Designer’s Guide Community

aaron_do

Senior Fellow

Offline

Posts: 1398

Accurate duty cycle for sampling clock in ADC
Jun 16^th, 2014, 10:07pm

Hi all,

I need an accurate duty cycle for the sampling clock in my ADC. I was thinking to design a simple DLL based on a string of varactor-loaded inverters, a PFD, a charge-pump, and a single capacitor loop filter.

I'm wondering if this is the typical approach or does anybody have any better ideas. I could just use a tunable delay line (string of inverters again), but then it would need to be calibrated.

Any input is welcome.

thanks,
Aaron

Back to top

there is no energy in matter other than that received from the environment - Nikola Tesla

IP Logged

raja.cedt

Senior Fellow

Offline

Posts: 1516
Germany

Re: Accurate duty cycle for sampling clock in ADC
Reply #1 - Jun 17^th, 2014, 3:33am

hi Aaron,
I haven't designed any DCC for ADC but have done few for high speed clock duty cycle corrector. basically many people use analog loop, but I am not sure about DLL based. The basic idea would is low pass filter the clock so you would end up with dc level above or below vdd/2 based on your duty cycle now correct Duty cycle with current starved inverters until you see vdd/2 dc after low pass filter. However there is even simple idea with high pass filter. Please check fig 5 in the following pap.

http://www.ece.tamu.edu/~spalermo/ecen689/2012_28Gbps_4tapffe_15tapdfe_xcvr_bulz...

Thanks,
Raj.

Back to top

IP Logged

aaron_do

Senior Fellow

Offline

Posts: 1398

Re: Accurate duty cycle for sampling clock in ADC
Reply #2 - Jun 17^th, 2014, 5:41pm

Hi Raj,

thanks for the suggestion. Actually I had thought about something like that, but the only problem is that I also need a clock signal which goes low just before my main clock goes high again. Also, it honestly seems as complicated as a DLL :P

I will keep your suggestions in mind and maybe think of some alternative.

thanks,
Aaron

Back to top

there is no energy in matter other than that received from the environment - Nikola Tesla

IP Logged

loose-electron Senior Fellow Offline Best Design Tool = Capable Designers Posts: 1638 San Diego California	Re: Accurate duty cycle for sampling clock in ADC Reply #3 - Jun 20^th, 2014, 7:51pm how many clock edges? what frequency? what alignment between edges? what accuracy? whats available as a reference clock source? define the question please
Back to top	Jerry Twomey www.effectiveelectrons.com Read My Electronic Design Column Here Contract IC-PCB-System Design - Analog, Mixed Signal, RF & Medical IP Logged

RobG Community Fellow Offline Posts: 569 Bozeman, MT	Re: Accurate duty cycle for sampling clock in ADC Reply #4 - Jun 21^st, 2014, 10:57am Why don't you create 2x the clock frequency and divide by two with a toggle flop?
Back to top	IP Logged

raja.cedt Senior Fellow Offline Posts: 1516 Germany	Re: Accurate duty cycle for sampling clock in ADC Reply #5 - Jun 21^st, 2014, 11:43am @RobG, Divide by 2 might be easy and low power at ADC clocking frequencies but don't you think 2X is difficult. Very options like 2X with XoR gate or is there any other method. I know this is very common at VCO level but I haven't seen any other places. Thanks, Raj.
Back to top	IP Logged

RobG

Community Fellow

Offline

Posts: 569
Bozeman, MT

Re: Accurate duty cycle for sampling clock in ADC
Reply #6 - Jun 21^st, 2014, 12:03pm

raja.cedt wrote on Jun 21^st, 2014, 11:43am:

@RobG,
Divide by 2 might be easy and low power at ADC clocking frequencies but don't you think 2X is difficult. Very options like 2X with XoR gate or is there any other method. I know this is very common at VCO level but I haven't seen any other places.

Thanks,
Raj.

I'm guessing I'm missing something about your application that prevents you from using 2x frequency to begin with. I assume you know that you can divide by two with a flip flop like this:

What am I missing?

Back to top

IP Logged

aaron_do

Senior Fellow

Offline

Posts: 1398

Re: Accurate duty cycle for sampling clock in ADC
Reply #7 - Jun 22^nd, 2014, 7:56am

Hi all,

I'm trying to get a duty cycle of ~36%. When you mention divide by 2 I assume you're talking about getting 50% duty cycle?

As for the number of clock edges, I also need one clock edge just before the end of one period, with no overlap.

As this is just a test chip, right now the scheme I'm using is basically just using tunable delays and some logic to prevent overlap. I'm mainly wondering now what is the standard method for dealing with this? So is it common to see a delay-locked loop on an ADC chip? That seems like the best solution to me.

thanks,
Aaron

Back to top

there is no energy in matter other than that received from the environment - Nikola Tesla

IP Logged

RobG Community Fellow Offline Posts: 569 Bozeman, MT	Re: Accurate duty cycle for sampling clock in ADC Reply #8 - Jun 22^nd, 2014, 10:23am aaron_do wrote on Jun 22^nd, 2014, 7:56am: Hi all, I'm trying to get a duty cycle of ~36%. When you mention divide by 2 I assume you're talking about getting 50% duty cycle? Oh, wow, I knew I must be missing something. Yes, I was thinking you were after 50%.
Back to top	IP Logged

loose-electron Senior Fellow Offline Best Design Tool = Capable Designers Posts: 1638 San Diego California	Re: Accurate duty cycle for sampling clock in ADC Reply #9 - Jun 24^th, 2014, 3:08pm loose-electron wrote on Jun 20^th, 2014, 7:51pm: how many clock edges? what frequency? what alignment between edges? what accuracy? whats available as a reference clock source? define the question please One more time...
Back to top	Jerry Twomey www.effectiveelectrons.com Read My Electronic Design Column Here Contract IC-PCB-System Design - Analog, Mixed Signal, RF & Medical IP Logged

aaron_do

Senior Fellow

Offline

Posts: 1398

Re: Accurate duty cycle for sampling clock in ADC
Reply #10 - Jun 24^th, 2014, 6:22pm

Hi loose-electron,

Quote:

One more time...

1) I need three clock edges. The first two form the sampling clock, and the last is to trigger an event.
2) 160 MHz. 40nm CMOS
3) alignment? Do you mean the spacing? The first two form a roughly 36% duty cycle (DC) clock. The last edge triggers just before the 36% DC clock.
4) Accuracy is not important (DC could be from 30% to 40% for example), but jitter should be around 1ps rms or less.
5) As this is a test chip, the clock source will be a signal generator.

you know, despite not giving all of the information you asked for, I got some very useful answers. Raj. mentioned sensing the duty cycle, and RobG mentioned the use of dividers (although div 2 won't give the edges I need, div 3 or more might). I just need some general guidelines, and I'm intelligent enough to sort out what can and what can't be used in my system. Also, even if a method can't be used in my work, it may still be interesting to hear...

regards,
Aaron

Back to top

there is no energy in matter other than that received from the environment - Nikola Tesla

IP Logged

loose-electron

Senior Fellow

Offline

Best Design Tool =
Capable Designers

Posts: 1638
San Diego California

Re: Accurate duty cycle for sampling clock in ADC
Reply #11 - Jun 26^th, 2014, 5:51pm

aaron_do wrote on Jun 24^th, 2014, 6:22pm:

you know, despite not giving all of the information you asked for, I got some very useful answers. Raj. mentioned sensing the duty cycle, and RobG mentioned the use of dividers (although div 2 won't give the edges I need, div 3 or more might). I just need some general guidelines, and I'm intelligent enough to sort out what can and what can't be used in my system. Also, even if a method can't be used in my work, it may still be interesting to hear...

When shooting at a target, the probability of success is much higher when you know what the target is and where it is placed.
Better yet, if you can see the target, so much the better.

You can use a DLL to subdivide the clock (or generate defined delays) to get what you want. Other options are possible as well.

Back to top

Jerry Twomey
www.effectiveelectrons.com
Read My Electronic Design Column Here
Contract IC-PCB-System Design - Analog, Mixed Signal, RF & Medical

IP Logged

aaron_do

Senior Fellow

Offline

Posts: 1398

Re: Accurate duty cycle for sampling clock in ADC
Reply #12 - Jun 26^th, 2014, 8:43pm

Quote:

When shooting at a target, the probability of success is much higher when you know what the target is and where it is placed.
Better yet, if you can see the target, so much the better.

So the question is whether I provided a visible enough target for the level of accuracy that I required......

thanks anyways for the help,
Aaron

Back to top

there is no energy in matter other than that received from the environment - Nikola Tesla

IP Logged