Low rout on folded cascode

I am a 3rd year UG student and as part of my VLSI design course I have to make a folded cascode opamp. I am very new in this field and had a couple of questions.

1) I have biased my OTA but the gain is insanely low (6 Db). The reason is that the rout is very low. For example, theoretically in one NMOS, Id=25uA so ro should be 200k, it is coming to be 50k. All the MOSes are in saturation.

2)Will it be easier to make a single-ended diffamp first, then convert it to fully differential? Or does it not work that way?

1) You already mentioned that the gain is insanely low. You should solve that first.
And if it's really due to low rout, but even if rout is increased by 10 times, that would only translate to a 20dB increase in gain.
Still "insanely low". So chances are low rout is the problem. You probably have MOS not in saturation.

2) It is not easier, but you are new! Understanding one more circuit always helps!

Hello getusoon,

If you don't mind what is your power supply voltage (Vdd) and which design kit do you use to implement your amplifier?

For the first question,as far as the gain of the FC is concerned,i would suggest that you study well the equation that gives the low frequency gain of the amplifier as a function of the DC calculated parameters (transistors' gm,gds) and try to size the active devices in such way that will give you the desired gain.

Another approach is that you can break the total output resistance in a combination of two parallel output resistances (one corresponds to the Rout due to cascode current mirror for a single ended implementation and the other Rout corresponds to the cascode devices,the current sources underneath and the diff input pair) and try to get the desired gain from the general equation Av=-gm*Ro=-gm*(Rout1//Rout2).I think it would be more tough than the first approach that i suggested you.

If you still cannot get enough gain due to low power supply voltage-so you can't bias all the transistor to desired point-you can apply gain boosting technique.

For the second question,if you are not much experienced,it is a good idea to start with the single ended version and then it is easy to convert to fully differential.For the fully diff OpAmp do not forget to design the respective CMFB circuitry (initially you can use an ideal one).

Hope the above answered your questions.

Regards,
Jimito13

@checkmate: Haha. I will try it out someday. I have till the end of this month to hand my assignment in, however, so no time right now.

@Jimito13:Vdd=3.3V, Vss=0V. Thanks a lot! I tried your approach but the problem was a very simple, stupid one.The Vin was biased out of ICMR. Am getting a 45 db gain now with decent UGB

However, the specs given to me are insane. 120 db DC gain with 180 db CMRR and only 0.5mA total current budget! For the present I am trying to use gain-boosting. So, yeah, again a couple of questions

1) I saw lots of different looking gain boosting circuits online. Any idea which gain-boosting circuit could be best for fully differential folded cascode?? A simple diagram would be more than enough.

2)How high can gain-boosting boost my circuit actually?

3)To get my insanely high CMRR, increasing the Rss is an option, using cascaded MOSes. But this will eat into my voltage headroom and kill my voltage swing. Also, my MOSes are all very precariously biased. Any other way except the "Rss way" to increase CMRR?

Hello getusoon,

Ok,your power supply is quite large,so there will be no problem to bias the folded cascode and reach the specs.

I will answer your questions underneath :

1.My suggestion is to use the same circuit as the main amplifier for the gain boosters.

2.You will choose the gain for the boosting amplifiers.Let's say that you seek a total gain of 120dB as you say and you manage to achieve 50dB from the main folded cascode,then you are going for 70dB for the boosters.

3.Take a look at Kayal Stefanovic's book "Structured Analog CMOS Design",if i remember well there are suggestions in there for your problem about CMRR.

Regards,
Jimito13

Now I've almost got most of the specs (110 db gain coming, CMRR is very high-some 220dB :O ) but now I have to attach a CMFB circuit to the output to keep the o/p at 1.65 V (currently it is at 1.89 V with 3.3 V supply). But no CMFB I use (I found a couple of circuits on the net) is working!!! I cant use two Resistors as my o/p resistance is quite high now and that would kill my gain. SO ONCE AGAIN, PLEASE HELP by providing a good CMFB circuit that I can use. Mine is an NMOS-input Folded Cascode OTA.

Hi getusoon,

What did you do to fix the CMRR of the folded cascode topology if you don't mind?

You can use the two resistors to sense the voltage of the outputs of the fully differential OTA BUT you must inject a buffer between the OTA and the CMFB amplifier in order not to kill the gain ;-)

The choise of the CMFB circuit depends on the gain and the speed of the CM amplifier so that the full amplifier (OTA+CMFB) will be robust over all corner situations.An initial idea is to use an ideal CMFB and make some simulations to determine the needed gain and speed and after specifying that info you will make a choise among various existing topologies.

Regards,
Jimito13