0

I want to redesign my precharge circuit. Currently I am using a bulky chassis mount resistor, which is adding an additional step in production.

Instead I have seen some precharge PCBs use a string of SMD resistors in series to achieve the same effect has the chassis mount resistor.

In addition, this also makes production much cheaper and quicker.

My precharge I am currently using has a 20 Ω 50 W chassis resistor (link)

enter image description here

Question:

  1. How do I calculate the resistor values I need and specced them in terms of their wattage?

  2. How should I lay out these SMD resistors on the PCB?

winny
  • 13,064
  • 6
  • 46
  • 63
JoeyB
  • 1,804
  • 13
  • 34
  • How long does the precharge stage take? How much energy is transferred to the resistance during that time? – Andrew Morton Feb 26 '23 at 10:19
  • "precharge" to me means capacitor, and I don't see any high capacity caps on the board. Maybe you mean something else? – Maple Feb 26 '23 at 10:32
  • The silkscreen says it's a 5R6 resistor, not 20R. Also, is the silkscreen marking "U1" in somewhat the wrong location? – Andrew Morton Feb 26 '23 at 11:08
  • @AndrewMorton I need to dig up my calculations (I did the calculations about a year ago). This value of resistance is correct for my application. As my question states I would like to see if I can move a more cost-effective method in terms of production, by using SMD. – JoeyB Feb 26 '23 at 11:13
  • @Maple the capacitor in question is off the PCB – JoeyB Feb 26 '23 at 11:14
  • 1
    Does the creepage across F1 with your track very close to the input side meet IEC requirements? It looks very close. – winny Feb 26 '23 at 16:22

1 Answers1

1

50W is quite a lot of power. If you are applying this for an extended period of time (as opposed to a pulse of < a few ms) you are going to need to carefully think about how you are going to cool them. This will probably involve some amount of either thermal modelling or trial-and-error with a thermal camera.

In terms of power handling, if you split the resistor into a series chain of resistors, then power dissipated splits between then proportional to resistance (\$P = I^2R\$, where \$I\$ is the same for all resistors as they are in series).

For example, you could go for two \$10\Omega\space25W\$ resistors to get nominally the same rating as your \$20\Omega\space50W\$ currently being used. Alternatively go for four \$5\Omega\space12.5W\$ resistors to split the dissipated power over a larger physical area.

Of course that is in an ideal world. In practice SMD resistors are usually physically smaller and so harder to keep cool. Their power ratings tend to be heavily proportional to temperature, so with worse cooling and higher operating temperature you end up needing resistors with higher power ratings to compensate.

For example, just from a quick cursory search on a well known distributors website, Bourns PWR263S-35-10R0F is a \$10\Omega\space35W\$ rated resistor, but by \$60^\circ C\$ it's rated power has dropped to barely above the required \$25W\$, so unless you can keep it cool enough, then the rating would be too low. Instead the WR263S-35-5R00F from the same series is \$5\Omega\space35W\$, which with four of them would allow you to operate up to \$100^\circ C\$ while still having a small rated power margin.


TL;DR; Using more resistors allows splitting the power over a larger area. If all resistors have the same value, then both the resistance and power rating for each is the overal required divided by the number of resistors.

But remember that the power rating is temperature dependent, so make sure you can keep them cool enough. In addition time is an important factor as many resistors have a pulse power rating allowing them to tolerate much higher than their nominal power for very short periods.

Tom Carpenter
  • 63,168
  • 3
  • 139
  • 196
  • Precharge is a pulsed application; and, without heatsink on the metal-case resistor, we can safely assume it's much less than 25W steady state. It might even be a few 100s of W in the ~ms or whatever it takes to charge. Mind this is contingent on the load not being shorted and never finishing charging -- protective logic might be recommended here. – Tim Williams Feb 26 '23 at 12:32