need help with an 1G GBW op design!

[airboss]

Hi,

I'm designing an op amp. Here's the spec.

GBW = 1G
power: <20mw, vin_cm =0.9v, input swing: 0.5v p-p, output swing: 1v p-p

I used folded cascode as my 1st stage and CS as the 2nd.

************
Overall AC response without loading cap
DC gain: 48.7dB
GBW: 5.53G
PM: 0

Overall AC response with 1pF loading cap and 10pF compensation cap
DC gain: 48.7
GBW: 726.2 M
PM: 6.2
(Even if I place a 20uF compensation cap, the PM is only 16)

Individual performance before combination.
folded cascode: DC gain=27.95dB, GBW=2.92G, PM=71.4
CS : DC gain=26.78dB, GBW=12.27G, PM=85.1


Please give me some suggestion. Thanks in advance.

 

[shaq]

Re: need help with an op design!

Hi,

I'm designing an op amp. Here's the spec.

GBW = 1G
power: <20mw, vin_cm =0.9v, input swing: 0.5v p-p, output swing: 1v p-p

I used folded cascode as my 1st stage and CS as the 2nd. The individual performance
is (before I combine them):

folded cascode: DC gain=27.95dB, GBW=2.92G, PM=71.4
CS : DC gain=26.78dB, GBW=12.27G, PM=85.1

However, the overall performance is poor. DC bias condition is correct. I got DC gain=48dB, GBW<=800MHz. Even if I use a large compensation CAP, the PM is still less than 15. Please give me some suggestion. Thanks in advance.

Can you tell me what process do you use?

TSMC 0.18um? TSMC 0.35um?

 

[airboss]

Re: need help with an op design!

Hi,

I'm using tsmc018 technology.

 

[DenisMark]

need help with an op design!

I think ur cascode correction is wrong. U connect miller capacitors only to pull-down circuit, and miss miller capacitors to pull-up part of circuit (to net0222 nad net0218).
I see Rout isn't a goal. Try to realise ota only on the basis of folder-cascode stage.
CMFB is bad, where u get resistance in 1T? Try another technique.
As rule of thumb in analog design take L>=3Lmin. And readability of circuit us poor i think.

 

[sankudey]

Have u gone thru this topic......may be help full....



sankudey

 

[airboss]

I think ur cascode correction is wrong. U connect miller capacitors only to pull-down circuit, and miss miller capacitors to pull-up part of circuit (to net0222 nad net0218).
I see Rout isn't a goal. Try to realise ota only on the basis of folder-cascode stage.
CMFB is bad, where u get resistance in 1T? Try another technique.
As rule of thumb in analog design take L>=3Lmin. And readability of circuit us poor i think.

Hi,

Question:
What do you mean, L>=3Lmin.

For the readability issue, do you have any suggestion? Are there any rules of thumb how we can connect the elements? I meant to fit the entire schematic in a window view so I sort of "squeeze" my schematic....

Added after 8 minutes:

Have u gone thru this topic......may be help full....

**broken link removed**

sankudey

So I guess the information you posted is about gain-boosted cascode.
Actually, I posted my spec a while ago. A few guys on this forum suggested me to use gain-boosted cascode. But my, "supervisor" wanted me to use folded-cascode + CS. So....

 

[DenisMark]

For ur process Lmin=0.18um. When u use minimum L for given process u have very small ro. It's known that gain for cascode stage is gm*gm*ro*ro. The open loop gain about 60 dB easy achive with only one cascode stage.

 

[zeeshanzia84]

Well, i didn't have time to go thru all of the replies. But this is my OTA which operates with 1.44 GHz GBW, 59 dB DC gain, 28 deg phase margin, and pretty less power requirement (designed for .13u MOSIS process)

GBW ≈ gm1 / (2 * pi * (Cc + Cgs11))

Cgs for hand calculation maybe assumed to be Cgs = 2 W fF

You can derive the gain eqn yourself and then work backwards....

and yes... the fnd = freq. of non-dominant pole = gm11 / (2 * pi * Cload * (1+Cgs11/Cc)) = 3 * GBW

keeping fnd = 3 * GBW ensures about 70 deg phase margin.

 

[pfd001]

why not use a single cascode stage? u have two high resistance node, it is not easy to compensate it

 

[airboss]

why not use a single cascode stage? u have two high resistance node, it is not easy to compensate it

ummm,

you mean, i can obtain a stable op with 1GBz GBW using single folded cascode?? can you give me some clues?....

 

[pfd001]

yes. the load capacitance of the single folded casscode is also the compensation cap. That will be easy for you to make your amplifier stable