Hobby-grade LTE reception testing

[davegravy]

I have an LTE-M shield I designed and built for an SBC. I wasn't sure if I should use an SMA connector with external antenna because I was rather constrained with placement options for an SMD chip antenna but the space saving factor of the latter made me choose it anyways. I also did this because I mainly want to use it in urban environments where towers are typically close.

Now I want to do some testing to see if it's getting reasonable performance, as best I can with a hobbyist budget.

My thought as a strategy was to collect RSSI, RSRQ, RSRP, SINR data and compare to a commercial product (e.g. cellphone) in a variety of RF environments, expecting to get a few dB worse performance.

Does this seem like a reasonable approach?

Here is the antenna FWIW: Molex 1462000001

There's a few non-idealities such as the fact the SBC when plugged in sits over one end of the antenna, and a big metal screw is immediately adjacent the antenna.

[Salitronic]

When using a chip antenna it is very important that you follow the manufacturer's recommended design as closely as possible. Many seem to make the mistake in thinking that the antenna is just the SMT part on its own. That's not correct, the antenna consists of the SMT chip part (with a specific placement), the recommended copper clearance around it and the ground plane of a specific size. You can change some of these parameters but the performance that is specified by the antenna manufacturer is based exclusively on the recommended design. As you depart from the recommended design you will start losing performance. With simulation capability and proper test equipment it is often possible to find other configurations that work but doing this blindly on a hobby budget is not easy so ideally stick as much as possible to the manufacturer's recommended design.

For example the recommended design has the longer arm of the chip antenna at the PCB corner, you have it the other way round and moreover there the copper clearance on that side does not continue across the full width of the board. I would rotate the chip to follow the recommendations. That screw in the antenna clearance zone is what first caught my eyes, that should be avoided (at least try using a Nylon screw, not metal). Having part of the antenna under the SBC is not ideal either.

It is likely that this will "work" as it is and for hobby purposes this would probably be enough but you'll not get the performance specified by the manufacturer.

Comparing to cell phone is not easy, chip antennas are typically relatively low in radiating efficiency whereas phone antennas are highly optimized. Also LTE-M uses different modulation, bandwidth, data rate, power levels and possibly also frequency bands, so if anything you should compare with similar LTE-M devices.

A low cost SDR might be useful to do some comparative radiated power measurements.

[davegravy]

When using a chip antenna it is very important that you follow the manufacturer's recommended design as closely as possible. Many seem to make the mistake in thinking that the antenna is just the SMT part on its own. That's not correct, the antenna consists of the SMT chip part (with a specific placement), the recommended copper clearance around it and the ground plane of a specific size. You can change some of these parameters but the performance that is specified by the antenna manufacturer is based exclusively on the recommended design. As you depart from the recommended design you will start losing performance. With simulation capability and proper test equipment it is often possible to find other configurations that work but doing this blindly on a hobby budget is not easy so ideally stick as much as possible to the manufacturer's recommended design.

For example the recommended design has the longer arm of the chip antenna at the PCB corner, you have it the other way round and moreover there the copper clearance on that side does not continue across the full width of the board. I would rotate the chip to follow the recommendations. That screw in the antenna clearance zone is what first caught my eyes, that should be avoided (at least try using a Nylon screw, not metal). Having part of the antenna under the SBC is not ideal either.

It is likely that this will "work" as it is and for hobby purposes this would probably be enough but you'll not get the performance specified by the manufacturer.

Comparing to cell phone is not easy, chip antennas are typically relatively low in radiating efficiency whereas phone antennas are highly optimized. Also LTE-M uses different modulation, bandwidth, data rate, power levels and possibly also frequency bands, so if anything you should compare with similar LTE-M devices.

A low cost SDR might be useful to do some comparative radiated power measurements.

I had many of these thoughts while designing that I brushed aside but reading it coming from you I'm now coming around. I'm about to place an order for the next prototype revision and I'm thinking I'll just replace the chip with SMA, that way I have flexibility to change antenna gain and position if I ever need to.

[Nominal Animal]

I wasn't sure if I should use an SMA connector with external antenna

Could you fit an u.FL/IPEX connector (like CONMHF1-SMD-T from Mouser, 4mm × 4.1mm total footprint area with a 2mm × 2.1mm central region without traces)?

These are extremely common in miniPCIe LTE modems, with u.FL/IPEX to SMA/RP-SMA coax "pigtails" easily available in various lengths.  I prefer the RG178 ones myself.

FWIW, I often use M2 and M2.5 nylon screws and standoffs for my PCBs, from cheap Aliexpress/Banggood kits.  They're surprisingly good for their price, in my opinion.  However, I'm just a hobbyist playing with 4G/LTE modems on Linux routers with miniPCIe support, and not an expert in any of this.

[davegravy]

I wasn't sure if I should use an SMA connector with external antenna

Could you fit an u.FL/IPEX connector (like CONMHF1-SMD-T from Mouser, 4mm × 4.1mm total footprint area with a 2mm × 2.1mm central region without traces)?

These are extremely common in miniPCIe LTE modems, with u.FL/IPEX to SMA/RP-SMA coax "pigtails" easily available in various lengths.  I prefer the RG178 ones myself.

FWIW, I often use M2 and M2.5 nylon screws and standoffs for my PCBs, from cheap Aliexpress/Banggood kits.  They're surprisingly good for their price, in my opinion.  However, I'm just a hobbyist playing with 4G/LTE modems on Linux routers with miniPCIe support, and not an expert in any of this.

If I'm removing the chip antenna there's loads of space for a connector like that, but also room for an end-launch SMA. Is there a benefit of the u.FL over the SMA other than size and cost? I'm inclined towards SMA since this is an outdoor device that might get thrown around a bit, so ruggedness is valuable.

[Nominal Animal]

If I'm removing the chip antenna there's loads of space for a connector like that, but also room for an end-launch SMA. Is there a benefit of the u.FL over the SMA other than size and cost? I'm inclined towards SMA since this is an outdoor device that might get thrown around a bit, so ruggedness is valuable.

If you have other connectors poking out of the PCB at fixed locations, then direct SMA may make more sense.

The u.FL-to-SMA pigtail allows one to use many different enclosures, as the SMA connector can be placed anywhere within reach.

Also note that the PCB antenna can be on a separate PCB (connected via flexible u.FL-u.FL coax cable), as that's how some Mikrotik and Teltonika 4G/LTE outdoors routers do it.

I think the Mikrotik R11e modems are a typical example:

^{(Click to embiggen)}

Because the position of the miniPCIe modem varies even among Mikrotik products supporting it, the main and diversity antenna connectors (top left and top right corners, just inside from the mounting/bracket holes) are u.FL ones so a short pigtail can be used to route them to SMA connectors on any enclosure used.

In comparison, I do have seen 4G/LTE routers that incorporate the corresponding circuitry on their main board, and they do use direct 90-degree through-hole SMA connectors to the external antennas.  It is more cost-effective, and probably has a fraction of a decibel less insertion losses.

I like to also do custom enclosures (right now I have a 25cm tall birch half-stump I want to make into a wall enclosure; looks like a nature art, maybe with a round display as if it was a birdhouse), so I personally do prefer the u.FL ones allowing me more leeway with custom enclosures.  Most users won't care, so I'd say using direct SMA connectors is almost as good a solution.

Why almost?  Because of the enclosure-internal PCB and ceramic antennae typically come with u.FL connectors, allowing you to choose between an external SMA or RP-SMA connector (via u.FL to SMA/RP-SMA pigtail), or an internal (PCB or ceramic) antenna (via u.FL to u.FL pigtail).

For an experimental prototype, I'd definitely use the u.FL connector, because of the additional options and testing different antenna solutions it allows.  For a proper product, I have no idea: me hobbyist, not product designer.

[Nominal Animal]

For what it is worth, I recently got an ADT-Link WS17 USB adapter for USB 2.0 -based miniPCI-e 3G/4G/LTE modems like Huawei ME909s-120, Mikrotik R11e-LTE, and so on, from AliExpress.  It is easily the best of all USB 2.0 adapters I've seen, and does not suffer from overheating; well worth the ~20€ price (including shipping and VAT).  For USB 3.0 based ones, use WS16 instead; I probably should have too, considering the schematic indicates it should be compatible with USB 2.0 ones as well.  The structure is very nice layered construction, too, making it, uhm, adaptable for shenanigans involving SBCs and small enclosures.

I deliberately used a power-hungry MIMO (two-connector) planar antenna (deliberately badly oriented) for testing, and the modem definitely generated some heat, but the supply circuitry on the adapter barely got warm.