Issues concerning designing current mirrors

[srivatsan]

Current Mirrors

In designing current mirrors:
If I can use only one 20kohms resistor in the whole circuit for current mirrors, then it becomes really tough to construct W/L. I am not able to find answer to this question. Assume small current like 10uA and large current 100uA. How does the design change for these cases? thanks for replies...
srivatsan

 

[hmsheng]

Re: Current Mirrors

Can you discribe your question in more detail?

 

[frankon_o2]

Current Mirrors

It seems not nessencary to use resistor in a normal current mirror. Are you asking about how to generate a current bias with some transistors and only one resistor?

 

[alanbrooke]

Re: Current Mirrors

For the current mirror, I think no resistor is needed, but for the current Bias, you may need it.
You can simply multiply the current by multiply the W/L ratio.

 

[srivatsan]

Re: Current Mirrors

Well, I do need the current bias in the current mirror circuit. First of all: L is fixed at 0.6um and is not changed in the whole circuit. Then, resistor used in bias circuit is atmost 20k and definitely not more than that. Now the only parameter that can be changed is W to get current from 10uA to 100uA. Using equations (taken from allen and/or baker) to calculate the W has been proved futile because when simulated in BSIM3V3.2 SPICE, the circuit failed to the intended purpose. Now the answer is to use combination of the transistors to get the answer. How to do it? If ye'll know, let me know.
srivats

 

[rajath]

Re: Current Mirrors

Your question is still not very clear! can you also attach the circuit which
you are attempting to build?.. even a rough sketch will do!
If you want to generate a host of currents, the best thing would be to generate the 100uA
and then use current splitters for smaller values.

 

[StevenL]

Current Mirrors

Or if you require a host of currents, you can perhaps determine one common denominator and add unit cell current mirrors. (because of matching issues)

Since the current (at fixed L and Vds) should scale linearly (roughly) with W, I do not understand why the circuit would not operate in your simulations. could you elaborate?

 

[srivatsan]

Re: Current Mirrors

Apparently, everything works when it is atleast 1 um tech. The moment you go downwards, things are not that clear when deciding analog systems. I am not looking for creating host of currents from single 100uA because of current hogging. So I am trying to design the bias and therefore mirror and other components from scratch for each and every op-amp and therefore ADC and etc..... The idea is to develop a procedure which can help me realize analog components which can be (+ or -) 10 percent of the final answer. If you can design the bias and therefore mirror circuits from 10uA to 100uA., that will be cool for 0.6um. If you refer to the discussion in this same forum, I have asked about current mirrors. There I have the code. Simulate it. Let me know.
Regards,
srivats