Dealing with LDO heat

[bryce1]

Hi guys,

I am using this one IC on my board that draws about 200mA at 3.3v. The rest of the circuit runs at 5V, and I have problems dealing with the heat of the LDO I am using (for 5V to 3.3V).

The reasons mainly being the density of the board and my lack of experience. Since cost is not a big issue, should I just go for a switching regulator like the SC189ZSKTRT? Do you have any layout / calculation advice?

Thanks!

[T3sl4co1l]

Only 0.34W... what, is it SOT23 or something?  If you can reserve some area for heat spreading copper and vias, that's plenty fine.  Maybe risky for automotive or something.  If you can't do that, a larger device like SOT89 or -223 would do better.

Tim

[mij59]

Hi,

What's the 3.3V IC, LDO?

The inductor for the SC189ZSKTRT would also take up space.

[bryce1]

Right now, I am using a NCP1117ST12T3G. It is a SOT-223 package.

[mij59]

Right now, I am using a NCP1117ST12T3G. It is a SOT-223 package.

Take a look at page 8 of the data sheet fig. 21, minimum pad size is about 3mm.

[T3sl4co1l]

How hot is it getting, or, how is it being a problem?

[bryce1]

I think it might not be the cooling after all. It gets untouchable after milliseconds. I checked, it is corrected the right way, so I have no idea what is going wrong.

[T3sl4co1l]

Milliseconds?  Like hundreds of them, or..?

Now I'm curious, as any kind of human response takes a very long time, electrically speaking.  If you're accidentally measuring temperatures within (single digit) milliseconds, that would actually be pretty exciting..

Tim

[bryce1]

Maybe I am that good  :-D
Seriously, there must be something wrong with the part or the PCB. I'll try to figure it out tomorrow.

[David Hess]

If you want the higher efficiency needed to lower the power dissipation, then a switching regulator is the way to go.  Using a linear regulator to drop 5.0 volts down to 3.3 volts yields 66% efficiency which is not bad but a typical switcher with an efficiency of 83% would drop the heat requirements down to half which may or may not be worth it.  A good switcher could drop the losses by half again.

This is probably about perfect if your layout skills are good:

http://www.linear.com/product/LTC3542

A larger SO-8 package and lower operating frequency might be easier to deal with:

http://www.linear.com/product/LTC1627
http://www.linear.com/product/LTC1707

[tszaboo]

I think you should be able to handle that amount of heat without too much issue. Typically a SOT223  has <150K/W thermal resistance.
You either underestimated the current consumption, you have a short on the PCB or a component is toast. Also 1117 doesnt like ceramic capacitors directly on the output, it needs something bigger and with more ESR. I would measue the resistance between 3.3V and GND (while off) to start with.

[Neilm]

Personally, I'd take it off the board and put a PSU in. That will allow you to measure the current so you will be able to see what is going and what the current draw actually is.

[bryce1]

Also 1117 doesnt like ceramic capacitors directly on the output, it needs something bigger and with more ESR. I would measue the resistance between 3.3V and GND (while off) to start with.

Why?

datasheet:

"The capacitor type can be ceramic, tantalum, or aluminum electrolytic as long as it meets the minimum capacitance value and ESR limits over the circuit’s entire operating temperature range".

[T3sl4co1l]

Yes, and what limits are those?

[mariush]

lm1117  http://www.ti.com/lit/ds/symlink/lm1117-n.pdf page 9

Output Capacitor
The output capacitor is critical in maintaining regulator stability, and must meet the required conditions for both
minimum amount of capacitance and ESR (Equivalent Series Resistance). The minimum output capacitance
required by the LM1117-N is 10?F, if a tantalum capacitor is used. Any increase of the output capacitance will
merely improve the loop stability and transient response. The ESR of the output capacitor should range between
0.3? - 22?. In the case of the adjustable regulator, when the CADJ is used, a larger output capacitance (22?f
tantalum) is required.

ncp1117-d http://www.onsemi.com/pub_link/Collateral/NCP1117-D.PDF

See graphs on page 6
Also see page 8.

Frequency compensation for the regulator is provided by
capacitor Cout and its use is mandatory to ensure output
stability. A minimum capacitance value of 4.7 uF with an
equivalent series resistance (ESR) that is within the limits of
0.25 ohm to 2.2 ohm is required. The capacitor type can be
ceramic, tantalum, or aluminum electrolytic as long as it
meets the minimum capacitance value and ESR limits over
the circuit’s entire operating temperature range. Higher
values of output capacitance can be used to enhance loop
stability and transient response with the additional benefit of
reducing output noise.

I typically use 33-47uF 35-50v rated electrolytics , they usually have 0.45-0.7 ohm ESR and the can size is just as large as caps rated for lower voltages (so i don't have to stock this capacitance in several voltage ratings and can sizes).  See Panasonic FC or Nichicon PW for example, good series for linear regulators. 
If you want something more modern, you could go with 68-100uF 16v or 47uF 25v Panasonic FR  which are rated for 5000h@105c but the esr @ 100kHz is already down to about 0.3 ohm so in some extreme cases the regulator may not like it.

[tszaboo]

Also 1117 doesnt like ceramic capacitors directly on the output, it needs something bigger and with more ESR. I would measue the resistance between 3.3V and GND (while off) to start with.

Why?

datasheet:

"The capacitor type can be ceramic, tantalum, or aluminum electrolytic as long as it meets the minimum capacitance value and ESR limits over the circuit’s entire operating temperature range".

There are various sources on TI, AD or Linear website explaining the phenomenon. The loop stability is not good with low ESR caps. But, this is from TI's datasheet:
"The output capacitor is critical in maintaining regulator stability, and must meet the required conditions for both
minimum amount of capacitance and ESR (Equivalent Series Resistance). The minimum output capacitance
required by the LM1117-N is 10?F, if a tantalum capacitor is used. Any increase of the output capacitance will
merely improve the loop stability and transient response. The ESR of the output capacitor should range between
0.3? - 22?. In the case of the adjustable regulator, when the CADJ is used, a larger output capacitance (22?f
tantalum) is required"
Typical ceramic cap goes down to 0.01Ohm ESR easily. Source: www.avx.com/docs/techinfo/mlc-tant.pdf Fig. 1.

I know, that you dont necessarily use TI LDO, or that capacitor from AVX, but as a general rule of thumb, if they dont say in the LDO datasheet, that it is stable with ceramic, than don't use it with ceramic. It is a feature of the newer ones, and I'm sure there are several thousands of engineering work-hours making newer LDOs stable with them.

mariush: did you just qoute the same thing as me like five seconds before me?

[David Hess]

Low dropout regulators rely more on the output capacitor for frequency compensation because the common emitter or common source output stages have voltage gain.  The negative regulator version of the LM317, the LM337, suffers from the same problem despite not having low dropout because it also uses a common emitter output stage which is why it needs an output capacitor for stability unlike the LM317.

[nitro2k01]

Before we go to self oscillation as an explanation, let's look at how you have connected the chip, in particular the tab. The tab is connected to pin 2. In fact, pin 2 and the tab are likely the same metal piece, which is used as a holder for the die, as shown in the attached image. Pin 2 on your particular device is the output, meaning that if you connected the tab to ground for heat dissipation, you are shorting the output to ground. What would happen then is that device would heat up rapidly, and likely shut off due to thermal overload protection.

Another thing, NCP1117ST12T3G is a 12 V regulator, not a 3.3 V regulator. Maybe you used that as a generic name for the whole range of those regulators, but if you used an actual NCP1117ST12T3G, you've probably fried every 3.3 V chip on the output side, and maybe some of the capacitors as well if 12 V is out of range for them, and possibly created a short somewhere inside a chip in the process. The good news is that the regulator likely survived, so if you need a 12 V regulator in the future, you can probably salvage that one.

[bryce1]

it is the 3.3v version. so higher board density => smd tantal?

[mariush]

Did you bother to read our comments?

Want to use smd tantalum capacitors?  => select your capacitor => read datasheet => is the esr above the minimum value recommended?  => if yes, you can use it.

[bryce1]

No need to get angry. I'll rephrase the question: If electrolytic is not an option due to the density and ceramics all have too low ESR, is my best bet a tantalum capacitor?

[David Hess]

No need to get angry. I'll rephrase the question: If electrolytic is not an option due to the density and ceramics all have too low ESR, is my best bet a tantalum capacitor?

Yes, tantalum capacitors work well for this.  Alternatively a ceramic capacitor may be used with a resistor in series.  Some low dropout regulators are better than others as far as tolerating output capacitor characteristics.

[george graves]

Milliseconds?  Like hundreds of them, or..?

Now I'm curious, as any kind of human response takes a very long time, electrically speaking.  If you're accidentally measuring temperatures within (single digit) milliseconds, that would actually be pretty exciting..

Tim

Come on - you know what he meant.   

[sleemanj]

No need to get angry. I'll rephrase the question: If electrolytic is not an option due to the density and ceramics all have too low ESR, is my best bet a tantalum capacitor?

You can use a ceramic with a low value resistor in series instead of a tantalum, see http://www.ti.com/lit/an/slva214a/slva214a.pdf

[tszaboo]

No need to get angry. I'll rephrase the question: If electrolytic is not an option due to the density and ceramics all have too low ESR, is my best bet a tantalum capacitor?

I wouldnt use tantalum in the 21 century (unless no other choice). There is a high chance (over 50) it was sold to buy weapons to make genocide in Congo.