The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> how to bias constant gm rail-to-rail OTA
https://designers-guide.org/forum/YaBB.pl?num=1331722050

Message started by cktdesigner on Mar 14^th, 2012, 3:47am

Title: how to bias constant gm rail-to-rail OTA
Post by cktdesigner on Mar 14^th, 2012, 3:47am

Hi,

Can someone please explain how to choose Vb1 and Vb2 in the constant-gm rail-to-rail OTA in the attached schematic?

Title: Re: how to bias constant gm rail-to-rail OTA
Post by loose-electron on Mar 14^th, 2012, 4:00pm

Interesting architecture - you got a source for that?

Title: Re: how to bias constant gm rail-to-rail OTA
Post by HdrChopper on Mar 15^th, 2012, 10:17am

hi,

Set Vb1 to Vonp and Vb2 to Vonn, where Vonp and Vonn are the input CM levels at which PMOS and NMOS diff pairs turn on respectively.
If you do so then,

for vicm < vonn , IM1,2 = 0. thus IM5 = 2IR2 and IM8=0 (since M3,4 sources turn M8 off)
for vicm > vonp, IM3,4 = 0, thus IM8 = 2IR1 and IM5=0 (since M1,2 sources turn M5 off)

for vonn < vicm < vonp, both IM8 and IM5 are OFF; otherwise, since the current loop is positive, adaptive bias circuit would not be controlled. In this case IM1,2 = 2IR2 and IM3,4 = 2IR1

Hope this helps
Tosei

Title: Re: how to bias constant gm rail-to-rail OTA
Post by RobG on Mar 15^th, 2012, 10:20am

You have a lot of flexibility -- you could probably connect both to mid rail so both diff pairs would only operate together at the mid-point. Whatever you do, the idea is to choose Vb1 so that M7 is still saturated when M8 is diverting all the current (i.e. Vin+ at top rail). Same thing with Vb2. You might, for example, create Vb2 with a diode connected mosfet so that M5 acts as a low-voltage cascode when it diverts the current.

[edit, Tosei posted while I was typing... ]
rg

Title: Re: how to bias constant gm rail-to-rail OTA
Post by HdrChopper on Mar 15^th, 2012, 10:25am

Rob,

I think if you choose mid rail for Vb1 and Vb2 then M8 and M5 will both be on and there is a +ve loop gain in your adaptive biasing circuit with those two guys on. I think you are losing control of such loop if both are on

Tosei

Title: Re: how to bias constant gm rail-to-rail OTA
Post by RobG on Mar 15^th, 2012, 12:16pm

HdrChopper wrote on Mar 15^th, 2012, 10:25am:

You may very well be correct as I haven't tried it. I don't think there's a problem with them both being on. It looks to me like there will be a steady state operating point where the "catch" devices have the same current density as the diff-pair devices. For example, if the gates of M5, M8 and M1-M4 are all tied to midrail, then Vgs5+Vsg8 = Vgs1+Vsg3 so all of the devices will be biased at 2*I_R2/3. -- I guess that would make the gm not-quite constant if both M5 and M8 are on, but other than that I think the operating point is defined.

Title: Re: how to bias constant gm rail-to-rail OTA
Post by HdrChopper on Mar 15^th, 2012, 12:54pm

Rob,

I think you are correct. I've seen this one with ratios higher than 1:1 in the adaptive biasing loop, in which case then I think I would be correct. But that ration being 1:1 I guess is safe and would operate as you pointed out.
Not sure how mismatches would play a role here if you bias everything at mid rail though. I'm think what if, instead of the ideal 1:1 ratio, you get 1:1.01...slowing drifting to rails?