How would you bypass DC supply with this LDO?

[thiagofinottimoraes]

Hello,

I am considering using LP2980 LDO to supply DC power to a thermistor excitation and reading circuit.

Although it is a very simple circuit, I am short on space and need tight voltage accuracy (tolerance +-0.5%). LP2980 seems quite attractive in its price range. However, LP2980 datasheet (pg 10, attached) suggests not using ceramic capacitors below 2.2 uF in the output, since low ESR values could unbalance frequency stability of the circuit.

I am also using OPA340 Op amp., and it requires a 0.01uF ceramic capacitor bypassing DC supply (datasheet attached, pg 8). Then I got trapped! Tantalum caps won't go down as 0.01uF and film caps have even lower ESR.

It looks like I got out of option in this configuration. So I pose two questions:

1) OPA340 datasheet asks 0.01uF bypassing DC supply, but how bad/unreliable would it be if I use 2.2uF?
2) LDO's can be quite ineffective for many applications if it really can't drive ceramic bypass caps in the 100nF range. Do you have any thoughts/experience on that?

EDIT: Sorry, maybe it would had been better to post this in Analog Electronics section!

 

[barry]

You require two, not one caps. One for the regulator >2.2uf and one for the opamp. Putting a 2.2uf cap at the opamp probably won't help the regulator, unless they are physically very close. If you don't have space to put a small 0603 (or even a 0402) cap, you've got problems.

 

[thiagofinottimoraes]

Berry,
Thanks for the reply!

Well, there is already a 10u tantalum in the regulator output. I agree I also need a cap in the op amp input, and I have the space for that. My question is if what would be the outcome concerning stability/reliability of the op amp of I can replace the 0.01uF for a 2.7uF ceramic. Both caps are the closest as possible to their respective IC's, about 0.85 mm between pads.

What do you think? This is one of those small details you better pay attention to closely, it may be a deal braker!

 

[barry]

Why do you not want to use a .01uF on the op amp? That cap is for decoupling, it doesn't affect stability of the op amp; it's there to minimize noise from the power supply lines getting into your signal.

A 2.7uF might work fine, but its high frequency performance may not be as good as the .01; it really depends on your specific application.

 

[thiagofinottimoraes]

Yes, I agree with you, the application is noise sensitive, since it will be operating close to a 7500W SMPS, and I am craving to use the 0.01u ceramic!!

The reason I am hesitant is that the LP2980 datasheet (pg 10, attached) suggests not using ceramic capacitors below 2.2 uF in the output, since low ESR values could unbalance frequency stability of the circuit. Although it is just a bypass cap, I am not sure of the stability of LDO if I place it there.

 

[barry]

Yes, I agree with you, the application is noise sensitive, since it will be operating close to a 7500W SMPS, and I am craving to use the 0.01u ceramic!!

The reason I am hesitant is that the LP2980 datasheet (pg 10, attached) suggests not using ceramic capacitors below 2.2 uF in the output, since low ESR values could unbalance frequency stability of the circuit. Although it is just a bypass cap, I am not sure of the stability of LDO if I place it there.

I think you should be ok. Keep in mind it's the ESR of the output cap that's the issue with LDOs, not caps elsewhere in the system. The ESR and capacitance form a pole in the response; if the ESR is too low, your pole frequency will be too high and you risk oscillation. Since you've already established a low frequency pole with your tantalum, there shouldn't be any problem.

 

[thiagofinottimoraes]

Barry, I am willing to think like that. But could you take a look at the attached paper? It is published in TI's LP2980 ldo website and disregards this last conclusion of ours that it should be ok if there is already a low pole from tantalum.

Paper: Engineers note: Capacitors are key to voltage regulator design
https://www.ti.com/litv/pdf/snoa842

LP2980: https://www.ti.com/product/lp2980-n


Please tell me your thoughts!

EDIT: Particularly page 8, in case it gets too long...

 

[barry]

Hmm, that's pretty interesting. And pretty scary. I've learned something new. I guess you've got several options: add a small resistor in series with your a opamp power supply(this will mess up your power-supply rejection a bit, but I've seen it done before); use ceramic caps with appropriate ESR; forget all this hardware stuff, and become a C++ programmer.

 

[thiagofinottimoraes]

Pretty much alarming: in simple words, it says you can't bypass anything powered by most (or all, if he is correct) LDO's with cap less than about 100n, that is the lowest tantalum goes, or without adding a series resistance to ceramic caps, which seems to be, first of all, odd - and detrimental if you are trying to provide a low impedance path to noise in that particular band related to ceramic caps!

 

[barry]

When you put a small resistance in series with the decoupling cap you are actually making a low-pass filter, which is a good thing. You just have to verify that the variation in voltage for the op-amp supply (due to the variation in op-amp current through that resistor) won't affect your output too much (look at the PSRR parameter).

But I agree, this makes this look like a pretty useless regulator.