A bandpass allows a certain electrical signal to pass while filtering out undesirable frequencies. In a speaker bandpass, the mid-range speaker doesn’t receive tones meant for the tweeter or woofer. Most of the time, this filtering is done with capacitors to remove low frequencies and inductors to remove high frequencies. In radio, the same concept applies except the frequencies are usually much higher. [The Thought Emporium] is concerned with signals above 300MHz and in this range, a unique type of filter becomes an option. The microstrip filter ignores the typical installation of passive components and uses the copper planes of an unetched circuit board as the elements.
A nice analogy is drawn in the video, which can also be seen after the break, where the copper shapes are compared to the music tuning forks they resemble. The elegance of these filters is their simplicity, repeatability, and reproducability. In the video, they are formed on a CNC mill but any reliable PCB manufacturing process should yield beautiful results. At the size these are made, it would be possible to fit these filters on a business card or a conference badge.
30 thoughts on “Bandpass Filters From The CNC Mill”
Microstrips scare me. They just remind me how easy it is to accidentally build a filter when all you meant to build was a connection between two points.
Any piece of wire is a low pass filter. You just try to put the cutoff frequency high enough for it to work in your application.
Wires don’t do what you think wires do.
You draw a line on a schematic and assume zero inductance, zero resistance, zero capacitance to ground, zero propagation delay and infinite frequency response. Real wires are different.
Wires are really a concept at this point. We’ll need to make some interesting latices before we actually make a real wire. Until then, we just got these shitty conductors.
The map is not the territory
“Buildings have walls and halls.
People travel in the halls, not the walls.
Circuits have traces and spaces.
Energy travels in the spaces, not the traces.”
– Ralph Morrison: Grounding and Shielding, Circuits and Interference. Sixth edition.
His poetry isn’t great, but the book is a worthy reference. The first few chapters cover EM fundamentals, up to about a 2nd year level.
(I actually prefer Ott, Henry: Electomagnetic Compatibility Engineering as a reference, it has much more detail and practical information, and is about five times the thickness of Morrison.)
the reference to arbitrary units of “mils” makes me laugh, I ran across this in a electronics component spec and it took me all day to discover that it’s a milli inch and also goes by thou.
Aren’t all units of measure arbitrary, they all come about to help one quantify something.
Natural units aren’t arbitrary, unless you argue that the universe itself is arbitrary.
Well, it *could* be. ;)
If you try to do a build in Planck lengths you’re going to have a bad time.
Actually I’d love to have a ruler marked with yottaplancks but i think that would still be several orders of magnitude too small
The US should switch to a smoot based system.
I strongly prefer “thou”, to avoid confusing spoken mils with (millimetre) spoken mils.
Yeah, I avoid “mils” now, even though I grew up with them, because I’ve noticed a lot of people call mm “mils”. I’ve also heard people call milligrams “megs”, which annoys me tremendously, because I csn’t help thinking, “mega-what?”
I’ve been making an effort to learn all of the units I come across in the various articles I have been reading across all sorts of construction topics. Mils is one that comes up a lot, especially as I learn about PCB design. So what do I do when I see a new unit? I Google it, find out how it converts to other measurements both metric and US systems and add it to my digital flashcards.
So that has worked great. Then I reached mils. Googling that I find mostly pages that say it is a thousandth of an inch but also several that say it sometimes refers to a millimeter. What does one do with that? I don’t care if it’s metric, US, Imperial or Martian but how is a measurement with two different meanings useful? If I see plans for something that are measured in mils can I be safe in assuming it is .001″ (or 25.4 micrometers)? Or do I have to consider that it might be measured in mm? How common is it to use mils as mm? Is that just a casual conversation thing or might I find it in plans and construction articles?
Not to mention the misspelling of MHz…
I’m genuinely curious: where is ‘mil’ not understood to be 0.001 inch?
I’ve seen it in 4 metric countries over my 3+ decade working life. Maybe considered outdated and even quaint, but understood.
It’s a useful unit: the thickness of household aluminum foil, about the precision easy to hit on a mill or lathe, and about the precision you need to specify for most mechanical assemblies.
I agree “thou” is less ambiguous though.
In some countries, mil is used to mean millimeter. In spoken language in the US in many fields, it means milliliter (ml).
I was not confused by the reference, since large plastic garbage bags in the US are often specified in mils of thickness.
But, as you said, less ambiguity is better.
Then there’s the circular mil on top of that. Once you get above 4/0 wire, you start ordering in terms of circular mils.
I’ve used mils (0.001″) for over 30 years at every job I’ve had too. I didn’t even realize it was a problem for anyone. It is a nice unit since you can look at the ticks on a caliper. You can describe most dimensions with an integer this way. Almost all pcb dimensions are described this way with everyone I’ve ever worked with.
Here in the UK, a “mil” can be millimetres for size or milillitres for liquids depending on context. It literally is only when talking about circuit traces might it actually mean “one-thousandth of an inch”. In metalworking it would still be a “thou”.
+1, eyes rolling… kids these days.
My dad was a.primier machiniat, who taught me about mils… later denying he had. But, you never heard if, mili-micro? Yeah… maybe before your time, before nano was common, much less pico, atto? and femto. That’s as far as I got in the milli direction in school, and only worked as far as nanoseconds and picofarads, I think.
Then you get into machining where “tenth” means a tenth of a thousandth.
Mils get encountered more than a little in RF/microwave work. This is due to a combination of their prevalence in machining (mentioned above), and the amount of RF work done by US defense contractors and such. I’ll admit I was one of those who used “mils” conversationally for both ml and mm units. I was in for a rude awakening. I was pouty for a minute, but I got over it. I’m still bitter about ounces being used to describe substrate plating thicknesses, though.
It’s still better than Chemical “let’s use every damn imperial unit we possibly can” Engineering.
Interesting article. Hope to read more about machining microwave systems components and the details regarding more in the future.
Reminded me of “chicken dots” Thought and the “DOPPLER MODULE TEARDOWN REVEALS THE WEIRD WORLD OF MICROWAVE ELECTRONICS” HaD article.
Here’s the link to the comment above I was mentioning: https://hackaday.com/2017/11/01/friday-hack-chat-fundamentals-of-rf/#comment=4177732
The link to the article I noted: https://hackaday.com/2017/09/11/doppler-module-teardown-reveals-the-weird-world-of-microwave-electronics/
Microstrip is so much fun.
I’ve been designing and assembling microstrip RF filters for about a year now. I don’t have access to a CNC mill or the like so I just pay ~$5 per 10 + $15 shipping to have boards made at PCB fab companies in China. It’s really, really useful to have access to a full-wave electromagnetic simulator to experiment so you don’t have to wait 2 weeks between every iteration and can just send the final one off to be made in physicality.
Here’s a few links to the ones I’ve made plus some new designs I’m working on for the next batch:
Looks like if the algorithms are available, we (or someone) can make an opensource simulator. Might even be able to decompile the best source, translate and re-code in another language to save time.
The best effort in this direction I’ve found is OpenEMS (http://openems.de/start/index.php) and particularly the pyopenems wrapper. It’s really rough setting up even the most basic simulation scenarios (ie, you don’t draw, you code, you don’t set config options you set config variables) and it’s not exactly made for doing planar stuff fast.
Please be kind and respectful to help make the comments section excellent. (Comment Policy)