older
hterboard, bend them over & solder. Now if you run those bent over ends alo ng the board & solder them somewhere else again you get a more pull-through resistant fixing. Or perhaps fold them over the daughterboard edges, not t ried that.
random idea: header pins & silicone
Don't have room for that on either board, and peak current will be about 15 amps with risetime a few ns.
I guess I'll just smash the boards together with six nuts and bolts, no solder masks.
-- John Larkin Highland Technology, Inc lunatic fringe electronics
On Apr 21, 2018, John Larkin wrote (in article):
This can work, but needs some elasticity to make up the difference in thermal and elastic coefficients between steel and circuit-board material. Use stainless steel spacers, not aluminum, to reduce the thermal mismatches. (Or do some mechanical tolerance computations to see if aluminium is OK.) The standard dodge is to put some Bellville Washers under the screw head and or the nut, so the clamping pressure is maintained over time and temperature. (Well, that needs a tolerance computation as well. May need belleville springs in series.)
.Do gold-plate the clamping areas. Beware fretting corrosion in tin electrical contacts.
Joe Gwinn
Take a look at these, might work for you
-- Chisolm Republic of Texas
fredag den 20. april 2018 kl. 05.51.29 UTC+2 skrev John Larkin:
why male threads, how about
far too expensive, but I'm sure there are similar
That suggests connecting the electrical and mechanical; six slots, six copper strips (something like .020") fitted tab-in-slot style, to both conduct heat and electricity, and serve as spacers. Both ends will have to be hand-soldered, though.
Multipart (screw/nut/washer/spring) fastenings can work, but depending on it for heatsinking in a small assembly is problematic. Finding room for the wrench to tighten nut #2, very close to nut #1, is another problem. Finding the #80 washer that dropped, is a third. Lifting the tiny washer with human fingers, is a fourth.
Helpful hint: have all your tools demagnetized before assembly if you go with the loose small hardware.
I mean that I may just bolt the boards flat against one another. No spacers, just six nuts and bolts crushing the boards together. Lockwashers of course.
Bellevilles usually have huge forces and sharp edges that can cut into PCBs. Wavy washers are much better against PC boards. But I'll just use ordinary 0-80 ss hardware and common split-ring washers.
As they say at Facebook, done is better than perfect.
-- John Larkin Highland Technology, Inc lunatic fringe electronics
That is interesting, sort of a massive thermal via. But I don't see any dimensions.
-- John Larkin Highland Technology, Inc lunatic fringe electronics
from
On Apr 21, 2018, John Larkin wrote (in article):
Umm. Split-ring spring lockwashers are very sharp, and will tear the board if one attempts disassembly. Reassembly may not achieve full performance due to the damage. A belleville washer is actually less damaging. Nothing wrong with wave washers, if they are stiff enough. In either case, one can have the spring rest on a flat washer on top of the board, rather than directly on the board.
Just bolting the circuit boards together does not in fact solve the thermal coefficient problem, which is due to the large difference in the linear coefficients of thermal expansion between G10 (fiberglass) Z-axis and steel.
G10 also creeps a bit, especially when warm. When warm, the G10 gets soft while it tries to expand. But the steel hardware doesn?t expand as much as the G10, which is thus squished a bit. When cooled, the G10 does not return to the original thickness. Over time, the joint loosens, contact resistance increases, and so on. If average currents are high, this can go into thermal runaway.
This mechanism is what causes aluminum wire to fail when used with electrical components intended for copper.
There is an ideal contact pressure, expressed in psi. (I don?t recall the numerical value offhand, but there are charts in books on electrical contacts.) One designs to maintain this minimum pressure over time and temperature cycles, for the design lifetime of the product.
I assume that it will be mechanically awkward to just bolt the two cards together, because of solder bumps and circuitry and components. A brass or stainless steel washer between boards will help. It?s just a very short spacer.
The best (most robost) answer is to separate electrical from mechanical: Use Mil-Max header pins and sockets soldered to the board for electrical, and three or four 0-80 screws to attach the boards mechanically.
.So long as the warrantee is short enough.
. Joe Gwinn
We have tens of thousands of units in the field that use ss hardware and split-ring lockwashers to mount boards, or to secure parts to boards. We don't see those problems.
The only time I have seen such a problem was a (thankfully) discontinued product that had 120 amps flowing from a PCB to an aluminum bracket. The connection got hot, the FR4 flowed, and the resistance went up, the rest is left to the imagination. Well, I was young and foolish.
Our warantee policy is that if anything fails from a design error or a software bug, we will fix it forever.
-- John Larkin Highland Technology, Inc lunatic fringe electronics
Tolerances on the part vs. staking die and forces on the part look iffy. Normal clearance hole for 0-80 is 0.073. You definitely want the disk deburred on the bottom or it won't sit flat, and if the top has a similar fillet it will tend to extrude into the die.
Might it be easier to make a plated part and solder it in?
--sp
The screw machine people would prefer to make a solderable part and not knurl something that small. That's OK with us... pressing in a part is about as much hassle as soldering.
I'm trying to talk my manufacturing people into attaching the baby board with six nuts; they prefer to plug pins into pin sockets. I'm looking at 3 or 4 amps peak per pin with nanosecond rise times, which is terrifying, to me, for small pin sockets.
-- John Larkin Highland Technology, Inc lunatic fringe electronics
On Apr 22, 2018, John Larkin wrote (in article):
Oof. A 120 amps current will certainly reveal any weaknesses. Quickly.
.OK. You can depend on being in shirt-sleeve environments, so transportation will be the only real test.
I come from the MIL-SPEC world, and our stuff has to survive severe thermal cycling and shake and bake testing, which split ring lock washers et al would not survive. Actually, they would also loosen and fall off unless they did dig into the board and tear it up. Nor do star washers work, and they corrode and fail fairly soon. In any event, we would never get into the warrantee period with such a design.
Joe Gwinn
We do some military stuff. The loophole is COTS. A customer wants something, and we design it, and then at their request we put up a web page and offer it for sale. Then it's COTS.
Star washers chew up PCB pads, but split rings don't.
How do you hold down a PCB?
-- John Larkin Highland Technology, Inc lunatic fringe electronics
If you want to go pin/socket try the mill-max pin receptacle.
is rated at 8 amps with a .040 to .050 pin. I've used a bunch of these in the past (smaller .025 pin size) and no problems.
Not sure what they will do to your rise times. Get a few samples and test.
-- Chisolm Republic of Texas
On Apr 24, 2018, John Larkin wrote (in article):
Yeah, I?ve done that too. Helped the vendor to design the product as well.
.Yep. Well, sorta. The split righs tear things up on disassembly, say for repair.
The typical service life requirement is 20 years, and some systems require 40 years, both with periodic maintenance and repair as needed.
.If air cooled, typically SS through screws into helicoiled bosses machined directly into the aluminum housing. In the old days, it would be metal standoffs and screws. These are never depended upon for electrical grounding.
It is never plastic snaps either - these wouldn?t last ten minutes, and the Customer community would murder any designer who proposed any such thing. Manufacturing does not get a vote here.
If conduction cooled, with wedgelock clamps on the two sides of the boards inserted into grooves machined into the sides of the aluminum chassis. Cooling is usually a big deal.
Joe Gwinn
Do you use lockwashers under the screw heads?
We assume that the screws ground the board to the enclosure or baseplate. Connector shells too.
I've got (totally unreasonable) pushback from manufacturing as regards bolting the boards together. I could pull rank but they would get even somehow.
So we'll put big mouse bites/castellations along two sides of the baby board and solder it onto the bottom side of the main board.
Small version:
New big version idea, 60 mil drills:
-- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
out using
a spring or
?l be
bolts,
ce in
matches.
OK.) The
head and
mperature.
eville
in
ointo
tboard
e due to
rong
have the
ly on
thermal
ear
d steel.
soft
and as much as
return
istance
thermal
?t recall the
nd
ards
ass or
very short
cal: Use
, and
ion
mal
would
id
rrode
eYou may think it is COTS, but if you don't sell any to someone else it's no t COTS and you have to obey the FAR. I've worked in defense contracting an d they can come back years later and cry 'foul' when there was no intent to cheat the government. The guy who will be prosecuting gets paid bonuses o n meeting his goals, so he doesn't care that you didn't intend to cheat the government, only that you broke the rules.
Did you get a lawyer to give you a legal opinion? At least that can keep y ou out of jail when they make your company pay the big fines.
Rick C.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.