Designer’s Guide Community

raja.cedt

Senior Fellow

Offline

Posts: 1516
Germany

Bandgap Frequency responce
Nov 05^th, 2009, 8:15am

hi,
i am new to bandgap design, i had couple of questions on that

1.Why none of the book is discussing frequency response, normally how much phase margin people target?

2.I feel op amp output is the best place to compensate, because by putting cap between vdd and opamp output any noise in the vdd won't enter in the circuit, but in one paper they claimed that this will cap will increase noise, can any one please explain this.

3. How to decide error amp bandwidth?
Thanks,
Rajasekhar.

Back to top

IP Logged

Mayank

Community Fellow

Offline

Posts: 334

Re: Bandgap Frequency responce
Reply #1 - Nov 5^th, 2009, 11:19am

hi raja,

Quote:

1.Why none of the book is discussing frequency response, normally how much phase margin people target?

Maybe, noone talks about bandgap frequency response because bandgap is a very slow circuit...

Quote:

2.I feel op amp output is the best place to compensate, because by putting cap between vdd and opamp output any noise in the vdd won't enter in the circuit, but in one paper they claimed that this will cap will increase noise, can any one please explain this.

No, i agree with the paper...Placing a cap b/w opamp o/p node and VDD will directly couple any high frequency Noise into the ckt...thereby degrading PSRR at higher Frequencies...

Quote:

3. How to decide error amp bandwidth?

I am myself not much sure about this....But i guess, you choose it depending on the noise profile of your supply, and the BW of the bandgap ckt.Maybe some senior can help...

--mayank

Back to top

IP Logged

Vladislav D

Community Member

Offline

Posts: 76
The Netherlands

Re: Bandgap Frequency responce
Reply #2 - Nov 5^th, 2009, 12:51pm

raja.cedt wrote on Nov 5^th, 2009, 8:15am:

hi,
� i am new to bandgap design, i had couple of questions on that

1.Why none of the book is discussing frequency response, normally how much phase margin people target?

Commonly, phase margin is a useless staff during design of a bandgap because the one is a more than two-pole system with two or more local feedbacks, positive as well as negative. It does not matter how circuit is slow. Any feedback loop should be stable. If your circuit is slow it only means you will see the lo-frequency oscillation. Only step response gives correct answer

Quote:

2.I feel op amp output is the best place to compensate, because by � putting cap between vdd and opamp output any noise in the vdd won't enter in the circuit, but in one paper they claimed that this will cap will increase noise, can any one please explain this.

Bandgap generate DC voltage not at the output of that OTA! So, ideally the reference voltage must be with zero bandwidth in order to exclude any output noise. Large cap - low BW - low noise, pure and simple.

Back to top

IP Logged

Mayank

Community Fellow

Offline

Posts: 334

Re: Bandgap Frequency responce
Reply #3 - Nov 6^th, 2009, 2:23am

Raja,
Did you refer to the book on Voltage References by Rincon Moora...A prof. for GaTech....The book's quite comprehensive....That could help you in BandGap design....

@ Vladislav : Suppose, you place a cap b/w two nodes....Impedance offered in 1/jwC ...So as the cap value increases, impedance b/w two nodes keep on decreasing....So, lower frequency signals will pass more easily through the cap....Isn't this correct ?

As for placing a cap as per raja's wish, The cap value required to produce some substantial coupling for even some noise of 10 MHz frequency would require very high cap value > some hundreds of nF .....So as long as your cap value is on pF, PSRR degradation will happen in some GHz range...If you are ok with that, then i think you can go for this scheme....

--Mayank.

Back to top

IP Logged

Vladislav D

Community Member

Offline

Posts: 76
The Netherlands

Re: Bandgap Frequency responce
Reply #4 - Nov 6^th, 2009, 3:06am

Mayank wrote on Nov 6^th, 2009, 2:23am:

Raja,
@ Vladislav : Suppose, you place a cap b/w two nodes....Impedance offered in 1/jwC ...So as the cap value increases, impedance b/w two nodes keep on decreasing....So, lower frequency signals will pass more easily through the cap....Isn't this correct ?
--Mayank.

true, but this is not a noise. There are many ways to do gorund and vdd �clean� but you can nothing to do with 4kTR, 4kTgm*gammar but decrease the BW. So, I think this is the trade-off

Back to top

IP Logged

raja.cedt

Senior Fellow

Offline

Posts: 1516
Germany

Re: Bandgap Frequency responce
Reply #5 - Nov 6^th, 2009, 6:15am

hi,
thank you very much for your replies...but you people are claming that band gap is slow, but if any supply transients then it has to settle in the specified time..so BW of error amp is matter i feel..is it correct?

manyak do you have that voltage reference book..if you have any e book please send it me

Thanks,
Rajasekhar

Back to top

IP Logged

HdrChopper

Community Fellow

Offline

Posts: 493

Re: Bandgap Frequency responce
Reply #6 - Nov 6^th, 2009, 6:45am

raja.cedt wrote on Nov 6^th, 2009, 6:15am:

hi,
� �thank you very much for your replies...but you people are claming that band gap is slow, but if any supply transients then it has to settle in the specified time..so BW of error amp is matter i feel..is it correct?

I actually do not agree the bandgap loop is slow...in a simple version (without an op-amp) the loop has a very wideband response....
the reason why regular stability parameters are not usually considered is that the loop is inherently positive and always with a gain less than 1 (otherwise it would oscillate)
More complicated circuits to make the bandgap reference more accurate might eventually slow down the loop, but still I think is not that slow.
I do agree we only care about very low frequency content of the bandgap signal (Actually its DC value only) though...

Best
Tosei

Back to top

Keep it simple

IP Logged

raja.cedt Senior Fellow Offline Posts: 1516 Germany	Re: Bandgap Frequency responce Reply #7 - Nov 6^th, 2009, 6:57am hi chopper, i agree with you but its not a +ve feedback loop,because there is high gain _ve FB loop also there, but finally we care about dc but ac oscillation may give problems in some cases.... so i just want to know is there any reference which is discussing noise performance and error amp design? Thanks, Rajasekhar.
Back to top	IP Logged

Mayank

Community Fellow

Offline

Posts: 334

Re: Bandgap Frequency responce
Reply #8 - Nov 6^th, 2009, 10:08am

Actually,
The error amplifier tries to maintain same voltage in the two branches on these resp nodes ---> one above the resistance, another directly above the BJT in a standard BG architecture....
There are two loops that are formed, one is a +ve FB loop, and another a -ve FB loop....As far as i know, you always keep -ve FB loop stronger than the +ve FB loop(usually by a factor of around 2x)....So that your overall BG ckt is stable.....

@ chopper : The BG startup is usually a very slow process....I meant it that way....

--Mayank.

Back to top

IP Logged

HdrChopper

Community Fellow

Offline

Posts: 493

Re: Bandgap Frequency responce
Reply #9 - Nov 6^th, 2009, 6:43pm

Mayank wrote on Nov 6^th, 2009, 10:08am:

Actually,
� The error amplifier tries to maintain same voltage in the two branches on these resp nodes ---> one above the resistance, another directly above the BJT in a standard BG architecture....
�There are two loops that are formed, one is a +ve FB loop, and another a -ve FB loop....As far as i know, you always keep -ve FB loop stronger than the +ve FB loop(usually by a factor of around 2x)....So that your overall BG ckt is stable.....

�@ chopper : The BG startup is usually a very slow process....I meant it that way....

--Mayank.

Hi Mayank,

I was actually referring to a simpler bandpag topology, like the one shown here:
http://www.designers-guide.org/Forum/YaBB.pl?num=1247576404/9#9

In this one the loop gain is positive and two local negative feedback loops: the beta helper and the translinear loop.

Nevertheless, when an error amplifier is used, its negative feedback loop gain is way higher larger than the positive one from the standard bandgap. In this case the stability is purely set by the amplifier's loop and therefore the bandgap has almost nothing to do with the loop stability.

Best
Tosei

Back to top

Keep it simple

IP Logged

Mayank

Community Fellow

Offline

Posts: 334

Re: Bandgap Frequency responce
Reply #10 - Nov 13^th, 2009, 6:32am

Hello everyone,
I have a question on Error Amplifier used in standard PTAT architectures, BG ckts or active current mirrors for vds matching.....

Here, I am talking wrt to Active Current Mirrors where unlike BG, the current through the FET connected to opamp o/p has to be altered considerably...

Should the o/p swing of the opamp be very high ?

-Mayank

Back to top

IP Logged

raja.cedt Senior Fellow Offline Posts: 1516 Germany	[]Y Reply #11 - Nov 13^th, 2009, 8:50am hi mayank, i didn't get your question, could you please explain. Are you talking about load current drawn from BG? Thanks, Rajasekhar.
Back to top	IP Logged

Mayank

Community Fellow

Offline

Posts: 334

Re: Bandgap Frequency responce
Reply #12 - Nov 15^th, 2009, 2:42am

Actually,
This isnt related to BG in any way..I was talking about active cascode config of current mirrors....Error amplifier was being discussed here, that's why i asked the question in this thread...I should be asking it in a separate thread....I would try to attach a snapshot of schematic, although i have always failed to upload any pic on this forum..

regards,
Mayank

Back to top

IP Logged