The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> RF Design >> op amp used in charge pumps for charge sharing / mismatch
https://designers-guide.org/forum/YaBB.pl?num=1358298571

Message started by mixed_signal on Jan 15^th, 2013, 5:09pm

Title: op amp used in charge pumps for charge sharing / mismatch
Post by mixed_signal on Jan 15^th, 2013, 5:09pm

Hi,
In charge pumps to remove charge sharing a unity gain opamp is used.
Again to minise current mismatch or increase resistance of current sources, a servo opamp can be used as shown in razavi books.

I want to know which method is more helpful in minimising spurs?
I have power constaraint and want to use only one

Title: Re: op amp used in charge pumps for charge sharing / mismatch
Post by raja.cedt on Jan 16^th, 2013, 2:58am

hello,
Can you please post schematic? Very Generic but famous charge pump attached, please check this.

Title: Re: op amp used in charge pumps for charge sharing / mismatch
Post by mixed_signal on Jan 16^th, 2013, 9:02am

Hi raj am talking abt the one u have attached for removing charge sharing.

I have attached the one in razavi which uses source switching and a servo opamp. Can you give me suggestions on the opamp?
Opamp has to have rail to rail i/p swing. so a complementary nmos+pmos diff pair solves the issue. But i am facing problem in gain stage since most of them use cascode and I have 1v vdd

Title: Re: op amp used in charge pumps for charge sharing / mismatch
Post by raja.cedt on Jan 16^th, 2013, 10:08am

Hello,
Need for rail to rail depends on your VCO, check across the frequency range of interest control voltage range. Some times it won't be rail to rail. Even if you need rail to rail it's not that tough to keep 4 transistors under 1v in this kind of applications. Here op amp don't have any signal swing at the output.

I have one Question in your configuration (which is not similar to one i posted), you have both +ve/-ve feedback here, so how do you ensure -ve dominates the other?

Thanks,
Raj.

Title: Re: op amp used in charge pumps for charge sharing / mismatch
Post by mixed_signal on Jan 16^th, 2013, 1:11pm

Hi,
The left & right strands are exactly symmetric. At DC the circuit is quasi stable or unstable with -ve=+ve. But at high operating frequency , the CPOUT node has a loop filter and hence is steady than the vref node or it is a low AC resistance. So, the loop with M2 is weaker.

My question:
My VCO needs control voltage in the range 0.4 to 0.8 voltage. But still, should not i design the opamp for rail to rail because before settling control voltage will traverse near 0 and vdd and it has to be made sure that the opamp has sufficient gain for the cp to be operational. How people usually do it ?

Title: Re: op amp used in charge pumps for charge sharing / mismatch
Post by Yutao Liu on Apr 18^th, 2014, 2:53am

Hello,
The source switching charge pump post above is attractive due to its less current consumption.
I try to design this charge pump. But its switching speed is not so high as current-steering counterpart.
In my case, Icp in DC simulation is 300uA, the bias current is 30uA. And the reset pulse from PFD is 2ns.
I ran transient simulation with no phase error between clk_ref and clk_fb.
The current flowing out from M2's drain did not reach 300uA within 2ns, but about 260uA. But the voltage at M2's source is lift to Vdd as soon as UP is low.
The same situation happened to NMOS counterpart.
why did it take so long for the current source(M2) to reach its steady value?

Best Regards,
Yutao

Title: Re: op amp used in charge pumps for charge sharing / mismatch
Post by Yutao Liu on Apr 18^th, 2014, 3:00am

During coarse tune (the loop is open), set the Vctrl on a value between 0.4~0.8. During lock (the loop is closed), release Vctrl. The input range of opamp will stay within 0.4~0.8.
Hope it is useful.

mixed_signal wrote on Jan 16^th, 2013, 1:11pm:

Title: Re: op amp used in charge pumps for charge sharing / mismatch
Post by loose-electron on Apr 20^th, 2014, 11:30am

Strictly speaking I think it is better called charge injection rather than charge sharing. That can be debated, if some academic has used the "sharing" name then more power to them...

One thing not mentioned in a lot of the academic books and is important to charge pumps -

Limit the voltage swing on the gates going into your current steering switches.

Generally:
1 - the switches end up being minimum geometry.
2 - the following op-amp helps a lot with the flipping back and forth of the charge pump
3 - reducing the control signal voltages (pump up, pump down) on the gates of the current steering transistors further improves performance by getting closer to the ideal switch performance that you want.