Post by thread:[ee] What's a reliable high-current connection?

I'm designing a PCB that needs to be able to handle at least 5 A, preferably up to 15 A. I may be insane. Right now I'm using PCB screw terminals, but the ones I'm using are rated at no more than 4 A. This is ok as in the prototype I'm only handling 1 A or so max. I thought of spade lugs, which can handle 15 A easily, but they tend to be somewhat bulky. Which are a better connector? What are people using for high current applications? Or, more likely, am I just insane here and need to use real wires? --Michael

> I'm designing a PCB that needs to be able to handle at least 5 A, > preferably up to 15 A. > Right now I'm using PCB screw terminals, but the ones I'm using are > rated at no more than 4 A. This is ok as in the prototype I'm only > handling 1 A or so max. I thought of spade lugs, which can handle > 15 A easily, but they tend to be somewhat bulky. There is a world full of high power connectors and a competent connector supplier will be able to advise. If it's for in house use or where specialists use the connector then an XLR connector may meet your need. Very available, OK price. eg the dread Maplin say http://www.maplin.co.uk/Free_UK_Delivery/3-Pin_XLR_Plug_1025/3-Pin_XLR_Plug_1025.htm They rate this one at 15A, 120 VAC. DC rating probably similar if not using as a power make break switch very often. Contact resistance of 5 milliohm max would give under 0.1v drop per contact at 15A and about 2.5 watts loss for two contacts. You can buy versions with more pins which would make it less likely that someone would plug a microphone into it :-). A standard IEC mains connector may work well for you. Again, beware what people may plug into it :-(. Do note that specs may vary with specific product and checking the proper spec sheet is always wise. RM

M Graff wrote: > I'm designing a PCB that needs to be able to handle at least 5 A, > preferably up to 15 A. I may be insane. > > Right now I'm using PCB screw terminals, but the ones I'm using are rated > at no more than 4 A. This is ok as in the prototype I'm only handling 1 A > or so max. I thought of spade lugs, which can handle 15 A easily, but > they tend to be somewhat bulky. > > Which are a better connector? What are people using for high current > applications? Or, more likely, am I just insane here and need to use real > wires? Try Phoenix Contact terminal blocks (page 204 of Digikey catalog). They have screw terminals rated at up to 24A. Best regards, Vitaliy

> I'm designing a PCB that needs to be able to handle at least 5 A, > preferably up to 15 A. I may be insane. > > Right now I'm using PCB screw terminals, but the ones I'm using are > rated at no more than 4 A. This is ok as in the prototype I'm only > handling 1 A or so max. I thought of spade lugs, which can > handle 15 A > easily, but they tend to be somewhat bulky. > > Which are a better connector? Are you sure the *connector* is your main problem? I am no high-current expert, but wouldn't the soldering of the connector to the PCB trace, and the PCB traces themselves, be a bigger problem? Wouter van Ooijen -- ------------------------------------------- Van Ooijen Technische Informatica: http://www.voti.nl consultancy, development, PICmicro products docent Hogeschool van Utrecht: http://www.voti.nl/hvu

M Graff wrote: > I'm designing a PCB that needs to be able to handle at least 5 A, > preferably up to 15 A. [...] Which are a better connector? What are > people using for high current applications? Conxall makes some nice connectors that can go up in those ranges. IIRC Digikey carries some models, otherwise they have a good rep that also sells smaller quantities (seems to be a general problem with connectors). > Or, more likely, am I just insane here [...]? That's definitely an unknowable, and as such not appropriate to discuss here :) Gerhard

Wouter van Ooijen wrote: > Are you sure the *connector* is your main problem? I am no high-current > expert, but wouldn't the soldering of the connector to the PCB trace, > and the PCB traces themselves, be a bigger problem? It's all solvable, and the only things that remain as problems are the ones you couldn't solve :) Seriously, of course you need to give special consideration to everything where such currents flow. Why do you think the soldering is a special problem, though? Of course assuming that the pins of a connector that's made for that current are thick enough to carry it. Of course the traces shouldn't be made to act as fuses :) Don't use 17u copper (seems to become standard for "normal" boards), use 35u or even 70u, depending on what else you have on the board. And keep them short; good placement is key. Gerhard

Xiaofan Chen wrote: > On 12/1/05, Vitaliy <spamKILLspammaksimov.org> wrote: >> Try Phoenix Contact terminal blocks (page 204 of Digikey catalog). They have >> screw terminals rated at up to 24A. >> >> Best regards, >> >> Vitaliy > > I think that is a good choice. Phoneix Contact terminal blocks are very good. Make sure to follow their datasheet instructions for torque -- have seen a few Phoenix connectors burnt. (-48 VDC @ 15A) Torque specs were not followed correctly and wires inserted into the connector not stripped to proper length during assembly. Nate

> Seriously, of course you need to give special consideration > to everything > where such currents flow. Why do you think the soldering is a special > problem, though? Of course assuming that the pins of a > connector that's > made for that current are thick enough to carry it. I am no expert at high currents (I am more into bits and bytes), so don't trust my intuition. My fear was: no matter how wide you make the PCB traces, all the current has to pass through each circle surrounding the connector pin (assuming the connection is a pin), as fars as these circles are not covered by solder (in which case the solder would probably take a large share of the current). When the soldered area is small the smallest such circle is also small. Am I seeing ghosts? Wouter van Ooijen -- ------------------------------------------- Van Ooijen Technische Informatica: http://www.voti.nl consultancy, development, PICmicro products docent Hogeschool van Utrecht: http://www.voti.nl/hvu

----- Original Message ----- From: "M Graff" <.....explorer-piclistKILLspam.....flame.org> To: "Microcontroller discussion list - Public." <EraseMEpiclistspam_OUTTakeThisOuTmit.edu> Sent: Thursday, December 01, 2005 5:49 AM Subject: [ee] What's a reliable high-current connection? > I'm designing a PCB that needs to be able to handle at least 5 A, > preferably up to 15 A. I may be insane. > > Right now I'm using PCB screw terminals, Either use multiple conectors as in car alternators (2 or more 15A spade terminals) solder the cables directly (make sure the PCB traces are wide and thick enough not to fuse!) or mount high curent components off-board and 'cold-switch' them, i.e run control wiring only from the PCB; this has the advantage of moving (usually) heavy and or hot running lumps off the PCB which can be made thinner.You'll also get less voltage drop in the current carrying cables...

Wouter van Ooijen wrote: >> Why do you think the soldering is a special problem, though? Of course >> assuming that the pins of a connector that's made for that current are >> thick enough to carry it. > > I am no expert at high currents (I am more into bits and bytes), so > don't trust my intuition. My fear was: no matter how wide you make the > PCB traces, all the current has to pass through each circle surrounding > the connector pin (assuming the connection is a pin), as fars as these > circles are not covered by solder (in which case the solder would > probably take a large share of the current). When the soldered area is > small the smallest such circle is also small. I assume those "circles" you are talking about are the pads of the connector pins? Let's say you need 10 A and choose 10 mm wide traces. With 35 um copper that's plenty and on the safe side. That trace has then a cross section of 0.35 mm^2. (Sorry, I can't help it: see why it's easier to use the proper unit for thickness instead of ounces of copper per square foot? :) Now just imagine that cross section. 0.35 mm^2 is not a whole lot, and actually the pads (with solder) and pins are a lot thicker than the trace itself. There may of course be a certain "current concentration" around the pad, where the effective cross section through which the current flows is a bit less. (Ah, now I think I understand what you meant with those circles: actually the cross-sectional cylinders through which the current flows when approaching a pin.) But a cylinder with 0.35 mm^2 area has a diameter of 3.2 mm (at a height of 35 um). That's usually where already the solder starts to increase the thickness of the conductive material and therefore the effective area. Also the whole thing is a thermal question. There may be a locally higher current density around a pin that creates more heat locally, but there's also enough copper around to carry that heat away. So in practice, this is not really a problem. That is, it may become a problem at /high/ currents (that I've never worked with), but not at 10 A :) Gerhard

Wouter wrote: > Are you sure the *connector* is your main problem? I am no high-current > expert, but wouldn't the soldering of the connector to the PCB trace, > and the PCB traces themselves, be a bigger problem? http://www.circuitcalculator.com/4pcb/trace_width_calculator.php I'm not an expert either, but plugging in 15A produces a trace width of 274 mils (for 2 oz copper). However, this is assuming long traces, so if you keep the traces short and connect them to large heat sinks, you won't have to make them as wide. One trick is to use large pads, and connect the pads with wire jumpers. Best regards, Vitaliy

On 12/1/05, Wouter van Ooijen <wouterspam_OUTvoti.nl> wrote: > > I'm designing a PCB that needs to be able to handle at least 5 A, > > preferably up to 15 A. I may be insane. > > > > Right now I'm using PCB screw terminals, but the ones I'm using are > > rated at no more than 4 A. This is ok as in the prototype I'm only > > handling 1 A or so max. I thought of spade lugs, which can > > handle 15 A > > easily, but they tend to be somewhat bulky. > > > > Which are a better connector? > > Are you sure the *connector* is your main problem? I am no high-current > expert, but wouldn't the soldering of the connector to the PCB trace, > and the PCB traces themselves, be a bigger problem? Not exactly, because all respectable power supplies have a doubled thick of copper. HP did that in the past by a quite thick gold layer. Chinese power supplies designs are tottaly upside down talking about the PCB traces. Those PCB traces are running usually hot, but there are very short and have up to 1mm thick of solder. The problems are indeed both the PCB traces but also the connector. Soldering of a group of wires directly in the PCB is the best choice. cheers, Vasile

On Thu, 1 Dec 2005, Vitaliy wrote: > However, this is assuming long traces, so if you keep the traces short and > connect them to large heat sinks, you won't have to make them as wide. One > trick is to use large pads, and connect the pads with wire jumpers. Some high current circuits are made by bolting or riveting strips of copper conductor to insulating board. The hollow bolts are used as holes to solder the parts into. This is really old technology, dates back to wooden and marble boards. Peter

Wouter van Ooijen 11:16 2005-12-01: >Are you sure the *connector* is your main problem? I am no high-current >expert, but wouldn't the soldering of the connector to the PCB trace, >and the PCB traces themselves, be a bigger problem? 12A is not extreme. I was once discussing a 800A 12V PWM design (yes, and worked and it had a 2kA cutoff function. I ended up doing it as an array of mosfets screwed on a coppper reail for bot power and cooling, and brass strips as both connections and curent sensing.) It did not go into production, but the PCB manufacturer and I talked about diung multilayera board with parallel traces on most layers, selecting thivkest availavle copper, and use the plating technique to make them even thicker. /Morgan -- Morgan Olsson, Kivik, Sweden