Really long input leads to a DC-DC converter

This question is for an industrial application where a hefty 12V source is distributed to numerous low power DC-DC Buck converters via a long cable. Each cable is approximately 100' in length AWG 22. Each converter is fed from it's own cable.

The converters output either 3.3V or 5V in the 1-15W range, and typically have a 10uF input capacitor. The DC-DC converter boards used are from numerous vendors and use various PMIC regulators, this is one example:
https://www.digikey.ca/product-detail/en/MP2451DJ-LF-P/1589-1118-2-ND/5298875

Should any additional conditioning circuitry be added to the input circuit of the converter for stability and reliability? Perhaps additional capacitance, transient protection, etc?

It depends on the inductance of the cable, it sounds like it could have a lot of inductance potentially, -that can cause an overvoltage ring at the first application of power. I would put a tvs at the converter end, make it as high voltage rated as you can, so it doesn’t usually conduct…and remember the tolerance on the breakover voltage.
10uF sounds a bit measly, can you spare some more?…and just to damp out vin ringing I’d put a RC series network across the input too…….hopefully you can fix r and c by scope and see, but start with say 1uF and 2 ohms, say.

I’d also put a diode across the input, just in case, to stop severe ringing causing the input to go reverse polarity…….obviously not a giant diode as most of the time it will not be conducting.

MPS2451 is max 36v at input so make the tvs stop it ever going above 36v, but not too near the 12v of the supply

MPS2451
**broken link removed**

- - - Updated - - -

Hopefully the little dcdc’s have some ceramic capacitors as close as possible to them, (if not put some there) so that the nasty high frequency harmonics don’t traverse a big loop…or else those long leads might become unwanted antennae

Never mind the inductance the max current is 3A, so whats the I X R drop down the wire, especially if there is a common return, up the return. The wire gauge seems to be too low to me.
Frank

15w max, means 1.25A per cable…..awg22 is 16mR /ft, so that’s 3.2 ohms per run
total……volt drop in wire is 4V…OOPS, yes, that’s a lot of voltage drop…but still youcan get your 5v out.

https://en.wikipedia.org/wiki/American_wire_gauge

...nice bit of damping in that cable!...not so much need for ringing protection at the input.

actually my above calcs need adjusting for the fact that with 15w, and that cable resistance, the input voltage at the smps will be lower , so the input current will be higher, so it mightn't be possible to make he 15w,5vout with that much cable drop

treez said:

Hopefully the little dcdc’s have some ceramic capacitors as close as possible to them, (if not put some there)

Click to expand...

I support this recommendation, which can probably be one of the most effective care to achieve a significant reduction of noise with high rates of dV/dt. Anyway, the Insertion of ferrite beads close to the distributed modules, and the use of shielded cabling properly grounded, can eliminate much of the inducted interferences.

For a 200' round trip of 22 AWG, your resistance will be ~3.2 ohms. If the source is 12V, then the load can only absorb up to 11.25 W. So you're screwed before even considering stability. Something about the design specs need to change.

I guess now that you might put a boost converter in front of the cable?..or just replace the cable

andre_teprom said:

I support this recommendation, which can probably be one of the most effective care to achieve a significant reduction of noise with high rates of dV/dt. Anyway, the Insertion of ferrite beads close to the distributed modules, and the use of shielded cabling properly grounded, can eliminate much of the inducted interferences.

Click to expand...

Andre,

Ceramic caps are 2mm from the input pin.

Would this be a suitable part for the ferrite bead?
https://www.digikey.ca/product-detail/en/NFZ5BBW310LN10L/490-12180-1-ND/5415425

Should I use one in each leg, or is one enough?

Thanks,

-by all means use a ferrite bead like you've shown, just one in the go line.
There is no quick way to calculate exactly what level of high frequency measures you should exactly take.......stuff like ferrite beads tend to get chucked in to the "chef's desire".
You could go the whole hog and use clip-on cable ferrites and common mode chokes too, but probably not needed.

My 2 cents, the beads you picked out are good for 2A, i assume that is peak. My best guess on the buck average current is 2.34A. I do not know how to calculate buck peak current but i am sure it is well over 2A.

From the buck chip data sheet 8V is min voltage for good efficiency for 5V output.

15W / .8 efficiency best guess = 18.75W
18.75W / 8V min input voltage = 2.34A average input current.

Now we need someone to calculate peak current.

I see what you mean about looking into the peak current, as its always important especially where there might not be enough input capacitance, but with enough input caps, most of your AC will be drawn from the caps, and youll get pretty pure dc going into the input caps from the cable, so the peak wont be so important in that case as you know.