[SOLVED] Gm-C Filters

[ahmedatef0]

I have an assignment to make a Gm-C low pass filter with a cut-off frequency of 0.8GHz. So I think as a rule of thumb the Gm-Cell should have a bandwidth of 8GHz (10*cut-off freq.)
My question is how do I make a Gm-cell? Is it like building an OTA? I've made OTAs before they never had this large bandwidth, in the OTA the output node had very large impedance thus it had a very small dominant pole so how will I get this BW, and If I decreased Rout the OTA will not act as an ideal current source.
It's also required that the filter should e DC coupled I don't know how to do this as well.
If anybody can help me understand the points that confuses me or can send me resources to understand better I would be very grateful.

 

[frankrose]

What is the purpose of this filter? TX output or RX input filter? Image reject filter, or you want to attenuate harmonic products?
Normally people doesn't use active filters on such high frequencies I think, why do you need complex pole pair on 800MHz? Or, are you sure you need active filter?
With passive RC or RLC circuits DC coupling is not an issue, but I think offset you can eliminate with trimming of OTA input, DC coupling is not an issue.

 

[ahmedatef0]

What is the purpose of this filter? TX output or RX input filter? Image reject filter, or you want to attenuate harmonic products?
Normally people doesn't use active filters on such high frequencies I think, why do you need complex pole pair on 800MHz? Or, are you sure you need active filter?
With passive RC or RLC circuits DC coupling is not an issue, but I think offset you can eliminate with trimming of OTA input, DC coupling is not an issue.

I don't know the purpose of the filter, it's an assignment I have to make and I didn't choose the topology

 

[sutapanaki]

Gm-C filters are used for pretty high frequencies but they don't provide good linearity, say in the order of 40-50dB. For example, in HDD, the filter can go up to 2GHz to 3GHz with 40dB of linearity.
The Gm cell is different than an OTA. OTAs usually work in a feedback configuration and cannot achieve very high (GHz) frequencies. Plus, the Gm cell for the filter will need to have certain finite input differential voltage and not a virtual short as is the case for OTAs. In this respect and for high frequencies a degenerated differential pair can be a good candidate.
As a rule of thumb if a Gm cell can drive the same Gm cell but diode connected without degradation of the frequency response up to the cut-off frequency of the filter, then it can probably be used to build that filter