how to measure the ouput impedance of cheap SAW transmitter modules

[gardner]

I have been experimenting off and on with little 433MHz SAW oscillator transmitters for remote data collection.  I've got several radios and the good ones have datasheets with useful characteristics documented.  The cheap ones that come in a bag of 10 for $5 have nothing.  I've noticed a good deal of variation in the different available units and I am able to characterize the power output, roughly, with a cheap AD8307 board from Amazon, but also using a reference antenna and an RF-Explorer.

What I am hoping to do is to come up with a way to characterize the output impedance of the little SAW boards so that I can devise a more ideal matching network than simply soldering a wire onto the ANT pin.  I am at a loss how to go about measuring the output impedance and I am looking for pointers how to do that.

My opening gambit is to make small resistive terminators of 10R, 20R, ... 300R and try each out whilst measuring the "power" output with the AD8307 board.  Its input impedance is 51.6R at 433MHz according to my NanoVNA, FWIW.

Is there a simple antenna tuner I could make up to allow a more fine-grained approach to finding the output impedance?  How about some other method to measure it that I don't know about?  I have a NanoVNA V2, an RF-Explorer, a 100MHz scope and an AD8307 board.  I don't have a lot of other fancy RF gear though.  At UHF, the inductive effects of breadboards and through-hole parts and so forth is also a factor.

Thanks.

[rf-fil]

Assuming your goal is to get the most range, so to get the most RF power out into the antenna.. it's most practical to do it empirically. Make up a little matching circuit, with a small inductor 10 - 20 nH range, and a trimmer cap, maybe 20pF. Then try different topologies of an L-matching circuit. The inductor would be in series with the antenna and the trimmer would go to ground on either side of the inductor. You might need to try different inductors. The end point is measuring the signal strength on the receiver that is at least a few wavelengths away, or a spectrum analyser, etc.

If you have the schematic diagram, you could build a harmonic balance model, but unless you have access to ADS or Microwave Office, that's not an option.

Why? Many of the cheap SAW modules are just an oscillator. Its output impedance is not well defined. It may actually be something weird, and one typically doesn't impedance match an oscillator output, because that may load it so much that it stops oscillating. If the transmitter has a little amplifier stage, it's the same story actually. An RF PA output impedance, even if it's tiny, is something very small and "weird". So one doesn't technically match the impedance of the amplifier. You provide a load to the amp (collector or drain) that produces the desired power output. The term to look up is "load pull", which is how you find the performance of a PA for different loads. I mean gain, output power, efficiency, for different load impedances. Then you chose the load impedance that produces your desired performance goal, and then design a matching circuit that transforms your antenna impedance (or 50 ohms), to the desired load impedance. Load pull is just a fancy way to empirically tune the output and observe the performance. Far too expensive for this use case.

Actually, if you want more range, it's probably easier to just add a small RF amplifier stage - as long as its output power is within the legal limits in your area.

[gardner]

Thanks for your ideas, they are helping me for sure.

I should have clarified that my goal is, yes, to optimize range probably via best power transfer, but I want to evaluate what antenna loads are safely driven with oscillator stability, noise and start-up characteristics.  One of the puzzling and frustrating problems with the cheapest radios is the oscillator start-up time which can be 5 us up to 25 us on some units, and this is one of the greatest sources of link reliability problems because it fairly arbitrarily changes the OOK signal timing.  I want to characterize whether any of those are a function of antenna load or matching impedance.

I think you're right that some sort of empirical approach is going to be the most useful.

As you say, the cheap modules are little more than an oscillator with the antenna just coupled onto the emitter of the oscillating transistor.  Someone posted this to stackexchange and I suspect its about what I am working with.  I have no doubt the output impedance is "weird" and loading could effect all sorts of properties from frequency to temperature and voltage stability.

[rf-fil]

One tip would be that oscillator startup time depends on the loaded Q. In your example schematic, there would be some degree of control by tuning C1 and C3. The loaded Q is a function of the SAW resonator Q, the circuits that couple the resonator to the transistor, and the (non-linear) impedances of the transistor itself.

[rf-fil]

Ultimately, it may be easier to grab one of these: https://www.analog.com/en/parametricsearch/12968#/. They're not that expensive, and will outperform the simple SAW based designs on basically every metric, including the cost of your time ;-).

[danymogh]

Hello,

I too am struggling with this design. I have to design the transmitter from scratch for my own application and can't use the ready-made modules.

I obtained some key FOBs from different vendors and compared them and they're all almost the same. I designed my own circuit based on one of them and what I could find online and I managed to only get it working. the output power is extremely low. (RX module doesn't receive anything more than 1 meter). here's my circuit diagram:



and here is my PCB



during my tests, i accidentally found out that if I stick a small wire in the Via left of C1 the output power increases significantly (almost like vendor FOBs) however I can't figure out why this is happening. I've read many datasheets and none of them point this out. I know the value of C1 and C2 is significant but whatever I try the output power is still low unless I stick the wire and hold it with my hand.

Also, I can't make sense of the length of the antenna. for 433Mhz it should be like 8.65 cm for 1/8 of the wavelength or 4.3 cm for 1/16 wavelength but none of the vendors have their antenna at this length. So I suppose the length is not related to antenna wavelength matching? most of the vendors are having a length of 53mm and a width of 1.7mm.

any suggestions?

[gardner]

I could be wrong, but I wonder about the physical placement of C3.  I think the thick C-shaped trace is meant to be the radiant antenna, and C3 should be physically connected to the opposite end of it near where L1 connects.  Instead it connects very close to the where C1 does, leaving only a very short section as a radiator.  OTOH, the trace may be meant to be some species of J-Pole with feeds at weird points along it.

It's all UHF RF magic, so it's outside my actual knowledge.

[danymogh]

You might be right. I'll test it tomorrow.

To my understanding, the Antenna is actually a very small inductor. all vendors design it that way. my design is very similar to a reference design of Microchip. Since holding the wire with my hand boosts the output power, I think it must be my body capacitance affecting the circuit. but which capacitor, I don't know.

[RFDx]

here's my circuit diagram:
and here is my PCB

Besides being unsuitable for a UHF circuit, the layout doesn't match the schematics. As an example, C3 and L1 are interchanged and C3 is shorting the antenna coil to ground. RF ground is just a thin wire. I'm surpised the oscillator works at all.

So I suppose the length is not related to antenna wavelength matching? most of the vendors are having a length of 53mm and a width of 1.7mm.

In this case the length of the antenna coil is not related to the wavelength. The coil inductance needs to resonate together with C1,C2 and the parasitic capacitances of the transistor at 433MHz.

[danymogh]

here's my circuit diagram:
and here is my PCB

Besides being unsuitable for a UHF circuit, the layout doesn't match the schematics. As an example, C3 and L1 are interchanged and C3 is shorting the antenna coil to ground. RF ground is just a thin wire. I'm surpised the oscillator works at all.

So I suppose the length is not related to antenna wavelength matching? most of the vendors are having a length of 53mm and a width of 1.7mm.

In this case the length of the antenna coil is not related to the wavelength. The coil inductance needs to resonate together with C1,C2 and the parasitic capacitances of the transistor at 433MHz.

@RFDx
I know there's something wrong but it's not C3 and L1 being interchanged I tested that. if I stick a wire to the Via left of C1 the output power is increased. if I hold the wire with my hand the power is even more increased. So since the body has capacitive effects I think it has to be from either a missing capacitor or inductor or untuned values.

can you clarify what should be a better layout?


@gardner
I tested that and the output shut down completely.

[RFDx]

I know there's something wrong but it's not C3 and L1 being interchanged I tested that.

I can only tell you what I see. In the layout you posted, C3 is connected to the tap point of the coil and is grounding it. L1 is connected to the end of the coil and is therefore in series with it. This is the exact opposite to what the schematic shows.

If you replace the SAW resonator with a 1nF capacitor the circuit must oscillate on its own at ~433MHz. Does it do that? The purpose of the SAW resonator is to stabilize the frequency by grounding the transistor base only at 433MHz.

can you clarify what should be a better layout?

You mentioned using a reference design from Microchip as a template. Did you study the layout of the reference design? A long and thin "ground" wire, that connects components together, is not appropriate at RF.

Take a look at the following link:

https://www.sunrom.com/p/rf-transmitter-433-mhz-ask

It's another small, simple, single stage, SAW stabilized grounded base Colpitts oscillator/transmitter design similar to yours. Instead of using the coil as an antenna, a small amount of power is extracted from the collector through a small capacitor for an external antenna. Connections between components are as short as possible. Every unused area on top and bottom of the PCB is flooded with copper to be used as ground. Top and bottom grounds are merged together with the help of a lot of vias.

[danymogh]

I know there's something wrong but it's not C3 and L1 being interchanged I tested that.

I can only tell you what I see. In the layout you posted, C3 is connected to the tap point of the coil and is grounding it. L1 is connected to the end of the coil and is therefore in series with it. This is the exact opposite to what the schematic shows.

If you replace the SAW resonator with a 1nF capacitor the circuit must oscillate on its own at ~433MHz. Does it do that? The purpose of the SAW resonator is to stabilize the frequency by grounding the transistor base only at 433MHz.

can you clarify what should be a better layout?

You mentioned using a reference design from Microchip as a template. Did you study the layout of the reference design? A long and thin "ground" wire, that connects components together, is not appropriate at RF.

Take a look at the following link:

https://www.sunrom.com/p/rf-transmitter-433-mhz-ask

It's another small, simple, single stage, SAW stabilized grounded base Colpitts oscillator/transmitter design similar to yours. Instead of using the coil as an antenna, a small amount of power is extracted from the collector through a small capacitor for an external antenna. Connections between components are as short as possible. Every unused area on top and bottom of the PCB is flooded with copper to be used as ground. Top and bottom grounds are merged together with the help of a lot of vias.

I see the problem now. thanks a lot for the tips. In layout, I was confused about the place of the inductor as the antenna. some vendors had it connect some components in the middle of it and I tried to copy them and it's evident that I've failed.

I tried to replicate your suggestions on the faulty layout as much as possible and the results got better. but I'll have to redesign the whole thing again to achieve better results.

the microchip layout didn't have any ground planes but the track length was a bit thicker.

[danymogh]

I made 2 other versions.

one is PLL-based and is working fine. the other is saw resonator-based and is not working at all.

I tried to copy the microchip design.




I'm not getting any output at all.
the previous design was at least giving me something.
any ideas?

[RFDx]

I'm not getting any output at all.
the previous design was at least giving me something.
any ideas?

According to the layout you unintentionally put ~10nH inductors (long, thin lines) in series with the transistor base and collector. As mentioned, you check/align the oscillator for ~433MHz by grounding the transistor base through a 1nF capacitor.