EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: max_torque on June 09, 2017, 11:02:10 pm
-
I have a 10mm diameter threaded stud, that will need to carry just over 100A to the pcb copper layers.
As FR4 has a long term creep / yield issue under compression loads, i plan to use a suitable stainless wave washer to apply a controlled / limited preload to a copper washer sitting on the copper of the pcb (pcb will be gold plated). The plan is to tighten the system until a measured crush is achieved on the wave washer, giving a controlled preload and low resistance connection to the pcb copper layer.
So far, so good. But........
I now need to pot the entire assembly in an epoxy potting compound to ensure environmental resistance. My worry is that this epoxy will effectively prevent the wave washer from continuing to apply a suitable preload if the FR4 material yields significantly?
Anyone done something similar or been bitten by FR4 yielding over time etc??
-
I'm curious why you are doing it like that?
Use the proper equipment:
(https://lugsdirect.com/images/SMTBOARDEXAMPLES.jpg)
(http://epsnews.com/wp-content/uploads/2014/05/ErniPowerElements.jpg)
(http://www.mouser.tw/images/microsites/Wurth7461148.png)
-
You really shouldn't bolt a ring terminal or the like directly to a PCB as the epoxy will progressively crush over time leading to less and less contact force. It might take weeks in the case of only using flat washers, or it might take months to years using conical washers (e.g. - Belleville) but eventually the connection will fail due to high resistance.
So I would definitely consider using a board mount terminal like the kind @CM800 suggested, or else solder a bus bar to the board so that the wire to board connection is made against solid metal, not the board. Also keep in mind you'll need to go up substantially in copper plating thickness on the board to carry 100A with a reasonable trace width and temp rise; probably 0.35mm (aka "10oz" in infernal US units).
-
You really shouldn't bolt a ring terminal or the like directly to a PCB as the epoxy will progressively crush over time leading to less and less contact force. It might take weeks in the case of only using flat washers, or it might take months to years using conical washers (e.g. - Belleville) but eventually the connection will fail due to high resistance.
So I would definitely consider using a board mount terminal like the kind @CM800 suggested, or else solder a bus bar to the board so that the wire to board connection is made against solid metal, not the board. Also keep in mind you'll need to go up substantially in copper plating thickness on the board to carry 100A with a reasonable trace width and temp rise; probably 0.35mm (aka "10oz" in infernal US units).
Design has short term tested fine with 2 Oz copper because the traces can be massive (no space concerns) The pcb starts off as 1.6mm thick, so as long as i have more than 1.6mm of spring compliance then the pcb can never be completely unloaded (i've currently spec'd wave washers that have 3mm of total crush) Obviously, as the washer extends the total clamp load falls of course.
Unfortunately i have use this type of arrangement because the design has to be a seamless swap into an existing system that uses a bolted stud, otherwise i would have used a press fit / solder power stud of some type (i've used these many times on high current inverter pcbs etc)
The question really is around the potting preventing the wave washer from expanding? I might need to use a soft silicon potting around the washer to allow for expansion before then hard potting the rest of the device (it needs an epoxy hard pot for mechanical support in the end application)
-
Unfortunately i have use this type of arrangement because the design has to be a seamless swap into an existing system that uses a bolted stud, otherwise i would have used a press fit / solder power stud of some type (i've used these many times on high current inverter pcbs etc)
How about seamless-ish?
http://www.we-online.com/web/en/wuerth_elektronik/news_weg/news_archiv_2014_weg/news-detail-we-gruppe_52610.php (http://www.we-online.com/web/en/wuerth_elektronik/news_weg/news_archiv_2014_weg/news-detail-we-gruppe_52610.php)
https://powerelement.we-online.de/fullsearch;jsessionid=117CE9C109ECE5964FFC18A65E81EF0B?productGroupId=22 (https://powerelement.we-online.de/fullsearch;jsessionid=117CE9C109ECE5964FFC18A65E81EF0B?productGroupId=22)
Wurth has some nice press fit (and solder) terminals that are pretty small. Also, you can abuse some of them a bit and use them as solderable parts.
-
I wish i could use a 'proper' solution but i can't. My client has said i must use the existing power stud, and that my Power Distribution Unit has to be an 'invisible' swap for the existing h/w (which is a mess of ring terminals, wires and separate fuse holders etc).
So, to fly, i need to get about 125A off an M10 brass threaded stud and onto the copper of a pcb.
So, back to the original question :-DD Anyone had long term issues with potting spring or clamping components?
-
FR4 will reflow until it has no force on it.. If need be, it will cut through the board..
I would go with a brass insert that goes trough the board and is soldered to the copper.
That way force would not be against FR4..
I had problems with this on a measly TO3 case with two M3 screws... A bit of heat and washer ended embedded in the PCB..
M10 will apply much bigger force...
-
The question really is around the potting preventing the wave washer from expanding? I might need to use a soft silicon potting around the washer to allow for expansion before then hard potting the rest of the device (it needs an epoxy hard pot for mechanical support in the end application)
It seems that you have a solution. What is wrong with this?
-
Couldn't you just fit a brass nut (maybe a half thickness nut) onto the bolt first and solder that to the PCB? Then fit ring terminals followed by another nut.
-
FYI,
(https://www.anochrome.com/wp-content/uploads/2011/12/table_torque_tension_control.jpg)
check out these clamping forces...
-
FR4 will reflow until it has no force on it.. If need be, it will cut through the board..
Given enough clamp load maybe. But, i have a brass (not steel) coarse thread, with a maximum 8Nm tightening torque limit. Putting in place the wave washer limits the max clamp load, and that load is spread over around 122 mm^2, so we are not talking massive clamp loads here!
I've got a test sample done up to 20Nm (over 2 x max recommended) without any force limiters and i'm going to keep an eye on the creep of that sample over the next few weeks (with an elevated temp cycle at the end if it passes room temp tests)
The mechanism for slow FR4 creep is that the glass rovings slide sideways as the epoxy that holds them in place slowly creeps, but with the surface area of my geometry i can't see how those glass rovings can completely escape sideways, the distance is just too large (20mm dia washer), so i'm not expecting the clamp load to drop to nothing?
-
The question really is around the potting preventing the wave washer from expanding? I might need to use a soft silicon potting around the washer to allow for expansion before then hard potting the rest of the device (it needs an epoxy hard pot for mechanical support in the end application)
It seems that you have a solution. What is wrong with this?
It's a major PITA to actually do! Either it needs a hand operation to cover the fasteners with soft silicon before the main potting process, or i have to arrange some sort of "dam" to allow that soft silicon to cure first etc. Either way a messy, difficult, and somewhat hit-n-miss operation
-
Does the customer understand they are sacrificing reliability for only a minor change? (for all intensive purposes, no change)
-
Since you cant use a stud to screw into, can you afford a bit of increase in thickness? if so you can tru and get away with a solder in PEM of the shortest height you can. If using the pem's threads still are not an option then oversize it and the bolt can still be passed through and tightened down.
http://www.pemnet.com/fastening-products/fasteners-for-mounting-into-printed-circuit-boards-2/ (http://www.pemnet.com/fastening-products/fasteners-for-mounting-into-printed-circuit-boards-2/)
-
Another option would probably be a heavy copper PCB and some creative milling.
http://hackaday.com/2016/12/16/massive-20-oz-copper-pcb-enables-electric-racing/ (http://hackaday.com/2016/12/16/massive-20-oz-copper-pcb-enables-electric-racing/)
-
Put lots of plated through holes around the stud to take the end load, then Belville washers, nothing else. These can be stacked in parallel or, as here, in series to increase the compliance distance.
I have disintegrated endless bits of electronics that try to take high current loads off a PCB, only Belville washers haven't burnt the PCB over the years of use. Incidentally those terminals shown by CM800 have not really been successful either, for some reason the solder crystalises and fractures. Any ideas why? My thoughts were that tin at high temperatures will crystallise, and having mechanical stress as well, a 35mm cable attached to the PCB really doesn't help. Could also be the PCB gets hot and flows, causing more mechanical stress on the soldered joints.
It isn't an easy problem to solve. I personally would try to fan the cable out to multiple connections so each one only has to carry 5A or so. Just a thought.
-
I have disintegrated endless bits of electronics that try to take high current loads off a PCB, only Belville washers haven't burnt the PCB over the years of use. Incidentally those terminals shown by CM800 have not really been successful either, for some reason the solder crystalises and fractures. Any ideas why? My thoughts were that tin at high temperatures will crystallise, and having mechanical stress as well, a 35mm cable attached to the PCB really doesn't help. Could also be the PCB gets hot and flows, causing more mechanical stress on the soldered joints.
The majority of the terminals I shown there actually are press-fit into the PCB without solder.
They are very reliable.
Also I can't imagine you're actually meant to put any mechanical stress between connector & PCB via the cable. I'd say that you should always tie a cable down in a neutral position if it's high current.
-
Yup, i've already got around 60 via's around the main stud hole, in order to try to share current between the top and bottom layers, so these will also act as a forest of supporting copper pillars for the clamp loading. As the entire structure is hard potted there shouldn't be any significant mechanical strain on the pcb or stud where it penetrates the pcb. My main concern is simply maintaining the clamp load over time.......
-
On some of the 100-200A servo drives I see being sold, they are connected to using a heavy copper pcb with the terminals stacked like so:
NUT
SPLIT LOCK WASHER
(CABLE CRIMP)
COPPER PLATE
PCB
COPPER PLATE WITH BOLT FACING UP.
These are mil-spec drives so evidently go just fine.