WEBVTT 00:01.000 --> 00:03.230 I hear that you got a software-defined radio. 00:03.230 --> 00:08.310 In this presentation, you will learn how to make one of your own antennas to use with your new SDR. 00:08.310 --> 00:10.950 Welcome to the DEF CON RF Village presentation. 00:10.950 --> 00:19.830 I'm Erwin Karinczik, a cybersecurity researcher, and I will show you several methods to make an antenna using easily accessible materials and open source software. 00:20.310 --> 00:35.350 Previous researchers with published works in the wireless and software-defined radio, or SDR domain, have all expressed that using a higher gain directional antenna would improve distances at which the desired application could be executed. 00:35.350 --> 00:42.830 The antennas that these researchers have used were mainly commercially purchased monopoles and dipoles, which have a relatively low gain. 00:42.950 --> 00:55.650 Other antennas that were used are common TV antennas called Yagi-Uda antennas, which have a much higher gain, but they have impedance of 75 ohms, while most SDR hardware has impedance of 50 ohms. 00:55.650 --> 01:06.990 Without a specific impedance matching circuit, this mismatch causes a significant amount of reflection, and because of that, original high gain of the antenna is drastically reduced. 01:06.990 --> 01:09.990 This will be discussed in the later part of this presentation. 01:09.990 --> 01:20.030 In the meantime, the focus of this presentation addresses a lack of readily accessible, high gain directional antennas operating at frequencies of interest. 01:21.630 --> 01:28.930 SDR is a radio communication system which uses software for modulation and demodulation of radio signals. 01:28.930 --> 01:34.590 SDRs can be reconfigured for a wide range of frequencies as noted in this table. 01:34.790 --> 01:48.490 Despite the software reconfiguration ability, all applications using SDRs require different antennas to operate efficiently at specific frequencies, which are the subset within the supported range of the SDR. 01:48.490 --> 02:00.590 However, a problem occurs when the researcher performs an assessment of a device and requires the use of a specific frequency for which an antenna with high gain is not readily available. 02:02.630 --> 02:11.410 Antennas included in SDR kits such as the HAKRF by Great Scott Gadgets and the USRP B210 from Edis are monopoles and dipoles. 02:11.410 --> 02:16.590 We can see that they are omnidirectional and that they have about 3 dBi gain. 02:17.210 --> 02:20.610 Now we will discuss what exactly that means in the following two slides. 02:20.850 --> 02:27.110 In order to understand what omnidirectional is in terms of gain, we have to first cover directivity. 02:27.110 --> 02:35.710 Directivity is a ratio between an antenna's radiation intensity in a given direction and radiation intensity average over all directions. 02:35.710 --> 02:37.990 On the screen, we can see two antennas. 02:37.990 --> 02:43.150 One is omnidirectional, which radiates in all directions around the antenna in a donut shape. 02:43.150 --> 02:51.130 On the right hand side, we see a directional antenna that primarily radiates in one direction, which is the z direction in this case. 02:51.750 --> 02:54.310 Next, we will discuss what gain means. 02:54.310 --> 02:58.610 Gain is a measure of power radiated in a particular direction. 02:58.610 --> 03:02.950 The decibel is a logarithmic unit used for describing a ratio. 03:03.110 --> 03:07.590 The increase of every three decibels means that the value has doubled. 03:07.590 --> 03:10.550 In this case, we have DBI, which is slightly different. 03:10.550 --> 03:17.250 It shows a value in decibels by which the antenna gain is higher than that of an isotropic antenna. 03:17.290 --> 03:26.590 The isotropic antenna, shown on the right, is a hypothetical antenna that has the same radiation in all directions from one point, the red point in this case. 03:26.590 --> 03:34.690 This is different from omnidirectional, since this one radiates perfectly in all directions, while omnidirectional has a donut shape. 03:34.690 --> 03:42.550 DBI is used as a simplified estimation analysis and provides a reference point for comparison of different antennas. 03:43.030 --> 03:50.710 On the top of this table, we see loops, dipoles, and monopoles, which have a gain of 5 DBI or less. 03:50.710 --> 03:55.110 A theoretical halfway dipole has a gain of 2.15 DBI. 03:55.110 --> 04:02.410 Then we have microscript patches, which are directional, and they have a gain between 5 and 8 DBI. 04:02.470 --> 04:08.750 Then, higher gain are antenna arrays such as Yagi-Uda, which provide 8 to 15 DBI. 04:08.750 --> 04:14.770 Lastly, we have horns and reflectors, which have higher than 15 DBI gain. 04:15.890 --> 04:33.810 Now that we know the basics of antennas, the solution to the problem of not having a high gain antenna at specific frequency is that if we can make patch antennas with directional gain between 5 and 8 DBI, which would be higher than that of any of the regular dipole antennas, 04:33.810 --> 04:45.910 this presentation focuses on bridging that gap by providing a way for researchers to create their own patch antennas without deep electrical engineering experience using the tool created as part of this presentation. 04:46.990 --> 04:50.810 So who am I and how can I help with this solution? 04:50.810 --> 04:57.110 I'm a senior security consulting engineer for Cisco Systems, primarily working on assessing security of IoT devices. 04:57.110 --> 05:00.570 I'm also a PhD student at Virginia Commonwealth University. 05:00.570 --> 05:08.770 I have prior experience in reverse engineering, exploit development, pentesting, TCP IP networking, and over the past few years, software-defined radios. 05:08.770 --> 05:20.610 I have a number of high-value certifications such as CCIE enterprise infrastructure and all of the five old offensive security certifications, OSCP, OSCE, OSWP, OSWE, and OSEE. 05:20.650 --> 05:24.190 I'm also part of the WhatTheFreak CTF team. 05:24.190 --> 05:26.050 We won last year's DEFCON RCTF. 05:26.050 --> 05:28.010 Hopefully we win this year again. 05:29.790 --> 05:37.110 Special acknowledgments for this presentation go to Jonathan Lundquist and Lauren Lincolns, who were instrumental in assisting me with fabrication. 05:37.410 --> 05:39.550 We all work under our advisor, Dr. 05:39.550 --> 05:43.210 Erdem Topsikal, who provided expert guidance on this project. 05:45.980 --> 05:50.820 So now the first question that we need to ask is, what even is an antenna? 05:50.820 --> 05:56.400 An antenna is a device that is used to efficiently transmit and receive electromagnetic waves. 05:56.400 --> 05:59.100 The key word here is efficiently. 05:59.860 --> 06:08.980 The history of antennas started with James Clerk Maxwell, who unified the theory of electricity and magnetism, formulating Maxwell's equations. 06:08.980 --> 06:18.920 The journey then takes us to Heinrich Rudolf Hutz, who demonstrated the first wireless system by observing a spark between a dipole and a loop antenna. 06:18.920 --> 06:33.180 Following that demonstration, a true advancement happened when Guglielmo Marconi sent the first transatlantic transmission between England and Canada in 1901, creating a monopole operating at 500 kHz. 06:33.180 --> 06:43.340 The World War II era introduced more antennas, many of which were aperture type antennas used for communication, radar, remote sensing on both earth and air platforms. 06:43.340 --> 06:50.220 Radar was discovered in 1904 by Christian Hutzmeier to detect ships, but it failed to generate interest. 06:50.220 --> 06:50.680 The U.S. 06:50.680 --> 06:59.080 Naval Research Laboratory, during testing in 1922, discovered that radio waves could be used to detect the presence of enemy ships. 06:59.080 --> 07:06.080 However, only in 1935, Robert Watson Watt built the first practical radar system. 07:06.540 --> 07:10.780 The 1960s had advancements in satellite communication. 07:10.880 --> 07:15.800 Microscript antennas became very popular in the 1970s for spaceport applications. 07:15.800 --> 07:19.540 Today, they are used for all kinds of commercial and government applications. 07:19.540 --> 07:30.700 The 1980s had a widespread use of cellular telephones which was enabled by a form of patch antenna called a planar inverted F antenna, or PIFA antenna. 07:31.580 --> 07:33.820 Common antenna types are following. 07:33.820 --> 07:43.340 Wire antennas such as dipoles, loops, and helix antennas are familiar to all people as they are seen everywhere, on buildings, cars, ships, airplanes. 07:43.340 --> 07:54.640 The microscript antennas were originally used for space applications in the 1970s, but due to the ease of analysis and fabrication, they are used on satellites, airplanes, ships, cars, and mobile devices. 07:54.640 --> 08:03.560 Antenna arrays use multiples of other antenna types to create higher gain and directivity compared to just one antenna. 08:03.940 --> 08:09.940 Aperture antennas such as waveguides and horns are seen on most space and airplane applications. 08:10.160 --> 08:13.100 Reflector antennas look like common satellite antennas. 08:13.220 --> 08:16.960 They were the reason for major success in communication with outer space. 08:20.640 --> 08:27.280 Dipole antennas, shown on the right side, consist of two equal length conductors fed by a transmission line. 08:27.280 --> 08:33.760 The half wave dipole is one of the most common ones and it is named such because it is half wavelength long. 08:33.760 --> 08:37.520 The wavelength for a certain frequency is calculated by the formula on the right. 08:37.520 --> 08:43.080 It is equal to the velocity, which in this case would be speed of light, divided by desired frequency. 08:43.240 --> 08:49.260 Dipoles have an omnidirectional radiation pattern and their communication could be described by the diagram shown on this slide. 08:52.630 --> 08:57.670 Next, we will talk through the design of a dipole antenna for a local TV. 08:57.670 --> 09:11.250 This includes looking up the frequency of local broadcast, which is 560 megahertz, then using antenna calculator tool and specifying half wave dipole and frequency of 560 megahertz. 09:11.250 --> 09:21.350 We get a total dipole length of 26.79 centimeters and each element is half of that, which is 13.39 centimeters. 09:21.350 --> 09:29.310 On the right, we see an antenna that was built and attached to the TV and that local signal was received. 09:29.510 --> 09:41.930 The part right below the dipole elements is a ballon, which is an electrical device that allows balanced and unbalanced lines to be interfaced without disturbing the impedance arrangement of either line. 09:43.270 --> 09:50.750 Dipoles are normally low gain, but if they are set up in an array, then they can have higher and more directional gain. 09:50.750 --> 09:56.150 Most people can recognize Yagi-Uda antennas due to their frequent use as a TV antenna. 09:56.150 --> 10:06.430 The reflector, of which there's usually only one, is made 5% longer than the dipole and its goal is to reverse direction emitted from the rear of the antenna. 10:06.430 --> 10:12.210 Directors, which are 5% shorter than the dipole, focus energy from the front of the antenna. 10:12.210 --> 10:16.650 Adding more directors is effective, but eventually the antenna can become too large. 10:16.650 --> 10:24.990 The interesting note with dipoles and Yagi-Uda antennas is that their impedance is 75 ohms compared to the 50 ohms of software-defined radios. 10:24.990 --> 10:31.910 Some people have used Yagi-Udas in their experiments directly, but the issue there is with mismatch impedances. 10:31.910 --> 10:39.410 The gain will be much lower and the equipment can be damaged with bad mismatch if too much power is reflected back to the SDR. 10:39.630 --> 10:48.970 The reflection coefficient with the 50 ohm transmission line outlined by Z0 and 75 ohm antenna outlined by ZL would be 20%. 10:48.970 --> 10:54.090 This means that the gain of 8 and 12 decibels will be lower. 10:54.090 --> 11:01.750 This issue can be solved by matching transmission lines with various methods, but that will not be discussed in this presentation. 11:01.750 --> 11:04.850 Our main demonstration of this presentation are patch antennas. 11:04.850 --> 11:11.910 Patch antennas are low-profile, simple, inexpensive to manufacture, and have high gain and directionality. 11:11.910 --> 11:20.090 There are many types of patch antennas and this presentation will cover a rectangular patch antenna as that is the simplest to calculate and fabricate. 11:20.090 --> 11:27.690 This section will cover how to mathematically calculate width and length of the patch as well as the feed point of the SMA connector. 11:27.710 --> 11:38.350 We will be covering the transmission line model which is the easiest to calculate but is slightly less accurate compared to more complex models that require the use of very powerful simulation computers. 11:38.350 --> 11:44.670 It does, however , get relatively close to the required design and requires only a few small changes to get desired results. 11:44.670 --> 11:49.550 We will be given a frequency, the electric constant of the substrate, and substrate height. 11:49.550 --> 11:54.770 The most commonly used substrate is FR4 which is used for PCB applications. 11:54.810 --> 12:00.930 It has a dielectric constant of 4.4 and a typical height of 1.6 millimeters. 12:01.030 --> 12:09.010 The electric constant is defined as the ratio of the electric permeability of the material to the electric permeability of the free space. 12:09.270 --> 12:16.070 Other than the material, the only part that one has to choose as part of the design is the desired frequency. 12:17.690 --> 12:22.970 Important mention is that the fields at the edge of the patch undergo fringing. 12:23.050 --> 12:27.530 While the effects are limited, this needs to be accounted for in the calculations. 12:28.010 --> 12:31.870 Next, we can calculate the widths of the patch with the following formula. 12:31.870 --> 12:38.250 V0 in most cases will be speed of light, ER is the dielectric constant, and FR is the frequency. 12:38.250 --> 12:54.210 Effective dielectric constant is the dielectric constant of the uniform dielectric material so that the following figure outlining the effective dielectric constant has identical electrical characteristics as the actual line shown below. 12:55.390 --> 13:01.970 Due to the fringing effects mentioned earlier, electrically the patch looks greater than the physical dimensions. 13:01.970 --> 13:05.150 Next, we will cover strip line width calculation. 13:05.150 --> 13:09.050 First, we calculate A and then use it in the equation on the right. 13:09.150 --> 13:14.930 And if WS over D is less than 2, then we use that strip width outlined by WS. 13:14.930 --> 13:18.970 Otherwise, we would calculate B and use it in the longer equation below. 13:18.970 --> 13:22.170 One of them would be true, but never both. 13:22.590 --> 13:25.870 Now that we know the width, we just need to calculate the inset. 13:25.870 --> 13:28.890 This is the same for both microstrip and probe fed antennas. 13:28.890 --> 13:35.430 This is the spot where impedance is 50 ohms so that it will match the transmission line which is also 50 ohms. 13:37.010 --> 13:46.130 We went through the math just to have complete understanding, but the math in previous slide was automated in an antenna calculator on the rectangular patch. 13:46.290 --> 13:52.450 In this tool, the required antenna type specifications at this time can be either microstrip or probe fed. 13:52.450 --> 13:56.830 Next, we specify the desired frequency and substrate material properties. 13:57.370 --> 14:06.550 Using an example of 2.4 gigahertz with FR4, we get all of the necessary values to design a patch shown in the bottom right-hand corner. 14:06.550 --> 14:16.450 If we use the verbose option, then additional calculations are shown which are intermediate calculations based on the formulas outlined in the earlier part of the presentation. 14:17.490 --> 14:21.390 In this presentation, following manufacturing methods will be outlined. 14:21.390 --> 14:27.930 Copper tape method, chemical etching, PCB milling, CNC method, and lastly professional fabrication. 14:28.430 --> 14:31.750 The first method of manufacture is the simplest one. 14:31.750 --> 14:37.350 The main materials needed are copper tape, single-sided FR4, and an SMA connector. 14:37.510 --> 14:44.690 Using the antenna calculator with PNG output will provide a picture in the size of a letter paper that can be printed out. 14:44.690 --> 14:46.210 It is also to scale. 14:46.210 --> 14:55.710 Taping the printed patch to the copper tape and then cutting them both will save a lot of time since one does not need to draw the calculated measurements on the copper tape before cutting it. 14:55.930 --> 14:59.470 Once the tape has been cut, it can be attached to the substrate. 14:59.490 --> 15:06.690 An SMA connector is attached to the bottom of the the center pin has to touch the microstrip line and it should be soldered. 15:07.890 --> 15:10.170 Next method is chemical etching. 15:10.170 --> 15:15.570 This is very dangerous and it is important to follow all necessary precautions when working with acid. 15:15.570 --> 15:23.870 This includes the use of gloves for handling hazardous chemicals and the use of a fume hood or do the experiment outside with appropriate respiratory protection. 15:23.870 --> 15:37.710 The materials needed for this method are muriatic acid, hydrogen peroxide, double-sided FR4, SMA connector, iron, laser printer, vinyl backing and optionally nail polish and a thermometer or thermal camera. 15:38.130 --> 15:44.450 The first step is to use the antenna calculator with PNG output in the same way we have just used it before. 15:44.610 --> 15:55.050 The patch is first printed on regular paper then the vinyl piece is removed from the backing and the backing is attached glossy side up on masking tape where the antenna was printed. 15:55.050 --> 16:02.770 The paper is then placed back in the printer's import tray and it is reprinted when the printer prints directly to the vinyl backing. 16:02.770 --> 16:07.830 A new template is then removed from the paper and cut down to appropriate size. 16:08.430 --> 16:12.430 This part provided the material for all fabrication methods. 16:12.430 --> 16:14.370 Double-sided FR4 has been purchased. 16:14.370 --> 16:15.550 The thickness is measured. 16:15.550 --> 16:22.930 This is very important since the nominal 1.6 millimeter is not always accurate due to manufacturing tolerances but in this case it was accurate. 16:22.930 --> 16:28.070 The pieces for various experiments were measured and cut with large metal shears. 16:28.570 --> 16:31.230 After that an iron was turned on to preheat. 16:31.230 --> 16:39.210 Using a thermostat or a FLIR camera iron temperature can be checked and it should be between 140 and 158 degrees Celsius. 16:39.210 --> 16:41.530 The temperature does not need to be exact. 16:41.530 --> 16:46.770 Attempt the process at the middle range temperature and if it doesn't work then make the iron hotter. 16:47.670 --> 16:50.810 Next we take the printed patch to the FR4. 16:50.810 --> 16:56.790 We then place it under the iron then place the parchment paper on top and iron for about two to three minutes. 16:56.790 --> 17:03.650 During those two to three minutes pressure should be applied across all of the backing so that it all transfers evenly. 17:05.010 --> 17:08.670 Once the time has passed the iron can be removed. 17:08.670 --> 17:15.870 A flat surface should be placed on top and a weight should be placed on that flat surface to hold the backing down while it cools. 17:15.870 --> 17:22.850 After about two more minutes the weight can be taken off and vinyl backing can be peeled which should reveal that the transfer succeeded. 17:23.290 --> 17:31.310 The same exact method can be used for microstrip patch printing into the vinyl backing and ironing and then completing the design. 17:31.310 --> 17:33.370 Next is the back of the substrate. 17:33.370 --> 17:39.150 We can either use the same method to transfer the black coat or we can just use nail polish. 17:39.150 --> 17:43.630 It should be just a single coat but at the same time no copper spots should show through it. 17:43.630 --> 17:49.050 The nail polish that is necessary for this application is a nail enamel which means it is just acrylic. 17:49.050 --> 17:52.890 It is important because some nail polish needs to be UV cured to set. 17:52.890 --> 17:55.530 Usually cheap nail polish is the one that we need. 17:55.530 --> 17:57.310 Next we get to the chemicals. 17:57.310 --> 18:02.050 To reiterate again it is very important to take safety precautions when working with chemicals. 18:02.070 --> 18:10.490 The most important parts are to use gloves for handling hazardous chemicals, also use goggles and have good ventilation as well as respiratory protection. 18:10.490 --> 18:15.970 It is also highly recommended to have a person spotting in case a spill happens and a medical emergency occurs. 18:15.970 --> 18:24.010 For the actual experiment use a plastic container and put one-third muriatic acid and two-thirds hydrogen peroxide. 18:24.130 --> 18:31.130 The mixture is clear but as soon as copper is placed it will start eating it away and become blue. 18:31.130 --> 18:37.810 If the mixture becomes green that means it is becoming saturated and more mixture should be added. 18:37.810 --> 18:41.930 In about two minutes all of the unprotected copper will be removed. 18:41.930 --> 18:49.510 The acid should be rinsed well with water from both the patch and the gloves at which point gloves should be removed and thrown away. 18:50.030 --> 18:54.570 The nail polish and toner should be removed using acetone which will expose the copper. 18:54.630 --> 19:01.670 After that one should cut two of the four SMA connector legs and the SMA connector should be attached as shown in the picture. 19:01.670 --> 19:08.050 After that solder should be applied to the two legs on the background plane and the center pin on the front. 19:08.050 --> 19:16.030 It is important not to solder center pin to the edges of the SMA connector or you will electrically short the connection. 19:17.450 --> 19:20.970 For probe fed antennas a hole needs to be drilled into the antenna. 19:20.970 --> 19:24.330 Using a drill and a marked spot a hole should be drilled. 19:24.330 --> 19:28.790 After that SMA connector should be connected and soldered on both sides. 19:28.790 --> 19:31.010 The top of the pin should be cut off. 19:31.810 --> 19:36.650 PCB milling method involves the use of a PCB prototyping machine. 19:36.650 --> 19:41.670 Additional parts needed are double-sided FR4 and SMA connector. 19:42.290 --> 19:50.750 This specific mill supports DXF format for importing files and works best with millimeters which is why that is the specified as the unit. 19:50.750 --> 19:59.370 Different mills have different requirements and not all of them support a wide range of units so one needs to verify what exactly is supported and export with supported units. 19:59.370 --> 20:05.530 Using LibriCAD we can see that both designs in DXF show both of the patches accurately. 20:05.950 --> 20:11.370 We start the LPKF CircuitProbe program which initializes the connection to the mill. 20:20.670 --> 20:25.410 Here we go to templates and specify single-sided top. 20:28.580 --> 20:37.980 We then go to file import and specify our DXF file in millimeters. 20:39.180 --> 20:48.500 Under layer templates we change it to top layer and we verify that the imported units are indeed in millimeters. 20:51.330 --> 21:03.250 We then select the patch and convert lines to a polygon followed by assigning the object to layer top layer. 21:03.530 --> 21:09.810 After that we click on the outline and assign objects to layer and specify board outline. 21:13.890 --> 21:18.210 Next we go to toolpath technology dialog. 21:18.930 --> 21:25.850 Under isolate we specify complete rubout and under contour routing we specify corner gap. 21:25.850 --> 21:27.270 We click start. 21:27.350 --> 21:34.830 That calculates how to complete the milling process and it determines which tools are required and whether they are in the machine at that time. 21:35.010 --> 21:41.510 We can see here that different tools will be used for different parts of the process indicated by different colors. 21:43.310 --> 21:46.810 Now we go to the wizard's board production wizard. 21:52.840 --> 21:57.900 At this point we mount the FR4 in the mill and secure it with tape. 22:03.260 --> 22:11.940 On this screen we set copper thickness to 75 micrometers which is a bit higher than the actual but that will account for any imperfections. 22:12.120 --> 22:20.300 The copper thickness needs to be set deep enough that the end mill will remove all of the excess surface copper. 22:20.640 --> 22:29.100 At the end the mill will go through the FR4 and into the sacrificial board underneath to ensure a clean cut around the edges. 22:29.100 --> 22:30.540 We click continue. 22:33.040 --> 22:40.000 On the next screen we have the center placement which in this case is exactly where the specific point is located. 22:40.080 --> 22:44.080 So we just click set center to ensure it still starts there. 22:47.190 --> 22:51.550 We'll also save the document in case we need to redo this part at some point. 23:08.240 --> 23:14.200 Now the mill does the tool exchange and the tool adjustment followed by a warm-up of the spindle motor. 23:14.540 --> 23:18.460 At this point mill goes through the process and completes the milling. 23:30.890 --> 23:37.110 At the end we just need to solder the SMA connector using the same method as in the chemical etching process. 23:38.070 --> 23:45.710 In order to have a variety of tests we found that the university workshop in the library has Inventables X-Car CNC machine. 23:45.710 --> 23:49.030 We asked if we could use it for a conference to make an antenna. 23:49.030 --> 23:53.750 The manager Ken Hobson was very helpful and assisted us with a very short deadline. 23:53.750 --> 23:58.410 This machine uses DeWALT 611 and it can certainly be used for this purpose. 23:58.410 --> 24:04.850 End mill bits are different from common drill bits in that they are used on top of the material rather than drilling straight through it. 24:04.850 --> 24:11.290 The cost of this mill is about $2,600 and cost of the enclosure in the vacuum is between $500 and $1,000. 24:11.490 --> 24:17.050 All of it is open source and a machine like this could likely exist in a local makerspace. 24:17.750 --> 24:26.210 The ESOL software used with the CNC machine uses inches as units so DXF file was generated with inches as the unit. 24:26.210 --> 24:33.930 This machine is slightly different from the specialized PCB milling machine in that it only uses one tool for the entire design at a time. 24:33.930 --> 24:41.850 Multiple tools can be used but the design would have to be broken up in multiple projects which use different end bits for different parts of the project. 24:41.850 --> 24:51.750 In this case due to limited time availability before the conference only the probe fed design was milled out with the CNC machine which did not require different end bits. 24:52.230 --> 24:54.550 This is the ESOL software. 24:57.060 --> 25:02.180 It uses a DEWALT 611 router as shown here. 25:03.200 --> 25:08.000 Under general settings we can see different settings for the CNC machine. 25:08.800 --> 25:13.740 Under material we specify PCB which is FR4. 25:13.740 --> 25:19.880 The cut FR4 was measured and the units were placed under width length and thickness. 25:19.880 --> 25:29.020 A comment on the 1.5 mm thickness is that it was found to work better for this specific setup even though the material was measured to be 1.6 mm. 25:29.020 --> 25:32.760 The bit used here was 1 16th of an inch. 25:36.880 --> 25:41.600 The following settings were experimentally determined for the feed rate and plunge rate. 25:41.600 --> 25:58.180 Now the copper thickness is in reality closer to 0.04 mm but on this setup a higher number was used due to an older drill bit so maximum in this case would be 0.25 mm and the x-axis option was found to work best and it was used for the fill method. 25:58.320 --> 26:03.180 We then import the DXF file with lines joined and on path. 26:10.590 --> 26:12.990 We then move it slightly offset 26:22.680 --> 26:27.280 selecting the outline and specifying depth as 0.2 mm. 26:32.530 --> 26:38.790 Making a clear out pocket as well as sending it to the back will ensure that the copper is removed. 26:38.790 --> 26:45.550 Next is the patch which will also clear out a pocket but with 0 mm as we do not want that part touched. 26:50.120 --> 26:52.380 Last part is the probe feed. 26:52.380 --> 26:59.300 It currently does not show up so it needs to be adjusted to 0.6 by 0.6 mm so we can see it. 27:06.590 --> 27:11.890 But as our bit is a bit bigger than that hole we'll remove it and drill it manually. 27:22.920 --> 27:25.440 Now we can start the process of carving. 27:25.440 --> 27:30.200 We measure the thickness and clamp down the material as well as confirm the bit. 27:30.300 --> 27:39.300 In order to confirm a new work zero we use the keyboard and move the router with appropriate intervals between 0.01 and 10 mm. 27:45.520 --> 27:48.680 The bit needs to touch the substrate. 28:11.230 --> 28:17.290 A trick is to use a piece of paper and adjust by 0.01 until the paper cannot be moved. 28:17.290 --> 28:22.850 We then raise it by 0.01 and lower it down again with the paper removed. 28:36.960 --> 28:40.340 We then confirm the new position and raise the bit. 28:47.470 --> 28:51.670 Once the bit was raised we can turn on the spindle and inspect it. 28:55.240 --> 28:58.300 We say that it's on and we let it carve. 29:02.350 --> 29:05.810 Now we can see that it did not drill deep enough. 29:11.720 --> 29:21.340 So we stop it and adjust the depth by 0.1 mm which is the minimum that we can adjust it by on both cut of the substrate and cut settings. 29:38.220 --> 29:42.840 We go through the same process and use the last position for the home location. 29:46.570 --> 29:50.930 Now this one did not do a good job as well since it missed a part. 29:50.930 --> 29:54.770 We stop it again and adjust it by 0.1 mm again. 30:32.220 --> 30:34.040 Now it does a really good job. 30:34.040 --> 30:36.280 It completed the milling process successfully. 30:36.280 --> 30:46.660 The part here that I wanted to show is the troubleshooting process rather than trying a bunch of times until it worked perfectly with 0.1 or 0.2 mm which is the ideal sort of part. 30:46.660 --> 30:56.660 The issues can be uneven sacrificial boards underneath, uneven clamping, potentially a different bit that was required and many other issues that are commonly encountered when using a mill. 30:56.680 --> 31:01.700 The last fabrication method is the professional fabrication such as one offered by Oshbar. 31:01.700 --> 31:04.500 They offer three copies for $5 per square inch. 31:04.500 --> 31:07.760 The supported files are KiCAD, EagleCAD, and Gerber. 31:07.760 --> 31:09.000 I'm not sponsored by them. 31:09.000 --> 31:11.780 I'm just showing one of the examples that was recommended to me. 31:12.280 --> 31:18.500 These formats are currently not supported due to time limitations but it is planned in future releases of this tool. 31:18.500 --> 31:24.220 A workaround is that DXF files can be converted using KiCAD 6 PCB editor. 31:24.220 --> 31:30.240 Using the antenna calculator by specifying meter as the unit of measurement is how the DXF file is generated. 31:30.480 --> 31:36.500 Within the PCB editor under file import graphics is where DXF file is imported. 31:36.500 --> 31:42.600 Interactive placement is selected and import scale is one and all of the other units are in millimeters. 31:42.600 --> 31:46.520 This will appropriately import the patch with accurate millimeter dimensions. 31:46.520 --> 31:53.940 From here we can export it as a KiCAD model or Gerber which can be used at the website previously mentioned. 31:54.980 --> 32:00.180 After the antenna is built it should be measured and evaluated whether it matches the original design. 32:00.180 --> 32:09.220 As mentioned previously the transmission line model that is used here is the easiest to calculate but it is slightly less accurate compared to more complex models. 32:09.380 --> 32:15.680 This is where for a given frequency small changes may be necessary to get it to perform exactly as expected. 32:15.680 --> 32:19.940 On the left we have the S11 parameter which shows the reflection coefficient. 32:19.940 --> 32:23.580 The lower the number the better the antenna performs at that frequency. 32:23.580 --> 32:28.400 In order to be considered an antenna at that frequency it has to be at least negative 10 dB. 32:28.400 --> 32:36.060 On the right hand side we have measured the gain of the antenna in an anechoic chamber and found it to be 5.12 dBi. 32:36.120 --> 32:44.280 This same measurement can be done using two nano VNAs with identical patch antennas separating them exactly by a meter and using Friis equation to calculate the gain. 32:44.280 --> 32:50.940 If both antennas are exactly the same an assumption can be made that both of them have the same gain which will simplify the equation. 32:50.940 --> 32:58.180 In this case you can use the calculator and match both antenna gains in order to get the observed power received which will be known. 32:58.280 --> 33:01.400 It is important to check your local laws for transmission rules. 33:01.400 --> 33:08.080 The ISM bands can be used without a license so one can transmit and receive at those frequencies in most countries. 33:08.080 --> 33:14.960 If one is attempting to test a device at very specialized frequencies then the use of an anechoic chamber is suggested. 33:14.960 --> 33:28.620 An anechoic chamber is a shielded room that has radio wave absorbing material applied to wall ceiling and floor so all the transmissions stay inside the chamber and at the same time the chamber is shielded from outside interference. 33:28.820 --> 33:35.360 In conclusion the presentation showcased an antenna calculator that can help antenna design and several fabrication methods. 33:35.360 --> 33:38.540 Rectangular patch antennas are simple and easy to manufacture. 33:38.540 --> 33:42.520 I encourage you to try some of these manufacturing methods and make your own antennas. 33:42.520 --> 33:53.800 The only thing that I ask is that you share with the community and if throughout the process you have any additional requests or ideas feel free to make an issue or a pull request and I will be happy to look into it and further improve it. 33:53.800 --> 34:04.340 Here are some of the references and some specific books that I used to generate some of this presentation and of course if you have any questions feel free to post them in the comments and I will try to answer them to the best of my ability. 34:04.340 --> 34:05.220 Thank you so much!