How to sum two opposite polarity currents in CMOS 0.18u technology?

[thanu]

I am trying to obtain the temperature invariant current reference using basic BGR circuit, but now i'm stuck at this part. Exactly how to implement this part of the circuit when i have two opposite polarity currents

 

[dick_freebird]

Can you guarantee that one is always larger than the other?
If so then this is basically like a folded cascode type problem,
the stronger should draw from the cascode source and the
weaker only subtracts some of it.

Bipolar operation is trickier. But I don't think this is needed
in this case.

 

[thanu]

Thanks for the reply freebird.
Actually I am using a BJT to obtain the CTAT current, but I hope that will not be a problem to implement the current summing thing which i'm looking for. And yeah the currents are not exactly the same but nearly the same but with opposite polarity.
What u r seeing in the below picture is, i'm obtaining the PTAT current from regular BGR circuit, and from the small circuit at the right side with a BJT and load is to obtain a CTAT current. So I should manage to bring in both the currents to obtain a current reference.

 

[Warpspeed]

The trick might be to fold one of the currents back using a current mirror.
Then just sum the two currents directly.

 

[thanu]

Hi Tony, Thanks for the reply. How exactly i should fold one of the currents using current mirror and how to sum them together? can u just elaborate in a pictorial form. Is the below picture is something similar to what u r trying to say? I have a major doubt in this part, is it this easy to sum two currents and is it a right way of implementation?

 

[erikl]

... is it a right way of implementation? View attachment 122077

I wouldn't sum on a resistor, but on the (master) transistor of a current mirror. Here is a short note on a CMOS current reference, which uses such a method: View attachment CMOS_current_reference_with_supply_and_temperature_compensation.pdf

 

[thanu]

Hi Erikl, I've implemented this circuitry and sadly i'm not obtaining the current reference output, take a look on these images:

 

[erikl]

I've implemented this circuitry and sadly i'm not obtaining the current reference output ...

Hi thanu, I just wanted to show you the summing principle. Perhaps I should have warned you about the reproduction of the full circuit: my result (in a 0.18µm process, too) wasn't much better - and this after a lot of resizing trials on transistors and resistors:



It was, however, good enough between 0 and 85°C for our purpose.

 

[thanu]

Erikl, is it true that not all ur transistors are in saturation region? Since i'm seeing the current response what u have got is in terms of nAs.

- - - Updated - - -

Erikl, by the way i'm trying to obtain the reference current in the range of 10uA with just 0.1% of variation for pretty good temperature range (atleast till 100 deg cel.). I know it's too much to ask for, but it's my requirement.

 

[erikl]

is it true that not all ur transistors are in saturation region? Since i'm seeing the current response what u have got is in terms of nAs.

Of course they all were in sat. region. A 0.5µA current reference was our requirement.

by the way i'm trying to obtain the reference current in the range of 10uA with just 0.1% of variation for pretty good temperature range (atleast till 100 deg cel.). I know it's too much to ask for, but it's my requirement.

For a 1000ppm/100°C = 10ppm/°C drift forget about this circuit. You'd need one of the best available bandGap references for this (check e.g. this Vita/Iannaccone design), and a resistor with an even lower temperature drift - probably an external one.

 

[thanu]

Thanks for the advice Erikl, I'll go through this design, will implement and share the response I obtained from this.