Designing cavity filter, may I know more about yours?

[quote=“david.baker”]The FA external filter is a LC ladder circuit.

Cavity filters in China are in the same price range of the FA filter if they have your frequency.
Custom filters are much more expensive.

I have also seen DIY cavity filters on RTL-SDR.com including design files.
Some are just a plain metal box and others are machined out of solid metal.
[/quote]

Would it be possible for FA to offer for purchase a 1090 mhz cavity filter at a reasonable price?

Cavity filters are hard.

Lumped-component filters such as the very good Flight Aware filter use individual inductors (L) and capacitors (C). The PCB has to be carefully designed for operation at 1090 MHz, and normal variation in component values will skew performance. That’s one of the usual tradeoffs in deciding how many sections you’re going to make in an LC filter – how many sections can I get away with given maximum variation in component values? But you figure it out and it’s surface mount parts on a board. Easy and cheap. Getting the connectors on and sealing it in the can is more involved.

But you can design one, build a bunch of prototypes, and be confident of filter performance. I don’t know how much testing the manufacturer Flight Aware does – individual, statistical, or none at all. The important thing is you have zero adjustments.

Cavity filters on the other hand – dimensions are critical. Individual alignment is required. This requires good equipment, and while the process can be automated, it’s usually cheaper to have people doing it.

Even PCB filters for this frequency are hard, if you’re going for performance. FR4, the usual fiberglass substrate, doesn’t hack it for performance – you’re looking at a PTFE substrate such as Rogers Duroid (more expensive than FR4). (FR4 doesn’t hack it for a number of reasons, including aging, and performance variation from batch to batch). PCB filters for 1090 MHz are larger than the LC FLight Aware filter – one of those physics things with feature size depending on wavelength.

You can make cavity filters at home. You can get them to work. It takes skill and the right equipment.

bob k6rtm

1 Like

Thanks Bob, for the detailed explanation!

I was, somewhat, disappointed in the performance of the Uputronics 1090Mhz ADS-B Filtered Preamp which I purchase a few months ago and am now hesitant to fork down nearly 100 bucks for the Jetvision.de 1090 MHz 3 Pol cavity filter.

I’m an extreme novice at this stuff and lack the equipment and technical expertise to verify that the cavity filter remains calibrated to specifications, especially after exposure to the rigors of transoceanic shipment.

Do cavity filters require periodic re-calibration to keep them in spec?

Again, thank you for your help!

If the settings are locked and the environment is somewhat controlled, cavities shouldn’t require periodic re-alignment. Shipment shouldn’t change things.

The SDRs we use make for versatile and inexpensive test instruments – using one of the unfiltered ones and a wideband noise source (available on eBay for around $40 IIRC), you can put together an inexpensive network analyzer, for example comparing the output of the noise source swept over a frequency range of interest with and without the filter. Such results won’t be carefully calibrated as they would be with a commercial test setup, but they’re better than a poke with a stick and very useful.

Cavity filters are widely used in radio communications systems in setups called duplexers, where you have a transmitter and a receiver operating on close frequencies and connected to the same antenna through a set of interconnected cavities. In the ham radio world, spacings of 600 KHz and 5MHz are common for systems running at 144 and 440 MHz respectively. The 144 MHz cavities are somewhat large, with the 440 MHz ones smaller than 2 liter pop bottles.

Duplexer alignment is a real PITA, requiring skill and the proper equipment. I know of one instance where a cavity required re-alignment. Well, it required evicting a large number of insects, and then re-alignment…

–bob k6rtm

Thanks Bob!

I knew I should have gotten that HAM license years ago! I’m sure a lot of this stuff would make better sense.

You’ve given me some impetus to possibly try out that cavity filter built in Germany. :laughing:

DVB-T RTL 820 as 1090 Mhz Signal Generator

.

Cool!! I’ll give it a try.

g8jnj.net/softwaredefinedradio.htm

We spec out or design the parts we sell. Prosticks, antenna, filter, etc.
The factories then produce the parts for us.
We then do statistical testing for all the FlightAware parts after we get them from the factory.

The FA filter are made by a well known USA company (in their asian factory) that makes filters, splitters, and RF cables. They haven’t sent us one bad part that we know of.

You are right about the filter having zero adjustments and about using PCB that are specifically made for RF. I can take a picture of the insides if you are interested.
Seems there is a lot more interest in the details than I thought.

We did look at cavity , stripline, saw , and ceramic filters. Every filter has a trade off.
Cavity filters have a very low noise figure but are quite a bit larger. The prices for cavity are just a tad bit higher in quantity than the FA filter.
The labor involved is probably enormous vs the cost of the final product. But this what we are usually find in China where low tech labor intensive products are actually very competitive price wise with higher tech products.

We do order the FA filter by the pallet size and we pass through the parts at very little markup.
I am not sure if the market volume is high enough to get a good price on the 1090MHz cavity filters.

Cool!! I’ll give it a try.

g8jnj.net/softwaredefinedradio.htm

@chrisjohnston50

Have you tried this yet? I just had a go and saw nothing on the receiver.

I tried SDR# with both frequencies (266.5MHz and 268.9MHz) The antenna was the little stubby that came with the dongle. At the receiving end I’m using a tablet running SDRTouch attached to another stubby antenna. The receiver was set to 1090 MHz. Even putting the two stubby antennas about 1cm apart I didn’t see any change in the waterfall.

I’ll make a dipole for the transmitter end in the next couple of days and report back.

Unfortunately, it didn’t take long before it became painfully obvious that this stuff is WAY above my pay grade. :confused:

Unfortunately, it didn’t take long before it became painfully obvious that this stuff is WAY above my pay grade. :confused:

Me too, but I muddled through and drew a blank :unamused:

Time for abcd567 to do one of his tests I think :wink:

I’d be interested…
I spent about a decade and a half working for HP - Agilent Labs, most of that time working with the RF groups in Santa Rosa and Spokane. I don’t consider myself an RF expert, but will claim to know a lot more than the average bear… For a lot of the designs I worked with, Duroid was used for the low end, with the really high frequency stuff being stripline and slotline on more specialized substrates.

And thanks again for the fantastic job FlightAware is doing in making these products available – as you described, the design decisions are complex.

cheers–

bob k6rtm

HINT–

Use the cheapest SDR you have! The FlightAware ones won’t work, as they are too good, with pre-amps and filters in the front end!
This trick is betting on a cheap SDR front end re-radiating harmonics of the local oscillator.

Or go to eBay and look for RF Noise Source – lots of folks (in China and HK) selling the same PCB for $10 - $20. The boards are good to 3 GHz or so. Don’t run them at 12 Volts – run them at 9 or 10 Volts and it will last a lot longer with only a slight drop in output level.

bob k6rtm

1 Like

As we used to say in the military, “RF is FM” (for the non-military, I’ll give you a hint, M does not equal Modulation)

Here is a picture of the FlightAware ADSB filter.

http://imgur.com/a/SSkpT

[quote=“david.baker”]Here is a picture of the FlightAware ADSB filter.

http://imgur.com/a/SSkpT
[/quote]

Thank you, David! Very interesting!

Even at this price level, looks like they’re adjusting the inductors – when you look at the pic, you can see that the turns spacing on many of the inductors is uneven – that’s the result of someone taking a plastic (or even wood) tuning stick and spreading turns to adjust the inductance and the filter performance. Some inductors are untouched – compare turn spacing in L8 on the bottom part of the image to L9, L10 and you can see clearly where the turns have been spread out. That tweaking probably takes no more than 20 - 30 seconds per unit, if that.

(You want to or have to adjust the inductors as there are manufacturing tolerances in the capacitor values, probably a few percent, and also variations introduced by using FR4 as the substrate for the board – at 1 GHz, those tolerances add up and some parameters could swing as much as 10%. That leaves you with the inductors being the easiest (only) thing to adjust to tune up performance.)

Also interesting mix of automated and hand assembly – I’m guessing based on the look of the solder joints, but it makes sense that all the capacitors are surface mount, and some of the inductors on that side of the board look to have been reflow soldered as well, with the inductors on the bottom and of course the connector pins done by hand (and trimmed by hand, that’s another give-away). The duller, grey-ish joints are typical of lead-free reflow, where parts are loaded on to the board and run through an oven that takes the board (and its components) through a well defined heating profile which causes the solder paste to go molten, releasing the flux, and also pulling components into alignment on their pads by surface tension, then cooling with a really uniform look.

Other solder joints, such as the connector pins at the end, are more silver and don’t have the uniformity of a reflow connection. Looking at the top connection for L8, and the connection for L9, those look as if they might have been done by hand. I’ve done double-sided reflowed boards with surface mount components on both sides plenty of times. Never tried to reflow a board with through-hole components such as those inductors on both sides.

Thanks for an up close view of a successful product – and a good choice on vendor!

Anyone want to try their hand at designing a little beast such as this, and taking it into volume production? It’s a specialized business!

bob k6rtm

1 Like

And thank you Bob for the detailed chip-english translation :slight_smile:

looking at the pictures … there’s space for an additional row of capacitors, inductors.

Anyone hazard a guess why?

My guess is that it’s a general purpose board that can be stuffed to order. Didn’t need those parts for this one!

bob k6rtm

This board seems to be capable for making both the Balanced (more components) and Unbalanced (Less components) Filters.
The Flightaware model seems to be Unbalanced one.

LC LADDER FILTER - UNBALANCED
https://c5.staticflickr.com/1/626/31759098036_d3dbd743a4.jpg

LC LADDER FILTER - BALANCED
https://c3.staticflickr.com/1/421/31648947882_c15ed69f61.jpg