Low Pass Sallen Key Filter design

[mfhanif]

Hello Everyone, I want to design a 4th order Butterworth LP filter of Sallen Key type. The 3dB frequency is 750MHz....can someone give me ideas, references, etc. to achieve this high frequency filter design

thanks

 

[FvM]

Not a promising idea. It can at best work inside an IC implementing a very fast transistor technology. With discrete transistors, even of sufficient bandwidth > 10 GHz, the electrical delay will be too large to achieve an acceptable filter characteristic or just stability.

Simply go for passive LC filters.

 

[Milad-D]

At this frequency you can use gm_c filter. Designing at such frequency might be difficult with active rc filter or switched capacitor filter. But the linearity if gm-c filters is inferior compared to the other ones.
Conventionally, you design it with a cascade of two bi-QUAD stages

 

[love_analog]

Agree with Milad-D. I implemeted 6th order filter this way in 1995. This was upto 4MHz. I don't know why FvM is saying use discrete components.

 

[FvM]

This was upto 4MHz. I don't know why FvM is saying use discrete components.

Do you consider, that there may be a difference between 4 MHz and 750 MHz?

I any case, a filter different from a simple cascade of first order will need feedback. Then, also a gm-C filter has to struggle with stability issues. Thus I'm curious to see a 4th or 6th order gm-C filter with 750 MHz characteristic frequency outside an IC. But I'm willing to learn.

 

[Milad-D]

Dear FvM,

In the old tech 0.35um CMOS, they have achieved 550MHz -3dB bandwidth.

Pavan, S.; Laxminidhi, T.; , "A 70-500MHz Programmable CMOS Filter Compensated for MOS Nonquasistatic Effects," Solid-State Circuits Conference, 2006. ESSCIRC 2006. Proceedings of the 32nd European , vol., no., pp.328-331, Sept. 2006

Therefore, i think it will be very easy to get f-3dB=750MHz in 0.18um or 0.06um. Why gm-C is better than Active-RC? Because you can use much less complex structure for gm cell than Op-amps which means that the frequency capabilty of the gm-cell is higher.


By the way, Passive filter is the best one, but it consumes a lot of area in chip.

 

[FvM]

Therefore, i think it will be very easy to get f-3dB=750MHz in 0.18um or 0.06um. Why gm-C is better than Active-RC? Because you can use much less complex structure for gm cell than Op-amps which means that the frequency capabilty of the gm-cell is higher.

I don't contradict. But the discussion hasn't been about chip design, seems like you didn't read my initial post exactly.

I said, that I only see a chance to implement the filter on a chip, without considering particular topologies. But there's no indidcation, that the original poster has been talking about chip design. Because a discrete active implementation won't work, he should better use a passive LC filter.