Hello,

I am trying to implement  a gm-c filter with low input impedance (the previous stage is current). Does anybody have any idea how to do that? I saw all of Gm-C filter is deal with voltage.

Thanks

Hi Loose-electron,

The reason I don't want to convert it to a voltage is to keep the previous output stage's node have small voltage swing. Ideally, I would like the Gm-C filter act like a transimpdeance device.

mx

gmC filters are generally used for signals of such high BW that op-amp methods are not viable.

That said, you can process the signal as a current, as suggested above.

buddypoor - can you suggest a paer he can read as a starting point for research?

- jerry

Hi loose-electron,
Hi nxing,

here are some information on current-conveyor (CC) based filters.

* The second generation CC (called CCII) is a current source (output z) with two inputs:
  positive high resistance: y
  negative low resistance (<50 ohms): x
*  Two types: CCII+ and CCII-  (depending on output current ditection)

* Commercially available (CCII+): AD844
* Note: The AD844 has also an output buffer on-board; therefore, it can also be used as current-feedback amplifier (CFA) and very often it will be used as such. But the high-impedance node is accessible and the AD844 can be used also as CCII+.
* Unfortunately (I don't know why) a lot of publications for CC-filters with the AD844 are called "CFA-based filters", although it is used as CCII+.
* Therefore, in case you google for those filters, it is useful to search for "CFA based filters". You can be sure, it is an CC-filter (very often with AD844).
* You can find a lot of articles in IEE and IEEE publications as well as in "Int. Journal of Electronics".
* Unfortunately, I only have one article available as a pdf file (see attachement).
* Some references (out of many):
- Comparison of RC opamp and RC current conveyor filters (J.A.Svoboda), Int. J. Electronics, 1994, vol 76, 615-626;
- New configuration for single CCII first order and biquadratic current-mode filters (Hou, Wu, Liu), Int. J. Electronics, 1991, vol 71, 637-644;
- Application of a commercially available current conveyor (Svoboda, McGory, Webb), Int. J. Electronics, 1991, vol 70, 159-164;
- Insensitive voltage-mode and current-mode filters from commercially available transimpedance opamps (Fabre), IEE proceedings-G, 1993, vol 140, 319-321.
- Book: The Current mode approach (C. Toumazou)  contains a lot of papers dealing with CCII.      

* Internet search for "Toumazou" leads you to other relevant links.
_____________
All for now.
Regards

nxing wrote on Jun 21^st, 2011, 5:46pm:

If you are operating at that low a frequency I think you may want to consider op-amp architectures.

Why? As a general rule, a high gain amplifier used with feedback (classic op-amp) will have much better linearity.

On CMOS processes, from, roughly 0.25um and smaller, I would not start using gmC structures until you get above 10MHz BW or so.

Those number may vary some, but if you are under 5MHz you should be able to implement with op-amps.

A gmC filter may require a sizable DC bias current to avoid linearity problems, and may require scaling up the output current to minimize second order effects due to the non ideal nature of the transconductor.

Op-amp structures can run with Class AB outputs and less current.

I need the voltage output

Hi,

A good idea for gmc filter design is to start with LC prototype. Mostly such LC filters are doubly terminated, because this leads to lowest sensitivities on element values!
However, you can drive the filter also with a voltage or a current, but then termination with a load resistor is NEEDED (otherwise Q will be infinite)!
You can also add an op-amp to get a current input and use no gm=voltage-controlled current source, but a current-controlled voltage source based filter.

Bye Stephan