ASIC design of a transimpedance amplifier for a high input current dynamic range(50n-10u and output is 5V) working in a very low frequency/DC mode

[VisheshG24]

I am trying to design a transimpedance amplifier to fabricate on silicon without using any IC with the following specs:

1. Input current range: 50nA to 10uA.
2. No bandwidth requirement.
3. I want the output voltage to be max 5V.
4. I have a vdd supply of +5V.

I am not able to find any work similar to the specs that I mentioned. Most of them focused on bandwidth, but I have no bandwidth requirement. Additionally, the work area is mostly a small signal for the existing work that I can find. I just want to input the current from a sensor and convert it into voltage. Can someone help me guide how to approach this problem or share any pre-existing work? I would appreciate any help.
Thanks and Regards, Vishesh

 

[dick_freebird]

You need to figure out some stuff like the source impedance
and return. Does sensor deliver a value thaty is common-
mode-invariant or is its fidelity only had against a (virtual or
otherwise) ground?

A simple resistor properly scaled could fit the bill, limited
as the "ask" is.

But your low end will suffer for noise and low deflection
against a varying Vio (either true differential or implicit
difference-to-ground).

For many low current applications a CTIA is used, cap
is reset and then connected for a calibrated interval, and
your value is in dV/dt_sample=I/C. You can get a scale
factor option simply by altering the sample pulse width.
Simple ones are not very linear but you could cal-map that.

 

[KlausST]

Hi,

I see nothing in your requirements that excludes a simple 500kOhms resistor.
It will "translate" 10uA to 5V

Klaus

 

[VisheshG24]

But your low end will suffer for noise and low deflection
against a varying Vio (either true differential or implicit
difference-to-ground).

Hi, thanks a lot for the reply. Can you please elaborate more on this? Can you provide any way to still use the resistor and reduce the problems that you mention? Can you please share any work that you can find for the CTIA?

 

[VisheshG24]

Hi,

I see nothing in your requirements that excludes a simple 500kOhms resistor.
It will "translate" 10uA to 5V

Klaus

Hi, Thanks a lot for the reply. What about the input from the sensor? Like if the Rin of the current-to-voltage converter is greater than the Rout of the current sensor, the current would not be able to get in very well?

 

[KlausST]

What about the input from the sensor?

You have to tell us. We can not know the values, function and requirements of your application.

Like if the Rin of the current-to-voltage converter is greater than the Rout of the current sensor, the current would not be able to get in very well?

Rout (Zout) of a current transducer is expected to be close to infinity.
It will regulate output current .. for the specified output conditions.
--> read the datasheet

Klaus

 

[VisheshG24]

You have to tell us. We can not know the values, function and requirements of your application.


Rout (Zout) of a current transducer is expected to be close to infinity.
It will regulate output current .. for the specified output conditions.
--> read the datasheet

Klaus

Hi, Thanks a lot for the reply. The current is supplied through the sensor, which is a FET-type sensor that gives the current output through the drain. So, if I make sure that the Rout of this sensor is much much greater than the Rin of the current-to-voltage converter( now, in my case, it will be a simple resistor), then it is perfectly fine to use it. Do you know any other problem that might arise with the use of the resistor besides this input resistance problem?

Best Regards,
Vishesh

 

[KlausST]

Hi,

this is a physical / mathematical problem. It can´t be solved with text.

Datasheets give values. That´s what I asked for. Indeed, it´s not me who needs the values. It you / the designer who needs the values.

***
With your values I calcualted 500k. This is good for an "on PCB" solution. But if there are longer traces, wires, cables involved .. they will pick up noise.
We can not know about your situation. We can not know whether you use shielded wires, whether there are noise sources nearby. We don´t know whether you intend to use any low pass filters.

You say you have no bandwidth requirements. This is not true. A signal is expected to settle within a given time .. to be measured.
It may be several milliseconds, seconds, minutes .... even hours .. but it relates to a bandwidth. There is no signal with no bandwidth.

Just because we are used to frequencies in the GHz, MHz, kHz range ... does not mean slower signals need no bandwidth .. to be measured.
Let´s say you have a signal with rather low frequency. Let´s say 0.0000116Hz .. and you want to measure it ... you need a much higher bandwidth for it to be read with an expected accuracy. Maybe you expect to measure this signal with a phase accuracy of 0.25°. Then the "measurement" needs to be processed with a much higher bandwidth than this "fundamental" 12 microHertz signal.
This example is not unrealistic. The frequncies are not unrealistic, nor is the phase accuracy.
The frequency is "a day" ... and the phase accuracy equals 60 seconds. A quite usual daily measurement problem: You want to read the time of the day with an accuracy of one minute.

Klaus

 

[VisheshG24]

Hi Klaus, Thanks a lot for the explanation; I really appreciate it. I am a beginner in designing, and I am not sure what the best way to approach these design challenges is. The current sensor used is not a publicly available device, so I cannot get the data. I will try to get the data from the people that made the device.
I agree with what you said about BW. So, I think assuming a 5KHz BW is reasonable for a low-speed requirement.

But my direction of making this current-to-voltage converter will be using a TIA or a simple resistor, I am still confused. I understand you need more data to give an answer(I'll do it asap). But how, as a designer, this goes into your mind and how you solve this would be really insightful for me. You can assume a resistance or any other parameter of the current sensor and tell me how to approach it.

Thanks and Regards,
Vishesh

 

[KlausST]

Hello.
The sensor is the one side ....but your application is the other side that needs to be specified.
Accuracy, precision, resolution...you need to decide numbers. Nobody can do this for you.
Also: (repeated) cable length, cable type (shielded, twisted pair...) trace lengths, noise sources around.

There is no general rule. It depends on your application. Then one can calculate. And calculators work with numbers only. No calculators work with textual informations.

Klaus

 

[VisheshG24]

Hello.
The sensor is the one side ....but your application is the other side that needs to be specified.
Accuracy, precision, resolution...you need to decide numbers. Nobody can do this for you.
Also: (repeated) cable length, cable type (shielded, twisted pair...) trace lengths, noise sources around.

There is no general rule. It depends on your application. Then one can calculate. And calculators work with numbers only. No calculators work with textual informations.

Klaus

Hi,
Thanks a lot for the reply. You are right; nobody can do this for me, and I also want to do it myself. I just need some guidance on what the thinking process should be. I don't need exact answers, just how to start approaching the problem.

For cable length etc., I understand. But when you start making your design you deal with them later(as far as I know)? and you can always adjust your layout accordingly.

My application concerns the information I gave you; I want to integrate this current-to-voltage converter with the current sensor. Please tell me what else you usually require (I am sorry if I am asking you again; I am not very experienced).

Thanks and Regards,

Vishesh

 

[KlausST]

design you deal with them later(as far as I know)?

No. You first need to estimate how much noise you are likely to pick up.
I hate to repeate. If the node is short ... it may be high impedance. But if it's long you need to lower the node impedance.
If you don't care from the beginning you do it wrong, don't get the expected quality... and have to design it again.

But enough repeated. I wiil be back after you provided useful numbers. Everything else is a waste of time. You don't do electronics design by writing text.

You did not provide a single number. If you are the " trial and error designer" I'm not designing this way. I use science, physics, math.

Klaus