How does LTSpice measure the inoise in noise simulations? [Ouput dependant result]

[frilance]

Hi everybodoy,

I've got a doubt concerning the use of noise simulation in LTSpice. As far as I know, the input current noise density (inoise in LTspice) is the equivalent noise at input coming from every noise source in the circuit.

I'm dealing with one schematic representing a TIA photoreceiver circuit, formed by a common transimpedance first stage, a then 2 different paths, one for the AC signal and other for the DC signal (different amplifications). I was surprised seeing that the inoise of the circuit (using a input current source replacing the photodiode performance) is different depending where I place the output of the circuit. That's something I don't get. Shouldn't be the same always? Taking into account every noise source is affecting always, no matters where i want to place the output.

Thank you in advance.

Regards.

 

[keith1200rs]

The equivalent input noise is calculated by taking the output noise and applying the transfer function from input to output to find out what it WOULD be if it had all originated at the input. So, if you add a stage which either alters the noise or the frequency response and treat that as the output, the EQUIVALENT input noise will be different. The actual input noise hasn't changed but you will have either added some noise or removed some (by filtering) or a bit of both.

Keith.

 

[frilance]

Thank you so much for your answer Keith. It helped me with the understanding of equivalent noise calculation.

I would like to ask just one more thing... if I got it right... In the case I'm using as output the one from a first stage (instead of a secondary/last one).... Are the noise effects of the secondary/last stage included in the noise for the inoise calculation? They shouldn't be included, right?

Thank you again.

 

[keith1200rs]

Correct. If you select the output node as the first stage then everything after that point is ignored (unless there is some weird feedback to the earlier stage). In general you should see the equivalent input noise reducing as add more stages (unless you have made a bad choice of device or design) because the first amplifier should generally be the one where noise is highest (as the signal is smallest).

Also, you will see weird things happening when looking at circuits with many stages. The input noise can climb at high frequencies when it didn't when you just had one stage. This is again due to the way the equivalent input noise is calculated. If the front end has low bandwidth and a later stage has a high bandwidth but with some high frequency noise then when you try to plot that as an EQUIVALENT input noise the noise must increase at high frequencies. It can be confusing to begin with until you get to know what is important.

I am actually currently working on a TIA and so this is the sort of thing you will see even from one stage:



Keith.

 

[FvM]

If you manage to add relevant noise contributions in other circuit stages, the input referred noise will increase.