Cellular and WiFi

With the advent of MSM8974 comes the third generation of Qualcomm’s LTE modem, in this case what 8974 brings over the second generation is new Category 4 LTE (up to 150 Mbps on the downlink) over the previous generation’s Category 3 maximum, 3GPP release 10 carrier aggregation, up to 3 carriers of HSPA+ aggregated, and of course the requisite CDMA 1xAdvanced/EVDO, GSM/GPRS/EDGE.

Of course it’s up to each operator and handset maker to make their own decision regarding what features they want implemented via the choice of frontend RF components, and in the case of the Note 3 those insights will really have to wait until someone tears one down. I suspect however that with the Note 3 Samsung marches even closer to a single worldwide SKU or at least fewer SKUs for each region. In the case of the T-Mobile Note 3, there’s disclosure for the banding that I would expect in the FCC documents, Bands 2, 4, 5, and 17 LTE are supported, at minimum, I wouldn’t be surprised to see 7 and 20 also supported in the same hardware. I strongly suspect we’re looking at a situation similar to the SGS4 where there was shared platform and front end between AT&T, T-Mobile, and Canada. I haven’t had time however to go through and look at all the Note 3 variants and their banding, and as usual Samsung doesn’t disclose their band combinations up front.

Without 20 MHz FDD LTE to test on T-Mobile, there’s not much we can really say about Category 4 in the context of the Note 3 review. In other markets with 20 MHz FDD LTE however this will open up additional throughput headroom that some already are up against.

The Note 3 continues to have Samsung’s great ServiceMode option for looking at signal characteristics like RSRP, channel bandwidth, and forcing whatever bands you want. However Samsung did remove the unlock shortcuts that were popularized for the previous Galaxy S series and Notes through the debugging screens and phone control menu. This also happens at the same time Samsung has started region locking devices.

As devices move closer to being truly universal with more LTE bands, I wouldn’t be surprised to see more of this kind of arbitrary locking being used to keep operator verticals in place and protect markets.

I had no issues with the Note 3 in both 5 MHz FDD LTE markets like NYC where I did some of my testing, and my home market in Arizona which is 10 MHz FDD LTE on T-Mobile.

Antenna placement on the Note 3 is pretty pragmatic as well, at least on this variant. One big main antenna for transmit on all the bands supported by the hardware, and a second diversity antenna for LTE and WCDMA.

A big new thing in the Note 3 is the inclusion of a component from Qualcomm’s RF360 portfolio, the QFE1100 envelope power tracker, making it the first device to do so. This reduces overall thermal footprint and power consumption on the RF front end by varying the voltage across the power amplifier (which previously was a fixed DC voltage) to match the waveform of the transmitted signal. The envelope tracker essentially is a power supply with a closed loop between the baseband and the power amplifier, which enables the power amplifier to closely follow the desired transmitted waveform. Envelope trackers help mitigate the power waste that comes with high peak to average power ratio inherent to OFDMA (LTE) by closely matching the power amplifier voltage supply to the output signal. This allows the power amplifier to work much closer to the peak of its power added efficiency (PAE) curve, especially when the input signal envelope is low, by tracking the amplitude of the input signal.

 
ET (Envelope Tracking), and APT (Average Power Tracking) on Note 3

Watching ServiceMode on the Note 3 is a real treat since you can see PA State change between a few different modes, APT/PT/ET which I suspect translate to average power tracking, power tracking, and envelope tracking modes, again confirming the presence of the power tracker. Qualcomm has a post up about the QFE1100 and its use in the Note 3.

GNSS

With 8974 also comes another newer version of Qualcomm’s GNSS (Global Navigation Satellite System) software suite. The previous generation (GNSS Gen 8/8A) included support for GPS and Russia’s GLONASS system. The new one onboard 8974 and forward moves to GNSS 8B which includes support for GPS, GLONASS, and China’s Beidou constellation.

At present there’s no way for me to validate whether Beidou support is working on any of the 8974 phones I have, because the coverage for that constellation is not global but rather limited to a slice in the Asia-Pacific region. I’m not aware of any other improvements inside GNSS 8B, but Qualcomm’s GNSS has always been excellent for me, locking quickly even indoors or in urban canyon scenarios. The Note 3 performs very well and locks quickly and I don’t have any issues to speak of.

WiFi

I didn’t expect much to change on the Note 3 from SGS4 in terms of WiFi connectivity, but I got an interesting surprise. Instead of the BCM4335 combo which was ubiquitous in this last round for WiFi/BT/FM radio, there’s the newer generation BCM4339 combo which integrates the necessary power amplifiers and front end components for a lower cost solution.

BCM4339 is still a single spatial stream part with support for 802.11a/b/g/n/ac, and on 802.11ac support for up to 80 MHz channels with 256QAM under MCS9, for a PHY rate of 433.3 Mbps, which is what we’ve seen before with 4335. Again what’s new is the integration of the 2.4 and 5 GHz power amplifiers for transmit and low noise amplifiers the receive chain, as well as the other necessary front end components, although handset makers can still choose to support their own external ones. This drives integration cost down and hopefully means 802.11ac will find its way into more places.

WiFi Performance - iPerf

I tested the Note 3 using iPerf the same way I have other phones for a while now and saw some weird behavior. WiFi throughput during my test jumped around quite a bit and even on the 433 Mbps PHY rate the maximum sustained rate I could get was around 135 Mbps. Still very fast, but not quite up as high as some of the other 802.11ac phones I’ve tested.

I don’t like that there’s no band preferences option under the advanced menu anymore, although this seems to be a thing operators want removed (which makes no sense to me, but whatever). The Note 3 does add an auto network switch toggle which seems to be a feature Qualcomm has added to its BSP lately to allow for intelligent switching between WiFi and cellular when the WiFi network you’re attached to isn’t delivering any packets. I also noticed that the Note 3 will shut down the data session for cellular when you’re attached to WiFi as well to cut down on power use, something a few other vendors have done for a while.

Speaker and Noise Rejection

With the Note 3, Samsung changes up the normal speakerphone location dramatically, from the back side coming out of the battery cover to a new place blasting out the bottom lip of the Note 3 through a grille. I tested speakerphone on the Note 3 the same way I have other devices, by placing a call, cranking volume up to maximum, and using a data logging sound pressure meter to record and average the entire call.

Speakerphone Volume - 3 inches Away

Oddly enough, speakerphone is very quiet on the Note 3 and isn’t nearly as loud as SGS4. More puzzling, tapping the extra volume button doesn’t seem to do anything at all if you’re attached to WiFi, however the button works if you’re not attached to WiFi (and thus the background data connection is enabled). This seems like a bug to me, as it’s very quiet compared to the competition, and unfortunate since part of the Note 3 is that expectation of a great multimedia experience with the big display.

A big change in the Note 3 is the inclusion of a new generation of Audience voice processor, the Audience eS325 which we’ve talked about before, which includes support for using three microphones for ambient noise reduction. Interestingly it turns out the SGS4 had 3 microphones and this same solution, but only ever used them in pairs (the third was next to the earpiece and only used during speakerphone calls). The Note 3 has three very visible microphones, one at top, one at bottom, one on the bottom right side.

Audience enables noise reduction on narrowband (8kHz), wideband(16kHz), and super wideband (24kHz) calls with this new generation with the three microphones. This new solution also adds de-reverb which helps reduce the reflections audible when calling from an enclosed room, something the previous generation didn’t have.

To test I ran through the same test we’ve done in the past, I speak into the handset in front of speakers playing a babble distractor track up to a certain sound pressure level, and then back down. I added in a speakerphone mode (far talk) mode as well just for comparisons, with the noise reduction toggle that Samsung makes available in the phone dialer toggled on and off, for comparisons. I called between a T-Mobile Galaxy S 4 and the Note 3 on T-Mobile, which gave us 12.65 kbps AMR-WB (“HD Voice”) voice call, in a spectogram of the recording we can also see components above the 4 kHz filter for a normal narrowband voice call (narrowband is again 8 kHz sampling, AMR-WB uses 16 kHz sampling, after nyquist we pass 4 kHz and 8 kHz respectively).

 

The eS325 does a good job rejecting the babble track without rejecting a lot of the incoming voice or making me sound distorted. I’m always impressed that anything is audible at the maximum volumes as these are literally so loud I can’t hear myself talking into the handset and probably not conditions normal people would call under anyways. Noise rejection on these larger phablet or Note sized phones is also a challenge since the microphone pairs are far apart and the mouth is physically further from the microphones than they normally would be, so I’m impressed with overall performance considering those constraints.

Camera Conclusions
Comments Locked

302 Comments

View All Comments

  • doobydoo - Saturday, October 19, 2013 - link

    'The chipset IS in fact performing as the benchmark indicates'

    No, it isn't. The chipset CAN'T reach the same speeds for any non-benchmark application, for reasons such as battery life and heat.

    Your argument that bickering about performance is redundant is also stupid, because if it was, Samsung wouldn't feel the need to cheat them.
  • esterhasz - Wednesday, October 2, 2013 - link

    I would be much in favor of standardized qualitative testing. Have a set five people panel use the phone for a day in their normal workflow and use a questionnaire for performance rating. Sure it's subjective, but users are subjects last time I looked.
  • Demigod79 - Thursday, October 3, 2013 - link

    Although I too would like to a cheat vs non-cheat result in benchmarks (perhaps mark out cheat results in a different color or something), Anand did state clearly that this was cheating. There was no glossing over this fact, he laid it out explicitly and said that he wanted this practice to stop (for all OEMs that do it).

    He also mentioned that it's unlikely that OEMs will stop doing this. It's easy for the OEMs to do and makes their products look better (and frankly, it's only technical geeks who care about things like this, and we only represent a tiny segment of smartphone buyers). If it sells more products, then they will do it (although I find it about as frivolous as the Nvidia and then-ATI battle to have the fastest GPU, simply for the sake of wearing the performance crown for a couple of months).

    That being said though, this is benchmarks we're talking about here. Benchmarks do not represent real-world usage, and never have. All you have to do is look at 3DMark, which was criticized for some time as being too artificial (CPU speeds hardly mattered, whereas in real life CPU speed matters greatly). Benchmarks are, by nature, highly artificial tests meant to measure performance in a specific area. Although you can complain that cheating in benchmarks give a false impression of performance compared to other devices, you cannot say that such cheating misrepresents real-world usage since it doesn't represent real-world usage in the first place.
  • DanNeely - Tuesday, October 1, 2013 - link

    "The impact is likely small since most of these tests should drive CPU frequencies to their max state regardless (at least on the CPU side), but I'm going to make it a point to call out this behavior whenever I see it from now on."

    Unfortunately this isn't the case. By decompiling benchmarks and changing package names to disable the cheat function Ars Technica discovered that the GN3 is inflating benchmark scores by 20-50%. Most got a 20% boost; Linpack was an outlier at 50%.

    http://arstechnica.com/gadgets/2013/10/galaxy-note...
  • Anand Lal Shimpi - Tuesday, October 1, 2013 - link

    This is unfortunately something we've seen on a lot of devices, not just Samsung. Google Experience devices aren't affected, but we've seen it on the SGS4 and HTC One among others.

    Linpack isn't a very consistent test and it's too short to drive frequencies up consistently, which is why I'm guessing it's an outlier. The 20% end is higher than expected, it's entirely possible that Samsung is lifting a thermal limit as well as driving CPU frequencies up.

    I don't like any of it and I do want to see companies stop doing it. I was hoping we would see an end to it with the Note 3 but it looks like that was wishful thinking.

    Take care,
    Anand
  • Wojciech - Tuesday, October 1, 2013 - link

    Have you thought about doing an article about 'fixing' benchmark scores by other OEM's?
    If you're saying that HTC is doing the same with One then maybe LG is doing something similar and maybe even Sony.
    Normal behavior by Google experience devices would explain often lower scores than customized devices running on the same hardware platform.

    Don't you think that would be an interesting topic to examine?
    Right now I fear that more and more OEM's are going to start doing the same thing and the whole 'benchmark to determine real life performance' will be completely lost.
  • xype - Wednesday, October 2, 2013 - link

    You don't like any of it? But you still put up the graphs and numbers with an "Oh my."? People come here because AnandTech has a reputation of providing in-depth, honest reviews. Most people scan the text and go right to the graphs. Their takeaway will be a marketing lie that you didn't bother to correct because "A lot of companies do that."? Seriously?
  • Spunjji - Tuesday, October 8, 2013 - link

    If you skim the text and go to the graphs you will never, EVER get a representative review of anything. People come to Anandtech for analysis and they got that with this review. If they missed that then they might as well have gone to any of the other sites.
  • doobydoo - Saturday, October 19, 2013 - link

    Na, most people come to Anandtech because they know the graphics will have been performed in an objective and logical way. I would bet that the vast majority of readers don't read the text associated with such images.

    And that doesn't mean that they should go to other sites.
  • Squuiid - Tuesday, October 1, 2013 - link

    "It's also interesting to note that the Galaxy Note 3 appears to outperform all other Snapdragon 800 smartphones we've tested thus far. There's a couple of potential explanations here."
    You missed an explanation: Samsung cheat.
    From Ars:
    "The two functions applied to this list seem to be "PACKAGES_FOR_BOOST_ALL_ADJUSTMENT" which is no doubt the CPU booster, and "PACKAGES_FOR_LCD_FRAME_RATE_ADJUSTMENT" which makes it sound like they are also changing the display frame rate."

Log in

Don't have an account? Sign up now