Choosing the Placement and Voltage of a Filtering Capacitor

Thread Starter

Max Holdcroft

Joined May 12, 2024
37
Hi All,

This is probably a very easy question, but if I have a 3.7v lipo battery attached to a voltage regulator outputting 3v, and I have a 100nF filtering capacitor to filter out noise, should I put the capacitor before or after the voltage regulator. Also, would I just want the voltage rating of the capacitor to be 3v, or something else?
Screenshot 2024-05-13 at 12.30.15 pm.png
Thanks,
Max
 

nsaspook

Joined Aug 27, 2009
13,578
1715568589585.png
https://ww1.microchip.com/downloads/en/DeviceDoc/21373C.pdf
3.1 Output Capacitor
A 1F (min) capacitor from VOUT to ground is required.
The output capacitor should have an effective series
resistance greater than 0.1 and less than 5, and a
resonant frequency above 1MHz. A 1F capacitor
should be connected from VIN to GND if there is more
than 10 inches of wire between the regulator and the
AC filter capacitor, or if a battery is used as the power
source. Aluminum electrolytic or tantalum capacitor
types can be used. (Since many aluminum electrolytic
capacitors freeze at approximately -30°C, solid
tantalums are recommended for applications operating
below -25°C.) When operating from sources other than
batteries, supply-noise rejection and transient
response can be improved by increasing the value of
the input and output capacitors and employing passive
filtering technique
 

dovo

Joined Dec 12, 2019
72
The 3V TC1262 output voltage can vary +/-2.5% giving it a probable range of 2.925 to 3.075 volts. Given a dropout voltage of 30 to 650 mV (depending on load current) the battery voltage must remain above 3.205 V at 100 mA to 3.725V at 500 mA as it discharges.

As nasapook shows from the TC1262 datasheet the recommended output capacitor is 1 uF with an ESR of 0.1 to 5 ohms. To select a capacitor there are other parameters to consider:

Operating temperature range
Lifetime
Voltage (4V or higher)
Voltage coefficient of capacitance (we don't want the capacitance dropping much below 1 uF at 3V)
Capacitance, >=1uF
Capacitance temperature stability (we don't want the capacitance dropping much with temp variation
ESR, 0.1 to 5 ohms
ESL, <25 nH (from the 1 MHz max SRF spec and this means it could have up to 1" of total component + lead length)
Package
Price

Selecting a capacitor often takes some work. Yes, many engineers would simply drop in a 1 uF cap of whatever technology and call it good. Is the IC oscillating? It is stable over temperature as the cap parameters change. Is it stable over time as an electrolytic capacitor loses capacitance and its ESR climbs?

Tantalum (no)
I am not a fan of tantalum capacitors and using one to establish a specified ESR range is problematic. For example, none of the caps in the datasheet (link below) have an ESR of 5 ohms or less until we get to 22 uF. Perhaps this is fine for the TC1262 as it does not specify a maximum output capacitance. And, tantalum caps are expen$ive at over $2. To me these things are so last century.

Kyocera https://datasheets.kyocera-avx.com/TAC.pdf


Ceramic (yes)
I prefer to use X7R ceramic caps for low voltage coefficient and low enough temperature coefficient, a known low ESR and low cost. Looking at the Murata GRM155R70J105KA12 (0402, 1uF, 6.3V, $0.10) the 1 kHz DF is 0.125 max giving us an ESR = 20 ohms max. Too high. Let's try the Murata GRM21BR71C475KE51 (0805, 4.7uF, 16V, $0.15) it has a DF of 0.125 for a 1 kHz ESR = 4 ohms. This meets the upper end of the TC1262 allowable ESR spec. The minimum ESR is not specified and to be on the safe side you can place a 0.1 ohm resistor in series. Although I might be the only one doing this, it works. One caveat for the series resistor is the capacitor charging stress on the resistor as it must absorb energy equal to the charged capacitor during whatever charging period there is. For this location at the regulator output I would guess that is much greater than a few microseconds. It also must hold up to the AC ripple current.

https://search.murata.co.jp/Ceramy/image/img/A01X/G101/ENG/GRM155R70J105KA12-01.pdf


Electrolytic (no)
I did some looking and it didn't look good so I stopped.

Conductive Polymer Aluminum (okay but expensive)
The Panasonic 10SVP4R7M might work. Downsides are price ($1.14), short life and possibly a too-low of ESR at higher frequencies.

https://industrial.panasonic.com/ww/products/pt/os-con/models/10SVP4R7M
 

Thread Starter

Max Holdcroft

Joined May 12, 2024
37
The 3V TC1262 output voltage can vary +/-2.5% giving it a probable range of 2.925 to 3.075 volts. Given a dropout voltage of 30 to 650 mV (depending on load current) the battery voltage must remain above 3.205 V at 100 mA to 3.725V at 500 mA as it discharges.

As nasapook shows from the TC1262 datasheet the recommended output capacitor is 1 uF with an ESR of 0.1 to 5 ohms. To select a capacitor there are other parameters to consider:

Operating temperature range
Lifetime
Voltage (4V or higher)
Voltage coefficient of capacitance (we don't want the capacitance dropping much below 1 uF at 3V)
Capacitance, >=1uF
Capacitance temperature stability (we don't want the capacitance dropping much with temp variation
ESR, 0.1 to 5 ohms
ESL, <25 nH (from the 1 MHz max SRF spec and this means it could have up to 1" of total component + lead length)
Package
Price

Selecting a capacitor often takes some work. Yes, many engineers would simply drop in a 1 uF cap of whatever technology and call it good. Is the IC oscillating? It is stable over temperature as the cap parameters change. Is it stable over time as an electrolytic capacitor loses capacitance and its ESR climbs?

Tantalum (no)
I am not a fan of tantalum capacitors and using one to establish a specified ESR range is problematic. For example, none of the caps in the datasheet (link below) have an ESR of 5 ohms or less until we get to 22 uF. Perhaps this is fine for the TC1262 as it does not specify a maximum output capacitance. And, tantalum caps are expen$ive at over $2. To me these things are so last century.

Kyocera https://datasheets.kyocera-avx.com/TAC.pdf


Ceramic (yes)
I prefer to use X7R ceramic caps for low voltage coefficient and low enough temperature coefficient, a known low ESR and low cost. Looking at the Murata GRM155R70J105KA12 (0402, 1uF, 6.3V, $0.10) the 1 kHz DF is 0.125 max giving us an ESR = 20 ohms max. Too high. Let's try the Murata GRM21BR71C475KE51 (0805, 4.7uF, 16V, $0.15) it has a DF of 0.125 for a 1 kHz ESR = 4 ohms. This meets the upper end of the TC1262 allowable ESR spec. The minimum ESR is not specified and to be on the safe side you can place a 0.1 ohm resistor in series. Although I might be the only one doing this, it works. One caveat for the series resistor is the capacitor charging stress on the resistor as it must absorb energy equal to the charged capacitor during whatever charging period there is. For this location at the regulator output I would guess that is much greater than a few microseconds. It also must hold up to the AC ripple current.

https://search.murata.co.jp/Ceramy/image/img/A01X/G101/ENG/GRM155R70J105KA12-01.pdf


Electrolytic (no)
I did some looking and it didn't look good so I stopped.

Conductive Polymer Aluminum (okay but expensive)
The Panasonic 10SVP4R7M might work. Downsides are price ($1.14), short life and possibly a too-low of ESR at higher frequencies.

https://industrial.panasonic.com/ww/products/pt/os-con/models/10SVP4R7M
Hey Dovo,
Thank you very much for your detailed response! As you could probably tell, I'm a complete noob in PCB design, so I'm a little confused with where these capacitors go in my schematic. Do I need to add extra filtering caps to my schematic, aside from the ones that are included on the TC1262 datasheet? Are these the capacitors that you're referring to in your response? Again, thank you so much for helping me out with this.

Max
 

dovo

Joined Dec 12, 2019
72
Hey Dovo,
Thank you very much for your detailed response! As you could probably tell, I'm a complete noob in PCB design, so I'm a little confused with where these capacitors go in my schematic. Do I need to add extra filtering caps to my schematic, aside from the ones that are included on the TC1262 datasheet? Are these the capacitors that you're referring to in your response? Again, thank you so much for helping me out with this.

Max
Yes, the capacitor I commented on is placed right at the TC1262 output. As indicted by the SRF > 1 MHz spec and 1 uF it must have less than 25 nH inductance between it and the TC1262 output. Given a rule-of-thumb of 10 nH/cm of wiring the total path can be 2.5 cm = 1 inch. That is worse case. Given PCB traces over a GND plane the path might be closer to 50 ohms which gives us an inductance of ~10 nH/inch.

L = td x Zo or 17 ps x 50 ohms = 174 ps/in. x 50 ohms = 8.7 nH/in.

The power rail capacitance to place at the part(s) being powered is often stated in its datasheet. This is where things can get messy when we want to place a significant amount of low ESR capacitance (think ceramic caps) at the load yet this violates the LDO output capacitor spec. High C with low ESR at the load may or may not cause the TC1262 to become unstable and oscillate. As with other drain-output LDOs the TC1262 has a naturally high output impedance that is lowered by a lot of feedback. Having a shunt AC resistance on the output of 0.1 to 5 ohms (per the datasheet) provides a zero in the LDO feedback loop that renders it stable.

The TC1262 datasheet says " Aluminum electrolytic or tantalum capacitor types can be used." These will work as long as the ESR is between 0.1 - 5 ohms over temperature and time. Time being a factor especially with the aluminum electrolytic because its ESR rises over time. So, datasheets must be consulted to determine if the ESR is the 0.1 - 5 ohm range and that the capacitor life is suitable for the product life. As we saw earlier, tantalum caps are expensive. For these reasons I favor ceramic caps for price and almost infinite life. If a resistor is placed in series with the cap to establish a certain minimum ESR that adds the cost of a resistor plus parts placement cost yet can still beat the price of even an aluminum electrolytic cap.


Microchip TC1262 datasheet
https://www.researchgate.net/figure...th-an-alumina-surface-Electrode_fig17_3165903
 

Thread Starter

Max Holdcroft

Joined May 12, 2024
37
Yes, the capacitor I commented on is placed right at the TC1262 output. As indicted by the SRF > 1 MHz spec and 1 uF it must have less than 25 nH inductance between it and the TC1262 output. Given a rule-of-thumb of 10 nH/cm of wiring the total path can be 2.5 cm = 1 inch. That is worse case. Given PCB traces over a GND plane the path might be closer to 50 ohms which gives us an inductance of ~10 nH/inch.

L = td x Zo or 17 ps x 50 ohms = 174 ps/in. x 50 ohms = 8.7 nH/in.

The power rail capacitance to place at the part(s) being powered is often stated in its datasheet. This is where things can get messy when we want to place a significant amount of low ESR capacitance (think ceramic caps) at the load yet this violates the LDO output capacitor spec. High C with low ESR at the load may or may not cause the TC1262 to become unstable and oscillate. As with other drain-output LDOs the TC1262 has a naturally high output impedance that is lowered by a lot of feedback. Having a shunt AC resistance on the output of 0.1 to 5 ohms (per the datasheet) provides a zero in the LDO feedback loop that renders it stable.

The TC1262 datasheet says " Aluminum electrolytic or tantalum capacitor types can be used." These will work as long as the ESR is between 0.1 - 5 ohms over temperature and time. Time being a factor especially with the aluminum electrolytic because its ESR rises over time. So, datasheets must be consulted to determine if the ESR is the 0.1 - 5 ohm range and that the capacitor life is suitable for the product life. As we saw earlier, tantalum caps are expensive. For these reasons I favor ceramic caps for price and almost infinite life. If a resistor is placed in series with the cap to establish a certain minimum ESR that adds the cost of a resistor plus parts placement cost yet can still beat the price of even an aluminum electrolytic cap.


Microchip TC1262 datasheet
https://www.researchgate.net/figure...th-an-alumina-surface-Electrode_fig17_3165903
Hey Dovo,

I added the capacitor as you recommended. This is the current schematic I'm working with for the voltage regulator. Screenshot 2024-05-14 at 9.35.38 am.png
I've probably missed something, but I really appreciate you helping me out with this!

Thanks,
Max
 

dovo

Joined Dec 12, 2019
72
Okay, C3 and C4 are correct. C1 and C2 are not because they block the Vin and Vout DC path. Remove C1 and C2 in your schematic above.

1715657355857.png
 

dovo

Joined Dec 12, 2019
72
The capacitor on the battery side is not to filter any noise from the battery line, it is to establish a suitable impedance for the LDO to draw current from. The LDO does not want to see too much inductance, or too much resistance. The inductance of the 10 inches of wire mentioned in the LT1262 may be as high as 250 nH and placing a capacitor near the LDO helps isolate it that.

Standard engineering practice is to place capacitors at the input and output of voltage regulators. We see engineers sprinkling decoupling caps quite freely, often without regard to whether or not they are actually needed.

From the LT1262 datasheet page 3, "A 1uF capacitor should be connected from VIN to GND if there is more than 10 inches of wire between the regulator and the AC filter capacitor, or if a battery is used as the power source"

https://ww1.microchip.com/downloads/en/DeviceDoc/21373C.pdf
 

DickCappels

Joined Aug 21, 2008
10,246
The purpose of the regulator is that of making the output voltage stable and in general, low noise. Noise on the voltage source feeding the regulator is reduced as the regulator works to maintain the intended output voltage.
 

Thread Starter

Max Holdcroft

Joined May 12, 2024
37
The purpose of the regulator is that of making the output voltage stable and in general, low noise. Noise on the voltage source feeding the regulator is reduced as the regulator works to maintain the intended output voltage.
So are you saying I don't need to place any filtering capacitors in the circuit?
 

dovo

Joined Dec 12, 2019
72
Standard engineering practice is to place decoupling caps on a part whether it actually needs it or not. I would add a 1uF at the LDO input and at the output. I am concerned about the 1 uF capacitor ESR falling below the minimum specified 0.1 ohms. Kemet just happens to have a capacitor simulator we can consult for some of their ceramic caps. The Kemet C0603X105K9RAC plot shows ESR dropping below 0.1 ohms above 32 kHz. Kyocera also has a capacitor simulation tool that shows ESR. To ensure this isn't an issue with the LDO spec a 0.1 ohm resistor can be placed in series the LDO output 1 uF ceramic cap.

I know this looks like a lot of picky detail work to select a capacitor but taking the time to select the right parts can reduce parts cost, reduce field failures and increase manufacturing yield.

1715733432728.png


KEMET simulator https://ksim3.kemet.com/capacitor-simulation
 

atferrari

Joined Jan 6, 2004
4,798
Standard engineering practice is to place decoupling caps on a part whether it actually needs it or not. I would add a 1uF at the LDO input and at the output. I am concerned about the 1 uF capacitor ESR falling below the minimum specified 0.1 ohms. Kemet just happens to have a capacitor simulator we can consult for some of their ceramic caps. The Kemet C0603X105K9RAC plot shows ESR dropping below 0.1 ohms above 32 kHz. Kyocera also has a capacitor simulation tool that shows ESR. To ensure this isn't an issue with the LDO spec a 0.1 ohm resistor can be placed in series the LDO output 1 uF ceramic cap.

I know this looks like a lot of picky detail work to select a capacitor but taking the time to select the right parts can reduce parts cost, reduce field failures and increase manufacturing yield.

View attachment 322352


KEMET simulator https://ksim3.kemet.com/capacitor-simulation
Link not working for me.
 

dovo

Joined Dec 12, 2019
72
Here is how I have it set up.

1715803881811.png

Let's use the KEMET tool to find a suitable tantalum capacitor. First I looked at Digikey to see what they had in stock for Kemet tantalum caps and found the T491A105K016AT. Plugging this number into the Kemet tool it gave this plot. That parts ticks off all the boxes for the LDO and I think is a good choice. Digikey price is $0.18 in quantities of 100.

1715804516141.png
 
Top