I need to test and maintain a batch of devices which contain GPS modules. I can work on the rest of the device, but when it comes to test the GPS part I have no signal on my desk, no matter how near I try to get to the window. I just need to check that the receivers get proper signal, acquire position and get a S/N Ratio I can compare with other working units.

I could place an antena on the roof (I've actually got one), but the devices have no antenna input (and they don't support active antennas anyway). Since I'm going to do this task for quite some time, I've thought a reradiating GPS antenna is what I needed. However, here in Argentina they're practically unheard of. No specialized shop has them, so I'll have to build my own. Even worse, in Argentina is quite difficult to get most of the specialized electronics components (think of a list of no more than 200 most common transistors), and importing from overseas is practically impossible. Hey, even 1/8W size through hole resistors are getting difficult to find! (only 1/4W 5% are available).

So, what can I do? I've got a couple of commercial PCB active and passive GPS antennas. I tried feeding the active antenna with 5V through an inductor, and connecting the output to the passive one through a small capacitor (1nF). The active antenna is supposed to have a gain of 28 dB, and I've got (by spec) at most 2 dB of losses in the cables and connections. I'd thought it would work but I cannot see any improvements in the received signal at all. The active antenna is on the roof, completely obscured from the room I'm into, so I don't think the antennas could interfere with each other. The power supply is a 7805 fed from a 12V battery (so I know that no strange noises are coming through the power rail).

The active antenna is not the problem. I can make it work connecting it to a receiver that does support active external antennas and it acquires position all right.

Perhaps I should install a second LNA stage, near the radiating antenna? What should be its gain? I have barely any measurement equipment. I certainly have nothing useful to measure a 1.5 GHz signal at all, so I'm working blind here. A sample circuit with very few components would be specially useful.

Any help would be appreciated.

EDIT2: Added crappy photo of my board

• You can try antenna magus (software for calculating antennas) and make it. Feb 12 '15 at 23:20
• Mmmm..... I could start studying quantum physics and deduce the rest from there. Sorry, I couldn't help myself. I think that option is out of my reach, but thanks. Feb 12 '15 at 23:33
• What do you mean by feeding the antenna with 5V? Please post schematics. Feb 12 '15 at 23:37
• The 5V are meant for the LNA in the active antenna. Here's the schematic. As you see, it's pretty basic... Feb 13 '15 at 0:30
• What you need is called GPS repeater. If you google it you'll find several available commercially. Feb 13 '15 at 2:12

A GPS repeater is what you need, and you are most of the way to making one that works.

# Overall design of the repeater

will be something like this:

• Patch antenna outdoors, to receive nice strong GPS signals
• Amplifier (in the patch antenna)
• Cable to bring signal down to your workbench
• Bias T to inject DC for the amplifier

The devil is in the detail, and I might be able to help with a some detail. Starting with an example circuit and refining it.

## First calculate the path loss

Compare the two scenarios:
1) Your Device, outdoors: Satellite --> Device
2) Your home-made repeater: Satellite --> Patch --> LNA --> Cable --> Patch --> Free Space Loss --> Device
The difference is of course the antenna, amplifier, antenna and free space path loss, marked in bold. The first step is to add up the impact that has on the signal strength seen at the receiver.

Antenna gain, for a good quality, fairly large (25 mm), passive ceramic patch, is about 3 dBi on boresight. The more compact 10 mm patches, and the ceramic F antennas are not great for this application.
The outdoor antenna has a built-in LNA and filter, you only need to supply power on the cable.

Path loss is wavelength^2 / (2*pi*r)^2.
At 1575 MHz, it is as follows:

 0.1 m : 16.4 dB
0.25 m : 24.3 dB
0.5 m : 30.4 dB
1.0 m : 36.4 dB


Cable loss will be about 5 dB for a 5 m length of RG174, the usual cable on these antennas. You can do better with a thicker cable...

The indoor patch antenna can be the same type, but the LNA and SAW filter are not required. Carefully de-solder them and bridge across with some thick wire or a bit of copper tape.

Assuming you keep the re-radiating patch about 25 cm from the device under test, the sum of the losses and gains above is then

• Gpatch +3 dBi
• Glna +27 dB
• Gcable -5 dB
• Gpath -24.3 dB
• Gpatch +3 dBi

This means that if all the assumptions above are correct, the signal level seen by your device under test will be slightly stronger than what it would see outdoors.

Any deviation (or mistakes) will quickly push the signal level down below where the receiver can cope with it. This could be due to a longer cable, lower gain antenna, mismatch or loss in the bias-T, or longer range from transmit patch to receiver.

Possible improvements

If you have an unavoidable loss in the circuit, or you need a much greater distance from transmit patch to DUT, then you'll need a second LNA, perhaps scrounged from the second patch antenna, powered with a second bias-T.

This will get very messy unless you have a very good clean bias T. Any RF leakage into the power lines, or radiation, could couple back into the LNA input (where the first bias T is), causing feedback, oscillation, and complete failure.

A bias T looks like this: but MUST have a capacitor to ground from the inductor. (I'll update with a better circuit later.)

You've made a good attempt at the bias circuit, but I fear that the inductor and particularly the capacitor are too far away. We'd generally use a chip inductor and capacitor, 2 mm long, which doesn't disrupt the transmission line too much. I'm also not sure that the impedance of the line is close enough to 50 ohms, it might be costing you a few dB.

What to do? Just use a resistor.

The LNA in a typical GPS antenna I see says Power Supply : 1.8-5.0 VDC at 10mA. So if you supply it through a 220 ohm 0805 chip resistor, the voltage drop will be only about 2.2 V. Just make sure you have a few volts more than the minimum required, and the LNA will be fine. The resistor will absorb some of the GPS signal, but not enough to change anything here. You still need a bypass capacitor, C2 in your circuit, but make it a 10n ceramic and keep it as close to the resistor as possible. It should all fit between the legs of the SMA connectors.
Test the circuit DC voltages when powered and running a GPS antenna, and adjust your source voltage accordingly.

Solder the two SMA connectors back-to-back or use a short run of PCB. Try to keep the impedance close to 50 ohms, that is, a 3 mm microstrip track on a 1.6 mm FR4 copper board, with ground underneath.

Note that this circuit also powers both antennas at the same time. This shouldn't be a problem, as long as the indoor antenna is not DC grounded. Check this first, if it's a short, you need to add a capacitor in series with it somewhere.

Having built this bias circuit, you can test it with a regular GPS receiver that has an SMA connector. Simply compare the signal strength of the GPS satellites in the two cases:

• with the GPS antenna directly connected to the receive
• with the GPS antenna connected to the receiver through your bias circuit

Of course you don't add any power to the bias circuit, let the GPS power the antenna.

# Final Thoughts

Experiment a bit - put the transmit patch touching your circuit, then move it away slowly.

Don't expect this circuit to give you good-quality GPS signals for characterising the device's own GPS antenna - for that you probably need to be outdoors. The environment is different, all signals come from above, and there is a warm room not cold sky contributing noise in a different way to the receiver. This is more for a pass-fail test of the GPS, and the circuits that depend on it.

• Of course! I guess I underestimated the cable and air path loss. I'll see what I can do about it, but it will not be very easy. Thanks for a good explanation that gives me something solid to work with. Feb 22 '15 at 21:01

is your building a metal building? if it is, getting the native gps signal in maybe impossible.

If your building is not metal framed, and the window is not metal framed. you can try this and it's pretty cheap and might just work.

get 3-5 pieces of large cardboard and cover them up with aluminum foil, and place them outside the window and in different angles so that they reflect the overhead signals into your window. they need to be spaced loosely apart and aiming different angles of the sky. if you want to make them more effective, try to make them into parabolic shape.

• Thank you, that's good out-of-the-box thinking, but not at all possible in my case. Feb 17 '15 at 23:19

There are both active and passive GPS repeaters available from many sources. The main downside as user3453518 said, is that you will measure the roof antenna's coordinates.

An active GPS repeater will give you anywhere between 12 and 28dB gain, and if you use a stationary antenna, you will get much better performance than with a small patch antenna.

• As I said, buying a repeater is out of the question. They're not available in Argentina and Argentina has serious limitations regarding the import of goods (of any kind). Feb 14 '15 at 1:48

I would decrease the gap between the two SMA to board launch connectors as much as possible. You want to get the distance between the two to be much less than 1/10 of a wavelength below which the near-field effects dominant; at 1.5 GHz this is approximately 2cm. Judging by your thumb in the picture I'm guessing that the internal distance between the two SMA launchers is approximately 2cm.

I would go so far as cutting off the pins on the SMA launch so that there is just enough of a stub to attach the RFC inductor and the blocking capacitor. Furthermore I would get rid of the blocking capacitor if the indoor antenna is already open-circuit at DC. If possible, I would also advise moving the RFC closer to the signal trace, while it should look like an open-circuit at 1.5 GHz transmission line effects could modify the effective impedance; for example, an open-circuit seen through a quarter-wavelength long transmission line will look like a short-circuit.

I would considering adding an RFC to the ground return on your 5V supply, it is possible that due to the lack of a good ground plane this is influencing the circuit. Adding a grounded metal shield box around the whole contraption will also help limit the possibility of an external source adversely coupling into the transmission line.

This wont work as you are expecting.

It might work from a electrical standpoint. But having the satellite signals travelling a cable and being reiradiated will introduce error into the GPS read signal. Cant grasp all possible problems from pure deduction, but, you will probably make a fixed measure from the position of the active antenna, not from the position of the gps device. As the active antenna is powered, it will overcome the signal from sats in a certain radius. In other words, you end up making a certain area give a fixed result. Or something like that.

As time of arrival is crucial to GPS positioning, you need the antenna to be in the same place as the receiver, move with it. If you use an external antenna in a car, this will not interfere with the result, because you expect to measure the position of the car, not of the device. But on a building, whats the porpuse ?

You might very well go to the roof, read the position, write it into a piece of paper and stick it to a wall, it wont change, the building is going nowhere, it wont move around. By having an antenna on the roof, you achieve nothing.

• I don't think this is a problem. As far as I have understood, he just wants to test if the GPS works, it does not matter if it is a few meters wrong. Feb 13 '15 at 23:21
• This makes a lot more sense now. Feb 14 '15 at 1:45
• Gleison Storto is right. I don't care about position accuracy at all. I just want to test if the receivers acquire any position at all. I know where my desktop is located. ;) Feb 14 '15 at 1:46
• Well, hi doesn't give mucho info about his first inductor, but the inductor in the next page (gpsnuts.com/mygps/gps/LEI%20tips%20tricks/Active%20antenna/…) is well specified and the online calculator gives me around 10 µH. It must not match the capacitor since the only use for it is to prevent the RF to divert into the power stage. Feb 14 '15 at 13:06
• @user3453518 I think you should be more specific when you say a capacitor in series may "alter the signal". It would be possible if he used a capacitor with long terminals, which would have a large parasitic inductance. As he is using a SMD capacitor, I don't think it is a big deal at all. Even the distance traveled between the cap terminals is much smaller than the signal's wavelength. Feb 15 '15 at 16:56

Another solution, but perhaps not feasible for you based on your statement about the availability of electronics, would be to use a GPS simulator. This is a test instrument that radiates the GPS signal that you would receive at some position or set of positions that you specify.

• I think these are impossible to get. You need clearance from US Military, AFAIK. Anyway, they're not available in Argentina. Feb 20 '15 at 23:16