Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Filter caps for LM2940CT?

Status
Not open for further replies.
I'm looking at using an LM2940CT to power some LEDs off my car's accessory socket. Now these LEDs are supposed to have some onboard regulation, but I know the car is a very electrically noisy environment so I want the input to the LEDs to be as clean as possible for maximum longevity (when it takes 2-3 weeks to get a replacement off the slow boat from China, replacing them is a nuisance!).

The datasheet shows
1: 0.47µF cap ("Required if regulator is located far from power supply filter") on the input,
2: 22µF cap (Cout must be at least 22µF to maintain stability. May be increased without bound to maintain regulation during transients") on the output. Locate as close as possible to the regulator. This capacitor must be rated over the same operating temperature range as the regulator and the ESR is critical; see curve.

My questions are thus:
  1. Would it be beneficial to put a larger cap (e.g 47µF) in parallel with the 0.47µF input filter cap?
  2. What (if any) would the drawbacks be to using a larger cap (47-220µF) on the output?
  3. Would there be any benefit to putting a small-value low-ESR cap in parallel with a larger electrolytic cap?

Thanks!
 
3. Generally, there are 0.1uF capacitors placed on the input and the output for high frequency spikes.
2. A larger output capacitor means the output will stay up longer when you lose input power. This may damage the chip, although the data sheet does say can increase without bound. Generally, a diode is placed from the output pin (anode) to the input pin (cathode) reversed bias. This has no effect on power up, but when you lose power, it will help to rapidly discharge the output cap.
1. yes, the more capacitance you have, the less spikes affect your circuit.

Automotive power is really dirty. You should also put some resistor in series with your power feed (like 50-100 ohm, 1-3watt, depending on how much current you are using), and a TVS (transient voltage suppressor) like a 1.5ke18 or .6ke18. This will clamp spikes over 18v before it even reaches the regulator, helping your line regulation.
 
For the TVS, would this work?

What would putting a resistor in series with the power feed accomplish? I assume this is effectively creating a capacitive low-pass filter? My expected current draw is 1.2A, with plans to expand to 1.8A (and the input is fused at 3A), so how would I figure out what sort of resistor (value & wattage) I need?

50Ω+ doesn't seem to make sense to me, unless I'm doing the math wrong:
[LATEX]P = I^2R = 1.2^2 \times 50 = 1.44 \times 50 = 72{\text{W ???}}[/LATEX]

I think a 1Ω/10W resistor would make more sense...
[LATEX]P = I^2R = 1.2^2 \times 1 = 1.44 \times 1 = 1.44{\text{W}}[/LATEX]
[LATEX]P = I^2R = 1.8^2 \times 1 = 3.24 \times 1 = 3.24{\text{W (for expansion)}}[/LATEX]
and
[LATEX]V = IR = 1.2 \times 1 = 1.2{\text{V drop}}[/LATEX]
[LATEX]V = IR = 1.8 \times 1 = 1.8{\text{V drop (for expansion)}}[/LATEX]
Does that sound right?

Is this the general circuit you're suggesting (see attached)?

If it makes a difference, I'm looking at running 3x LM2940 off the input line (which is how I'm getting to 1.2A/1.8A -- two of the LM2940's will be driving 1x 300mA LED with spare connectors for a second each, and the third LM2940 will be driving 2x 300mA LED). I'm currently looking at having all the input-side circuitry be "common" between the three regulators, with separate output caps and diodes for each regulator.

P.S.: for those who may be concerned, I'm heatsinking each LM2940 with 13.4°C/W heatsinks that are 1.38"x0.5"x1.0" :)
 

Attachments

  • LM2940circuit.png
    LM2940circuit.png
    5.7 KB · Views: 652
Last edited:
The TVS would go on the right side of the resistor, but that is the general idea. I'm not sure 1 ohm would do you much good, so you could probably nix it. The 2940 is an automotive device, so on overvoltages during load dumps it will shut down (this is where choosing your cap values really comes into play). This is what the TVS is trying to help alleviate. The max input of the regulator is 26v, and the max clamp of the 18V TVS is 25.2V.

Heatsink would be the next question. Since you are using 3 separate regulators, that will help. Going 12V to 5V on a linear regulator is tricky, especially when you get above 1/2 amp, but 24 to 12 is a lot worse. Did you do the thermal calculations, or did you just grab a heatsink? I can help you with those as well.

OK, at this point I should take a step back. I am assuming you are using a 5v output... are you? Or are you using a 12V and just letting the regulator clip the 2v off the top? What LED are you driving?
 
If it shuts down "gracefully" (i.e. not letting the smoke out) I'm okay with that... 99% of the time I expect the LEDs to be used when the car is off anyways; I just want to make sure that IF they're used when the car is on, I don't smoke the LEDs or regulators. If an overvoltage just results in a little LED flicker I can live with it, with the idea that the filtering should make it infrequent enough to not really be noticeable.

I tried to do the thermal calculations to the best of my meager ability, but I'm also trying to keep this all in a small(ish) ventilated enclosure. Since I'm just using the regulators to "take a little off the top" (~14V -> 12V) I'm figuring the per-regulator thermal dissipation should be no more than:
[LATEX]P_D = IV = 0.6{\text{A}} \times 2{\text{V}} = 1.2{\text{W}}[/LATEX]
If I'm reading the datasheet right, and assume that the temperature in the car will very seldom exceed 140°F (especially while we're actually using the lights), I should be able to go by the following:
[LATEX]T_{R(max)} = T_{J(max)} - T_{A(max)} = 125^{\circ}{\text{C}} - 60^{\circ}{\text{C}} = 65^{\circ}{\text{C}}[/LATEX]
thus, the maximum allowable junction-ambient thermal resistance is
[LATEX]\theta_{(J-A)} = T_{R(max)} / P_D = 65^{\circ}{\text{C}} / 1.2{\text{W}} = 54.17^{\circ}{\text{C}}/{\text{W}}[/LATEX]
Since the datasheet says "If the maximum allowable value for θ(J−A) is found to be ≥ 53˚C/W for the TO-220 package [...] no heatsink is needed since the package alone will dissipate enough heat to satisfy these requirements" any heatsinking is really just superfluous. Of course I like a little margin of error, so I threw in some heatsinks :)

I got two sets of LEDs because they're cheap enough, and if one looks like crap or fries on the first test run, I have a fallback. One set of LEDs are 12V/300mA units, and the other are "3 watt", so I assume 12V/~250mA. Both sets are allegedly designed as automotive lamps, but given that they're Chinese eBay specials, I don't want to assume too much about any built-in noise/voltage/current regulation. The second set actually appears to have some filtering/regulation hardware built in (I can spot an inductor, what looks to be a quad set of diodes as a bridge rectifier, and what I think is a small SMT regulator with attendant resistor and capacitor); the first set though appears to just be a bunch of LEDs stuck on a board with the connector attached. Even if the second set has all the necessary regulation built in, I'd rather have something a little more robust than whatever they managed to fit on a 0.5" x 1.5" PCB... call me crazy :p
 
Last edited:
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top