Hi, all, So we're doing this multispectral sensor for finding smouldering fires in heavy agricultural equipment. It has to operate outdoors in a high-vibration environment over a case temperature range of about -20C to +70C. The board is conformal-coated, but we need to avoid condensation to keep the window from fogging up and avoid board leakage that could trash the performance of the NIR channel, which uses a very high-Z TIA. Accordingly we're using IP68-rated M12 connectors and customized enclosures with fancy O-ring seals, self-sealing screws, and carefully-placed filtered vents to prevent pressure differences that could transport liquid water to the interior of the box. Because of the air exchange due to the vent, we're filling a major fraction of the interior volume with Linde 5A molecular sieve in a Tyvek envelope. (The envelopes are actually intended for use in steam autoclaves for sterilizing surgical instruments, but they're just right for this job.) We calculate on fairly pessimistic assumptions that the interior ought to stay adequately dry for 25 years of service, which is fine. There's a T/H sensor on the board, so the firmware can keep track of whether there's any danger of condensation and report a fault if there is. So far it tests out well, but there are one or two loose ends that need to be taken care of before we deploy the pilot run. Which brings me to the question: what connectors to use in hot/cold and very dry environments with lots of vibration. We need 8 positions including power, ground, half-duplex RS485 MODBUS, and some analog signalling outputs to talk to a PLC. Nylon famously gets brittle and fails in that situation, so we're looking at these: <http://www.jst-mfg.com/product/pdf/eng/ePA-F.pdf> Specifically, we're tentatively planning to use the ones with glass-filled polybutylene terephthalate housings: through-hole jacks and crimp-on plugs (not the insulation-displacement ones). The solder will be normal Sn63 because they don't grow cotton in Europe. ;) This is not our usual corner of the design space, so I'd appreciate comments from folks with relevant experience. Thanks Phil Hobbs -- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics Briarcliff Manor NY 10510 http://electrooptical.net http://hobbs-eo.com
Connectors for high-vibration environments with extremely low humidity
Started by ●August 11, 2020
Reply by ●August 11, 20202020-08-11
On 11/08/2020 06:45, Phil Hobbs wrote:> Hi, all, > > So we're doing this multispectral sensor for finding smouldering fires > in heavy agricultural equipment. It has to operate outdoors in a > high-vibration environment over a case temperature range of about -20C > to +70C. The board is conformal-coated, but we need to avoid > condensation to keep the window from fogging up and avoid board leakage > that could trash the performance of the NIR channel, which uses a very > high-Z TIA. > > Accordingly we're using IP68-rated M12 connectors and customized > enclosures with fancy O-ring seals, self-sealing screws, and > carefully-placed filtered vents to prevent pressure differences that > could transport liquid water to the interior of the box. > > Because of the air exchange due to the vent, we're filling a major > fraction of the interior volume with Linde 5A molecular sieve in a Tyvek > envelope. (The envelopes are actually intended for use in steam > autoclaves for sterilizing surgical instruments, but they're just right > for this job.) We calculate on fairly pessimistic assumptions that the > interior ought to stay adequately dry for 25 years of service, which is > fine. There's a T/H sensor on the board, so the firmware can keep track > of whether there's any danger of condensation and report a fault if > there is. > > So far it tests out well, but there are one or two loose ends that need > to be taken care of before we deploy the pilot run. Which brings me to > the question: what connectors to use in hot/cold and very dry > environments with lots of vibration. We need 8 positions including > power, ground, half-duplex RS485 MODBUS, and some analog signalling > outputs to talk to a PLC. > > Nylon famously gets brittle and fails in that situation, so we're > looking at these: <http://www.jst-mfg.com/product/pdf/eng/ePA-F.pdf> > > Specifically, we're tentatively planning to use the ones with > glass-filled polybutylene terephthalate housings: through-hole jacks and > crimp-on plugs (not the insulation-displacement ones). The solder will > be normal Sn63 because they don't grow cotton in Europe. ;) > > This is not our usual corner of the design space, so I'd appreciate > comments from folks with relevant experience. > > Thanks > > Phil HobbsGlenair connectors are often specified for high vibration/temperature downhole use, a bit like the old saw "nobody ever got sacked for buying IBM". MWDM2L-9SCBRP-.110-513 for example, that may be over the top but they do several other types. Check your bank balance first, and lead-free solder is tougher. -- Cheers Clive
Reply by ●August 11, 20202020-08-11
A client has the same application but in rugged automotive conditions (mining equipment, etc.). They use a thermopile sensor. (They're detecting direct flame; sounds like your case needs more sensitivity, so that probably wouldn't go, and explains the TIA, with, I guess, a long wave photodiode or pyroelectric sensor.) I forget what the minimum temperature is (maybe it was more or less than your requirement?), but they use M-series Turck connectors throughout. Pigtail type, leads soldered directly to the board. Don't remember what plastics they're made of or if they say, but they're rated for the service in any case. And as much potting as possible, no direct route for moisture to get in. They do have problems with vibration of connectors causing dropouts of power or signal. A recent change has been to put an MCU inside the boxes, so instead of a DC signal, a serial keepalive message is sent, and the system can be more aware of its integrity as well as more forgiving of momentary dropout. Not sure what you're showing with the rectangular header -- internal board-to-board wiring? External connections?? Tim -- Seven Transistor Labs, LLC Electrical Engineering Consultation and Design Website: https://www.seventransistorlabs.com/ "Phil Hobbs" <pcdhSpamMeSenseless@electrooptical.net> wrote in message news:2746d762-6fd6-b65d-69e4-0d6946a084b7@electrooptical.net...> Hi, all, > > So we're doing this multispectral sensor for finding smouldering fires in > heavy agricultural equipment. It has to operate outdoors in a > high-vibration environment over a case temperature range of about -20C to > +70C. The board is conformal-coated, but we need to avoid condensation to > keep the window from fogging up and avoid board leakage that could trash > the performance of the NIR channel, which uses a very high-Z TIA. > > Accordingly we're using IP68-rated M12 connectors and customized > enclosures with fancy O-ring seals, self-sealing screws, and > carefully-placed filtered vents to prevent pressure differences that could > transport liquid water to the interior of the box. > > Because of the air exchange due to the vent, we're filling a major > fraction of the interior volume with Linde 5A molecular sieve in a Tyvek > envelope. (The envelopes are actually intended for use in steam > autoclaves for sterilizing surgical instruments, but they're just right > for this job.) We calculate on fairly pessimistic assumptions that the > interior ought to stay adequately dry for 25 years of service, which is > fine. There's a T/H sensor on the board, so the firmware can keep track > of whether there's any danger of condensation and report a fault if there > is. > > So far it tests out well, but there are one or two loose ends that need to > be taken care of before we deploy the pilot run. Which brings me to the > question: what connectors to use in hot/cold and very dry environments > with lots of vibration. We need 8 positions including power, ground, > half-duplex RS485 MODBUS, and some analog signalling outputs to talk to a > PLC. > > Nylon famously gets brittle and fails in that situation, so we're looking > at these: <http://www.jst-mfg.com/product/pdf/eng/ePA-F.pdf> > > Specifically, we're tentatively planning to use the ones with glass-filled > polybutylene terephthalate housings: through-hole jacks and crimp-on plugs > (not the insulation-displacement ones). The solder will be normal Sn63 > because they don't grow cotton in Europe. ;) > > This is not our usual corner of the design space, so I'd appreciate > comments from folks with relevant experience. > > Thanks > > Phil Hobbs > > -- > Dr Philip C D Hobbs > Principal Consultant > ElectroOptical Innovations LLC / Hobbs ElectroOptics > Optics, Electro-optics, Photonics, Analog Electronics > Briarcliff Manor NY 10510 > > http://electrooptical.net > http://hobbs-eo.com >
Reply by ●August 11, 20202020-08-11
Phil Hobbs wrote:> Hi, all, > > So we're doing this multispectral sensor for finding smouldering fires > in heavy agricultural equipment. It has to operate outdoors in a > high-vibration environment over a case temperature range of about -20C > to +70C. The board is conformal-coated, but we need to avoid > condensation to keep the window from fogging up and avoid board leakage > that could trash the performance of the NIR channel, which uses a very > high-Z TIA. > > Accordingly we're using IP68-rated M12 connectors and customized > enclosures with fancy O-ring seals, self-sealing screws, and > carefully-placed filtered vents to prevent pressure differences that > could transport liquid water to the interior of the box. > > Because of the air exchange due to the vent, we're filling a major > fraction of the interior volume with Linde 5A molecular sieve in a Tyvek > envelope. (The envelopes are actually intended for use in steam > autoclaves for sterilizing surgical instruments, but they're just right > for this job.) We calculate on fairly pessimistic assumptions that the > interior ought to stay adequately dry for 25 years of service, which is > fine. There's a T/H sensor on the board, so the firmware can keep track > of whether there's any danger of condensation and report a fault if > there is. > > So far it tests out well, but there are one or two loose ends that need > to be taken care of before we deploy the pilot run. Which brings me to > the question: what connectors to use in hot/cold and very dry > environments with lots of vibration. We need 8 positions including > power, ground, half-duplex RS485 MODBUS, and some analog signalling > outputs to talk to a PLC. > > Nylon famously gets brittle and fails in that situation, so we're > looking at these: <http://www.jst-mfg.com/product/pdf/eng/ePA-F.pdf> > > Specifically, we're tentatively planning to use the ones with > glass-filled polybutylene terephthalate housings: through-hole jacks and > crimp-on plugs (not the insulation-displacement ones). The solder will > be normal Sn63 because they don't grow cotton in Europe. ;) > > This is not our usual corner of the design space, so I'd appreciate > comments from folks with relevant experience. > > Thanks > > Phil Hobbs >Sn63 is distinctly Europe non-RoHS!
Reply by ●August 11, 20202020-08-11
>Sn63 is distinctly Europe non-RoHS!Hence my comment about nobody in the EU growing cotton. For this one I don't have to care about Ro so-called HS. Cheers Phil Hobbs
Reply by ●August 11, 20202020-08-11
On 2020-08-11 03:29, Clive Arthur wrote:> On 11/08/2020 06:45, Phil Hobbs wrote: >> Hi, all, >> >> So we're doing this multispectral sensor for finding smouldering fires >> in heavy agricultural equipment. It has to operate outdoors in a >> high-vibration environment over a case temperature range of about -20C >> to +70C. The board is conformal-coated, but we need to avoid >> condensation to keep the window from fogging up and avoid board >> leakage that could trash the performance of the NIR channel, which >> uses a very high-Z TIA. >> >> Accordingly we're using IP68-rated M12 connectors and customized >> enclosures with fancy O-ring seals, self-sealing screws, and >> carefully-placed filtered vents to prevent pressure differences that >> could transport liquid water to the interior of the box. >> >> Because of the air exchange due to the vent, we're filling a major >> fraction of the interior volume with Linde 5A molecular sieve in a >> Tyvek envelope. (The envelopes are actually intended for use in steam >> autoclaves for sterilizing surgical instruments, but they're just >> right for this job.) We calculate on fairly pessimistic assumptions >> that the interior ought to stay adequately dry for 25 years of >> service, which is fine. There's a T/H sensor on the board, so the >> firmware can keep track of whether there's any danger of condensation >> and report a fault if there is. >> >> So far it tests out well, but there are one or two loose ends that >> need to be taken care of before we deploy the pilot run. Which brings >> me to the question: what connectors to use in hot/cold and very dry >> environments with lots of vibration. We need 8 positions including >> power, ground, half-duplex RS485 MODBUS, and some analog signalling >> outputs to talk to a PLC. >> >> Nylon famously gets brittle and fails in that situation, so we're >> looking at these: <http://www.jst-mfg.com/product/pdf/eng/ePA-F.pdf> >> >> Specifically, we're tentatively planning to use the ones with >> glass-filled polybutylene terephthalate housings: through-hole jacks >> and crimp-on plugs (not the insulation-displacement ones). The solder >> will be normal Sn63 because they don't grow cotton in Europe. ;) >> >> This is not our usual corner of the design space, so I'd appreciate >> comments from folks with relevant experience. >> >> Thanks >> >> Phil Hobbs > > Glenair connectors are often specified for high vibration/temperature > downhole use, a bit like the old saw "nobody ever got sacked for buying > IBM". > > MWDM2L-9SCBRP-.110-513 for example, that may be over the top but they do > several other types. > > Check your bank balance first, and lead-free solder is tougher. >Thanks. I was perhaps unclear--the connector I'm concerned about is inside the case, connecting the short wiring harness from the M12 connector to the board. We're expecting to put some heat shrink tubing on the wires to damp out vibration. The issue we're concerned with at the moment is preventing the connectors from crumbling into dust, which is the likely fate of nylon connector shells in this environment. Thanks Phil Hobbs -- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics Briarcliff Manor NY 10510 http://electrooptical.net http://hobbs-eo.com
Reply by ●August 11, 20202020-08-11
On 2020-08-11 07:16, Tim Williams wrote:> A client has the same application but in rugged automotive conditions > (mining equipment, etc.).� They use a thermopile sensor.� (They're > detecting direct flame; sounds like your case needs more sensitivity, so > that probably wouldn't go, and explains the TIA, with, I guess, a long > wave photodiode or pyroelectric sensor.)We have a visible channel, a SWIR channel (0.9-1.7 um), and a MWIR channel (3-5 um). The SWIR detector is 0.3 mm in diameter, so it needs a very high-Z TIA. (It uses two 100M resistors in series, shunted by a series pair of very low leakage diodes so that it doesn't rail in bright light.)> > I forget what the minimum temperature is (maybe it was more or less than > your requirement?), but they use M-series Turck connectors throughout. > Pigtail type, leads soldered directly to the board.� Don't remember what > plastics they're made of or if they say, but they're rated for the > service in any case.� And as much potting as possible, no direct route > for moisture to get in.I don't have any experience using potting, so I'd be reluctant to use it for this one. We do have to use goop between the board and box to keep it from vibrating much. Mining is a very dirty business indeed. I've been down a few working mines--underground, a mercury mine in BC (Pinchi Lake) and a gold mine in the Northwest Territories (the Con) as well as an open pit lead/zinc mine also in the NWT (Pine Point). My Dad used to work for the old Cominco [not the company currently trading under that name], and he took me with him on a tour of their operations in the North. It was pretty cool--I was 12 at the time, and got to hold an 80-pound gold bar. (It was 75% gold and 25% silver iirc.)> They do have problems with vibration of connectors causing dropouts of > power or signal.� A recent change has been to put an MCU inside the > boxes, so instead of a DC signal, a serial keepalive message is sent, > and the system can be more aware of its integrity as well as more > forgiving of momentary dropout.Yeah, ours have an LPC845 Cortex M0+ in them. Nice part--32 MHz, with a gigantic pin mux so you can have almost any function on (almost) any pin.> Not sure what you're showing with the rectangular header -- internal > board-to-board wiring?� External connections??It's to go from the pigtailed M12 to the board. We've been going back and forth about soldering the wires directly to the board, but have heard stories about vibration causing fatigue failure at the stress concentration point where the tinned region ends, as well as anecdotal wisdom about the goodness of crimped connections in environments like that. Cheers Phil Hobbs -- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics Briarcliff Manor NY 10510 http://electrooptical.net http://hobbs-eo.com
Reply by ●August 11, 20202020-08-11
On 11/08/2020 19:01, Phil Hobbs wrote: <snip>> > Thanks. I was perhaps unclear--the connector I'm concerned about is > inside the case, connecting the short wiring harness from the M12 > connector to the board. We're expecting to put some heat shrink tubing > on the wires to damp out vibration. > > The issue we're concerned with at the moment is preventing the > connectors from crumbling into dust, which is the likely fate of nylon > connector shells in this environment. > > Thanks > > Phil HobbsGlenair do that sort of thing. 'Latching Microstrips' they call them. Good quality stuff. https://www.glenair.com/micro-d/q.htm -- Cheers Clive
Reply by ●August 11, 20202020-08-11
On 11/08/2020 19:23, Phil Hobbs wrote: <snip>> > We've been going back and forth about soldering the wires directly to > the board, but have heard stories about vibration causing fatigue > failure at the stress concentration point where the tinned region ends, > as well as anecdotal wisdom about the goodness of crimped connections in > environments like that.Standard way is to loop the wire through holes adjacent to the solder pads, so through the board with insulated wire, then double back through a pad to solder. That is reliable. If you have heavy parts - eg large through hole MOSFETs, then don't solder them to the PCB, use looped wires soldered to their pins. /Extreme/ vibration calls for potting, but a conformal silicone coating is fine otherwise. And lead-free solder makes a stronger joint. Place small parts along the least flexy direction, use Flexicap capacitors, expect microphony. No BGAs, and be more generous with the solder than normal. -- Cheers Clive
Reply by ●August 11, 20202020-08-11
"Phil Hobbs" <pcdhSpamMeSenseless@electrooptical.net> wrote in message news:7023e4dd-38c1-165a-2aed-75078864178c@electrooptical.net...> I don't have any experience using potting, so I'd be reluctant to use it > for this one. We do have to use goop between the board and box to keep > it from vibrating much.Yeah, automotive does that a lot. Not uncommon to see modules with electrolytics sticking up from the muck. Might do, then, to have just a little more, enough to cover where the wires come out -- or put some extra goop around them? And yeah, NASA doesn't like tinned wire either, indeed they have a whole methodology for doing that correctly (which must be pretty inevitable for the popularity of solder-cup MIL terminals?). If they can do it, it's definitely possible; whether it's worth following that procedure, or maybe a watered-down version of it, is a potentially costly question though. Alas, I don't have that experience to offer... Also a possibility, solder-in crimp termals. Seen 'em on ATX PSUs plenty of times. Probably more of an assembly expedient than for reliability though? (The crimp is basically just the barrel, no socket/pin. The barrel and wire is what goes into the board to get soldered. There's usually a retention tab, I assume to hold it in place for wave soldering.) I think in my example, the connections are all hand soldered. Which has to be a big adder to their assembly cost. It's not mass production quantities, and they don't seem to have much competition (and for an end product that doesn't have much competition in turn), so I guess they don't mind. YMMV. Oh also, I forget what they do with the IR sensors, if they're fully potted, or partial, and if they have a seal around the optics. Might be the latter. Yet another thing to think about...> Mining is a very dirty business indeed. I've been down a few working > mines--underground, a mercury mine in BC (Pinchi Lake) and a gold mine in > the Northwest Territories (the Con) as well as an open pit lead/zinc > mine also in the NWT (Pine Point). My Dad used to work for the old > Cominco [not the company currently trading under that name], and he took > me with him on a tour of their operations in the North. It was pretty > cool--I was 12 at the time, and got to hold an 80-pound gold bar. (It was > 75% gold and 25% silver iirc.)Very cool. Unless you were in one of those super deep mines that's over 40�C and full humidity, but other than that... :^) They also like to wash down the trucks, and I mean using whatever is handy: steam, seawater... Electrolysis in connectors (even if they're IP67) is sometimes a problem too. At least you wouldn't have that problem I guess. :)> Yeah, ours have an LPC845 Cortex M0+ in them. Nice part--32 MHz, with a > gigantic pin mux so you can have almost any function on (almost) any pin. >Ah, yep, cheap too. Though I've heard they're Microchip levels of buggy. (Haven't looked at them in enough detail to know that myself.) We use a ton of STM32s: affordable enough, easy to use (even supported by Arduino, should you happen upon that route), not too many bugs. Personally, I like AVR XMEGA, charming 8-bit machines, pricey though they are. Fortunately, as a hardware guy, I just play with them; I'm in no danger of putting one into production. :^) Tim -- Seven Transistor Labs, LLC Electrical Engineering Consultation and Design Website: https://www.seventransistorlabs.com/
