14.4V ground reference for ADC

[Simon]

That will probably be enough.

[Arjunan M R]

one question, the datasheet of the vref says to put a low esr ceramic cap at the supply input for bypassing.
Do i need to care about low esr ??

[Simon]

Yes ESR is very important. A real capaocitor is modelled in a theory by a perfect capacitor with a resistor in series. this resistor is the ESR. Equivalent series resistor. If your ESR is high it slows the capacitors response to demand. This is why often you see an electrolytic with a ceramic in parallel because the ceramic has near 0 ESR but they are not large and bigger ones are expensive, so a compromise is made with enough capacity helped by some fast acting capacitance.

[Arjunan M R]

here is the datasheet of the capacitor i am using.
I ca't find any values for ESR in it.Please have a look at it.

[Arjunan M R]

Is this the ESR?

[Simon]

If it's a ceramic it's low ESR, ESR is a problem for electrolytics or niche applications with ceramic.

[Arjunan M R]

If it's a ceramic it's low ESR, ESR is a problem for electrolytics or niche applications with ceramic.

So what are the parameters i need to look when selecting the ceramic capacitor rather than voltage and capacitance

[Simon]

The dialectric is an indicator of the capacitors quality and tolerance. The best is X7R, the next best is X7S. All ceramics will have good enough low ESR for what you want unless you are dabling with EMC which you are not here.

[Siwastaja]

Standard ceramic capacitor all have very low ESR, the lowest you can get practically. For a typical 0805 size, it's roughly around 10 milliohms. Compare this to around 5-10 ohms, three orders of magnitude more, for similar capacitance electrolytics, tantalum or aluminium. It's often not specified, as it's irrelevant whether it's 0, 10 or 20 mOhm, it can be approximated zero in 99% of the cases. Note that ESR has both good and bad sides to it, sometimes you want to minimize it, sometimes you absolutely need it.

The most important single parameter for ceramic capacitors is actual capacitance versus DC bias voltage characteristic. This is widely misunderstood subject. Opposite to common misbelief, you can't reliably deduce this characteristic from the dielectric type (unless it's C0G/NP0); you need the actual data from the manufacturer. Sometimes the spec doesn't exist; if the actual capacitance value matters at all, ignore such capacitors. It's fair to expect it can go down to 10% of the rated C, even for "good" X7R type caps.

A further complication is that often the datasheets that instruct you to use a, say, 4.7uF capacitor, already assume you use such a capacitor type with poor DC bias characteristics, but fail to tell you how much actual capacitance they actually require. So requiring "4.7uF ceramic" in a datasheet can mean less than 1uF actually, but it can mean 2uF as well, so you need to guess, or play safe and use a bigger package with known good DC bias characteristics.

[Arjunan M R]

The dialectric is an indicator of the capacitors quality and tolerance. The best is X7R, the next best is X7S. All ceramics will have good enough low ESR for what you want unless you are dabling with EMC which you are not here.

I am not bothered with EMC.

[Arjunan M R]

Standard ceramic capacitor all have very low ESR, the lowest you can get practically. For a typical 0805 size, it's roughly around 10 milliohms. Compare this to around 5-10 ohms, three orders of magnitude more, for similar capacitance electrolytics, tantalum or aluminium. It's often not specified, as it's irrelevant whether it's 0, 10 or 20 mOhm, it can be approximated zero in 99% of the cases. Note that ESR has both good and bad sides to it, sometimes you want to minimize it, sometimes you absolutely need it.

The most important single parameter for ceramic capacitors is actual capacitance versus DC bias voltage characteristic. This is widely misunderstood subject. Opposite to common misbelief, you can't reliably deduce this characteristic from the dielectric type (unless it's C0G/NP0); you need the actual data from the manufacturer. Sometimes the spec doesn't exist; if the actual capacitance value matters at all, ignore such capacitors. It's fair to expect it can go down to 10% of the rated C, even for "good" X7R type caps.

A further complication is that often the datasheets that instruct you to use a, say, 4.7uF capacitor, already assume you use such a capacitor type with poor DC bias characteristics, but fail to tell you how much actual capacitance they actually require. So requiring "4.7uF ceramic" in a datasheet can mean less than 1uF actually, but it can mean 2uF as well, so you need to guess, or play safe and use a bigger package with known good DC bias characteristics.

Then why do manufacturers says to use lower value ceramic and also higher value tantalum together  for bypassing??

[Arjunan M R]

The dialectric is an indicator of the capacitors quality and tolerance. The best is X7R, the next best is X7S. All ceramics will have good enough low ESR for what you want unless you are dabling with EMC which you are not here.

I am going to use X5R.

[Simon]

Standard ceramic capacitor all have very low ESR, the lowest you can get practically. For a typical 0805 size, it's roughly around 10 milliohms. Compare this to around 5-10 ohms, three orders of magnitude more, for similar capacitance electrolytics, tantalum or aluminium. It's often not specified, as it's irrelevant whether it's 0, 10 or 20 mOhm, it can be approximated zero in 99% of the cases. Note that ESR has both good and bad sides to it, sometimes you want to minimize it, sometimes you absolutely need it.

The most important single parameter for ceramic capacitors is actual capacitance versus DC bias voltage characteristic. This is widely misunderstood subject. Opposite to common misbelief, you can't reliably deduce this characteristic from the dielectric type (unless it's C0G/NP0); you need the actual data from the manufacturer. Sometimes the spec doesn't exist; if the actual capacitance value matters at all, ignore such capacitors. It's fair to expect it can go down to 10% of the rated C, even for "good" X7R type caps.

A further complication is that often the datasheets that instruct you to use a, say, 4.7uF capacitor, already assume you use such a capacitor type with poor DC bias characteristics, but fail to tell you how much actual capacitance they actually require. So requiring "4.7uF ceramic" in a datasheet can mean less than 1uF actually, but it can mean 2uF as well, so you need to guess, or play safe and use a bigger package with known good DC bias characteristics.

Then why do manufacturers says to use lower value ceramic and also higher value tantalum together  for bypassing??

Low ESR is good but too much capacitance with a low ESR can cause instability. For reasons of cost and stability usually you use a large high ESR capacitor with some ceramic. The ceramic acts fast and by the time it is depleted the electrolytic is supplying power. It's about the frequency response. A fast edge has a high frequency component although the actual main waveform frequency may be low. It's that fast edge that the ceramic deals with.

I have looked at a SMPS controller chip and it specifies that the output capacitor of 100µF must NOT be low ESR as the capacitor reacts so fast that it will make the control loop go unstable. But for good performance it wants low ESR capacitors on the input of only 2x2.2µF. You can pay £0.60 for a 100µF capacitor with 200mOhm ESR, a 28mOhm ESR capacitor is larger and costs £2.60.

[Arjunan M R]

I am using a 100nF ceramic and 10uF tantalum for bypassing ADC , DAC and VRef.
Is that enough?

[Arjunan M R]

 So the manufacturers says to use 10uF tantalum for frequency response and stability.

[Simon]

I am using a 100nF ceramic and 10uF tantalum for bypassing ADC , DAC and VRef.
Is that enough?

Do you mean you have one 100nF ceramic and one 10uF tantalum for all 3 or that they have a pair each?

[Siwastaja]

Then why do manufacturers says to use lower value ceramic and also higher value tantalum together  for bypassing??

Higher value tantalum has more inductance due to larger package, which, together with high ESR, prevents it from reacting to high-frequency noise.

In parallel, they filter different frequency ranges.

I comment about the inductance and package, because to be really effective, the small 100n cap has to be in the right place.

When paralleling low-ESR capacitors of different size, there is a risk they form oscillating resonance frequency ranges, together with stray inductances. But when the larger capacitor has some ESR, it dampens this effect, like a shock absorber does with mechanical springs. Hence it's very common to see a combination of a small, low-ESR capacitance, combined with a much larger capacitance, but with considerable ESR. Tantalum is a very effective type for providing well controlled ESR. Of course, a MLCC + separate series resistor can be used as well.

[Arjunan M R]

I am using a 100nF ceramic and 10uF tantalum for bypassing ADC , DAC and VRef.
Is that enough?

Do you mean you have one 100nF ceramic and one 10uF tantalum for all 3 or that they have a pair each?

A pair of 100n and 10uF for 3 of those.

[Arjunan M R]

which capacitor i need to use in here?
electrolytic, tantalum or MLCC ?

[Arjunan M R]

which capacitor i need to use in here?
electrolytic, tantalum or MLCC ?

i will use tantalum.

[Arjunan M R]

Is adding a TVS diode to the input and output a good idea?

[Simon]

Is adding a TVS diode to the input and output a good idea?

Input of what?

[Arjunan M R]

Is adding a TVS diode to the input and output a good idea?

Input of what?

input of psu

[Simon]

which capacitor i need to use in here?
electrolytic, tantalum or MLCC ?

No idea, consult the datasheets of the parts in question. 10µF looks quite big there.

[Simon]

Is adding a TVS diode to the input and output a good idea?

Input of what?

input of psu

If you are powering it off a mains transformer it should not be neccessary. No harm in using one if you want to. Be sure to understand all the specs of the TVS, they don't trigger on/off at a single threshold.