and move right. It uses a servo with its own controller, a belt drive system I have designed and machined, and 24v battery operated with floating charger. The batts will keep it running for 24 hours in power failure. We are installing heat, smoke, and possible gas sensors for overkill safetey since it is very cheap to add the sensors. In an alarm mode, the door will open if programmed to do so, and send the alarm state to a hub in cases where there are more than one unit. Out goal is hotel installs, mostly likely just bathrooms, but we are shooting for the main entry to each room, with optional thumbprint scanners, plus card readers. With the look of the colored glass, the flawless accel/decel motion, the precicion movement with rails and slides as the rail system, it is quite impressive. We have opportunities already to show it to some major developers of vegas type hotels, so we need to get a plan together of how hundereds of systems can be monitored and remote controlled in an emergecny. Lets say there is a know fire, all doors can be opened remotely, if that made sense. Also, killing the servos remotely so the door can freewheel if the controller dies is required as well. There are sonar motion above for safety(retraction), the toucvh sensitive glass cause retraction(or can be used as themain open/close feature- if you like fingerprints). The motor has a current sensing output for retraction should ll else fail.

Thats the idea in a nutshell. We want to get a concept together for the software to monitor and administer info. Basically exactly what you said-- poll to see if everybody is working, then have the ability to have any number of units spit out alarms as needed without waiting for polls.

I like the wireless idea, all battery backup, no chance of wires going down! Almost failsafe too. Transmission through a building? Not sure if that is doable, maybe the length of a halway to a floor level hub?

Hey thanks again for the detailed breakdown. When we get that far I'll see if you have time to help us put it together as it will be a major undertaking.

As it is, I am planning for the home systems with ethernet hookups: remote locking/unlocking of the house, remote monitoring of sensors. Possibly even a java based website for each door that can be monitored in real time with images, entry data, ability to lock it remotely. With the Parallax system, it will even send you an email on alarm or input. Say you are get an alarm a theif just broke in your office, lock them in so they can do more damage! Kids trying to sneak out? Busted, you have detailed report emailed daily!

We are also doing a skylight, with LCD or similar variable light adjustment, and a retractable glass system.
 
Ah man... That sounds sweet! I'll take a dozen!
Thanks for the clear detailed description.

A central floor router would be a great idea. They have about 10 channels of selection; first floor would be channel 1, second 2, tenth would be 10, eleventh could be 1 again. (or some such organization)

You also encode a bunch of info in each packet so a query could tell all about a door. You could also just have all the doors radio their full state each time they operate and otherwise only poll each door once an hour or day about its last hour history, etc.

Hey you could probably do the full servo, all the safeties, fingerprint, etc on a DSPic. And save a bundle, over the Parallax!

Keith Cress
Flamin Systems, Inc.-
 
yeah I am sure there is a better way to put it all on a processor, and lose the $400 proprietary controller I am using now. For today though, it's about making a few led's blink from the micro when I push a button! I'll keep you posted.
 
Be careful about how/where you market this (LOVE the idea, BTW), especially when it comes to safety (e.g., fire). For example, you may not want the doors to open automatically if a fire is detected... sometimes leaving the door closed offers better protection from smoke/heat. Imagine a faulty sensor opening all doors, leaving occupants in awe and/or unprotected from outside harm, like when they're asleep. Also, battery backups fail, so make sure you can force the doors open in case of an emergency, or some other manual override.


Dan - Owner
Footwell%20Animation%20Tiny.gif
 
I knew a guy who was "too dumb to understand" that you can't do multidrop with RS232, and managed to actually do it, controlling up to 32 logically addressable Stamp based boards from a PC serial port. He managed to sell a few, too.




Mike Halloran
Pembroke Pines, FL, USA
 
I always appreciate ideas and suggestions

"you may not want the doors to open automatically if a fire is detected"

Those are decisions that I would not have make fortunately. Also, the system can be locked from the inside and disabled from opening remotely or via sensors, if desired to set it that way. Any option is possible, even to having an internal alarm go off to give the occupant time to make his own decisions as well before sending an alarm out.

There are manual and remote overides for cutting power to the servo motor so it can freewheel, and in that case, the door can easily be slid open by hand.

 
Hey Mike;
I have multi-dropped RS232 sorta. You put 2 transceivers on each of the boards. The data comes onto the board and back off at each board then they all ease-drop at the TTL point. Diode OR the transmit stuff. Works fine but I wouldn't want to go more then a few boards 10 maybe 20.

Keith Cress
Flamin Systems, Inc.-
 
That's how most any EE would do it.

This guy was an ME; OR'ed it on the wire.



Mike Halloran
Pembroke Pines, FL, USA
 
Hiya-

You mentioned:
"As it is, I am planning for the home systems with ethernet hookups: remote locking/unlocking of the house, remote monitoring of sensors. Possibly even a java based website for each door that can be monitored in real time"

And I guess more specifically, if you are doing rework to
an existing structure......

I would look at using existing house wiring (AC to pass a
modulated signal through the AC wiring). You are going
to have to make a connection to it at each room anyway, so
you might as well use it for the signaling as well. At least to my feeble little mind.

Power companies are now delivering broadband over their
circuits today, so the bandwidth can be quite high, so
that should not be a problem. Seems to me that Radio
Shack used to sell little modules that would control lights
and things that would plug into the wall and be controlled
by a central little box that also plugged into the wall.
I also think that there used to be Popular Electronic's
articles on homebuilt ones. A high voltage blocking
cap to get rid of the 60Hz. Pass the control signals as
"high frequency" AC.

You might find that "ethernet" and here actually, I'm
referring to the upper layer protocols TCP/IP that sit on
top of the ethernet, might prove to be a bit much for a
stand alone PIC. The TCP/IP stack takes up quite a bit
of buffer space which is somewhat limited in the PICs.

I agree with "itsmoked" that, considering the number that
you plan to build that you could save over the Parallax.
I would suggest looking at a small ARM processor. And
here's where my bias will kick in. Using ucLinux. It
has quite a good TCP/IP stack on it.

If you want to prototype, you might want to look at a
gumstix (about $90.00 OEM version). Slap a PIC I2C in
the front end of it for the signaling over the AC lines
and you are ready to rock and roll in prototype form very
quickly. Enough for a dog and pony show or proof of
concept.

A couple of other thoughts, you could use the "MAC"
address scheme for the signalling just like ethernet, and
maybe use the broadcast address from IP for talking to
all the controllers at the same time. Continuing the
scheme a little further, you could also leverage the
collision detection scheme from ethernet, however you might
have to change some of the speeds. Ethernet 10baseT is
limited to it's 100yards (or so) due to the fact of the
propogation delays for packets coliding on the wire. The
sending station has to be able to determine that the
packet it sent got steped on. Or you could use a Token
Ring topology and just pass the little token from door
to door as it were.

Alternatively, with a "home" version with just a few
stations (not hundreds), I could easily perceive a system
with just a PIC and a custom protocol for station to
station transmission. You could still have one gumstix
in there acting as the web host for the LAN of the
household if desired.

Hope that these thoughts help. Best of luck with your
efforts and let us know how it goes!

Cheers,

Rich S.
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor