3
\$\begingroup\$

I'm designing a board that I want to initially manufacture using JLCPCB for both the board fabrication and assembly. The pricing approach at JLCPCB separates available components into commonly used "basic" components, and less common components labeled "extended" which involve an extra fee to use. Pretty much every power inductor in their library is in the extended category, and the fee is high enough that the price difference between different inductors is insignificant. My board also doesn't have tight space constraints, so to keep costs low I'm considering using a single size inductor that can meet the requirements of the three switching DC/DC converters on the board.

There's a 24v/5v -> 3.3v regulator (TPS54331) which uses an 18uH inductor, a 24v -> 10v regulator (also TPS54331) that uses a 4.7uH inductor, and three constant current LED drivers (TPS92515) that all use a 47uH inductor. I'm thinking of using a 47uH inductor for all three, which mean I would only pay the extended part fee for a single item. My understanding is that using a larger inductor that required lowers the frequency of the switch, and reduces ripple current, both of which are relatively positive. Is this a reasonable cost-optimization for the board, or will I run into trouble doing this?

\$\endgroup\$
4
  • 1
    \$\begingroup\$ You'll have to state which converters/LED drivers you're using, including their data sheets. Otherwise no-one can know if these devices would be tolerant of changes in the inductor value. \$\endgroup\$
    – Unimportant
    Commented Jan 8, 2022 at 20:45
  • \$\begingroup\$ Ah OK, I sort of figured that buck switching converters had similar calculations for the relationship between inductance, frequency and ripple current. Regardless, I'll add the specifics of the parts. \$\endgroup\$
    – flimsy
    Commented Jan 8, 2022 at 21:03
  • \$\begingroup\$ Since you are using TI parts, have you simulated the parts on TI website if they all work as expected with the 47uH inductor? If the inductor is not compatible then it isn't. \$\endgroup\$
    – Justme
    Commented Jan 8, 2022 at 21:10
  • \$\begingroup\$ I was actually playing around on the TI simulator, but from what I could see switching out the inductors doesn't change the simulation. I even tried some really wild values and the the bode plot was identical, which makes me kind of suspicious of the simulator generally. \$\endgroup\$
    – flimsy
    Commented Jan 8, 2022 at 21:19

1 Answer 1

Reset to default
5
\$\begingroup\$

Inductor rationalisation for BOM consolidation can and does reduce manufacturing costs . If you use a larger value than recommended the ripple current is lower but you can sometimes get instability on these jellybean peak current mode hardswitch chips . check for this ,you may have to delve into the control loop .

\$\endgroup\$
4
  • 1
    \$\begingroup\$ OK, so it sounds like this plan is at least feasible, but I will have to do some testing. When you say "instability", can you describe what I might see on a scope that would indicate that? \$\endgroup\$
    – flimsy
    Commented Jan 8, 2022 at 21:14
  • 1
    \$\begingroup\$ @flimsy the regulator might not start at all, may not produce correct output voltage, may turn into oscillator, or just respond slowly to sudden changes in the current draw, which then causes voltage undershoots and overshoots at transients. \$\endgroup\$
    – Justme
    Commented Jan 8, 2022 at 21:19
  • 1
    \$\begingroup\$ Look at the switching node voltage on the scope when running into a resistive load .Compare your coils .You do not want subharmonics ,you want even pulses that are not bunched .You dont want shaking on the trailing edge .Many chips are crap but you do not want your coil to make things worse.90 % of the time it will be alright. \$\endgroup\$
    – Autistic
    Commented Jan 9, 2022 at 3:34
  • \$\begingroup\$ Super helpful, thanks so much! \$\endgroup\$
    – flimsy
    Commented Jan 9, 2022 at 18:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.