"Moding" in microwave transmission lines

We recently did a project, a time-domain optical modulator using custom dual-stage lithium niobate modulators and expensive distributed amplifiers.

The two legs of this mach-zender interferometer are unequal length. That's OK if the light is very coherent, but it makes for a bad tempco. So we built a gigantic, massive oven for the modulator.

Distributed amps are, like all RF parts, poorly characterized for time-domain use. One has to experiment.

On 2023-12-04 13:05, John Larkin wrote:> On Mon, 4 Dec 2023 06:42:02

-0800 (PST), Fred Bloggs > snipped-for-privacy@gmail.com wrote: >

Oops, that's an expensive one. :(

Fortunately not fatal. On the plus side, unbalanced MZs make very nice delay discriminators for FM. I've used that effect a lot in lidar designs.

Cheers

Phil Hobbs

Got whacked back down again pretty sharpish by the paper contract dumpers, though! You might almost think they're rigging the market. ;->

What's goin' on here, John? You always said you hated RF and regarded it as a black art and now you seem to have at least become very well-acquainted with it - and at microwave frequencies, too! Even the likes of Phil Hobbs haven't found fault with your pronouncements. Jolly well done. Autodidactic approach to the subject, I'm guessing?

I do despise sine waves, s-params, noise figures, and Smith charts. They are all legacies of the slide rule days.

I do use "RF" parts in fast, wideband, time domain products. The data sheets are often deceptive, so I test and characterize the parts myself, and sometimes hack Spice models.

Two annoyances:

When a frequency response graph stops at some low frequency, you can expect that something weird is being hidden, like a slow internal bias loop. We used one absorptive RF switch part whose claimed LF response is "DC" and actually didn't work right below about 100 MHz. I spoke to the chip designer about the LF behavior and he wouldn't discuss it because it's "proprietary". That was Maxim.

Many RF devices assume that some Vcc is connected to their output through an inductor. So in reality they expect 2*Vcc as an instaneous sine peak voltage, double the abs max spec. I test them to destruction (at $300 each!) and back off some.

Many others.

Decent test equipment for those kind of frequencies is horrifically expensive, so I have no idea how you can justify forking out that kind of money given that this kind of development is outside your usual line of activity. Do you just hire some in when you need it on an ad hoc basis?

Our usual line of activity is picosecond electronics.

500 MHz oscilloscopes, and 30 GHz sampling scopes, are affordable. I don't think we have ever rented test equipment. We don't work in frequency bands, so don't need a range of things.

It's annoying that part and equipment suppliers expect their customers to be working in one frequency band. I'm seeing that a lot lately in ICs and even fets. I guess they are tuning wire bonds and such for one market.

People are labeling the pins of fets RF IN RF OUT GROUND instead of gate, drain, source. DC specs are "adjust the RF port bias trimpot until it works."

Bell Labs spent decades understanding and {theoretically} perfecting waveguides.

Their 4 GHz {later 6 & 11 as well} Long Lines network {1951~1985} had 3000+ sites. They would feed their KS-15676 horns with a series of traps, filters, and combiners to accomplish there. There was LOTS of copper used in same. Modes are discussed in the following:

But they didn't quit while they were ahead. They wanted to put in waveguide from NYC to DC,

The field test was a fustercluck; I've chatted with one of the engineers. when it concluded, the project wrote a glowing report, and WT died a quiet death.

Soon, MCI was busy burying that fiber stuph, and Ma never caught up.

It's pretty incredible that optical fibers can be made as transparent as they are. A few dB/km, utterly amazing.

Jeroen Belleman

Waveguide highpass (scale model of above):

Not sure how much of that is "moding" exactly vs. it just gets dispersive and/or the relatively inexpensive SMA cal standard used wasn't a very good reference past a few GHz for wave guides.

The low frequency cutoff of an X-band waveguide that size is about 7 GHz for the TE10 mode and the next highest propagating mode is at 13 GHz.

Also not entirely sure about the performance of the coax-waveguide coupling up there..

Don't. Zerohedge is exactly the kind of publication that would take money to help a pump-and-dump scheme defraud gullible punters.

Sorry, Bill, but you obviously don't know what you're talking about. ZH has saved my arse many a time with its various heads-ups on all sorts of matters and not all related solely to money by any means. Remember it was ZH who first broke the news of the Coronavirus outbreak in China - and the first to make public the increasingly accepted 'lab leak theory' - which the MSM immediately decried (of course - as did you yourself). ZH was also the first to caution against the mRNA vaccines and highlight the risk of blood clots in the Astrazenica (and other) vaccines - again when the MSM was touting them as an essential panacea and even calling for compulsory vaccination of the masses. I could go on and on about how much money ZH has saved me over the years as well, but I know it would all fall on deaf ears, so I'll save myself the bother...