Designer’s Guide Community

rajesh14 New Member Offline Posts: 6 tuni, andhra pradesh	why not ft as bandwidth Oct 23^rd, 2009, 10:35am how Ft (unity gain frequency) is different from Fh(3db frequency)? why we did not considered ft in frequency reponse analysis ? thank you all....:-)
Back to top	IP Logged

Mayank

Community Fellow

Offline

Posts: 334

Re: why not ft as bandwidth
Reply #1 - Oct 23^rd, 2009, 11:17am

hmm,
A pretty Newbie question ;) ...But since it's basic...it's a good one..I like basic questions :)

Ft is generally referred to the transit frequency of a MOS Transistor...(Hope you are not confusing it with UGB of a ckt)...Its the frequency where short circuit, common-source current gain falls to unity...

Generally speaking, UGB is the frequency where gain falls to unity.
But Most of the times, we are concerned with the frequency till which we can get ideal behaviour from an opamp....That's where -3dB BW comes into Picture....We assume that till -3dB point, Opamp behaves like an ideal opamp with infinite(or very high) gain. The choice of -3dB point is upto you. In cases, where you require more accuracy, we consider -1dB Compression Point rather that -3dB.

But in the Frequency Response, UGB is equally important as F_-3dB because UGB sets your Gain and Phase Margins which are generally the most important criteria to define the stability of your ckt.

So, we do consider fT in the Freq. resp. Further, fT for a given circuit topology is constant for a give process, while F_-3dB will vary according to the Low Freq. gain of the circuit.

please further this if you have any doubts....it's good to revise basics.

regards,
Mayank.

Back to top

IP Logged

raja.cedt

Senior Fellow

Offline

Posts: 1516
Germany

Re: why not ft as bandwidth
Reply #2 - Oct 23^rd, 2009, 6:48pm

hi

manyak: i want to add one more point to your ft definition..its not only current gain but also voltage gain.

rajesh: ft related to process where as 3db related to circuit. So in a process you can create many circuits with different bandwidthd ,otherwise you can think like you can't design circuits faster than ft.

Thanks,
Rajasekhar.

Back to top

IP Logged

buddypoor

Community Fellow

Offline

Posts: 529
Bremen, Germany

Re: why not ft as bandwidth
Reply #3 - Oct 24^th, 2009, 1:37am

May I add some remarks/corrections to MAYANK�s contribution?

Quote:
Ft is generally referred to the transit frequency of a MOS Transistor..
I know about several opamp data sheets in which the UGB is called transit frequency (open loop gain=1) ; this holds only for universal - compensated opamps.

Quote:
But Most of the times, we are concerned with the frequency till which we can get ideal behaviour from an opamp....That's where -3dB BW comes into Picture....We assume that till -3dB point, Opamp behaves like an ideal opamp with infinite(or very high) gain.

I don�t agree with this. The -3dB point has no practcal meaning - as far as the open loop opamp gain is concerned. Nobody really assumes the opamp as "ideal" only in the range up to this frequency.
(An opamp with 120 dB max. is ideal until 117 dB ; and another one with 80 dB max. is ideal until 77 dB ?? Sounds not logical, does it? ).
The -3dB oint comes into the play only if you speak about the gain of an amplifier with feedback, because we consider this range as the "pass region" (especially for filters).

Quote:
But in the Frequency Response, UGB is equally important as F-3dB because UGB sets your Gain and Phase Margins which are generally the most important criteria to define the stability of your ckt.

This sounds a bit misunderstanding. The UGB does not "set" the margins. Instead, the PHASE margin is measured at the UGB frequency. BUT: This applies only for the UGB of the LOOP GAIN and not for the UGB of the opamp alone!

Regards

Back to top

LvW (buddypoor: In memory of the great late Buddy Rich)

IP Logged

HdrChopper

Community Fellow

Offline

Posts: 493

Re: why not ft as bandwidth
Reply #4 - Oct 28^th, 2009, 8:07pm

I agree completely with buddypoor. Concerning the PM and GM may be Mayank described the scenario keeping in mind that there is a way to express the PM as a function of the GBW but not of the UGB .
In any case, PM and GM are measured at UGB as buddypoor stated, and are set (mainly) by the distance between the first secondary pole and the GBW.

Best
Tosei

Back to top

Keep it simple

IP Logged

RFICDUDE

Community Fellow

Offline

Posts: 323

Re: why not ft as bandwidth
Reply #5 - Nov 1^st, 2009, 4:49pm

The difference or relation between ft and the 3dB bandwidth has to do with the definition of both.

Ft is the unity gain frequency for current input versus short circuit output current in contrast to the 3dB voltage gain which is, most likely, not measured using a short circuit load.

Transistors inherently control an output current in respect to either a input voltage or current (depending on how you want to look at the problem). Ft is an appropriate measure of "potential" high frequency performance since at high frequencies the input impedance is low due to the capacitance. So, ft is a measure of how well you can generate an output current when the input is low impedance due to the input capacitance.

Enough about ft, onto the 3dB response. The frequency response is not limited by the input capacitance (like ft is); rather, it is limited by the output resistance and capacitance or perhaps the input may become dominant due to the Miller effect if both the input and output impedance is high. It all depends on the resistance and capacitance where as ft depends on device capacitance and transconductance. As soon as you add resistance for a load the bandwidth decreases do to the capactance of one device and the output resistance of another. Input and output poles are related, but may be separated by many octaves or decades depending on the resistances at the input and/or output. To make matters worse, there may be many poles which impact the overall frequency response making it impossible to define a meaningful 3dB frequency response.

In summary, ft tells you how fast the devices are in the circuit and this tells you something about the maximum possible frequency of operation. On the other hand, the frequency response has more to do with the gain and capacitance that determine the dominant pole response (longest time constant) usually dependent on the output impedance of one device (or circuit) and the capacitance of another; however the frequency response may also be influenced by the feedback from the output to input as in the case of Miller multiplication.

Now I'm not so sure I answered the question ...

Back to top

IP Logged

Mayank

Community Fellow

Offline

Posts: 334

Re: why not ft as bandwidth
Reply #6 - Nov 1^st, 2009, 9:08pm

Hi all,
Seems there were a lot of misunderstandings from my post :(

1st of all : Tosei, Can you please explain what you meant by Quote:

there is a way to express the PM as a function of the GBW but not of the UGB

I beleive GBW is equal to UGB of any ckt...I refer GBW commonly as UGB of the ckt...UGB == BW at which Gain is equal to1, ie, GBW product at Gain=1....But i beleive GBW product remains same for a given topology & process....It's just the gain and -3dB freq that you can alter once you fix your Topology & Process. Either my beliefs are highly mistaken, or i understood your post wrongly... Pls explain....

BuddyPoor, :
POINT 1 : I said Ft is generally referrered to the Transit Frequency of a MOS....UGB on the other hand, is generally referred to the Transit Frequency of a Ckt, & both of them are usually not the same...

POINT 2 : By the word Ideal Behaviour, i meant a frequency till which you can consider opamp to have high enough gain required for your application....Don't take it quantitatively....You set the DC Gain of the opamp to be high enough to satisfy your Gain Error specs...F-3dB is a point till which Gain falls to Adc/sqrt(2) As for Quote:

The -3dB oint comes into the play only if you speak about the gain of an amplifier with feedback, because we consider this range as the "pass region" (especially for filters).

, F-3dB doesnt define the pass region for all Filters...It depends on the Filter Approx.,(like for chebyshev, we use -1dB point)...Filter cut-off is generally equal to PB ripple size..

POINT 3 : Here, of course, i didnt mean UGB decides the magnitude of PM/GM....I wanted to profess that PM is measured at UGB of the LG response...Of Course, it's decided by the Poles & Zeros of the Response Function....I am sorry here if i conveyed the wrong meaning...

thank you all,
Mayank.

Back to top

IP Logged

raja.cedt Senior Fellow Offline Posts: 1516 Germany	Re: why not ft as bandwidth Reply #7 - Nov 1^st, 2009, 9:13pm hi maynak, one point i would like to add to your post is ''you called gain bandwidth is ugb and it is constant for the circuit provided your circuit well compensated'' Thanks, Rajasekhar.
Back to top	IP Logged

Mayank Community Fellow Offline Posts: 334	Re: why not ft as bandwidth Reply #8 - Nov 1^st, 2009, 9:43pm Yes raja, I called GBW product of a ckt as UGB of that ckt....and UGB for a given process & ckt topology is a constant....Compensation is a passive(extra) addition/shifting of poles/zeros, which can modify the UGB of a ckt....Is that what you wanted to assert ? --mayank.
Back to top	IP Logged

raja.cedt Senior Fellow Offline Posts: 1516 Germany	Re: why not ft as bandwidth Reply #9 - Nov 1^st, 2009, 9:55pm yes
Back to top	IP Logged