High GBW OTA amplifier

[AMSA84]

Hi everyone.

I'd like to know which circuit is more suitable to have a high GBW frequency (> 300MHz) and high gain (> 60dB).

.3-current Mirror Cascode;
.Folded Cascode;
.Other?

Regards.

 

[Dominik Przyborowski]

The state of the art designs with highest FoM are developed from Recycled Folded Cascode amplifier.


Check this papers:
https://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=5226701
https://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=5585038&navigation=1
https://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=6731196&navigation=1
https://link.springer.com/article/10.1007/s10470-011-9762-y
https://link.springer.com/article/10.1007/s10470-012-9843-6
https://iopscience.iop.org/1674-4926/33/2/025002/

 

[nitishn5]

I do not think you would need a Recycling Folded Cascode (RFC) to achieve your specifications unless you are targeting some high FoM or great numbers..!

A simple Folded Cascode (FC) would be able to meet them.
You just have to burn a little more current than the RFC to get the required bandwidth.

 

[tompham]

Hi Amsa84
From my view, if you use vdd = 1.8v or above you can use telescopic opamp. It limits output swing but high gain, high GBW and less power compare to others

 

[AMSA84]

Hi tompham,

Thanks for the input. I thought that too, that's why I have implemented a Folded Cascode. However I think I am facing one problem here.

I think that I have 84mV of offset voltage.

My Folded Cascode has something around 80dB of gain, 85MHz of GBW and 61º PM.

When I ran a simulation to see how is the overshoot and so on, I saw this:

**broken link removed**

If you notice the wave form in the output doesn't start from 0V but from 80mV. I have configured the OTA as a unity gain buffer and inserted a vpulse source to simulate the response of the amplifier.

That 80mV is the offset voltage? How can I solve that?

Can someone post the correct schematic to measure the offset of an OTA amp?

 

[dick_freebird]

That offset is from the fundamental limitation of the output's
self-impedance, against the lower device dropping out of
saturation. gm*Ro is all there is. It's not an offset issue per se
(see none, at the 2V position). More like a gain error because
gain drops like a rock at Vout<VT(N). You might add a follower
and lower impedance shunt, but this kinda defeats the low
power interest. Question is, is 0.0V a valid voltage of interest?
Applications would typically stay away from either rail if the
accuracy is to be preserved.

This is not unique to any style of op amp. No free lunch,
nowhere.

 

[FvM]

As the gain of an OTA is measured in A/V (transconductance), the term GBW doesn't seem to make sense for it at all.

 

[AMSA84]

Hi guys, thank you very much for the replies.

freebird, it means that this is not a problematic issue? It is normal that something like that happens? By chance, I never did this kind of tests when I designed other amplifiers (at university). This is the first time, and because of that I am not familiar with this.

When you say that you don't see "none at 2V position" it's because it seems that the output put voltage stays at the same level as the input? Is that it?

I read some stuffs about how to measure the offset before posting here, but almost everything was supposing that the amplifiers supply voltage was between, -VSS and + VDD and not 0V and +VDD.

How can I test the offset in this case?

FvM, didn't understood your comment.

Regards.

 

[nitishn5]

If your design is for a supply of 0V to +VDD and your design is not a rail to rail amplifier, then you cannot expect a rail to rail response.

When you have -VSS to VDD supplies, 0V input is actually at the center of the region of operation of the amp. And hence you would see minimum offset there.

And even with a rail to rail amplifier, you would only get very close to the rails and not actual get outputs at the rail.

At the 0V input, what you are seeing is not exactly an 'offset' but actually the fact that your amplifier is effectively dead.
To actually measure offset, measure it in the region of operation of your amp.

 

[AMSA84]

Hi nitishn! Thanks for you information. Very helpful.

So one can say that the offset it is not an important thing here. Right?

What about the other characterizations that is done to op amps (OTA)? Are valid? Because of the power supply values.
the ICMR, PSRR, CMRR and the Output Resistance.

 

[nitishn5]

All your characterizations should be done in the region of operation of your amp.

This is where your amp is in the linear region and everything behaves as expected.

Beyond that region something would be non-linear and your performance would suffer.

Anyway, offset should be measured with a DC sweep and not using a Transient step since all settling crap comes into play.

 

[AMSA84]

Okay. It make sense.

I would like to measure the ICMR and Output Swing. For the ICMR, a put the OTA as unity gain (output to negative input terminal) and swept the input DC voltage source and plot the output vs input. I got this:

**broken link removed**

What can I say? The input accepts values from ~500mV up to ~2.6V?

 

[tompham]

Hi AMSA84

If you probe input and output signals same graph , it will tell you clear ICMR without asking anyone

 

[AMSA84]

Thank you very much for all the answers. I concluded that the problem here has to do with the gain-bandwidth product. I have to live with that.

 

[D.A.(Tony)Stewart]

There are some papers on

No-Capacitor Feed-Forward (NCFF) compensation scheme which uses a high-frequency pole-zero
doublet to obtain greater than 90dB DC gain, GBW of 325MHz and better than 70 ◦ phase margin

The NCFF compensation scheme enables both high gain and fast
settling time, resulting in accurate and fast step response.
LHP zeros are used to cancel the phase shift of poles to obtain
a good phase margin