Strange requirement in LDO from Infineon

[PartialDischarge]

I was looking for a 5V LDO with low quiescent current, more than 15V input voltage and enable input. Found this great part from Infineon, IFX20001 at only 0.21€/100 units. This is 1/3 of the cost of a similar TPS76050.

110uA quiescent current
45V max input
-40º +125ºC op temp

Then it has this strange requirement for the output capacitor,




So that you actually have to place a resistor in series with the C,

[helius]

The regulator has particular ESR requirements for its output capacitor to prevent oscillation. It isn't special to this Infineon part, but is characteristic of all LDO regulators.
See Texas Instruments Application Report slva115 for an explanation for why this requirement exists.

ESR between 2.5 and 8 Ω is pretty typical for small non-polymer tantalum or aluminum electrolytics, but if a MLCC is used, the series resistor is needed.

[mzzj]

Even the TPS76050 you mention says in datasheet:
output capacitor
Like all low dropout regulators, the TPS760xx requires an output capacitor connected between OUT and GND
to stabilize the internal control loop. The minimum recommended capacitance value is 2.2 µF and the ESR
(equivalent series resistance) must be between 0.1 Ω and 20 Ω. Capacitor values of 2.5-µF or larger are
acceptable, provided the ESR is less than 20 Ω. Solid tantalum electrolytic, aluminum electrolytic, and multilayer
ceramic capacitors are all suitable, provided they meet the requirements described above. Most of the
commercially available 2.2-µF surface-mount solid-tantalum capacitors, including devices from Sprague,
Kemet, and Nichicon, meet the ESR requirements stated above. Multilayer ceramic capacitors should have
minimum values of 2.5 µF over the full operating temperature range of the equipment.

This is very common pitfall with LDO regulators. I have been patching more than a few very expensive "professional" products with singing LDO's 

[PartialDischarge]

The regulator has particular ESR requirements for its output capacitor to prevent oscillation. It isn't special to this Infineon part, but is characteristic of all LDO regulators.
See Texas Instruments Application Report slva115 for an explanation for why this requirement exists.

ESR between 2.5 and 8 Ω is pretty typical for small non-polymer tantalum or aluminum electrolytics, but if a MLCC is used, the series resistor is needed.

I know that, how LDOs tend to oscillate. However the minimum of 2.5 ohm, I've never seen in other LDOs, that is what did strike me.
The TPS76050 for example is quite similar in specs and requires 0.1 minimum, LDK320 from ST no minimum ESR, TPS7B6350 no minimum, NCV8664ST50 no minimum...etc,etc...

[PartialDischarge]

Even the TPS76050 you mention says in datasheet:
output capacitor
Like all low dropout regulators, the TPS760xx requires an output capacitor connected between OUT and GND
to stabilize the internal control loop. The minimum recommended capacitance value is 2.2 µF and the ESR
(equivalent series resistance) must be between 0.1 Ω and 20 Ω.

0.1 is something met with any cap, so in practice no resistor is needed.

I should have been more specific. Show me other LDOs that have a minimum ESR requirement as high as 3ohm, cause the vast majority of the ones I've seen ever don't need any.

[mzzj]

0.1 is something met with any cap, so in practice no resistor is needed.

I should have been more specific. Show me other LDOs that have a minimum ESR requirement as high as 3ohm

>2.2uF  ceramics have ESR somewhere around 0.01 ohms so there is very common capacitor type that is below 0.1ohms.
TDK even has ceramics with added ESR https://product.tdk.com/en/products/emc/guidebook/eemc_product_02.pdf

Another example would be TI's LP2982 that requires 1 ohm ESR http://www.ti.com/lit/ds/symlink/lp2982.pdf

If you haven't seen minimum ESR requirement in LDO datasheets you haven't look enough hard. Worry not, you are obviously not the only one based on how many bad designs I have seen. 

[PartialDischarge]

Another example would be TI's LP2982 that requires 1 ohm ESR http://www.ti.com/lit/ds/symlink/lp2982.pdf

The LP2982 does not require a minimum of 1ohm, but around 0.03. You should look at the graphs better, not looking at them it's a common pitfall for young players...

If you haven't seen minimum ESR requirement in LDO datasheets you haven't look enough hard. Worry not, you are obviously not the only one based on how many bad designs I have seen. 

I've seen lots of LDOs datasheets and since I know about the oscillation problem I've always been looking at the output specs for the capacitance, and never had a problem with the LDOs in my designs.  However if you say otherwise then so be it, you surely know better than me.

Still looking for other LDOs with such a high esr requirement...

[David Hess]

The requirement is not unusual but some datasheets do not bother to mention it and may simply state something about a solid tantalum or aluminum electrolytic in the application notes without saying why.  Most "high" dropout regulators have a low output impedance so are more tolerant of output capacitor ESR but see below.  Low dropout regulators use a common emitter/source output transistor with a higher output impedance making output capacitor ESR more important.

This difference can also be seen between the positive and negative versions of old NPN process regulators.  317/7805 positive regulators have an emitter follower output and can tolerate a wide variation in output capacitance.  337/7905 negative regulators use the same NPN transistor so the output comes from the collector and they *must* be used with an output capacitor that has the proper ESR and their application notes reflect this.

There are ways to make a regulator work with an "ideal" zero ESR output capacitor:

1. Require a much larger capacitance for dominant pole compensation but this comes at the expense of closed loop bandwidth and large capacitors cost more and take up more space.

2. Add series resistance to the output and take AC feedback before the resistance and DC feedback after.  This comes at the expense of greater dropout voltage.

3. Tap the output transistor's structure for AC feedback to produce the same result as (2) above.  This is only available if the manufacturer controls their own process and the designers thought of it.

[capt bullshot]

David: Thanks for the insightful explanation. I'm not involved in chip design anyway, but it's always interesting to see what analog chip designers can achieve.

In general, Infineon stuff tends to be developed in another world, having other assumptions than most other chip manufacturers. One example is the Flash memory in certain Infineon microcontrollers, its erased stats is "all zeros" instead of the commen "all ones". This LDO just fits into that experience.
But nevertheless, they make quite rugged stuff for automotive use, but have the tendency to ignore other (industrial) customers and their habits and needs.

[PartialDischarge]

In general, Infineon stuff tends to be developed in another world, having other assumptions than most other chip manufacturers.

I have the same feeling when reading Infineon datasheets, I've mostly used their power mosfets, igbts and igbt drivers.

[T3sl4co1l]

I've seen some that show a graph of ESR vs. C.  TL431 usually does this, but I forget which LDOs I've seen it on.

You don't need to use an explicit resistor, but it is a good idea.  The alternative is to find an electrolytic in the right range (good luck with that, and woe be unto you if the temperature should ever drop), or use a tantalum (which has other problems: conflict minerals (if you mind that sort of thing) and sensitivity to excess voltage and current).

You don't have to tolerate a high supply impedance, just a modest peak impedance in a frequency range where it's not usually a problem (namely, at the LDO's crossover frequency).  You can use smaller bypass caps, without ESR, just fine, up to a total value maybe 1/5th of the bulk cap's value.  (Also consider a lossy bulk cap on the other end of the supply, if it is routed point-to-point like a filter or transmission line, to terminate it.  Very good results that way, without needing a 4 layer board!)

Tim

[741]

This "minimum ESR" point has frequently unsettled me. Surely, it is uncommon for capacitor DS to state a minimum ESR?

So the worry is "you get [un-]lucky, and buy an "unusually good" batch, or maybe the manufacturer improved the capacitor design to get a "typically lower on average" ESR.

So to guarantee good results without testing each capacitor, maybe you buy very low ESR capacitors and add a resistor. But in some cases, for larger capacitances (and working voltages?) a very low ESR capcitor is quite expensive.

The TDK "ESR Control" capacitors mentioned are the first time I've been aware of such a component.

[exe]

its erased stats is "all zeros" instead of the common "all ones"

May be it was different type of memory? Afaik, NAND and NOR memory have different clear state.