Please help me out for Gm-C(OTA-C) filter

Hi all,
Recently,I am designing a 7th-order elliptic OTA-C filter,I got the prototype from the software of filtersolution , then replaced the passive components (resistor and inductor)with active components(negative feedback Gm cell for resistor and gyrator for inductor).
The spec of the passive filter is passband=4MHz, stopband =5MHz(Transition-band =1MHz),attenuation=40dB@5MHz.
But,due to the simulation with spectre,the final active filter only has 3.2MHz passband, the attenuation at 5MHz is less than 30dB．
　ｓｏ，ａｎｙｂｏｄｙ　ｃａｎ　ｔｅｌｌ　ｍｅ　ｔｈｅ　ｒｅａｓｏｎ　ｏｆ　ｔｈｅ　ｒｅｄｕｃｉｎｇ　ｏｆ　ｔｈｅ　ｐａｓｓｂａｎｄ　ａｎｄ　ｔｈｅ　ａｔｔｅｎｕａｔｉｏｎ．
　　Ｉ　ｈａｖｅ　ｔｒｉｅｄ　ｆｏｒ　ｍａｎｙ　ｄａｙｓ，ｂｕｔ　ｆａｉｌｅｄ，ｔｈｅｒｅ　ｗｉｌｌ　ｂｅ　ｇｒｅａｔ　appreciation　ｆｏｒ　ｙｏｕｒ　ｈｅｌｐ．
　　Ｔｈｅ　ｐｉｃｔｕｒｅｓ　ｂｅｌｏｗ　ａｒｅ　ｔｈｅ　ｃｉｒｃｕｉｔ　ｏｆ　ｔｈｅ　ｆｉｌｔｅｒ　，ｔｈｅ　ｇｍ　ｃｅｌｌ，ｔｈｅ　ａｃ　ｒｅｓｐｏｎｓｅ　ｏｆ　ａｎｄ　ｔｈｅ　ｒｅｓｐｏｎｓｅ　ｆｒｏｍ　ｔｈｅ　ｆｉｌｔｅｒｓｏｌｕｔｉｏｎ　ｒｅｓｐｅｃｔｉｖｅｌｙ．

　　

Did you try out the filter by an ideal GM block? Make a macro single pole GM block using a VCCS and resistor capacitor at the output to model the output pole and see what is the bandwidth and gain required for making the system work. Then compare it with you Gm cell and see what is lacking.

Did you consider that the classical (simple) gyrator approach to replace inductors must be applied for grounded inductors only?

hi LvW, there are three active floating inductors in the circuit,four gm cells and on capacitor create a floating inductor.

OK, I just wanted to pick-up this point.
By the way, why didn't you use the Bruton transformation with the aim to get a topology based on grounded gyrator elements (FDNR approach) ?

hi aryajur, I replace the GM cell with ieal vccs, but the result become worse , I will upload the simulation later .

The gyrator adds a series resistance to the inductor is that true for your floating inductor gyrator? Maybe you could try simulating an ideal filter circuit with those series resistances added and compare both simulations.

I just replaced the Gm cell with the ideal VCCS,the response is as below,the 40dB attenuation frequency is 9MHz.

May be the dc gain is not sufficient,I will redesign the OTA