Thursday, September 27, 2018

iPhone XR trades 4x4 MIMO for greater antenna gain and higher EIRP

iPhone XR (model A1984, FCC ID BCG-E3220A) made a somewhat surprising early appearance in the FCC OET database this morning, since the phone will not be available yet for another month.  Typically, iOS devices have their authorization filings posted right before the devices go up for sale to the public.

Upon examining the certified lab tested RF data and crunching some numbers, what is immediately apparent is that the iPhone XR has -- as expected -- dropped the 4x4 MIMO antenna design from the iPhone XS and XS Max in favor of a standard 2x2 MIMO antenna.  This tradeoff of fewer antennas and MIMO channels, though, ostensibly has allowed for greater antenna gain and, concomitantly, higher radiated output power compared to those same metrics on the iPhone XS and XS Max.

Further analysis may be forthcoming, but for now, the graphs of the iPhone XR EIRP and antenna gain largely can speak for themselves.



For graphs and analysis of iPhone XS and XS Max lab tested figures, see my previous articles (1, 2, 3).

Source: FCC OET

Tuesday, September 25, 2018

Antennagate 2018: "You're getting my findings wrong"

Two weeks ago, I dug into the FCC OET authorization filing documents for the iPhone XS and XS Max the afternoon that they were made public, and I started to review and extract data on the certified lab tested RF performance of both forthcoming handsets.  This is something that I have done for dozens of devices over the last decade.  Some of the 2018 iPhone RF measurements struck me as comparably and unexpectedly low, worthy of some graphical and contextual analysis that might be of interest to a few hundred readers.  I did not expect another so called antennagate to emerge, nor for tens of thousands of average viewers to take note of what is a fairly technical topic.  And because my work (1, 2) has drawn an audience well outside its intended scope, I feel a need to editorialize a bit.

Correlation does not equal causation.  My goal long has been to present findings, raise questions, and posit possible explanations/solutions.  My goal has not been to assert that iPhone XS and XS Max low gain antenna designs are the exclusive or even partial causes of all the wireless signal ills that some new users are experiencing.  The correlation exists, but the cause(s) remain(s) undetermined.

Unfortunately -- though I do not think nefariously -- some forum posters and media outlets have run with my findings, questions, and explanations/solutions to form distorted or at least simplified conclusions that say something to the effect of the new 4x4 MIMO antenna array is to blame, the Intel baseband modem is not to blame, the Intel baseband is indirectly responsible due to design compromises that would not have been made for a Qualcomm baseband, etc.

In the interest of clarity, let us find some solid bedrock on which to stand.

What can be said definitively is the iPhone XS and XS Max exhibit lab tested RF performance for antenna gain and radiated power output (EIRP) that almost invariably measures lower than the same metrics on 2017 iPhone variants and other recent, comparable handsets.  A visual recap...



Additionally -- as I already have stated -- the Intel baseband categorically is not the source for conducted power output, radiated power output, or antenna gain.
...Apple's decision to forgo Qualcomm this year and source all cellular modems from Intel is not responsible for the RF power output limitations in the new iPhone models.  The cellular baseband modem is separate from and well upstream of the amplifiers that generate the conducted power and antennas that generate the radiated power being measured in lab testing.
Those two findings are demonstrable facts.

To follow up the latter of the two facts, I posted this on Twitter last night, though if you have been following just my blog pieces, you may have missed it.

The two 2017 iPhone X variants -- with Qualcomm baseband and with Intel baseband -- perform almost identically in RF testing submitted in their FCC authorization filings.  And I do mean almost identically, right down to the decimal place -- this speaks well to Apple's precision tolerances in design and manufacturing.  The lone deviations between the two variants are on bands 41 and 66, and only band 41 presents a difference worth discussing.  The maximum antenna gains across all bands are exactly the same for both Qualcomm and Intel variants, but for unknown reasons, the Qualcomm variant puts out 50 mW greater conducted power on band 41, hence also 50 mW greater EIRP on band 41.  That is absolutely it for the raw RF output differences.  Any other cellular performance differences likely can be attributed to the Qualcomm baseband versus the Intel baseband.  That controversy is well known.

2017 iPhone X with Qualcomm MDM9655 Snapdragon X16 LTE baseband:


2017 iPhone X with Intel XMM 7480 baseband:


So, a takeaway here is that Apple designed and manufactured the two 2017 iPhone X variants effectively the same -- just plopped a Qualcomm modem in one, an Intel modem in the other.  Now, definitely not a fact but informed conjecture, Apple ostensibly would have done likewise this year if there had been a Qualcomm variant.  In other words, a hypothetical iPhone XS or XS Max with Qualcomm baseband probably would have followed precedent with the same antenna array, thus the same radiated power levels as that of the current Intel model.  Take that for what you will.

To return to my soapbox for a time, many end users posting their experiences with signal bars and speed tests are not helping, actually are muddying the waters further.  Regarding signal bars and speed tests this is what I previously wrote:
...signal bars are just a coarse approximation of signal availability/quality and can be calibrated to almost any arbitrary standard. 
Comparing speed tests also is not very useful for gauging weak signal performance.  Unless a speed test fails or produces very slow speeds due to exceedingly marginal signal conditions, devices with 4x4 MIMO reception -- such as iPhone XS and XS Max -- likely will exhibit higher download speed test results under most good to normal signal conditions.
To expand on the above advisement, iPhone XS and XS Max possess capabilities -- aforementioned 4x4 MIMO reception, additional low band (bands 14 and 71), LAA (band 46), and downlink carrier aggregation combinations -- that are not available on previous iPhone models.  Some may argue that those are RF improvements, but again, they are capabilities or features that are not at parity with those of iPhone predecessors.

Greater signal bars, for example, could reflect signal strength on a new low band, thus presenting the illusion that the new phone has better reception than the old phone.  In actuality, the new phone might have worse reception than the old phone if both were on the same band.  The new low band could be thought of as a band aid -- pun intended -- that masks some of the RF deficiencies of the new phone.

Much the same, a faster speed test on a new phone, for example, could be the result of additional MIMO spatial channels or carrier aggregation SCCs that are not possible on the old phone.  Meanwhile, the old phone with measurably superior RF performance might provide faster speeds on a per spatial channel or per carrier basis.

That said, I get it.  To those who want to try to assure themselves that their new iPhones can meet their everyday needs and do not care whether that is accomplished through raw RF strength or new network capabilities, fine, okay.  Check signal bars and run speed tests.  Just understand that signal bars and speed tests -- absent thorough additional documentation from Field Test -- are not replacements for objective measurements. 

Source: FCC OET

Monday, September 24, 2018

Antennagate reduXS? If so, what can Apple and iPhone users do about it?

About 10 days ago, shortly after the Apple announcement event, I published an article that analyzed iPhone XS and XS Max RF performance as revealed in lab test measurements in the devices' FCC OET authorization filings.  Some of the lab test data raised potential red flags regarding the new iPhones' RF reception and transmission abilities in low signal strength conditions.  My analysis in that article concluded with...
...iPhone EIRP in the lab has not always been so compromised as it appears to be this year.  Real world RF performance comparisons when some users switch from the iPhone X and 8 generation to the iPhone XS generation, no doubt, will be interesting.
Well, over the weekend, more and more early adopters started getting their hands on their brand new iPhones, posting their real world impressions online.  While opinions on RF performance have varied from high praise to utter disillusionment, some anecdotal user reports and empirical signal comparisons at MacRumors forums and in more than one Reddit thread have been notable because they appear to corroborate how the aforementioned lab test results submitted to the FCC might extrapolate to less than stellar real world performance.

In many ways, this harkens back to iPhone 4 antennagate in 2010.  Without rehashing all of the details, iPhone 4 users just by holding the handset could bridge together unintentionally two antennas on the edge of the phone, thereby accidentally detuning both antennas and significantly degrading RF performance.  In response to the issue, Steve Jobs famously was misquoted as saying to users about the phone, "You're holding it wrong," which since has become a blame redirection meme unto itself.  Apple's official response was to provide all iPhone 4 purchasers with a free case to insulate the antennas from users' casual embrace.

But returning to the present, if the disappointing RF performance that at least some appear to be experiencing eight years later with iPhone XS and XS Max does boil down to another antenna issue -- weak antenna gain this time -- holding the handset differently or adding a case/bumper is not going to solve the issue.

Some have proposed that, if necessary, an Apple firmware update or two will sort this out.  Yes, iOS revisions and/or carrier updates can tweak the way the baseband modem interacts with wireless networks at large.  And, from an end user perspective, this can improve (or degrade) perceived performance.  However, what software cannot do is update physical qualities locked in during design and manufacturing.

An analogy for this could be the LCD or OLED display screen on your handset.  New firmware might improve image quality by way of adjusting gamma curve, color calibration profile, scaling algorithm, etc.  But that software update cannot change the screen size or pixel resolution -- both are physical characteristics.

In that spirit, if Apple acknowledges an issue and a fix, what software or hardware remedies could it propose?

Apple could submit to the FCC OET a Class II permissive change filing, which is an amendment to an original equipment authorization filing with additional lab testing to reflect a material change in the device.

Let us take a look at two possible Class II changes that Apple could implement, though for reasons to be explained, Apple probably would not pursue either one.
  • Increase conducted power via a firmware update
Conducted power is the input to the antenna, while radiated power (EIRP) is the output from the antenna.  But with increased conducted power, increased EIRP also would result.  This could help RF performance in situations in which the uplink hits its power limits before downlink reception drops below usability.

However, this is not a likely solution.  For an LTE device that operates in Power Class 3 (non HPUE) in a specific band, its target conducted power is 200 mW (23 dBm) +/-2 dB.  Similarly, for Power Class 2 (HPUE), target conducted power is 400 mW (26 dBm) +/-2 dB.  As noted in my previous article -- see quote below -- Apple already has applied conducted power somewhat liberally, pushing the +2 dB margin for conducted power across most bands.
...conducted power is not the issue.  The standard conducted power target of 200 mW (23 dBm) is +/- 2 dB.  And Apple is using the +2 dB margin to enhance its figures, pushing 250-320 mW (24-25 dBm) conducted power across many included bands.  This extends to band 41 HPUE, which has a standard conducted power target of 400 mW (26 dBm).  Again using the +2 dB margin, Apple has upped that ante to 500 mW (27 dBm).
Then, for visual reference, see a graph of iPhone XS Max conducted power.  Note that this graph is expressed in dBm instead of mW to reflect more directly the up to +2.5 dB boost Apple already has applied to conducted power above the standard 23 dBm and 26 dBm target levels, respectively.


Without HPUE across all bands, Apple has little, if any additional wiggle room to increase conducted power.  And, remember, bumping up conducted power helps only with uplink transmission.  It does not improve downlink reception.
  • Increase antenna gain via a hardware manufacturing revision
Increased antenna gain would cause increased EIRP -- with or without increased conducted power.  Additionally, as long as the same antenna arrays are used for both uplink transmission and downlink reception, improved antenna gain can assist in received signal strength, too.

This sounds like the ideal solution.  And, quite frankly, it is.  But for even more obvious reasons, Apple would be highly unlikely to go this modified hardware route.  This would be the nuclear option, so to speak, that would require some redesign and retooling of yet to be manufactured iPhone XS and XS Max units as well as possible replacement of extant phones for any dissatisfied users.  Instead, any antenna design changes probably will be saved for next year's iPhone generation.

So, what can iPhone XS and XS Max users do right now?

The new iPhones undoubtedly show some measured RF shortcomings compared to their predecessors and other comparable handsets.  However, the jury very much is still out on how much and how often this will affect typical use cases.

In the interest of investigation, many users already have jumped to comparing signal bars, though doing so is not very useful, since signal bars are just a coarse approximation of signal availability/quality and can be calibrated to almost any arbitrary standard.

Comparing speed tests also is not very useful for gauging weak signal performance.  Unless a speed test fails or produces very slow speeds due to exceedingly marginal signal conditions, devices with 4x4 MIMO reception -- such as iPhone XS and XS Max -- likely will exhibit higher download speed test results under most good to normal signal conditions.

Accessing Field Test to compare signal metrics (RSRP, RSRQ, CINR) across multiple devices actually is useful.  But this must be done with a watchful eye and solid understanding.  PCI, GCI, and EARFCN all must be the same in the same location to ensure that both devices are on the same PCC of the same site and sector.  Any difference among those matching signal characteristics eliminates a legitimate apples to apples comparison.  So far, Field Test reports are scant, though many more should be forthcoming as savvy users put their new phones through the paces.

Source: FCC OET

Wednesday, September 12, 2018

[UPDATED] iPhone XS and XS Max mostly fail to impress in lab tested RF power output

Following Apple's keynote presentation yesterday, authorization filings for many of its newly announced mobile devices started being made publicly available in the FCC OET database.  Among others, this included the US variant iPhone XS (model A1920, FCC ID BCG-E3218A) and iPhone XS Max (model A1921, FCC ID BCG-E3219A).  The US variant iPhone XR (model A1984, FCC ID BCG-E3220A) was not included.  Its authorization documents likely will be disclosed closer to its October release, thus may be the subject of a later update or a different blog post.

Before proceeding any further, RF power figures to come represent best averaged and rounded estimates of maximum uplink EIRP test results provided to the FCC OET in individual device authorization filings.  Or in the case of ERP low band test measurements submitted in the filings, that ERP has been converted manually to EIRP for level comparison purposes.  All EIRP figures are normed against a baseline of 200 mW (23 dBm), which corresponds to a standard conducted power target with unity antenna gain and generally represents good RF transmission performance.  Caveats about lab testing versus real world capability and uplink versus downlink always apply.

Now, let the EIRP graphs speak for themselves...



As the blog post title indicates, the lab tested radiated output from the two new iPhone XS models does not set the world on fire.  Nearly all bands on both handsets fall short of the 200 mW benchmark.  Only band 41 HPUE on iPhone XS looks pretty good on paper, and even that comes with a conducted power asterisk to be analyzed later.

But first to nip in the bud one potential conspiracy theory, Apple's decision to forgo Qualcomm this year and source all cellular modems from Intel is not responsible for the RF power output limitations in the new iPhone models.  The cellular baseband modem is separate from and well upstream of the amplifiers that generate the conducted power and antennas that generate the radiated power being measured in lab testing.

Furthermore, conducted power is not the issue.  The standard conducted power target of 200 mW (23 dBm) is +/- 2 dB.  And Apple is using the +2 dB margin to enhance its figures, pushing 250-320 mW (24-25 dBm) conducted power across many included bands.  This extends to band 41 HPUE, which has a standard conducted power target of 400 mW (26 dBm).  Again using the +2 dB margin, Apple has upped that ante to 500 mW (27 dBm).  That inflated conducted power is fine.  But bear in mind that it assists only in transmission, never in reception.  Plus, it also can be used to mask some antenna shortcomings.

Yes, with often greater than standard conducted power being generated -- rhetorical questions ahead -- where is all that power going?  Where is it being diminished?  The answer lies in antenna gain.

Indeed, deeper analysis of the FCC OET authorization filings shows the underwhelming EIRP figures to be almost entirely products of negative antenna gain.  For every 3 dB drop in antenna gain, 50 percent of conducted power is attenuated.

To illustrate visually, look at a graph of the iPhone XS Max antenna gain (Ant. 1) across its entire uplink low band, mid band, and high band frequency range.  Antenna gain inevitably reduces conducted power by about 5-7 dB.


Now, both iPhone XS and XS Max this year incorporate four antennas operational across many but not all bands.  That antenna diversity in and of itself is a good thing.  However, even with the four antennas -- and possibly because of the four antennas crammed inside -- antenna gain is universally negative.  And simultaneous transmission from multiple antennas is not possible due to a "break before make" switching mechanism among the antennas.

For a partial look at all four antennas, see a snapshot from the iPhone XS Max authorization filing:


Lastly, for an interesting comparison and stronger RF output, see lab tested EIRP figures from last year's iPhone X (model A1865, FCC ID BCG-E3161A) and iPhone 8 Plus (model A1864, FCC ID BCG-E3160A).



The takeaway is that iPhone EIRP in the lab has not always been so compromised as it appears to be this year.  Real world RF performance comparisons when some users switch from the iPhone X and 8 generation to the iPhone XS generation, no doubt, will be interesting.

Also, see follow up articles (1, 2).

Source: FCC OET

Wednesday, September 5, 2018

Moto Z3 Play hits a home run on band 41, swings big and misses on low band

The Moto Z3 Play (model variant XT1929-4, FCC ID IHDT56XE1) is sold directly by Motorola and via other third parties as an unlocked handset compatible with all North American cellular operators.  As such, it must support a wide swath of LTE bands.  And, indeed, the Moto Z3 Play is one of the first and only handsets thus far to support two nascent 600/700 MHz bands -- band 14, which AT&T is deploying for public-private FirstNet, and band 71, which T-Mobile and other smaller operators are deploying in more recently vacated UHF TV spectrum.  Somewhat unfortunately, though, that low band inclusion on the Moto Z3 Play seems largely nominal, since low band RF output across the board is rather meager.

Where the Moto Z3 Play shines is in its band 41 HPUE output.  Not only bumping up conducted power to better than 400 mW (26 dBm) as expected from HPUE Power Class 2 but also adding +2.0 dBi antenna gain makes a potent combination that peaks at over 700 mW EIRP on band 41.  This level of band 41 power and positive antenna gain is basically unprecedented and ostensibly should make the Moto Z3 Play a stellar performer on Sprint.  Almost all other antenna gain, however, is negative, even exceedingly negative -- all low band antenna gains run below -5.0 dBi.  High band antenna design, in this case, may have come at the cost of low band antenna optimization.  Questions arise how well the Moto Z3 Play will respond in low signal rural areas where low band or "Extended Range" LTE is supposed to be the coverage saving grace.

Before moving on to the graphs, let us run down the standard boilerplate.  RF power figures below represent best averaged and rounded estimates of maximum uplink EIRP test results provided to the FCC OET in individual device authorization filings.  Or in the case of ERP low band test measurements submitted in the filings, that ERP has been converted manually to EIRP for level comparison purposes.   Caveats about lab testing versus real world capability and uplink versus downlink always apply.   And for better visual comparison purposes, the vertical scale heightened to 800 mW needed to display Moto Z3 Play EIRP remains consistent across follow up EIRP graphs, too.


We would be remiss not to include at least a cursory juxtaposition with the VZW exclusive Moto Z3 (model variant XT1929-17, FCC ID IHDT56XJ1), which shares the very same external design with the unlocked Moto Z3 Play.  Note even the same basic model number between the two.  But, internally, the Moto Z3 carries a higher end Qualcomm Snapdragon 835 with X16 LTE modem chipset, and as the FCC OET lab testing figures show, it also offers more mundane yet consistent RF output across its disclosed bands.  Clearly, some internal RF hardware has been changed and/or radio software tweaked in the transition from the Moto Z3 Play to the Moto Z3.  This is corroborated by the many additional LTE bands that the Moto Z3 supports but are not similarly present in the Moto Z3 Play nor disclosed in the FCC OET documents -- because those bands are downlink (i.e. reception) only or software disabled for US operation on VZW.  Absent from the Moto Z3, though, is band 71 capability, and perhaps its inclusion in the Moto Z3 Play antenna design is at least partly responsible for that model's low band challenges.


For additional comparisons, look at two of my favorite Motorola handsets of the past year, both with which I have had ample user experience and both of which are compatible with MVNO Project Fi, the Moto X4 (model variant XT1900-1, FCC ID IHDT56WK1) and Moto G6 (model variant XT1925-6, FCC ID IHDT56XD1).  Neither has HPUE, yet each still manages to pump out some solid band 41 EIRP by way of positive antenna gain.  The Moto X4 provides superb +2.80 dBi antenna gain on band 7/38/41.  And my experience has been similarly positive -- the Moto X4 has an above average affinity for band 41.  Beyond that, the Moto X4 is about average.


On the flip side, the Moto G6 offers above average EIRP across most other bands, too.  And that is achieved not through conducted power inflated above 200 mW (23 dBm) but mostly good old fashioned positive antenna gain.  In particular, the low band antenna gain of +1.36 dBi for band 12/17 is impressive.


That covers the lab tested RF figures for the Moto Z3 Play along with several related or recent handsets from Motorola.  If you have requests for analysis of other devices, feel free to submit them in the comments.

Source: FCC OET

Monday, September 3, 2018

[UPDATED] Pixel 3 XL projects out to be mid band LTE powerhouse

Straight to the point, Pixel 3 XL lab test results show mid band RF power output that we rarely have seen from a handset during the LTE era.  Smaller sibling Pixel 3 does not reach quite the same RF power heights, though it holds minor advantages in some low band output.  And both 2018 Pixel handsets measurably outperform their Pixel predecessors of the past two years.

Pixel 3 XL (A4RG013C) and Pixel 3 (A4RG013A) authorization filings became available to the public at the FCC OET (Office of Engineering and Technology) database at the end of last week.

For the radiated power figures, caveats about lab testing versus real world capability and uplink versus downlink always apply.  But solid uplink ERP (Effective Radiated Power) or EIRP (Effective Isotropic Radiated Power) generally correlates with good overall RF transmission and reception.  Conversely, weak uplink output typically cannot be overcome for anything but middling to poor overall RF performance.

RF power figures below represent best averaged and rounded estimates of maximum uplink EIRP test results provided to the FCC OET in the authorization filings.  Or in the case of ERP low band test measurements, that ERP was converted manually to EIRP for level comparison purposes.

Before taking a look at the graphs, a standard baseline for good EIRP is 200 mW (milliwatt) or 23 dBm (decibels relative to one milliwatt).  This is the target conducted power -- the RF power generated before it reaches the antenna -- on most handsets for most bands.  Without delving into excessive explanation, if the 200 mW conducted power can be turned into 200 mW or greater radiated power by the antenna, then the antenna gain and design tend to be implemented well.  This is particularly important for mid band and high band, less so for low band, which can present a size challenge to antenna design in handheld devices.

First up, the star of this show, the Pixel 3 XL.  And it generates its massive mid band power from excellent antenna gain of +3.70 dBi (decibels relative to an isotropic antenna).  Band 41 -0.90 dBi antenna gain is less ideal, though HPUE target conducted power of 400 mW (or 26 dBm) makes up for some of that loss.  All figures depicted are for EIRP from the main antenna; the auxiliary antenna across the board is a lesser gain antenna.


Next is the Pixel 3, all figures again from its stronger main antenna.  Its mid band antenna gain of variably +0.30 dBi or +0.98 dBi is not quite so potent as that of its larger relative.  And some of Pixel 3's average to good power output comes from greater than 24 dBm low band conducted power, not from antenna gain.  However, band 5/26 has an ever so slightly above unity +0.03 dBi antenna gain, and that is good but rare to see any positive gain for low band.


Graphing just main antenna gain versus LTE band approximate center frequency for both Pixel 3 handsets, we can see visually that mid band positive antenna gain.  One theory is that the full glass, partial matte backplates on both 2018 Pixel 3 variants may be providing greater RF transparency -- as opposed to the partial glass, partial aluminum backplates on the 2016 and 2017 Pixel variants, a few of which we will offer up for comparison shortly.



Yes, as bases for comparison, take a look at those RF output figures from an LG manufactured Pixel 2 and HTC manufactured Pixel that have preceded this Google in house designed 2018 Pixel generation.  In these instances, both are XL variants.  To summarize, the 2017 LG Pixel 2 XL and 2016 HTC Pixel XL frequently fail to hit the 200 mW baseline, though they are mostly consistent across the board and never woefully deficient.  The LG acquits itself better with band 4/66 and band 41, while the HTC just seems to continue the company's passably average RF performance on LTE.  As an aside, the Pixel 2 XL measurements follow a Class II Permissive Change authorization filing that preceded the handset release and did adjust some uplink power output.  Thus, upcoming RF tweaks to one or both of this year's Pixel duo still are possible prior to street date.



Finally, to expand the scope outside the Pixel universe, Samsung Galaxy S9 is the current handset to which both 2018 Pixel handsets are most likely to be compared.  Anecdotally, Galaxy S9 has been said to be a good RF performer.  RF numbers wise, Pixel 3 and Galaxy S9 are quite similar, the biggest difference being band 41 HPUE in the latter.  As for Pixel 3 XL, its mid band output once more grabs the spotlight.  In that regard, Pixel 3 XL stacks up very well.


Field testing the latest Pixel pair in the hands of users and reviewers over the coming months will tell us more about real world performance, but as of now, the lab tested EIRP figures offer an auspicious peek at what project to be improved and flagship caliber RF.

Source: FCC OET