Author Topic: Small loop antennas @ 433 Mhz  (Read 26099 times)

joelucid

  • Hero Member
  • *****
  • Posts: 869
Small loop antennas @ 433 Mhz
« on: May 23, 2016, 03:52:58 PM »
As you know from this thread (https://lowpowerlab.com/forum/index.php/topic,1898.0.html) I've been toying with small antennas and in particular the small loop antenna. I think it holds a lot of promise for good efficiency at very small form factors.

Well now - after purchasing a small vector network analyzer - I have had the first successes today. The end result is ok, not great yet. But I think there is still a lot of room for improvement.

I went for an inductively coupled antenna, where a small non-resonant loop drives a larger resonant one:





This approach allows fixing the resistive impedance mismatch between the 50 Ohm radio module and the very low resistance resonant loop.

For the resonant loop I used 10mm copper tape. I left a gap which I first covered with adhesive tesa tape. I then covered the tesa film with a small piece of copper tape, effectively creating a capacitor:



It takes a lot of cutting and measuring to get the resonance of the resulting LC circuit just right onto 433 Mhz.

It's very possible to reach a return loss of -16dB here, which is very respectable. However getting the frequency right is not easy. The antenna bandwidth is just very small:



Lots of things change a bit when you disconnect the analyzer, so I think I need to do some further tuning based on RSSI values. Still - I already get reception all throughout the apartment with this setup. The RSSI is about maybe 15 lower than what I'd see with a Moteino with wire antenna. I suspect we can do much better than this.

Also this antenna still has a diameter of 4cm and I think one could go down a bit there to 3-3.5cm. However one problem is that the 2cm coin cell does block the magnetic field in a portion of the antenna. So going smaller loses gain in this setup even faster than with small loops by themselves (which already is with A^2). Possible solutions are going with the cr1632, which leaves more space for the loop - potentially even mounting the board perpendicular to the loop axis.

Still wanted to share this as a promising direction.

Joe 
« Last Edit: May 23, 2016, 03:57:43 PM by joelucid »

captcha

  • Full Member
  • ***
  • Posts: 108
  • Country: au
Re: Small loop antennas @ 433 Mhz
« Reply #1 on: May 23, 2016, 08:02:45 PM »
Very impressed with the direction you're heading. You mentioned that loop antenna's are very narrow banded and I'm wondering how much of the design can be easily reproduced by others. The antenna's performance may also be very much influenced by what's inside (and metallic objects just outside) the loop.

Comparing your design with common HF designs I noticed that your inner loop has a relatively large circumference. Also, the tuning capacitor would have to be tiiiiny and may already be detuned by changes in temperature or humidity.

One big benefit to the loops is that they are much better on signal to noise ratio, so a design well worth pursuing..

I look forward to further posts as I've been toying around with making dipoles on pcb's just so that they're reproducable and on a rigid material. Oh, well.. back to 4NEC2.. :-)

joelucid

  • Hero Member
  • *****
  • Posts: 869
Re: Small loop antennas @ 433 Mhz
« Reply #2 on: May 24, 2016, 03:07:12 AM »
Quote
You mentioned that loop antenna's are very narrow banded and I'm wondering how much of the design can be easily reproduced by others.

I agree. Kind of hard to tune it without VNA. On the other hand once I have a working design I could see others build it based on exact specifications. Maybe I would have to laser-cut the cap tape so it is just right.

BTW, I measured a monopole on the same Tiny board yesterday, too. And it behaves very similarly when you make it very short. It's a well know fact of antenna design that small efficient antennas are very narrow band. To really make them work well they need be tuned.

Loops are less susceptible to the environment since they are magnetic antennas. Which is one of the reasons why I think you'd rather want a tuned loop than a tuned stub.

Quote
Comparing your design with common HF designs I noticed that your inner loop has a relatively large circumference.

Yeah, typically one uses 1/5th of the greater loop size. Now whether I have to adjust for the object inside the loop - I'm not clear. I think probably not though. I do think that having the mote inside the loop will affect efficiency significantly. The changing magnetic field will induce eddy currents in the ground plane and the coin cell. Which is why I would prefer to mount the pcb vertically - but even with cr1632 that's 1.6 cm of height, yuck.

A big question is whether it's better to have a smaller loop with pcb outside or the larger one as I've done it now. The upside of the current design is that you use every square mm of area to maximize loop size. But eddy currents are going to be larger inside because of the larger field strength. It will require more testing.

Quote
Also, the tuning capacitor would have to be tiiiiny and may already be detuned by changes in temperature or humidity.


Luckily plastic film capacitors are among the most temperature stable, so hopefully at least on that front we should be good.

Being fixed on one frequency brings a lot of disadvantages obviously: No frequency hopping, no switching bands for listen mode or OTA etc. There are now some low loss digitally tunable capacitors (http://www.psemi.com/products/digitally-tunable-capacitors-dtc for example). But they are super expensive and still too high ESR for the low resistance resonant loop. So I just don't think they will work.

I also think these lastertrim caps are kind of cool (http://www.johansontechnology.com/lasertrim-rf-tuning). Although probably a bit too involved for a hobby setup and frankly also not high q enough.

You can always add some resistance to make the antenna more broad band. I think that's what Felix has done in his design. Unfortunately you pay for it in efficiency.

Quote
I look forward to further posts as I've been toying around with making dipoles on pcb's just so that they're reproducable and on a rigid material. Oh, well.. back to 4NEC2.. :-)

Definitely report on your results here! I use a dipole on the gateway and it works great. But I've also thought about making it on a PCB for manufacturability and as you say reproducibility. The antenna will probably become much smaller due to the dielectric constant of the pcb.

Quote
Comparing your design with common HF designs

BTW, that's a key point: consensus is that small loops are inefficient. Yet it is a antenna that HAM's build with an efficiency approaching that of a dipole in HF. Radiation resistance is proportional to A^2 / l^4, so as long as proportions are the same relative to wavelength it should not change in UHF. Only problem is skin-effect and proximity effect. They do cause losses that grow with sqrt( f ). This is why I think a good sized copper tape has a better chance of success than a pcb trace.

Joe

luisgcu

  • Newbie
  • *
  • Posts: 29
  • Country: us
Re: Small loop antennas @ 433 Mhz
« Reply #3 on: May 24, 2016, 07:17:56 PM »
As you know from this thread (https://lowpowerlab.com/forum/index.php/topic,1898.0.html) I've been toying with small antennas and in particular the small loop antenna. I think it holds a lot of promise for good efficiency at very small form factors.

Well now - after purchasing a small vector network analyzer - I have had the first successes today. The end result is ok, not great yet. But I think there is still a lot of room for improvement.

I went for an inductively coupled antenna, where a small non-resonant loop drives a larger resonant one:

***IMAGE(s) FROM ORIGINAL POST***

This approach allows fixing the resistive impedance mismatch between the 50 Ohm radio module and the very low resistance resonant loop.

For the resonant loop I used 10mm copper tape. I left a gap which I first covered with adhesive tesa tape. I then covered the tesa film with a small piece of copper tape, effectively creating a capacitor:

***IMAGE(s) FROM ORIGINAL POST***

It takes a lot of cutting and measuring to get the resonance of the resulting LC circuit just right onto 433 Mhz.

It's very possible to reach a return loss of -16dB here, which is very respectable. However getting the frequency right is not easy. The antenna bandwidth is just very small:

***IMAGE(s) FROM ORIGINAL POST***

Lots of things change a bit when you disconnect the analyzer, so I think I need to do some further tuning based on RSSI values. Still - I already get reception all throughout the apartment with this setup. The RSSI is about maybe 15 lower than what I'd see with a Moteino with wire antenna. I suspect we can do much better than this.

Also this antenna still has a diameter of 4cm and I think one could go down a bit there to 3-3.5cm. However one problem is that the 2cm coin cell does block the magnetic field in a portion of the antenna. So going smaller loses gain in this setup even faster than with small loops by themselves (which already is with A^2). Possible solutions are going with the cr1632, which leaves more space for the loop - potentially even mounting the board perpendicular to the loop axis.

Still wanted to share this as a promising direction.

Joe
« Last Edit: May 24, 2016, 08:32:41 PM by Felix »
..Knowing other is wisdom,
knowing yourself is Enlightenment..

emjay

  • Full Member
  • ***
  • Posts: 120
  • Country: nl
Re: Small loop antennas @ 433 Mhz
« Reply #4 on: May 25, 2016, 02:00:10 AM »
@luisgcu,

Out of courtesy to the author, IMHO the source of your download should be attributed.
http://www.hoperf.com/upload/rf/ANTENNAS_MODULE.pdf

WhiteHare

  • Hero Member
  • *****
  • Posts: 1298
  • Country: us
Re: Small loop antennas @ 433 Mhz
« Reply #5 on: May 25, 2016, 01:56:39 PM »
With or without the source, thank you both for posting it.  It nicely consolidates a lot of practical information.I

luisgcu

  • Newbie
  • *
  • Posts: 29
  • Country: us
Re: Small loop antennas @ 433 Mhz
« Reply #6 on: May 25, 2016, 09:16:30 PM »
@luisgcu,

Out of courtesy to the author, IMHO the source of your download should be attributed.
http://www.hoperf.com/upload/rf/ANTENNAS_MODULE.pdf
thanks for refreshing my mind.. I  just simply couldn't remember from  where  I got that file.. ;)
..Knowing other is wisdom,
knowing yourself is Enlightenment..

captcha

  • Full Member
  • ***
  • Posts: 108
  • Country: au
Re: Small loop antennas @ 433 Mhz
« Reply #7 on: May 25, 2016, 10:56:04 PM »
Quote from: joelucid
Quote
"The antenna will probably become much smaller due to the dielectric constant of the pcb."

Correct. With a pcb antenna, one half of the dielectic is the surrounding air, the other half will be the pcb substrate, so the pcb only partially affects the physical length of the antenna elements. After reading various articles on the matter I noticed that the dielectric constant of FR-4 (the common substrate for PCB's) can depend on:
 - pcb materials used (changes between manufacturers)
 - frequency of operation
 - thickness of the pcb (as a percentage of wavelength)

The design of the antenna also affects how to work out inter-element spacing (think yagi's and log-periodics). It's a puzzle with a lot of variables.

Dielectric constants at DC for FR-4 I've seen can range from 4.2 to 4.6 (velocity factor around 0.48). When you get into the GHz range this number can drop to 3.9 (Vf around 0.51) (i.e. requiring a physically longer antenna). They're not hugely different but it all matters when you're trying to design something that is flexible in terms of how to make one yet also performs well.

I am awaiting my first batch of dipole-PCBs from OSH Park. Because I want this antenna design to be accessible to the general public I kept it as simple as possible so it can be ordered as-is from most pcb shops (although I would recommend to stay with OSH Park to make sure the materials are not a factor to be worried about).

My design uses:
- 1 Oz copper
- 2-layer board
- 1.6mm thick pcb
- FR-4 pcb material
- 1mm wide copper antenna traces with solder resist on top
- SMA PCB edge connector pads (or just solder coax to it)

PCB dimensions: 1.6mm x 12mm x length (where 'length' will be determined by the target frequency, somewhere around 30cm for a 433MHz antenna).

joelucid

  • Hero Member
  • *****
  • Posts: 869
Re: Small loop antennas @ 433 Mhz
« Reply #8 on: May 26, 2016, 03:35:27 AM »
Quote
My design uses:
- 1 Oz copper
- 2-layer board
- 1.6mm thick pcb
- FR-4 pcb material
- 1mm wide copper antenna traces with solder resist on top
- SMA PCB edge connector pads (or just solder coax to it)

Sounds nice! Are you planning to add a balun to the design at some point? That might improve the efficiency of the antenna a bit further esp. with our rfm69 radios and with longer transmission lines.

captcha

  • Full Member
  • ***
  • Posts: 108
  • Country: au
Re: Small loop antennas @ 433 Mhz
« Reply #9 on: May 26, 2016, 05:47:43 AM »
I haven't got any data yet on impedance figures but if it behaves like a normal dipole it should be fine without a balun. Even at 1.5:1 SWR (73 ohm to 50 ohm mismatch) it would only be 4% reflected power and I'm not really losing sleep over that. Same goes for monopoles (quarter-wave verticals), their mismatch also produces around 1.5:1 (36 ohms into 50 ohms) and should also not worry most people. Okay, there's only 100mW to go around, now you get 96mW, big deal. The greatest concern should be lossy transmission lines at near-GHz frequencies and insertion losses.

One design I have been thinking about is to do a folded dipole and then we'd need the balun for sure. The folded dipole could go all on one copper surface, or it could be going through the pcb using through-hole pads (not sure if the distance between the elements is going to be large enough then). I like the folded dipole design as it has a wider bandwidth than the normal dipole and allows for much better matching to a 4:1 balun by varying element spacing and thickness. Although intriguing, this is waay down my list, hehe..


joelucid

  • Hero Member
  • *****
  • Posts: 869
Re: Small loop antennas @ 433 Mhz
« Reply #10 on: May 26, 2016, 07:26:14 AM »
Quote
Even at 1.5:1 SWR (73 ohm to 50 ohm mismatch) it would only be 4% reflected power and I'm not really losing sleep over that.

I was thinking a 1:1 balun in order to decouple the dipole from the feedline, not to transform impedance. HAM's seem to swear by that setup.

TomWS

  • Hero Member
  • *****
  • Posts: 1892
Re: Small loop antennas @ 433 Mhz
« Reply #11 on: May 26, 2016, 08:46:44 AM »
I am awaiting my first batch of dipole-PCBs from OSH Park. Because I want this antenna design to be accessible to the general public I kept it as simple as possible so it can be ordered as-is from most pcb shops (although I would recommend to stay with OSH Park to make sure the materials are not a factor to be worried about).
How much is the dielectric constant affected by the purple solder mask?  ;)

I'm looking forward to your results.

Tom

captcha

  • Full Member
  • ***
  • Posts: 108
  • Country: au
Re: Small loop antennas @ 433 Mhz
« Reply #12 on: May 26, 2016, 07:52:31 PM »

Quote
1:1 balun in order to decouple the dipole from the feedline

One could add a small binocular ferrite (mini-circuits has some nice ones) to the design but I feel that this is largely overkill with such small power levels. It adds additional losses and may try to solve a non-existing problem. Sure, itís probably not going to win the beauty contest but there are practically no reports on the forum about locked up or erratically behaving atmegas which leads me to believe that itís going to be just fine.

Iím always impressed by the level of success people are getting with antenna contraptions that in my view violate all design rules, but hey.. if those work and people donít report any significant issues, then Iím not too worried about using a properly designed dipole without a balun.

Itís different at HF where hams commonly use 100W and more. Adding even a slightly lossy balun there to protect the expensive radio equipment makes much more sense.


Quote
How much is the dielectric constant affected by the purple solder mask?

Iíve read that the Er of common solder mask is around 3.5 Ė 4.0 but as can be seen on the below picture itís not a layer that is applied equally in all areas. I expect it will make the antenna physically somewhat shorter but the layer is relatively thin. OSH Park lists an Er of 4.6 @1MHz for their FR-4, but doesnít list it for the solder mask. Iíll ask them and post the answer here.




joelucid

  • Hero Member
  • *****
  • Posts: 869
Re: Small loop antennas @ 433 Mhz
« Reply #13 on: May 27, 2016, 04:11:27 AM »
Quote
there are practically no reports on the forum about locked up or erratically behaving atmegas which leads me to believe that itís going to be just fine.

Oh that wasn't my concern at all. I was hoping to improve primarily antenna rx performance and in particular noise levels. Receiver noise reduction has been a recurring theme for me. When I researched and measured it, it seemed a lot came in via the antenna.

Then you read success stories like this: http://www.eham.net/ehamforum/smf/index.php?topic=65621.15 and wonder if that applies to our circuits as well. Of course the full recommended package in HAM land is a balun at the receiver, then shielded transmission line and another balun at the dipole.

Definitely not yet my area of expertise though. Just getting started in antenna land. But I do have cocoaNEC installed and running. And I'm starting to like my miniVNA Tiny  ;)

TomWS

  • Hero Member
  • *****
  • Posts: 1892
Re: Small loop antennas @ 433 Mhz
« Reply #14 on: May 27, 2016, 08:29:07 AM »

Quote
How much is the dielectric constant affected by the purple solder mask?

Iíve read that the Er of common solder mask is around 3.5 Ė 4.0 but as can be seen on the below picture itís not a layer that is applied equally in all areas. I expect it will make the antenna physically somewhat shorter but the layer is relatively thin. OSH Park lists an Er of 4.6 @1MHz for their FR-4, but doesnít list it for the solder mask. Iíll ask them and post the answer here.


Wow, thanks for the thorough answer!  My question was facetious, wondering whether the purple color made a difference or not, but I do appreciate having the answer. 

Thanks again,
Tom