LMP7721: Input noise prevention for trans-impedance amplifier

yunsik kim

Prodigy 130 points

Part Number: LMP7721
Other Parts Discussed in Thread: TLV9161

Hi, team.

I'm designing a trans-impedance amplifier using the LMP7721.

The C2 and IG1 is a model of sensor.

The sensor output is about 0.1~10 nA AC current.

The constant bias of 1.2 ~ 1.8 V to + input is necessary to read output with ADC.

I'm worried that the voltage divider affects the noise.

Is it best to use low noise voltage regulator to reduce input noise?

What should I consider when choosing LDO for this application?

Best regards.

over 2 years ago

0 kai klaas69 over 2 years ago

Guru 140200 points

Hi Yunsik,

this would be the "noise gain" of your circuit when using a 47µF aluminium electrolytic paralleled by a 470nF/X7R/0805:

yunsik_lmp7721_2.TSC

What LDO are you using?

Kai

0 kai klaas69 over 2 years ago in reply to kai klaas69

Guru 140200 points

You could decrease the "noise gain" this way:

yunsik_lmp7721_3.TSC

Kai

0 Michael Steffes over 2 years ago in reply to kai klaas69

Guru 47535 points

You should probably look at this recent discussion also,

https://e2e.ti.com/support/amplifiers-group/amplifiers/f/amplifiers-forum/1062671/lmp7721-tina-ti-simulation?tisearch=e2e-sitesearch&keymatch=lmp7721#

0 yunsik kim over 2 years ago in reply to kai klaas69

Prodigy 130 points

Hi Kai,

I am going to use a low noise LDO such as LP5907MFX-3.3.

Can I use a low-cost buffer amplifier for providing the bias of 1.65 V?

Yunsik

0 kai klaas69 over 2 years ago in reply to yunsik kim

Guru 140200 points

Hi Yunsik,

for lowest noise you should omit all this biasing stuff but should use a bipolar supply voltage and reference both the +input of OPAmp and the photo detector to signal ground. The reason is that voltage references and voltage regulators can show a huge increase of noise below 1Hz and you would need extreme low pass filtering time constants to suppress the noise to an acceptable level.

If you still want to go for this pseudo ground generation, you could try the TLV9161 or similar. But keep in mind that this OPAmp has a different common mode input voltage range. In any case you would need to run a phase stability analysis. Keep also in mind, that the OPAmp for the pseudo ground generation lies in the feedback loop of LMP7721 and should be almost as fast as the LMP7721. Otherwise the increasing (eventually inductive looking) output impedance could result in stability issues.

Kai

+1 Marek Lis over 2 years ago in reply to yunsik kim

TI__Guru 60610 points

Yes, you should always use a buffer to bias the reference voltage. The circuit is stable with 84 degrees phase margin - see below.

Half of the total noise comes from the LMP7721 integration of its input voltage noise across its bandwidth - see below.

While the rest comes from the conversion of the input current noise and the thermal voltage noise of the 250M resistor - see below.

However, the circuit will become non-linear for IG1 > 6.5nA - see below.

Thus, in order to measure IG1 range from 0.1nA to 10nA you should lower the Bias voltage to 200mV and you may increase the gain using 300M resistor.

Since 300M precision resistor would be very expensive, you should use the T-network to get the same overall transimpedance gain while using 10M feedback resistor together with two extra resistors - see below.

Amplifiers

Amplifiers forum

LMP7721: Input noise prevention for trans-impedance amplifier