Microwave Black Magic

[IntegratedValve]

http://en.wikipedia.org/wiki/Distributed_element_filter#cite_ref-1

[mikeselectricstuff]

 

[tom66]

I'm doing this in 2nd yr RF circuits and it's really not that hard. Microstrip technology is pretty cool. Passives are pretty much useless at 10GHz+...

[Dongulus]

Oh yea. I had to design a 1GHz Chebyshev LPF on a microstrip in college using Agilent ADS software. There was something like 7 elements to the filter that we needed to fiddle with their dimensions to meet specs for passband ripple, roll off, and minimum out of band attenuation. The software's optimization scheme couldn't seem to satisfy all specs simultaneously so it took some play to dial in the design to one that would fit the specs.

I still have no real understanding of how microstripping works, but I did learn that it's no fun trying to design something by randomly messing with many variables until the solution falls into my lap. It didn't help that the computers kept getting disconnected to the license on the network's servers every 5 minutes which froze the program. 

Experiences like this is probably why I moved into the digital regime. 

[IntegratedValve]

I still have no real understanding of how microstripping works, but I did learn that it's no fun trying to design something by randomly messing with many variables until the solution falls into my lap.

The second option is no fun either, fiddling with almost unsolvable partial differential equations.

[BBQ]

I can't grasp it.

On another note, I read somewhere about Khufus pyramid, a theory that it is some kind of RF lab with waveguides and resonance chambers and what not... Combined with some other fields black magic.

But I beg you not to treat me as a nutcase for just mentioning this in this topic, or at all. 

[TerraHertz]

I can't grasp it.

It's arcane and intimidating to me too. I wish I had these skills. But not yet. It's still Blinkenlights to me.

On another note, I read somewhere about Khufus pyramid, a theory that it is some kind of RF lab with waveguides and resonance chambers and what not... Combined with some other fields black magic.

But I beg you not to treat me as a nutcase for just mentioning this in this topic, or at all. 

Much fascinating mystery there. But as for waveguides... how is that supposed to work, with rock? It's not exactly electrically conductive, so would make piss-poor waveguides.
Otoh, I heard from someone who'd been there, that the Pine Gap facility does have some rather large underground tunnel structures, all lined with copper. That function as resonators and filters, for... something.

[IntegratedValve]

I would not believe in the pyramid RF mysteries, they did not have the math required. 

[c4757p]

God, I wish I had the equipment to play with this stuff. It looks like such fun

[don.r]

These things require grey beards that need to be stroked when working on them. Dave knows.

[KJDS]

These things require grey beards that need to be stroked when working on them. Dave knows.

What they actually need is an understanding of filter theory which is fairly straightforward and an understanding of microstrip which is easy enough until you have to deal with the fiddly bits such as  the dielectric constant being different for the X and Z directions. As frequencies increase, box modes and the problems of getting from coax to microstrip and back cause as much problems as the circuit design.

[TerraHertz]

Earlier this year I bought a HP 11600B transistor test fixture from a seller I was buying other stuff from. It was cheap, and I'd always wondered what the insides of these things looked like. Was not disappointed.
Only I can't figure out how to open the transistor socket part, without breaking it.

The substrate does not appear to be fiberglass. It's something homogeneous.
Also notice that HP have done scalpel blade adjustments to the end margins of the resonating structures, or guides, or whatever the isolated copper areas around the central tracks are for.

[mcinque]

Terrific!

Like TerraHertz said: "I wish I had these skills" 

[nctnico]

http://en.wikipedia.org/wiki/Distributed_element_filter#cite_ref-1

Probably the most useless Wikipedia article    It is better to get a couple of good books on the subject and learn how to use Sonnet. From there you'd need a spectrum analyser with tracking generator or an RF generator for real experiments. Working with microstrip elements requires some math but calculating stepped impedance filters from a lumped element (LC) circuit is very doable without fancy software.

[ConKbot]

I'm doing this in 2nd yr RF circuits and it's really not that hard. Microstrip technology is pretty cool. Passives are pretty much useless at 10GHz+...

Enthusiasm, Naivety 

[IntegratedValve]

http://en.wikipedia.org/wiki/Distributed_element_filter#cite_ref-1

Probably the most useless Wikipedia article    It is better to get a couple of good books on the subject and learn how to use Sonnet. From there you'd need a spectrum analyser with tracking generator or an RF generator for real experiments. Working with microstrip elements requires some math but calculating stepped impedance filters from a lumped element (LC) circuit is very doable without fancy software.

Wikipedia is the way to go. It will be the susbtitute for academia soon. Who knows...

Sonnet is fine...I usually use Eagles, but for designing microtrips I will stick with MS Paint, much more flexible.

RF generators and spectrum alanlyzers are expensive for me. 

[free_electron]

My introduction into 'real' RF was in 1989 by a finnish engineer at Salcomp. They had a spectrum analyser that took a waveguide as input....
The guy takes his calculator , types a few minutes, scribbles some numbers down, takes a tape measure , cordless drill and a drill bit and a pencil. Walks to the waveguide, measures some locations and marks em in pencil. Drills a few holes , and sticks in some self tapping screws.

He looks at the analyser screen and turns the screws deeper or shallower until he had the right bode plot.

There i was, 18 year old kid , looking at it with glazed over eyes realizing 'that stuff really is black magic'

[IntegratedValve]

My introduction into 'real' RF was in 1989 by a finnish engineer at Salcomp. They had a spectrum analyser that took a waveguide as input....
The guy takes his calculator , types a few minutes, scribbles some numbers down, takes a tape measure , cordless drill and a drill bit and a pencil. Walks to the waveguide, measures some locations and marks em in pencil. Drills a few holes , and sticks in some self tapping screws.

He looks at the analyser screen and turns the screws deeper or shallower until he had the right bode plot.

There i was, 18 year old kid , looking at it with glazed over eyes realizing 'that stuff really is black magic'

Looks like more technical/trade stuff than "black magic"

[peter.mitchell]

My introduction into 'real' RF was in 1989 by a finnish engineer at Salcomp. They had a spectrum analyser that took a waveguide as input....
The guy takes his calculator , types a few minutes, scribbles some numbers down, takes a tape measure , cordless drill and a drill bit and a pencil. Walks to the waveguide, measures some locations and marks em in pencil. Drills a few holes , and sticks in some self tapping screws.

He looks at the analyser screen and turns the screws deeper or shallower until he had the right bode plot.

There i was, 18 year old kid , looking at it with glazed over eyes realizing 'that stuff really is black magic'

Looks like more technical/trade stuff than "black magic"

he forgot to mention the guys eyes had rolled back into his head and he started making a gurgling noise.

[c4757p]

I still think it sounds much more fascinating than intimidating. I really need to find somewhere to learn this stuff.

[IntegratedValve]

I still think it sounds much more fascinating than intimidating. I really need to find somewhere to learn this stuff.

http://www.amazon.com/Microwave-Engineering-David-M-Pozar/dp/0470631554/ref=sr_1_1?s=books&ie=UTF8&qid=1388505716&sr=1-1&keywords=microwave+engineering

http://www.amazon.com/RF-Circuit-Design-Theory-Applications/dp/0131471376/ref=sr_1_3?s=books&ie=UTF8&qid=1388505692&sr=1-3&keywords=RF+circuit+design

http://www.amazon.com/Lumped-Elements-RF-Microwave-Circuits/dp/1580533094/ref=sr_1_sc_1?s=books&ie=UTF8&qid=1388505739&sr=1-1-spell&keywords=lumped+elements+for+micrwave

http://www.amazon.com/Microstrip-Slotlines-Artech-Microwave-Library/dp/1608075354/ref=sr_1_sc_1?s=books&ie=UTF8&qid=1388505778&sr=1-1-spell&keywords=microstriplines+and

http://www.amazon.com/Microwave-Filters-Communication-Systems-Fundamentals/dp/0471450227/ref=sr_1_2?s=books&ie=UTF8&qid=1388505841&sr=1-2&keywords=microwave+filter+design

[c4757p]

I'm going to have to do some reading! Thank you. Now, I can read all I want about anything, but that of course still leaves the problem of having absolutely no way to test my knowledge. How do people typically find themselves involved with this kind of work? Is this the sort of specialization that typically requires further study at university? (I'd have to find myself a better one... and a soul or three to sell to pay for tuition...)

[chscholz]

You'll need a vector network analyzer if you want to measure the phase response of your filter, can't get the phase with your spectrum analyzer/tracking generator.

Probably the most useless Wikipedia article    It is better to get a couple of good books on the subject and learn how to use Sonnet. From there you'd need a spectrum analyser with tracking generator or an RF generator for real experiments. Working with microstrip elements requires some math but calculating stepped impedance filters from a lumped element (LC) circuit is very doable without fancy software.

[IntegratedValve]

Electrical engineering is all about modeling physical systems that process electric current and voltage. This modeling can be done using Maxwell equations and electromagnetic theory, or a more simplified human friendly idealized model called lumped circuit model (AKA Electric Circcuit). For RF stuff you still use the lumped approach but a more general one called distributed element model.

Simulation/Approximation
Existing Electric Physical System  --> Lumped Model

Design
Lumped Model  --> Electric Physical System

Synthesis
Mathematical Equations/Block Diagram/Other models --> Electric Circuit

This is regardless of DC or RF.

Microstrip lines are a substitute to passive components: capacitance and inductance. They work because of high frequency wave nature, but they are still modeled as lumped elements using ideal capacitance and inductance.

[TerraHertz]

Electrical engineering is all about modeling physical systems that process electric current and voltage. And having the means to practically build circuits and use instrumentation to measure performance to verify the models.

Fixed that for you.
Modeling without practical experimental verification can be highly misleading, even utterly useless. In any field.

Ha ha... I can think of some fields of 'science' where the lack of experimental validation has led to particularly stupid situations.


It's frustrating that things like a modern, calibrated VNA are so far beyond my financial means. As an alternative I've been accumulating RF circuit analysis stuff from an earlier era, like slotted transmission line instruments. The aim is to set up a work area for RF experimentation. But progress on that is slow.

[nctnico]

Probably the most useless Wikipedia article    It is better to get a couple of good books on the subject and learn how to use Sonnet. From there you'd need a spectrum analyser with tracking generator or an RF generator for real experiments. Working with microstrip elements requires some math but calculating stepped impedance filters from a lumped element (LC) circuit is very doable without fancy software.

You'll need a vector network analyzer if you want to measure the phase response of your filter, can't get the phase with your spectrum analyzer/tracking generator.

True but there is another way. If the phase response is important you can use a filter with the appropriate phase response (Bessel, Butterworth, Cauer/elliptic, etc). If the frequency response of your microstrip circuit matches the (Sonnet) simulation then the phase respons is most likely the same to the one simulated. This leaves some room for speculation but Sonnet's simulation results are pretty much spot on.

[chscholz]

You'll need a vector network analyzer if you want to measure the phase response of your filter, can't get the phase with your spectrum analyzer/tracking generator.

True but there is another way. If the phase response is important you can use a filter with the appropriate phase response (Bessel, Butterworth, Cauer/elliptic, etc). If the frequency response of your microstrip circuit matches the (Sonnet) simulation then the phase response is most likely the same to the one simulated. This leaves some room for speculation but Sonnet's simulation results are pretty much spot on.

Good point. I guess as always, it depends on the application, too. If you build filters for a living you'll not get around fully characterizing your device of course. For active filters things might be a bit different, too. But we are not talking about any of these cases.
So, thanks for pointing this out.

[G0HZU]

Looking at the Wiki page content I'm a bit bemused when this stuff still gets described as magic/voodoo. Those all look like classic microstrip filter designs.

It gets much more interesting when you explore multi layer designs using defected ground technology up in the GHz region I've designed (and built/tested) various novel filter structures in this way using Sonnet. It really is remarkable how accurate and capable Sonnet is, even with quite complex 3D structures.

[VK5RC]

I like the quote I once read about waveguide was that is was calculated using a scientific calculator, build with a micrometer and tuned with a hammer!
RE the comments about VNA/Spectrum analysers, they make it a lot nicer but several hams have built and tested 100+GHz transceivers using only power meters.
For me the best reading was about TRANSMISSION LINE THEORY
Some of the literature aimed at amateur radio operators e.g. Microwave Encyclopaedia  http://www.microwaves101.com/encyclopedia/index.cfm may be a gentle and more practical introduction before hitting the hard theory later.