Slanted Edge Method with Non-Uniform Sampling

mawyatt · Post by **mawyatt** » Mon Sep 16, 2019 11:54 am

Fascinating discussion on utilizing the slanted edge method employing non-uniform sampling for lens system evaluation.

https://mtfmapper.blogspot.com/2019/08/ ... -what.html

https://www.dpreview.com/forums/thread/4421544

Buried in these long thread(s) is the analogy to Faster than Nyquist Sampling/Communication.

Also, an analogy to a Non-Uniform Sampling Analog to Digital Converter developed by Dr Mike Chen of USC back in 2013~2014. This ADC had the unique attribute of creating it's own "Nyquist Rate" based upon the input signal metrics.

Here's an image of this ADC chip from way back

Best,

Mike

rjlittlefield · Post by **rjlittlefield** » Mon Sep 16, 2019 1:32 pm

That first link as an awesome presentation! Best explanation of the slanted edge MTF method I've ever seen.

Thanks!!

--Rik

Smokedaddy · Post by **Smokedaddy** » Tue Sep 17, 2019 12:04 pm

I was never able to get MTF Mapper to work except with 1:1 magnification. Frans (the author of MTF Mapper) did give me a few pointers but unfortunately I lost interesting due to other projects. I've owned Imatest for several years but wanted to give MTF a try, have you tried using it?

Mike, I'm curious what you used to processed your image with. It looks more like a painting than a photographic image.

-JW:

mawyatt · Post by **mawyatt** » Tue Sep 17, 2019 1:37 pm

JW,

I don't have either software package.

This was a stack from long ago and processed with Zerene. Don't recall the details but I'm allowed to show it now. Using non-uniform sampling techniques way back in 2011 for ADCs!!

This chip was in a TSMC process which produced nice colors due to the surface passivation, which I enhanced a little with PS.

Best,

Smokedaddy · Post by **Smokedaddy** » Tue Sep 17, 2019 2:01 pm

... looks like it was processed with a program like Topaz.

rjlittlefield · Post by **rjlittlefield** » Tue Sep 17, 2019 2:08 pm

Ah yes, excellent technology, sampling multiple occurrences of a waveform so as to increase the effective sampling rate.

At the same time, I'm reminded of the analog sampling oscilloscopes that I lusted over when I was in high school, back in the late 1960's.

Of course nothing is exactly the same, but similar enough to prompt a sense of deja vu.

--Rik

mawyatt · Post by **mawyatt** » Tue Sep 17, 2019 4:57 pm

The Non-Uniform-Sampling ADC that Chen developed was rather unique in that the input band limited waveform created it's own Nyquist limit. So no anti-aliasing filter required and no aliasing occurs.

Remember those old Tektronix analog scopes well, they were (and still are) special. Some of the most advanced analog techniques and circuits during that era were developed for those scope. Waveforms were recorded with a Polaroid attachment and they had a 1GHz BW real time scope in 1969 that took hp 20 years to duplicate! They also had a special Micro-Channel-Plate CRT that could capture a very quick single event and hold the waveform on the screen for a long time.

Amazing times back then

Best,

mawyatt · Post by **mawyatt** » Sat Sep 21, 2019 10:42 am

Also along these lines of Non-Uniform Sequential Waveform Sampling was a new type RF and Microwave "Passive Mixer" developed back in ~2009, with original concepts dating back to ~1996 used in a baseband IQ demodulator.

On the surface, this "Passive Mixer" violated fundamental physics and Fourier Analysis with Noise Figures below 3.92dB (pi/2 the theoretical limit for Bi-Phase Mixing), with actual measured NFs below 2dB!! This falls into the "You Can't Do That Category!!", and created much interest in academia and researchers thought out with Cornell's Dr Molnar and Andrews leading the initial efforts in academia. These circuits were so intriguing as to cause DARPA to create a special workshop on such!

To help explain I've used the concept of the input RF waveform "Correlating with Itself" within the confines of the Baseband Bandwidth (Information BW) projected to the Mixer Input (antenna) as a means to describe this unique behavior.

Folks interested can look up PolyPhase Mixer (not filter!), N-Path Mixer and Mixer-First, these all mean the same thing.

Here's an old image I can show of the Broadband Tuning Reciever chip developed by Cornell's Molnar & Andrews utilizing the PolyPhase Mixer using the same TSMC process as the Non-Uniform Sampling ADC.

Best,