Designer’s Guide Community

AMSA

Community Member

Offline

Posts: 46

Differences between those two types of OTA's
Sep 11^th, 2014, 7:58am

Hi guys,

I want to design an OPAMP to drive a fast comparator. Actually the OPAMP that I want to design is to do a error amplifier.

I was searching for some topologies inside the OTA's topologies and I though that a good choice would be the two-stage miller OTA or the Symmetrical OTA (load compensated ota).

I don't have any specs restriction but I think that the most important parameters for the error amplifier would be high DC gain, nice PSRR, good slew rate and big bandwidth.

From what I read I think that the Symmetrical OTA has less gain when compared to the two stage miller OTA. However, the Symmetrical can be folded cascode, but I don't know what implication it will have.

Can someone help me here with the pros and cons? Which one would be best suitable?

Kind regards.

Back to top

IP Logged

raja.cedt

Senior Fellow

Offline

Posts: 1516
Germany

Re: Differences between those two types of OTA's
Reply #1 - Sep 11^th, 2014, 10:37am

Two stage op-amp: Potentially higher gain at the cost of compensation and rail to rail o/p capability.
Symmetric OTA: Less gain and single pole system and rail to rail capability.

Basically you will have to fix your gain and application. In 2 stage opamp also you can use folded cascade. Implication are function of many variables. We can't explain or guess until you fix an application or final circuit.

Back to top

IP Logged

AMSA

Community Member

Offline

Posts: 46

Re: Differences between those two types of OTA's
Reply #2 - Sep 11^th, 2014, 11:02am

Hi raja, thank you so much for your reply.

The application of this OP AMP will be to operate on a feedback system on a dc-dc converter. It will have to amplify the error difference at the input (there is a reference and an input signal that comes from the converter output). At same time it will have a feedback to compensate for the phase and gain margin.

I managed to design a symmetric ota where the gain I obtained was around 40dB with a current consumption around 680uA. I don't know if this consumption vs gain is a good indicator, that is, for that gain I have a good current consumption.

I don't know if I can get more gain from this amplifier, maybe at the cost of more current consumption. Another possibility is using cascode at the output. I think it would increase significantly the gain but I don't know what I would lose, maybe less voltage swing at the output - but I don't know if the voltage swing at the output is very important for this application.

Kind regards.

Back to top

IP Logged

SNIKE

Junior Member

Offline

Posts: 27

Re: Differences between those two types of OTA's
Reply #3 - Sep 11^th, 2014, 2:31pm

Gain doesn't depend on power consumption. [grain of salt notice].
Your unity gain bandwidth depends on power consumption.

How much gain do you require? What speed do you want this amplifier to operate? What is the load this amplifier is driving? Is it resistive load, capacitive load or active load? You have to figure out if your application requires large output swing.

"I want to design an OPAMP to drive a fast comparator. "
If that's the case check the Input common mode range and gain of comparator. This will answer the question if you really require large output swing.

Back to top

IP Logged

loose-electron Senior Fellow Offline Best Design Tool = Capable Designers Posts: 1638 San Diego California	Re: Differences between those two types of OTA's Reply #4 - Sep 11^th, 2014, 6:55pm everybody seems to be forgetting the important one: Offset Need I say more? A high gain amplifier into a comparator is useless if the offset is huge.
Back to top	Jerry Twomey www.effectiveelectrons.com Read My Electronic Design Column Here Contract IC-PCB-System Design - Analog, Mixed Signal, RF & Medical IP Logged

AMSA

Community Member

Offline

Posts: 46

Re: Differences between those two types of OTA's
Reply #5 - Sep 12^th, 2014, 2:15am

Hi all,

Gain doesn't depend on power consumption? Never heard that.

More current, more gm, more gain.

Regarding your questions: I require something like >60dB of gain, reasonable speed, load will be a fast comparator, which will compare the error that comes out from the error amplifier (the OTA that I am talking about) with a sawtooth wave.

I believe that the load that the error amplifier will be driving doesn't require a largeee output swing.

You can see here the application:

Regarding the offset. Making the L a bit larger than it should can attenuate the offset problem, I think.

Back to top

IP Logged

RobG

Community Fellow

Offline

Posts: 569
Bozeman, MT

Re: Differences between those two types of OTA's
Reply #6 - Sep 12^th, 2014, 9:13am

The main reasons to build a two stage (e.g. Miller) are to drive a load and to get a larger swing. The down side is that the maximum bandwidth is reduced and compensation is more complicated. A folded cascode followed by a second stage with Ahuja compensation is my go-to topology. Sorry I can't find a picture without too much effort but that is what I would start with.

Back to top

IP Logged

AMSA

Community Member

Offline

Posts: 46

Re: Differences between those two types of OTA's
Reply #7 - Sep 12^th, 2014, 12:18pm

Thanks for the suggestion.

I have designed a symmetric ota and I got 50dB of gain with a GBW~200MHz.

I was trying now to simulate the ICMR. I configured the OTA as a unity gain buffer and I have inserted a DC Voltage Source and did a sweep from 0V up to 3.3V.

I read, don't remember where, that we should plot the output voltage and the current on the transistor that bias the differential pair.

I got this:

From the picture we can say that the ICMR is from 2.26V to 2.85V?

Back to top

IP Logged

loose-electron Senior Fellow Offline Best Design Tool = Capable Designers Posts: 1638 San Diego California	Re: Differences between those two types of OTA's Reply #8 - Sep 12^th, 2014, 1:09pm ok so a comparator to a saw tooth wave - sounds like a duty cycle controller for a power converter or something similar. Offset may not be an issue there, because the offset in some situations just shows up as a time shift.
Back to top	Jerry Twomey www.effectiveelectrons.com Read My Electronic Design Column Here Contract IC-PCB-System Design - Analog, Mixed Signal, RF & Medical IP Logged

AMSA Community Member Offline Posts: 46	Re: Differences between those two types of OTA's Reply #9 - Sep 12^th, 2014, 1:14pm Yes, it is to control a duty-cycle.
Back to top	IP Logged

AMSA Community Member Offline Posts: 46	Re: Differences between those two types of OTA's Reply #10 - Sep 12^th, 2014, 2:25pm Anyone for the ICRM?
Back to top	IP Logged

SNIKE

Junior Member

Offline

Posts: 27

Re: Differences between those two types of OTA's
Reply #11 - Sep 12^th, 2014, 2:57pm

HI AMSA,

Gain doesn't depend on power consumption.

Let me give you an example.
If you bias your diff pair with just 1uA but have a cascoded output [like telescopic] you can easily acheive 60dB of gain.

I have many OTA's in silicon having gains > 80dB with less than 10uA bias.

But the bandwidth is going to be terrible. [ Bandwidth is gm/cl].

I think you need to find required speed of operation [ the error you can tolerate because of comparator delay].

Back to top

IP Logged

AMSA

Community Member

Offline

Posts: 46

Re: Differences between those two types of OTA's
Reply #12 - Sep 12^th, 2014, 3:26pm

What you mean by the required speed of operation? Regarding the consumption, it make sense what you said, however for typical opamps applications I think that what I said is valid.

Can you help me out with the ICMR?

Right now my symmetrical ota has 50dB of gain, 600uA of current consumption, GBW around 190MHz and a phase margin of 50� measured with STB in spectre and AC Analysis.

Right now I would like to make some charaterization on him but I don't have much knowledge on the required test benchs.

I have a report from a university where the author has some test benchs for each of the charaterization however I am not sure if I am doing correctly the configurations.

For examplem the case of the ICMR. I tried the OFFSET, connecting both input to ground but I don't know where to read the offset. If it is on the output, it has around 1V, but it seems to me that 1V is the VDS of the nmos transistor connected between the mid point on the output and ground.

Regards

Back to top

IP Logged