The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Mixed-Signal Design >> Fully Differential High Pass Filter
https://designers-guide.org/forum/YaBB.pl?num=1298365894

Message started by Rahul_K on Feb 22^nd, 2011, 1:11am

Title: Fully Differential High Pass Filter
Post by Rahul_K on Feb 22^nd, 2011, 1:11am

Hello Everyone,
I want to design an Fully Differential High Pass Filter which can give pass-band gain of 10dB at 5GHz, for some high speed circuit.
I have a Fully Differential Op-Amp with proper UGB, but the problem is the capacitor used for HPF, which comes at the input of an OPAmp is blocking the DC common mode voltage which needs to be maintained at the inputs. Can you suggest an HPF architecture which will remove capacitor from signal path. There are architectures like, Servo Feedback, which removes Capacitor frm signal path but they are more suitable for single ended architecture and fails for Differential i/p-o/p. I don't want to use Inductor as it makes circuit bulky.

Title: Re: Fully Differential High Pass Filter
Post by rfidea on Feb 22^nd, 2011, 4:10am

Hi Rahul!

You are saying that a capacitor is blocking the input of the opamp. Which topology are you using, Sallen Key? MFB? Or some other? Can you provide a schematic or a link to a picture of the topology.

Title: Re: Fully Differential High Pass Filter
Post by Rahul_K on Feb 22^nd, 2011, 4:36am

Hi,
refer fig.4 on page 4..Single pole HPF..
I am not able to provide the required DC voltage (Common Mode) at the inputs of OpAmp coz it gets blocked by Capacitor..So transistors in the OpAmp input stage are not geting proper bias...

Title: Re: Fully Differential High Pass Filter
Post by rfidea on Feb 22^nd, 2011, 9:25am

If the cm level at the inputs can be the same as the cm level at the output this is no problem. The opamp needs a cmfb loop at the output to set the output cm level. Then the inputs will be biased thru the feedback resistors R2a and R2b.

Title: Re: Fully Differential High Pass Filter
Post by Rahul_K on Feb 22^nd, 2011, 9:33am

When Opamp is tested without the filter, it works fine and CMFB loop works fine...But when filter is implemented as shown previously, because of improper input bias voltage, the input transistors have VGS less than Vth and hence remain in subthreshold and hence the output itself is not able to achieve the required cm level, so its not fed back thru CMFB.

Title: Re: Fully Differential High Pass Filter
Post by rfidea on Feb 22^nd, 2011, 11:48am

I do not think there is a way to get rid of the capacitor and getting the bias from preceeding stage. It is a high-pass filter and therefore the DC should be blocked.

If I understand it correctly the input is going low. This loads the cmfb internally so it can not lift the output high enough? It is hard to see what is causing this without looking at the circuit diagram. Maybe you can solve the problem by using dubble bias lines to secure the function of the cmfb even if the input stage is saturated. (I have done that once.) Is it a one-stage opamp? Maybe you need to use a two-stage to get better isolation between the dc levels at the input and output

Title: Re: Fully Differential High Pass Filter
Post by Rahul_K on Feb 22^nd, 2011, 9:45pm

Its a two stage Folded Cascode OPAmp with Lead-Lag Compensation and double bias..Still, I am not able to achieve proper HPF..

About OpAmp I can say:
It has Gain of 93dB
UGB: 400MHz.
PM:80deg

Don't you think its more like an ideal?

Title: Re: Fully Differential High Pass Filter
Post by rfidea on Feb 23^rd, 2011, 1:36pm

I was drawing your topology on a paper. Could the problem be that when the input is going low the current in the input diff pair is zero. Then the current sources on the drain side is sourcing all their current into the cascodes. It the cmfb can not handle all that current the output common source stage is saturated. and pulling the output low. Is this what happens?

One possible solution is increasing the maximum current from the cmfb.

Title: Re: Fully Differential High Pass Filter
Post by love_analog on Mar 3^rd, 2011, 11:46am

Rahul
The specs you have for the opamp are good. However, 2 things
a) The dc biasing you have when you test the opamp seperately is it the same as dc biasing when it is in HPF mode. Since the caps are open at dc, I would suspect that the output common-mode is the same as input common-mode. Is that what you see in your sims?
b) Is the specs tested with the same loading as the loading when you put it in HPF mode.

I would suggest you make sure all your dc biasing is correct.

Like others suggested you can't get rid of the opamp, however you can put a buffer between the opamp and cap and set the common-mode of the input differently than the output.

Title: Re: Fully Differential High Pass Filter
Post by Rahul_K on Mar 3^rd, 2011, 8:22pm

Hi,
I have given the same signal with same DC levels to both the tests.
The problem with HPF is its not allowing DC bias to reach at the OpAmp inputs its blocking the DC.
I don't understand what shall I do in order to get the proper DC biasing at the inputs.
The Caps and resistors used as feedback in HPF topology are creating problems for me, I guess. Coz without them the CMFB loop is able to maintain the required levels at the input as well as output.