Zum Inhalt der Seite gehen


guess i'll finally have to get into rf circuits and tuning (famous last words). also is there any workable foss 3d fem simulation thing that makes sense for playing with ghz antennas and enclosure geometry/materials? or is this a domain exclusive to expensive proprietary software? are these things even solvable purely in simulation with current computers? can someone just do it for me? :D

teilten dies erneut

Als Antwort auf lukas/lucie "minute" hartmann

i know i can probably buy this as a service from i.e. ezurio but it would be much better to understand more and also being able to share the techniques and insights openly
Als Antwort auf lukas/lucie "minute" hartmann

naively i imagine something like blender but instead of being constrained to visible light it's different EM frequencies
Als Antwort auf lukas/lucie "minute" hartmann

People at work who do research seem to tend to use COMSOL and similar (or their own code) but that's probably going to be way out of your budget and also probably more than you need.

A quick search does find some freeware, no-cost or oss tools for rf design. Edit: such as OpenEMS: openems.de/

Dieser Beitrag wurde bearbeitet. (1 Woche her)
Als Antwort auf Janne Moren

@jannem seconding OpenEMS. Kicad 9 brings some niceties to make the process less manual as well. I learned from youtube.com/@panire3, plenty of examples to get going
Als Antwort auf lukas/lucie "minute" hartmann

also, when it comes to HF, it might help to reach out to radio amateurs, some of whom have a ridiculous level of expertise in HF circuit design while being almost as far from monetarily motivated as you can get. They do this for fun and open hardware is also part of their culture...
Als Antwort auf lukas/lucie "minute" hartmann

Blender and graphics in general uses raytracing and the view of light as particle, while RF requires using the wave eq. If you discretize space and time this can in principle be simulated as a local operator, the field at time-space location (x,y,z,t) depends on the field near (x,y,z) at time t-1. I think the most challenging part would be to model the materials involved.
Als Antwort auf lukas/lucie "minute" hartmann

I was chatting about this with a physics postdoc wondering about a free software solution, and he suggested meep github.com/NanoComp/meep
This would be more like povray for RF. The simulations are slow, and require tons of memory because of the voxel representation, but you get an animation of the the waves propagating through space...
Als Antwort auf lukas/lucie "minute" hartmann

rf engineer for automotive applications here, simulation really depends on how complex the environment of the antenna is and how well the one doing simulations knows his handywork.
Als Antwort auf morl0ck

thinking about this, a fractal antenna design on the lid might be the best solution
Als Antwort auf lukas/lucie "minute" hartmann

I am not very well versed in fractal antenna design. But since this is a geometric repeating design, it shouldn't be too hard to match it to 2.4Ghz or dual band. First you would choose a shape like a triangle(or two) and then multiply it while shrinking at the same time(basically making cut outs in the big triangle). have a pcb mill at work wich isn't functional at the moment but maybe I can do some work on it in the next weeks. Others might jump on this idea as well.
Als Antwort auf lukas/lucie "minute" hartmann

@morl0ck I'm a EE working designing wireless modems and work closely with several antenna designers and have access to the sim tools and rf chambers. Even if you have the simulator the results will be garbage in garbage out. You need an expert operator and you need to correlate the model with real measurements. I would start by a clear goal definition with 'before' measurements, a review of existing/competitive/reference designs and literature. You may not need to design anything new.
Als Antwort auf poleguy

@poleguy half of that statement depends. Antenna simulation is quite good in certain applications, a big OEM car manufacturer has switched to simulation development because they achieved extremely good correlation.
But yes no need to reinvent the wheel here. Could use a simple wifi antenna with a plastic lid instead of the pcb one to improve reception.
But more fun would be if we could use the mnt logo as an antenna 😁 that's wers the fractal design could come into play.
Als Antwort auf morl0ck

@morl0ck @poleguy why would switching from copper-free FR-4 to plastic (which type of plastic?) improve reception?
Als Antwort auf lukas/lucie "minute" hartmann

I hadn't checked if it was completely copper free. If so it will not improve reception to change to a different material
Als Antwort auf morl0ck

@morl0ck no, not completely. but i need a heatsink, and the remaining copper square provides it. if switching the lid to aluminum, i guess it needs to be a hybrid of materials with one part for heatsinking and the other part for transmissivity. but how big does this "opening" for the wifi signals need to be? and is it enough if it's mostly on one side?
Als Antwort auf lukas/lucie "minute" hartmann

@morl0ck @poleguy this is the current "OK" setup. it's fine if you're in the same room as the router. but if i'm in the next room over, it only works ok if the antenna is pointing towards the router in the other room. (pointing through the back of the lid). a smartphone (fp5) at the same position does not suffer from this "directionality". or could it just be wifi 6 vs wifi 5?
Als Antwort auf lukas/lucie "minute" hartmann

did you use 2.4Ghz or 5Ghz WLAN? That should be the main problem. Could very much be that the heatsink blocks signal when you're not in front of the router. A network analyzer would help with finding out. Is there one in the space you have in berlin? Maybe some lab nearby?
Als Antwort auf morl0ck

@morl0ck 5GHz. i could probably get one, do you mean a VNA? i haven't used one so far. how would i go about analyzing this using a VNA?
Als Antwort auf lukas/lucie "minute" hartmann

@morl0ck So a VNA (I recommend the NanoVNA for a decent cheap/capable openish entry to the space) will more or less tell you if your antenna is “tuned” properly. A demo of this process can be seen here: youtu.be/xa6dqx9udcg

Now a VNA will not show you what the RF does after it leaves the antenna or how well it receives through the case. I’m not sure what the best way to do this is, but you could do rough measurements with a signal analyzer.

Als Antwort auf lukas/lucie "minute" hartmann

yes a vna. Easiest thing would be to set the frequency range from 2Ghz to 3Ghz and a similar span for the 5Ghz band later
Select S11 as measurement type and place a marker on 2.4Ghz
Connect the antenna and than place it on the normal backplate and after that (if you have) on a copper free pcb of the same size. The readings for the marker should gibe a clue how big the influece is
Als Antwort auf lukas/lucie "minute" hartmann

the higher the frequency the more it turns into magic and random chaos. 5Ghz can be very succeptible to influence.
Sidenot on smartphones: the antennas are shit. No space, a dozen of them in a tiny enclosure. They re not really matched or decoupled from ome another so smartphones are not good as a standard to compare to.
Als Antwort auf lukas/lucie "minute" hartmann

@morl0ck Did you review the application notes from the antenna you are using?

It goes into a lot of detail about efficiency loss with ground planes nearby in different orientations.

It also shows the polar patterns expected from that antenna, to give you an idea of which way it should be oriented.

That antenna is designed to be used far from any metal. Maybe better to go with an antenna that takes advantage of your large ground plane instead?

molex.com/content/dam/molex/mo…

Als Antwort auf lukas/lucie "minute" hartmann

The farther away you get this antenna from any metal, the better it will work. I've already had some experience with this exact type used in another project (ClockworkPi uConsole).

Its better in your case because it's not directly fixed to the metal but of course you'll always have shielding and reflections from nearby metal objects. In a compact device you have to live with that.

Measuring with a VNA (as mentioned below) will not give you a lot of info. It would tell you how good the antenna works generally (ideally measured as far from any metal as possible). In your enviriment, you already know that the situation is not ideal, so not a lot to gain from measuring it.

A fractal antenna design that could be used for Wifi, Bluetooth and more would be awesome. I have no Idea how to do this though and still consider it Magic ;-)

I think the easiest way forward would be to make the copper heatsink asymetricly smaller on the antenna side and to check if there is better (more RF transparent) PCB Material available. These specs should be available. There are certainly better ways, but this is easy and probably worth a try.

@minute @poleguy @morl0ck

Als Antwort auf AndiS 🌞🍷🇪🇺

@andi @poleguy @morl0ck thank you! i also reviewed the antenna datasheet/app notes again and agree it's worth to try moving the copper away even further. also bigger area antenna could be interesting?
Als Antwort auf lukas/lucie "minute" hartmann

@andi @morl0ck

I really think you should scrap the molex antenna as it is designed for mounting far away from a ground plane.

You should move to something like a PIFA antenna, like on the laptop teardown I sent a link to.

en.wikipedia.org/wiki/Inverted…

I also read the stuff here. I think this is the right idea, although I suspect the radiation pattern needs to be rotated 90 degrees so the null is not shooting out the side of your device:
mastodon.social/@pff/114178717…


@mntmn I shared the design files for the wifi antenna in my pocket reform lid in the forum
community.mnt.re/t/another-exp…
Als Antwort auf lukas/lucie "minute" hartmann

@andi @morl0ck Thought about this some more and did another search through what molex offers: Consider something like this maybe?

molex.com/content/dam/molex/mo…

Then you can have a full copper back plate except one corner. You'd have to mount this on the back plate and feed it with a coax. Maybe worth testing?

Als Antwort auf poleguy

This looks like it's exactly tailored for this application

Also mentioning @minute because Mastodon does not notify you about new thread items if you're not explicitely mentioned.

Als Antwort auf AndiS 🌞🍷🇪🇺

@andi I realize it's not an exact match. This is intended to be mounted on the top of a horizontal ground plane. In this case it would be rotated ninety degrees, so the polar pattern will be rotated incorrectly when the screen is open.
Als Antwort auf poleguy

@poleguy @andi thanks for sharing that, will look into it and also more chip antenna options vs flex pcb antennas
Als Antwort auf lukas/lucie "minute" hartmann

@morl0ck I'm not technically an RF engineer, but I'll try. You have to stop thinking of one piece as the antenna, and other things as things with openings or things that improve the antenna. Anything in the near field at the frequency you are using _are_ part of the antenna. So your antenna design includes the case, the heat sinks, the cables, the plastic, the FR4, and their relative positions when opened/closed rotated, etc.
FR-4 will have a dialectric constant different than plastic...
Als Antwort auf lukas/lucie "minute" hartmann

Free 3D EM modeling and simulation software recommendations are CST (non-commercial-use version) and ELMER. Unfortunately neither are easy to use for newer folks.

There are a few much easier to use tools for lower frequencies. I plan to learn and use MMANA-GAL (free version) for an upcoming #HamChallenge antenna simulation activity. This tool also has helper-apps for matching and other needs.

Have fun in your learning journey. #HamRadio

Als Antwort auf lukas/lucie "minute" hartmann

We will probably have a *@Metafunk ⚡ talk at @metalab this wednesday evening about antenna simulations. I'm not sure yet but if it happens, it will be streamed too.

It's mostly about sub-Ghz Frequencies but the principles still apply. Ralf, OE3RSA has written some new tool for the purpose (don't know exactly yet. have to see the talk first)

He does a lot of antenna simulation with genetic algorithms - pretty advanced stuff. There are some things to read about it on his website: runtux.com/

@minute

Metafunk ⚡ hat dies geteilt.

Als Antwort auf lukas/lucie "minute" hartmann

I think @azonenberg is in the side business of "has expensive software and reasonable rates for open projects" ?
Als Antwort auf Sven

@HeNeArXn I have a Sonnet pro seat including the far field radiation pattern option.

I'm not an expert at designing antennas - far from it - but if you have geometry you want me to try I can simulate it.

The one big limitation is that it's a planar solver which means that you can't model e.g. a parabolic dish, and it can only model radiation normal to the PCB surface (so a patch antenna is fine, but it wouldn't give accurate results for a PCB yagi that fires sideways).

But if you have a simple patch design or PIFA or something you want to tweak I can definitely help with that.

@Sven
Als Antwort auf Andrew Zonenberg

@azonenberg @HeNeArXn that's cool, thanks! but just to understand... you probably can't plop the 3d case part(s) surrounding it into that simulation to see how they affect the antenna, right? as i understood those parts become part of the antenna and detune it?
Als Antwort auf lukas/lucie "minute" hartmann

@HeNeArXn Yeah that would potentially be tricky. Sonnet is really meant for PCB or RFIC/hybrid modeling and when you try to do more system-level stuff it struggles.
@Sven