The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> how to digitally compensate for low order harmonic
https://designers-guide.org/forum/YaBB.pl?num=1136537025

Message started by svensl on Jan 6^th, 2006, 12:43am

Title: how to digitally compensate for low order harmonic
Post by svensl on Jan 6^th, 2006, 12:43am

Hello All,

I am currently concerned with sigma delta A/D conversion and have the following question. Is it possible to shape-out or cancel 2nd order harmonics in the output spectrum digitally (decimation filter maybe) ? For example, due to certain non-linearities in my circuit I not only obtain my signal at 5kHz, but also a dc signal and second order harmonics at 10kHz.
So is there a way to digitally compensate for low-order harmonics? If there are any references on this topic please cite them.

Regards

Title: Re: how to digitally compensate for low order harm
Post by sheldon on Jan 7^th, 2006, 1:02am

Sven,

Is this possible? How would the circuit know what the input frequency is?
Does your input frequency change? If the input frequency varies then the
filter would also need to be tunable and the filter would need to know, a
priori, what the desired frequency is.

Why does your circuit have such large second order distortion? In general,
differential design suppresses the even order harmonics so the odd harmonics
are dominant. It seems like it would be better to spend your effort eliminating
the distortion.

Best Regards,

Sheldon

Title: Re: how to digitally compensate for low order harm
Post by svensl on Jan 11^th, 2006, 12:52am

Thanks for your response. I know that it would be hard to design such a circuit. I was just wondering.

Speaking of non-linearities, how could I make a ring oscilator more linear?
Please consider a simple 15-stage ring oscillator with simple inverters in a 0.35um process.
The frequency is altered by changing the power supply of the inverters Vdd. I simulated the VCO and extracted the control voltage - frequency plot.
Please see figure which also includes a linear best fit to show the non-linearity.
I would like to make the ring VCO more linear. At the moment I can only change the widths and lengths of the inverters, so I guess I would have to consider a new design. From your experience which design would behave more linear.

Regards,
Sven

Title: Re: how to digitally compensate for low order harm
Post by sheldon on Jan 12^th, 2006, 5:32am

Sven,

A little outside of my area of expertise, have you looked at the
Stanford site, http://www-smirc.stanford.edu/papers? There
are several papers that may be useful, for example,

This paper includes discussion of how to design the delay elements and control loop
Low Phase Noise CMOS Ring Oscillator VCOs for Frequency Synthesis
http://www-smirc.stanford.edu/papers/iwdmic98s-raf.pdf

This paper includes discussion of how to optimize the number of stages in a VCO
Phase Noise in Multi-Gigahertz CMOS Ring Oscillators
http://smirc.stanford.edu/papers/CICC98s-ali.pdf

Good Reading,

Sheldon

Title: Re: how to digitally compensate for low order harm
Post by svensl on Jan 12^th, 2006, 8:30am

Thanks for your response. I briefly looked over the papers, but it seems that they're exclusively talking about phase noise and how to minimize it. I will have to search some more to find some publications that talk about VCO design concerning linearity.

Regards,
Sven

Title: Re: how to digitally compensate for low order harm
Post by vivkr on Jan 14^th, 2006, 5:57am

Hi sven,

It has been a while since I worked on ring oscillators, but it seems to me that for a
ring oscillator, the plot you show is still remarkably good. Normally, the V-F
curve of a normal oscillator is seldom very linear. For most cases, this is probably
good enough.

Why do you need it to be so linear anyway? Is there a special application or are you
trying to use your oscillator with some special loop configuration?

Regards
Vivek

Title: Re: how to digitally compensate for low order harm
Post by svensl on Jan 16^th, 2006, 2:25am

The VCO will be used in an open loop configuration. Thus, if excited with a sinwave for example, the FM output of the VCO will have the original sine wave plus non-linearities such as a dc component and a harmonic at twice the signal frequency. See picture.
That is why I would like to have the VCO behave as linear as possible.

sven