开云体育

With difficulty. I believe there's a method involving three sampling heads
looking at each other's kickout pulses and some maths which can be used to
evaluate their performance.

On 15 March 2019 at 14:03 "Reginald Beardsley via Groups.Io" <pulaskite@...> wrote:


I have one of Leo's 100 pS impulse units as a source for initial experiments. If I can narrow that to 10 -20 pS with discrete SMD parts I may be able to get the idea to work. There's a close connection to the Hilbert operator hiding in there somewhere.

I've been reading an excellent book on FPGA implementation of DSP by Roger Woods et al most of the afternoon. So I've not started on the dissertation yet. I can't really say much else until I understand the mathematics and physics of using the reverse biased diodes as capacitors. That should keep me entertained for a few weeks at least.

In any case, failure doesn't matter. It's a hobby project, not a client deliverable. I've been fascinated by the problem of getting a fast rise time edge for a long time. It's a bit of circular logic. My excuse for doing it is I bought an 11801 and my excuse for buying the 11801 is so I could try to do it. If I can narrow the 100 pS pulse by 5x I'll have an excuse to buy an SD-32.

I spent 3 years studying sparse L1 pursuits using "A Mathematical Introduction to Compressive Sensing" by Foucart and Rauhut as a text followed by reading the original papers by Donoho, Candes et al. Beats the hell out of watching television. I just wish I could find someone else who was interested in the subject to discuss it with. It's quite amazing what you can do with sparse L1 pursuits and really easy to do in practice. Understanding the math is not easy, but most people use FFTs without really understanding them.

Have Fun!
Reg

I have one of Leo's 100 pS impulse units as a source for initial experiments. If I can narrow that to 10 -20 pS with discrete SMD parts I may be able to get the idea to work. There's a close connection to the Hilbert operator hiding in there somewhere.

I've been reading an excellent book on FPGA implementation of DSP by Roger Woods et al most of the afternoon. So I've not started on the dissertation yet. I can't really say much else until I understand the mathematics and physics of using the reverse biased diodes as capacitors. That should keep me entertained for a few weeks at least.

In any case, failure doesn't matter. It's a hobby project, not a client deliverable. I've been fascinated by the problem of getting a fast rise time edge for a long time. It's a bit of circular logic. My excuse for doing it is I bought an 11801 and my excuse for buying the 11801 is so I could try to do it. If I can narrow the 100 pS pulse by 5x I'll have an excuse to buy an SD-32.

I spent 3 years studying sparse L1 pursuits using "A Mathematical Introduction to Compressive Sensing" by Foucart and Rauhut as a text followed by reading the original papers by Donoho, Candes et al. Beats the hell out of watching television. I just wish I could find someone else who was interested in the subject to discuss it with. It's quite amazing what you can do with sparse L1 pursuits and really easy to do in practice. Understanding the math is not easy, but most people use FFTs without really understanding them.

Have Fun!
Reg

Hi, Reg.

Regarding fast pulses and such, there is a wealth of information at Jim
Andrews' page here: He founded
Picosecond Pulse Labs in 1980 and sold it to Tektronix in 2014. I used
PPL products with a CSA803 back in the mid 1990's at Lockheed. I think
the sampling head I used was an SD-24, but since it was about 25 years
ago, my memory is a bit fuzzy.

These days I have a 7904, a 7603, a 7S12, a 7S11, two S-4 sampling
heads, S-53 and S-51 triggering heads, etc in my garage lab. Oh, and a
couple of 10:1 probes I made out of board-edge SMA connectors, 450 ohm
(470 paralleled with 1.8k ohm) 0402 resistors, and 3-pin right-angle 0.1
inch spacing headers. I've not completely characterized the probes, but
at least they pass up to about 12-13 GHz without much attenuation above
the expected 20 dB. Planning to use them to probe some digital RF
boards I'm going to build for work (Raytheon). Too much trouble to try
to get a fast enough sampling scope at work :(

Enjoy the PPL app notes! I've referred to AN-2a through -2e often.

Thanks for the literature references. I'll have to check them out
(someday).

Jim Ford


------ Original Message ------
From: "Reginald Beardsley via Groups.Io" <pulaskite@...>
To: [email protected]
Sent: 3/14/2019 6:03:56 PM
Subject: Re: [TekScopes] Test input for an SD-32 sampling head

I have one of Leo's 100 pS impulse units as a source for initial

experiments. If I can narrow that to 10 -20 pS with discrete SMD parts I
may be able to get the idea to work. There's a close connection to the
Hilbert operator hiding in there somewhere.

I've been reading an excellent book on FPGA implementation of DSP by

Roger Woods et al most of the afternoon. So I've not started on the
dissertation yet. I can't really say much else until I understand the
mathematics and physics of using the reverse biased diodes as capacitors.
That should keep me entertained for a few weeks at least.

In any case, failure doesn't matter. It's a hobby project, not a client

deliverable. I've been fascinated by the problem of getting a fast rise
time edge for a long time. It's a bit of circular logic. My excuse for
doing it is I bought an 11801 and my excuse for buying the 11801 is so I
could try to do it. If I can narrow the 100 pS pulse by 5x I'll have an
excuse to buy an SD-32.

I spent 3 years studying sparse L1 pursuits using "A Mathematical

Introduction to Compressive Sensing" by Foucart and Rauhut as a text
followed by reading the original papers by Donoho, Candes et al. Beats the
hell out of watching television. I just wish I could find someone else who
was interested in the subject to discuss it with. It's quite amazing what
you can do with sparse L1 pursuits and really easy to do in practice.
Understanding the math is not easy, but most people use FFTs without really
understanding them.

Have Fun!
Reg

---
This email has been checked for viruses by AVG.

Mark,

Thanks. I have to drive 150 miles round trip to access IEEE papers. I'll be collecting lots of NLTL papers the next time I go to Little Rock.

At the moment I'm working on deriving the equations for a transmission line with impedance steps designed to steepen the leading edge. Purely a passive linear pulse shaper using a coplanar waveguide with an air dielectric with no components if possible. After I have explored that I'll look into the magnitude of the improvement I can get using non-linear elements.

What I have in mind is a coplanar waveguide which starts with a 10 ohm impedance and then in a series of steps rises to 50 ohms. The concept is to apply a phase delay to the low frequencies which is greater than the delay applied to the high frequencies. My biggest concern at the moment is how to make the Q of the low frequency sections low enough to be physically tractable without resorting to coaxial cable.

Each step in impedance produces an infinite impulse response, so the mathematical notation can quickly get out of hand.

I have copied all the stuff from KH4HTV's website and read through most of it. Very nice stuff.

Leo Bodnar uses a Maxim 3949 LED driver in his pulsers. That is specified as 22 ps typical, 36 ps maximum Tr & Tf. That's already quite fast.

A fascinating aspect of this is that the mathematics are closely related to the normal incidence plane wave response of a layered medium in reflection seismology, the design of anti-reflection coatings in optics and to ultra wideband antenna design.


Have Fun!
Reg