Why are Radar and Lidar systems so expensive?

[MrOmnos]

We started using Radars in in 1940s and we have been using lidars for past 20 or so years. Price of everything that could possibly be used to make a Radar has decreased in past 10 or so years. Processing power is cheap, we have better educated community with great amount of experience from cold war era, use of radar and lidar are increasing in robotics and automated cars. But the prices of these things haven't dropped a bit. Why is that? I think that is partly because the hobbyist and opensource community doesn't seem to be interested in the field. It's not that the community is not there. I have been scouting the interwebs for few days and I found some people giving it a try and making their own stuff but it's few people that can be counted on fingers. Someone started a project called 'Lidarlite" but it's gone now, don't know why? Why are Radar and Lidars not so interesting for a hobbyist? If we look at the history and use of these things there should be a booming community based around them.

[Kilrah]

What would you do with one as a hobbyist?

[CatalinaWOW]

Lots of hobby uses for radar/lidar.  RC airplanes are one of my hobbies, the following are uses I have contemplated just in that area.  Altimeters for planes which is what the above mentioned project was good for.  Doing detail contour maps from a drone.  Tracking RC planes at the flying field to help people stay out of no fly zones (irritated/irritable neighbors).  Scanning for local manned traffic to aid in implementing see and avoid.

I don't know why prices have stayed high.  Lots of possibilities.  Even though there are many hobby applications, the market is still small.  Safety and other issues associated with high power radiators.  Technical challenges in obtaining range resolution and dealing with enormous dynamic ranges.   And finally, with my list as an example, there are often other ways to solve the problem that work "well enough".

[Radiohead]

Radar systems are not easy to make, since they are in the mega - gigahertz range, a lot of signal analysis, mechanical engineering and permit/paperwork is required I imagine. Neither can you expect to just build something and expect it to work like most common hobby projects. Besides every government on the world wouldn't be so happy if radar systems were easily available. Just imagine what a few nutjobs could do if they had easy access to a radar.

[Marco]

Dual use regulations probably prevent availability of low level COTS parts too.

[MrOmnos]

Well, same can be asked for Ham radios? What's the use of ham radios and why people spend 1000s of dollars on them, I mean there are better ways of communication available. Hobbyist may not be at the same level as Boston Dynamics but why should they have all the fun? I think we need to catch up. Everyone can make a drone these days and give it a bunch of GPS co-ordinates and tell it to fly, but very few can fly through an obstacle course on their own. I think in future everyone should be able to build their own house pet or upgrade their normal car into an street legal autonomous car by buying a kit online. May be my thoughts are far fetched. But still, Radars and Lidars are going to be widely used in autonomous vehicles and bots in future and I think we as electronics community are lagging far behind large companies in such fields. I mean it was hobbyist who brought PCs to everyone's home not IBM.

[C]

High frequency Radar is both Electronics and Mechanical Design
You could say the hard part is not the electronics but the Mechanical Design of the RF part.
For a Normal radio you tune the electronics. For Radar you tune the Mechanical part.
For example the Doppler radar might only have two electronic parts at RF with the remaining electronics us working in audio frequency range. The Unknown frequency of transmit is used to mix down the returning echo that only changes very little in frequency. With doppler the transmitter is at full power while receiving a very very small signal with small change in frequency, normally using same antenna.

You can use any frequency with radar. The hard part is the separation of the signals by direction of signal.
When you change to distance you have two options. Change transmitter frequency with time or pulse the transmitter.
Frequency change gives option of distance doppler.
Distance = time delay of echo return.
Doppler= Speed of movement to/from transmitter are in the return echo signal.

And remember that radar direction is via antenna pattern.

Run some numbers and see what you have to make function.

[Kilrah]

I think in future everyone should be able to build their own house pet or upgrade their normal car into an street legal autonomous car by buying a kit online.

There are many powerful entities that will do all they can to prevent you from doing it in order to keep their business though, same as it's always been.

But still, Radars and Lidars are going to be widely used in autonomous vehicles and bots in future and I think we as electronics community are lagging far behind large companies in such fields.

It may be feasible to make a lidar, but as mentioned they have limitations that don't really make it that interesting in a hobby context when the building complexity is taken into account so why bother with precision mechanics, optics and timing you can't really DIY. Where an industrial/security-versed entity would use a Lidar for super precise and reliable intrusion detection for which the $5k cost is peanuts compared to their usual expenses the hobbyist will take a $2 PIR detector, more than enough. Even if the Lidar cost only $200 they wouldn't take it.

Altimeters for planes which is what the above mentioned project was good for.  Doing detail contour maps from a drone.  Tracking RC planes at the flying field to help people stay out of no fly zones (irritated/irritable neighbors).  Scanning for local manned traffic to aid in implementing see and avoid.

All of that needs powers that are beyond what can be made generally available.

Lidarlite was purchased by Garmin who discontinued their products, I'd imagine it's probably either to keep the tech for use on their own products or to avoid any risk of lawsuits by people misusing them.

[Assafl]

RADAR and LIDAR are a very broad term.

RADAR based door openers are cheap (albeit mostly replaced by PIR except for Doppler reasons). A naval phased array 8 beam RADAR is not.

A laser based range meter is a very dumb LIDAR and is cheap. A car speed detector is regulated and calibrated and isn't.

I have a RADAR door opener (Gunn diode and IF and all) that I have lots of fun with. It was dismantled from a discarded door.

[Jeroen3]

If you had the knowledge and time to build radars, you'd be making it your job. And you won't be telling anyone, since it would cost you your job.

[System Error Message]

If i could have radar i wouldnt mind having one on the roof of my house to track aircraft or any space related things
And also to have drones go around and do things like drop cones to reserve parking and get mail, interfere with peoples 5Ghz wifi that use DFS.

[radar_macgyver]

As @Assafl pointed out, not all radars need be very complex. Look up Greg Charvat's work at MIT to build low-cost FMCW radar that can be used for ranging and some crude SAR work. He uses connectorized mini-circuits parts for everything, and coffee cans for antennas. You can get surprisingly good performance with a simple setup like this. The limiting factor is separating the transmit and receive signals. FMCW does this in the frequency domain, where the dynamic range is limited by the filtering one can do, and the receiver desensitization prior to filtering.

Where radar starts to get expensive is when you need to have better dynamic range, for which you need pulsed radar. Pulsed radar separates transmit and receive in the time domain, with theoretically infinite dynamic range. Traditionally, the only way to get both resolution and sensitivity required rather high power RF (+90 dBm = 1 MW) sitting next to very sensitive receivers (sensitive to -114 dBm). It's not so hard to get similar power levels (magnetrons are cheap and cheerful), and sensitive receivers are a dime-a-dozen. It does, however, take some work to get the sort of isolation required for pulsed radar. The circulators and receiver protectors needed are usually the expensive bits.

Pulse compression radar (in a way, a hybrid of FMCW and pulsed radar) takes an in-between path which means you don't need such high peak power. Signal processing recovers the necessary sensitivity and resolution. The FPGAs needed to do this are quite inexpensive these days.

Another thing that makes radar expensive is antennas. Antennas need to have at least a 0.5 sq.m aperture at ~10 GHz to have usefully narrow beams. This is what usually puts radar tracking applications out of reach of hobbyists. It's often a lot easier to use other methods like transponders instead.

If you had the knowledge and time to build radars, you'd be making it your job. And you won't be telling anyone, since it would cost you your job.

Yep. 

[CatalinaWOW]

Some simple numbers will explain much of this.

Radar and lidar have SNR that goes as 1/R^4.  If you want to track a satellite you need some combination of antenna gain (means size and precision which means dollars), the ability to point that antenna (again dollars), transmit power (again dollars.  Megawatt power levels are not unusual for people tracking satellites), and very sensitive receivers which also need to be broadband to get range resolution.  There are very good texts by Barton and Skolnik which don't require too much math to understand if you want to get a good basic understanding of what is going on.

Civilians generally aren't allowed to do megawatt level power transmission, though there are loopholes.  For example AFAIK the law hasn't caught up with technology, and optical frequencies are largely unregulated.  But if you hit someone with your megawatt lidar transmitter existing liability laws will be more than enough to cook your goose (just as you cooked someone else).   If you decide to play in the game be careful.   

So it becomes really obvious why amateur long range tracking of satellites and aircraft hasn't taken off.  (Though some googling on bistatic radar will give some ideas and options).

Short range radar is another story.  Seems like there are some opportunities here.  The first application of this sort of thing was the VT fuze back in the 1940s.  The radar part wasn't the hard part, it was making it survive being launched by a cannon.

Gigahertz parts are now easy to get and not horribly expensive.  Cars are starting to use them for braking and warning of lane changes so there are some hacking opportunities.  But there are real barriers too.  You don't just push this stuff together with some #22 wire, and the gear to actually look at signals is expensive.  Time to get really solid on your stripline design and figuring out how to infer what is going on up there at high frequency.

[System Error Message]

Best reason to have radar is to mess with peoples 5Ghz wifi.

[dmills]

Not sure there is such a thing as an Eye safe Q switched laser, but megawatt class peak pulses are easy if you go there, and narrow band filters are not that problematic for the common laser wavelengths.

I seem to remember a short ranged synthetic aperture experimental design being published as part of a course on radar systems somewhere (MIT?).

If I wanted to play, I would probably ignore the transmit side, there are plenty of transmitters having well defined pulsed signals (DAB has lovely convenient guard bands around each symbol, GSM has possibilities, Tetra...), a phased array of aerials and an accurate timebase would probably let you build an image by looking for correlation peaks between the transmitted signal and each of your beams, and modern graphics cards are good at that kind of maths.

Point one fairly narrow TX aerial at whichever transmitter you are using as your source, and point your phased array away from it so the transmitter is in a null, then digitize (A Rack full of those cheap dongles butchered to have a common reference clock), and get your math on.

Not everything interesting needs to be boring monostatic affairs, and the nice thing about bistatic is that sometimes other people have set up transmitters for you....

73, Dan.

[Marco]

Not sure there is such a thing as an Eye safe Q switched laser

Something like this?

Erbium doped phosphate glass rods and slabs and lasers build with them don't seem to be available as commodity products though. Boutique items with undoubtedly boutique pricing.

[amyk]

RADAR and LIDAR are a very broad term.

RADAR based door openers are cheap (albeit mostly replaced by PIR except for Doppler reasons). A naval phased array 8 beam RADAR is not.

A laser based range meter is a very dumb LIDAR and is cheap. A car speed detector is regulated and calibrated and isn't.

I have a RADAR door opener (Gunn diode and IF and all) that I have lots of fun with. It was dismantled from a discarded door.

Related video:

[XFDDesign]

There are plenty of people who do Radar work, and are not doing the work for Raytheon or similar operations.

Other people have gone into really odd directions on the matter as well.

There are a few key problems, some people have touched upon this already:

1. What is your purpose? - Altimeters have very different requirements than speed indicators. In short, Radar is very specific to the task. This is something of an antithesis to all hobby level stuff which aims to be so generic it isn't very good at anything (See: Modern state of hobbyist SDR systems)

2. What range are you after? - This dictates power levels. The FCC will love getting their collective dick in a twist if you're transmitting at any level sufficient to make the system useful.

3. How is it being mounted? - Someone else brought up some numbers I wasn't too hot with, but ultimately your frequency of interest will be dictated by how much surface area you can dedicate to antenna arrays.

Remember, Radar itself was discovered by reflected radio emissions at HF as a boat crossed the English channel. This isn't something revolutionary. If you want to just fuckabout, pickup a DirecTV receiver dish, a fixed 10-12GHz oscillator and a mixer that can resolve down to DC. You can get the received signal to beat against the LO by way of doppler shift and "listen to motion." A friend of mine who works for the NAIC Observatory did just this. Its range is only good to maybe 100m, however.

[Marco]

How does that oscillator in Clive's video work? Base seems to be DC (PCB capacitor directly connected to it) and the emitter is connected to an AC shorted transmission line ... does it oscillate at a wavelength equal to 4x the serpentine path?

[nowlan]

Hmm. Making me think about the Matel Radar Gun.

[rs20]

Supply and demand?

[Assafl]

RADAR and SONAR engineering are actually quite simple. I remember as a young SONAR engineer I met one of the older folk who designed naval SONARs (10's of kW to almost a MW)  - and asked him how one goes about designing such a complex system.

His answer taught me simplicity: you start with the SONAR (or RADAR) equation and work back.

TS (target strength), NL (noise level), DI (directivity index or gain if you want), and required SNR (signal to noise ratio) etc - all result in Needed SL (source level) per TL (transmission loss - i.e. Range).

And when set in logs - it is simple addition. Of course one can add processing gain (as in matched filters and the like) and some awful reverberation (noise from near field "blasting" by the high powered sonar) in near field (vs. noise).

Once you have the SNR and SL - the rest is simple electronics.

[Assafl]

My last answer (about the RADAR and SONAR equations) doesn't get to the cost.

So based on the equation results I may need a high SL (source) and high SNR (signal to noise ratio) to detect a small target (TS) at a high range TL (transmission loss).

If I have to build a safe, reliable, high power (Megawatt) transmitter with pulse compression and a cross correlation, low noise high dynamic range (for near field detection) receiver - that is a lot of engineering, takes up 10-20 cabinets and needs a tower to make it safe. All very expensive.

A door opener has a huge target, up close, walking towards the door (so I can leverage Doppler effects for simple and cheap SNR improvement). There fore a Gunn diode in a flared enclosure with a simple IF demodulator, pulse detector (with Doppler shift detection)  and filter - and I have a nice RADAR.

[Lukas]

Some people are building their own radars: http://hforsten.com/homemade-synthetic-aperture-radar.html It has become comparatively easy these days.

[rrinker]

 There's not a whole lot of work in this field outside military applications - high frequency rf and microwave is my best friend's specialty and he's been out of work, except for short consulting gigs, for a couple of years now. He has a pretty well equipped home lab and works on whatever he finds of interest in the meantime. I was recently there and he was showing me some of the devices he's made, it is really a different world for someone who comes from the digital world and may be used to analog in the audio ranges.