Hacker News new | comments | ask | show | jobs | submit login
6 GHz frequency modulated radar (hforsten.com)
261 points by paulgerhardt on Dec 3, 2014 | hide | past | web | favorite | 84 comments

Very impressive. I would have thought that this sort of RF work is outside the hobbyist realm.

The rest of the blog is equally impressive — 217-ball BGA soldered in a home-made reflow oven?

If you think thats cool, MIT students have been making Synthetic Aperture Radar (which uses "chirps" of radio waves and measures the time of flight) for the past few years, with enough resolution to image groups of push pins. Oh yeah, and it works through walls. Radar can seem like a black art but simulation software and a bit of knowledge makes simple stuff like this perfectly doable.


I think its awesome that GPS is regulated (Speed or Altitude) with export controls, but you can build high-quality, weapons-grade SAR at a college. The future is incredible.

cool footnote there:

"This work is sponsored by the Department of the Air Force under Air Force Contract #FA8721-05-C-0002."

well, you can build high-quality un-locked GPS receiver as a DIY too :)


When that link first hit HackerNews I thought, "There are some incredible people out there doing cool stuff".

Well, funny story. When North Korea put up their first satellite it used GPS. Which, when you think about it, is pretty funny because denying entities like the North Korean regime use of GPS in that use case is a major reason the export controls exist.

Which is very funny, as they could've used celestial navigation systems/star trackers (which are impossible to deny them from using) similar to that used with ICBMS.



I've always thought star trackers would make wonderful backups to GPS for high-altitude aircraft (they work even during daylight hours depending on atmospheric conditions).

"depending on atmospheric conditions" is a big IF in the polar regions.

A USAF U-2 on a high altitude polar air sampling mission accidentally wandered into Soviet airspace at the height of the Cuban missile crisis because auroras blocked the pilot's stellar fixes.

See: http://www.vanityfair.com/politics/features/2008/06/missile_...

Have we come far enough yet with mapping the earths magnetic field that you could use it to cross check your location with the star tracker's data?

Sort of. I know a little about this from working in the oilfield. If you think navigating in the sky is tricky, try navigating underground.

We had the BGGM[1], a model of the Earth's magnetic field, that gives us magnetic field strength and direction, given lat/long, altitude, and date. Accuracy was around 0.5deg in direction and I think 0.5% in strength. I'm not sure about the validity of it at more exotic locations and high altitudes, though. We used it as an accuracy check when taking field measurements to determine direction. I can elaborate if anybody's interested.

As for using the data to check aircraft location, if the model is reasonable valid in those conditions, it might work as a sanity check. I doubt it would be practical to reverse it and get location from field information, though. It's already tough to get really good field data in an environment filled with metal objects and electric currents, though.

[1] http://www.geomag.bgs.ac.uk/data_service/directionaldrilling...

It's extremely variable, and I assume unpredictable.

I seem to recall reading about magnetic field charts for aviation purposes, but their purpose was to translate from location to magnetic field orientation, not the other direction.

https://upload.wikimedia.org/wikipedia/commons/2/28/Mv-world... from https://en.wikipedia.org/wiki/Magnetic_declination

The P-3C Orion is fitted with a retractable sextant. Pre-GPS it was a primary navigation aid, but has been maintained post-GPS as an emergency system in case GPS fails. I don't know offhand of its replacement, the P-8 Poseidon, has any kind of celestial navigation system as a backup, but I would not be surprised if it did.

> they could've used celestial navigation systems/star trackers

Well, that implies they have the expertise to do so. Using a GPS is orders of magnitude easier.

Star trackers are available off the shelf, and aren't export restricted. They're essentially precision CCDs with on-board logic.

Disclaimer: I am a geo/navi/celestial positioning nerd.

the North Korea supposedly doesn't have keys to decode the military GPS signal content - thus they are limited to few meters precision which they can get from consumer GPS and from the un-decrypted military signal (and just by using the GPS satellites as close "stars" for old fashioned celestial navigation). If they managed to get military keys then of course they could get a few feet precision. Of course this difference in precision doesn't matter that much even when your warhead is only modest 5-10kt nuke :)

The point being that they had access to an illegal GPS receiver (which didn't have the limits of altitude and speed that all civilian receivers should have). But, of course, the civilian GPS signal is just a signal, and anyone can build a receiver capable of decoding it, so there's not a lot anyone can do to prevent that capability from getting into the hands of folks you don't want to have it.

>"illegal GPS receiver"

illegal where? for sale in US? I'd not be surprised that GPS gadgets produced in China for US market are just loaded with different version of firmware containing those limits while the same gadgets going somewhere else - with no limits firmware, or may be with even stronger limits (until pretty recently many "US legal" consumer GPS gadgets were illegal in Russia for being "too precise" :)

Hah, there's a lot more to weaponized/spy SAR than just the radar; you have to have very good measurements of your movement as you record. It can be improved algorithmically as well but data is king

However, export might be a different story.

Put the PCB board plan online, and it can be built/printed at almost any PCB fab shop.

Goodbye export controls. Its PGP all over again, just with atoms instead of bits.

He's an EE student and clearly a pretty good one, so IMO he is beyond the typical hobbyist realm. The nebulous space of the professionally trained tinkerer.

bunnie huang comes to mind as another example of this sort. When he cracked the Xbox, he had training & equipment that hobbyists don't have- but he did it in the hobbyist spirit.

Today you can get access to equipment thru hackerspaces, so its not even all that expensive.

It still depends on what you want to do- sniffing high speed memory busses and work in reflow ovens are still out of reach of most hackerspaces I have seen.

You can use a rework station (basically a low air speed hot air gun) and wave it around. You'll need one anyway. Maybe $100 minimum.

The biggest problem with building your own reflow oven is the conversion kit mfgrs want their profit off repacking and marking up retail parts (arduinos, SSRs, thermocouples, etc) and of course you need a donor oven (maybe $50 at walmart?) so you end up paying like $150 to bodge something together.... while over on ebay, you can get a brand new professional reflow oven shipped to you for a bit over $200 from the usual gray market sellers. Supposedly they work pretty well and I'll eventually buy one. Or maybe they're junk. If I get closer to buying one I'll research it in detail.

I've done the skillet / hot plate + pan thing. Make a sling of kapton tape (you do have kapton, right?) because otherwise you're going to melt the solder and then stand there thinking "um, ok, so now how do I remove a PCB thats too hot to pick up, and leaving it on the hot plate for 30 minutes while it cools down exceeds all sanity" Even with lead free solder paste I think its unacceptable to share hot plate with kitchen and lab. You can put the PCB on a steel spatula and the whole works on the hot plate and that works pretty well if you don't do kapton.

(edited to add and you can easily spend more than the cost of a reflow oven by trying to avoid buying a reflow oven, because a non-contact IR temp meter is a handy accessory to the hot plate technique, but if you need to "charge" the cost of a IR thermometer to the hot plate project, suddenly its cheaper to just buy an oven...)

Hand soldering SMD is very easy but slow. Faster than hand soldered thru-hole, but still pretty slow. I have done 0201 RF parts but it was not really any fun at all, the bigger stuff is much easier to hand solder. And no caffeine allowed.

I skipped out on making my own oven back in the day; results seemed spotty. I'm glad to hear they are more feasible now. I stuck with the reflow station I had access to. Nice to see those have come down in price too.

I still remember trying to do 0201 with the reflow station, accidentally blowing caps right off the board :)

Every Hackerspace I've seen has one of those modified pizza ovens with an Arduino controller attached. Its really cheap to do nowadays.

Yeah, kinda like that. If you're curious;



The PDF has the full story, the first link has some neat photos including die shots.

Then its easily in reach of average hobbyist. You dont need anything super expensive, the one I linked was done using $30 FPGA board.

Good lord, are FPGA's that cheap now? I remember the absolute bargain-basement slow-as-bricks FPGA board running $100.

one used in my link ~$35


smallerr one $16 free shipping


programming dongle ~$5 http://www.ebay.com/itm/USB-Blaster-Programmer-Cable-For-FPG...

We live in amazing times for hardware developers. Stuff that was out of reach 10 years ago is peanuts right now.

Kinda. There's a huge price split depending on if you just need GPIO or if you need transceivers (serial above 1GHz). If you just need GPIO $30 is reasonable. If you need transceivers, $300 would be a bargain.

Soldering BGA is not all that impressive, routing proper pcb for one is.

Contrary to popular belief soldering SMD in general is way easier than dip, all you need is an oven/hot plate/hotair gun and solder paste/flux. Magic happens due to molecular surface tension instead of your manual dexterity.

Yeah. I actually prefer prototyping with SMD these days, since everything is super cheap, easily available, small, and easy to solder. You basically apply flux liberally, heat up the pins, apply solder, and surface tension does the rest. (The whole "never melt solder with your iron, heat the pin and let it melt the solder" no longer applies. You can put a big ball of solder on your iron and just sweep it over the component you're soldering. There's plenty of flux in play because you drowned everything in flux.)

BGA is outside my abilities, however.

It's a shame the way the impedance-controlled traces on this board turned out all janky though. I get boards done at Sunstone and I guess the price might be higher but the results are good.

I think most people under-estimate the degree to which you can click a button on your computer and get manufactured goods in the mail. Even for small runs like ten boards I can cost-effectively get the boards fabricated and stuffed and delivered to my desk in a day or a few days depending on how much I'm willing to pay.

How much are you paying per in^2 to have Sunstone fab a board? Are you paying for assembly? OSH Park, as used in the article, is only $10/in^2 for 4-layer.

It strongly depends on how many you order, since nearly all the cost is in the setup. A 10 in^2 2-layer board is ~$300 for ten boards.

Don't forget: http://dirtypcbs.com/

There is too much interesting cool stuff in the world - I will never get to try out half of it !

It's like realising that willy wonka's chocolate factory is bigger than you can walk around in one lifetime.

Darn it.

Step 1: Life-extension

I really love that you can get a power amplifier for the 5Ghz RF for about $2.

Its about the right price for +20dBmW at 6cm. minicircuits.com has several MMIC in that freq range at that output power.

Transcom sells a nice well over 30 dBmW MMIC for that same freq band... you can buy a kit from minikits in australia, the EME141-5800, the bare chip is like $20.

Some ham radio work shows you don't need the fancy 4 layer weird oshpark material for workable stuff in that general freq range. http://www.w1ghz.org/MBT/multiband.htm

As an interesting extreme opposite comparison, there was a B-29-ish era chirp radar altimeter for aircraft landing which was almost entirely mechanical. A little buzzer/motor thing varied the size of a resonant cavity by a little bit, and the resulting audio signal was turned into a voltage / altitude using something like a capacitor/resistor bridge. So one vacuum tube, a resistor, a capacitor, a diode mixer, and a meter (and a lot of plumbing)

If you'd like better antennas around that range, yet again, just ask a ham radio guy. Its a very awkward freq band where dishes, and most importantly dish feeds, are huge compared to 10 or 24 GHz band, but a loop yagi like you'd use at 2 GHz or lower would be ridiculously small. The microwave ham radio contester guys seem to like horn antennas at that band so couple it into a moderately large waveguide and attach the (homemade) waveguide to a (homemade) horn antenna.

What costs money is the spectrum analyzer, network analyzer, and Microwave Studio license.


Testing RF is a PITA because the test equipment is so stupidly expensive.

Yes. Back in the 1990s I built a little LIDAR system using much the same principle as this system - modulate the output beam with a 20MHz carrier, down-convert at the receive end using the same local oscillator like an FM receiver, and measure the phase shift between the down-converted output and input. That transformation is phase-preserving, and the phase shift at the lower frequency represents the range.

It wasn't hard to build. But I didn't have the test equipment to adjust it.

The VCO (HMC431LP4) seems to be more like $20, though.

edit: should have read the rest of the article after the schematic where he says as much

I don't really see how the radar images at the bottom reflect the scene he showed. Any tips on reading those?

The radio waves are collected at points and don't have any x or y data, so making an image out of it is pretty challenging and usually requires moving the radar beam. He just kept the beam still, so the only things you can really see are things that move.

In the image, the bottom axis is distance(calculated from the time the reflection took/the frequency output by the mixer), the color is the strength of the reflection, and the left axis is the time. He walked away from the setup and then back, visible as the two lines, which fade out almost completely as he changes direction, because he is far away and therefore looks smaller to the radar.


From this data you could theoretically build an image of his upper torso and head, but it would be very blurry. His arms and legs were not static, and the radar doesn't know how they were moving, so they would just show up as blurs.

Cool, thanks! That helps a lot. Even though I knew what the axes were my intuition struggled a bit, I think partly because I'm so used to the bottom/X axis being used for time.

I'm impressed!

Since, I'm not an EE, is there a better antenna which could reduce the noise in the signal?

It looks as if the noise comes in a pattern and I was curious if that is because of the ripples in the antennas (i.e. cans)?

No, the ripples show up for very simple reasons. Each bar is a reflection at a distance, the wider bars represent wider objects(which will reflect over a larger range of distances, since the radar beam is not very straight). He guesses what a few of the reflections are and some of it is probably the ground.

The best antenna for radar is almost always a very directional one. Most radars use dish antennas of various designs. The very large majority of DIY microwave radios(more than you'd think) have horn antennas, which are desirable for their large bandwidth. You wouldn't use a yagi on a radar because its bandwidth is very limited, and bandwidth is directly related to resolution.

Most highly accomplished EEs can't answer most questions about fields, let alone antennae design detail.

Good engineers can do dipole approximations in their heads. This would require a computer- there are just too many internal reflections.

This really makes me wish I took Antennas as an elective in EE.

Are there any good books/resources to get up to speed on things like this?

The lecture notes on the MIT site these projects are inspired from has some useful PDFs: http://ocw.mit.edu/resources/res-ll-003-build-a-small-radar-...

When I was building my own version, Kai Chang's RF and Microwave Wireless Systems was a really great resource for the basics. Beyond that, you'll want to have a good foundation in signal processing to make use of the data from the device. The author doesn't mention it, but the radar he has is capable of creating SAR images. I'd love to see the results of a 6 GHz radar -- Dr. Greg Charvat (one of the MIT guys) has a demo on his website of a 10 GHz version that's pretty impressive. His thesis can help with some signal processing details, but the book Spotlight Synthetic Aperture Radar: Signal Processing Algorithms by Carrara, Goodman, and Majewski is even better.

Very impressed. Solid knowledge demonstrated. The blog makes it seem so easy, but imagine the years of trial and error needed to produce this knowledge (which of course the author gets "for free")

Nice work. Could ditch the Wilkinson though since your coupling to the mixer is -10 dB, so just use a resistive tap instead of 3 dB splitter and 7 dB attenuator.

I'm not sure this is legal in the US w/r/t FCC. I know the author isn't in the US, but pointing this out for US hackers.

It would be kind of grey. He was aiming for 5.8 which is an open band in the US, although his bandwidth was way above the width of that channel. Generally though unless you are actively impeding someone else or, god forbid, you're on government or medical channels, you're pretty okay and you are unlikely to get hunted down. People get nabbed for jamming cell phones and such but if you are using it rarely and not using a massive amount of power, nobody will notice.

Good practice is to stick to the open bands: http://en.wikipedia.org/wiki/Amateur_radio_frequency_allocat...

Also, he says he's aiming for a radar band: "The reason I chose to use 5 – 6 GHz band was because there is a band reserved for radars around 5.8 GHz and this range overlaps with 5 GHz WLAN."

Since 5.8GHz wifi devices are required to detect and avoid frequencies actively used by radar systems, it would be pretty straightforward to just use your laptop to check if it's safe to do a hobby radar project in your location. It's really easy to make this a very low-risk project.

The issue is not Licensing, but Type Approval.

Even on the ISM bands (http://en.wikipedia.org/wiki/ISM_band) any transmitter must still be submitted for Type Approval Testing.

The only exception to Type Approval is on the Amateur bands where "homebrew" is permitted (and encouraged). However the Amateur Regs specifically rule out any commercial use.

The only way this could legally be operated is in a professional quality screened room, or if the University has an "Experimental" class radio license.

It's in a legal band for any use, but you are correct, the output power is 300-600 times too high to be unlicensed.

> I'm not sure this is legal in the US w/r/t FCC.

What is the rationale for not allowing these?

It's not specifically not allowing these devices, it's broadcasting on a restricted frequency without the proper license. The FCC regulates who can broadcast on which frequencies. Getting a ham license can open it up a bit, but there are still regulations to follow.

I wish they would open up 214-400 THz range.

Then instead of charging everyone $99/month for poor cell phone service, we can charge $99/month just for living.


I think parent might have meant the Black-body radiation of a human [1], but was off by an order of magnitude... should be 14 um to 7.5 um wavelength = 21.4 THz to 40.0 THz.

[1] http://en.wikipedia.org/wiki/Black-body_radiation#Human_body...

He is not offering the product for sale. It is experimental and therefore allowed.

Actually no.

Any experimental device must be covered by the appropriate experimental class license.

Think about it. Any dumb CBer could claim that his homebrew Linear is "Experimental".

That's different. CBers are broadcasting.

FCC gives you exemption for up to five units [1]. You are also exempted during development; how else could you even build the thing. FCC of course reserves the right to come in and tell you that your device is messing up the neighborhood and to cut it out. [2][3] Your device can't be used explicitly for the purpose of broadcast or jamming.

The experimental license is for those who actually want to broadcast/communicate on those restricted frequencies.

[1] http://www.metlabs.com/blog/emc/some-electronic-devices-are-...

[2] http://www.law.cornell.edu/cfr/text/47/15.103

[3] http://www.law.cornell.edu/cfr/text/47/15.29

CB'ers are not "Broadcasting". Broadcasting has a specific meaning and it is not permitted under a CB license.

And yes, you are exempted for ownership while under development, but your organisation needs a written license or permit, plus you cannot transmit outside a screened room.

The Experimental License is for those who wish to Transmit outside a screened room.

Note that the O.P. is in Finland, not USA and the licensing details vary with each country.

Source: Half a lifetime as a Government Radio Inspector.

Thanks for that.

You are right. Broadcasting was the wrong word.

Given your extensive experience, then perhaps you can help clarify something for me: How do you reconcile the experimental license requirement with the FCC part 15 exemption I linked to?

For reference, my background is hardware development and I have done a fair bit of RF but they've all been in ISM band and we never had to get FCC approval until we intended to make/market the product and don't know of anyone who has ever obtained an experimental license, or a license of any kind.

Was this because of the ISM band or are people technically doing it wrong, but getting away with it because the device isn't radiating so high that it actually interferes with a neighbour and leading to an FCC complaint?

Firstly my training is not primarily in USA FCC regulations, so the terminology will vary. However the basic principles still apply, as all of this derives from the international ITU agreements.

The sources you link to refer to radiated emissions from non-transmit devices. All electronic equipment which is marketed requires a whole slew of Compliance testing.

This covers things like standards for connections to Power or Telephone, as well as Radiated and Conducted Emissions and RF Immunity. Various classes of equipment (eg Commercial, Industrial, Military, etc) have different standards to meet.

Part 15 makes some exemptions for low power Domestic devices. But you still require Type Approval even under Type 15, it's just that the requirements are somewhat relaxed.

If you are an approved developer, you are covered for Possession of non-approved equipment during development.

But this has nothing to do with equipment which can transmit. Transmitters require all of the above testing, but also a further round of testing to do with harmonics, stability, bandwidth, power levels, etc.

If you wish to test a non-approved Transmitter you must do so in a professional-grade screened room, or apply for an Experimental License to cover specific Frequencies and Locations.

Years ago an organisation could apply for a blanket Experimental License which basically covered any testing in the organisation. It's main function was to ensure that a nominated individual understood (and was responsible for) all of the relevant regulations. Typically the nominated individual would hold a First Class Broadcast or Marine ticket. More recently these Licenses have been supplanted by limited permits for specific frequencies and locations.

People think that it is fine to transmit on "unused frequencies". The problem is that there are virtually no frequencies which are not allocated to some organisation.

And people often don't realise how an incredibly low power transmission (or harmonic) can block a sensitive Emergency receiver on the other side of town.

I could tell you dozens of stories of clever people who thought they were using "unused" frequencies. Typically they were inadvertently transmitting on the input frequency of an Emergency Repeater and hundreds of users were slowly getting angrier and angrier until the Authorities managed to track them down. When convicted they not only paid large fines, but also the Investigator's costs and the downtime for the company affected.

The reason for the specific permits is that if things do go wrong, the authorities can quickly get in touch and fix the problem.

I wasn't aware that you'd need a license while developing an intentional radiator (what you're calling a Transmitter, I think). I'll have to look into it.

Thank you for that information.

This makes me wonder if you could map a building using existing 5ghz wifi n/ac signals.

The best antenna upgrade for least work would probably be to build a Yagi

Actually no, a yagi would be very bad. The Radar has a 700MHz bandwidth, almost 12% of the center frequency, and yagis work for a fraction of that. A horn is pretty simple to make.

Oh! I didn't see the bandwidth was so wide. You are correct. Wouldn't the resolution still be decent with a narrower bandwidth (i.e. the 10s of MHz that a yagi could manage)?

Also, as an alternative to a horn antenna, a parabolic reflector should work well in this band without being too large, right?

The range resolution would go from 8.5" to five yards if you changed the bandwidth to 30 MHz. Currently the radar could figure out the direction a person is facing to about 45 degrees. 5 yards would be good to find the orientation of a house.

Parabolic antennas are just horns with lenses. The bigger the parabola, the better the focus and the wider the bandwidth. A horn would be a few inches across, but a 700 MHz reflector would be a couple feet.

No, FMCW radar range resolution is proportional to RF bandwidth.

You would still need the horn as a feed for the dish, but yes, you would get more gain.

That would be a pretty tiny Yagi, too. :)

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact