The Qualcomm Shakes-up of RF is Beginning

It is early February, and in normal times, that meant companies would be gearing up for Mobile World Congress with a slew of press releases and product releases. The fact that MWC has been delayed this year did not stop Qualcomm from putting out some fairly impressive news. The company announced a new family of Snapdragon processors (an important update to a topic which has been on our minds lately) and a set of new Radio Frequency (RF) chips. The RF products are fairly minor commercially but they are an important harbinger of big changes coming to this corner of the market.

RF chips straddle the divide between real-world physical signals, in this case radio waves, and the digital components of the phone which do all the complex math for decoding signals into kitten memes and dancing TikToks on your phone. They are crucial for the operation of mobile phones, but are fairly difficult to build. The difficulty stems from the fact that RF products are not a single chip, they are modules filled with dozens, sometimes hundreds, of discrete components. The significance of yesterday’s announcements was that Qualcomm appears to be tying the performance of the RF components much more tightly to the digital components.

At heart, RF products are basically translators – taking analog signals (radio waves) and translating them into digital 1’s and 0’s. This is about the laws of physics more than the Moore’s Law. Historically, the RF suppliers have specialized in the analog side of the equation, and left the digital side to companies like Qualcomm who build the digital chips. Crucially, the Holy Grail of RF for years has been to find ways to use the digital components to help the RF chips do a better job of translation. Qualcomm appears to have actually captured, or least grasped, that Grail.

The telling product is the QET1700, a wideband envelope tracker. There is no simple way to explain what an envelope tracker does (this is draft seven of this paragraph), but it basically adjusts the phone’s power levels to best accommodate the RF signal. The easiest analogy is to think of the volume level on your headphones. When Spotify transitions from Bach’s Flute Sonata in B to Metallica’s Enter the Sandman the envelope tracker in this belabored analogy lowers the volume so that you can still capture the beauty of James Hetfield’s guitar chords without blowing out your headphones. There are ways to do this with entirely analog circuitry, but using a phone’s compute power allows for less circuitry and thus a smaller, more power efficient phone.

Qualcomm actually claims their tuning products use Artificial Intelligence (AI) to accomplish a 30% improvement in battery life. We are hugely skeptical of companies using the term ‘AI’ in their press releases, and Qualcomm is not entirely clear what they mean by this, but the point is that the company can credibly claim to use improved software techniques in their products now.

This is a minor product (it probably sells for a $1) and a 30% improvement in power for this one part of the phone is tiny when compared to the overall power budget of a device. That being said, it is hard to overstate the significance of this capability. For one thing, Qualcomm is likely just getting started. Their are many other ways they can make better use of a phone’s compute to improve RF performance. Envelope tracking is just an easy starting point. For another, none of Qualcomm’s RF competitors can match this. As noted above, the traditional vendors just lack much in the way digital capabilities. More to the point, to integrate with a phone’s digital circuits would require them to partner with providers of those digital circuits. Since Qualcomm is the one building those, the legacy RF makers are unlikely to get any support or even access to the digital side of the device. Not merely a flesh wound, competitively speaking.

Qualcomm has been slogging away at RF for more than a decade. We noted last year that since ~2019, they finally turned the corner and started producing real RF products. They have even started breaking the RF segment out in their earnings reports. Investors frequently ask us what has changed, and the mundane reality is that Qualcomm rationalized the management structure of their RF unit, after years of a very confused organization chart. Culture eats strategy for breakfast. So for several years Qualcomm has already been a real competitor in RF, and these new products hint at how much more formidable they are likely to become in the space.

7 responses to “The Qualcomm Shakes-up of RF is Beginning

  1. All this, and yet they still don’t have the high performance filters necessary to call it a “complete RFFE solution.” Using ceramics doesn’t cut it.

    • Yes, but
      Filters are a key value pool in RF, but Qualcomm doesn’t need them to capture share from other parts of the industry. They can win PAs, PADs, FEMs, and all the other modules, and source the filters from all the filter suppliers. They can still have a multi-billion dollar business without having their own filters. And they actually do have some filter capacity, so far they don’t have much to show for it, but there is reasonable chance that improves as well. Last, but not least, there is also a reasonable argument to be made that if they can build a fully integrated RF front-end, they can actually reduce filter content in a phone. That is probably still far from commercial reality, but it as absolutely possible.

  2. Pingback: RF Semis Update | Digits to Dollars·

  3. “…and source the filters from all the filter suppliers.” That’s the real issue, isn’t it? For 5G frequencies, you have AVGO and QRVO, and maybe SWKS(they never do say exactly). Those three won’t be selling discrete filters to QCOM and cutting their own throats. That leaves AKTS, who has successfully produced filters at all 5G, WiFi 6 and WiFi 6E frequencies. All they need is a strategic investment from QCOM to get manufacturing volumes up to Tier 1 levels. Resonant is far behind and locked in a deal with Murata, and they have a history of overpromising and never delivering. I don’t consider them a source until proven otherwise.

  4. Back in 2015, Qualcomm bought the assets of Nujira, a Cambridge UK-based envelope tracking system solution (analog+digital). I was consulting to Nujira at the time. Their technology was best-in-class; they had a PA characterization system that stressed the PA in various dimensions and created a pre-distortion lookup table for the PA, then wrapped their analog envelope tracker around it. The business model for envelope tracking without a PA is tough; a quality ET solution turns the GaAs PA into an op amp, and no one at SWKS or QRVO wants that…

    I’m guessing Qualcomm has incorporated the Nujira IP into the cellular platform. They have been using the AVGO ET IC for years to get their feet wet in the technology. As you say, Jonathan, this could be the tipping point for Qualcomm setting a new model in cellular chipsets where the RF and baseband are inextricably linked.

  5. I remember Nujira well. And I remember how frustrated they were in dealing with these issues with all the modem makers (obv. prior to their sale).
    But as long as we’re going down memory lane, Qualcomm actually
    had ET prior to Nujia. They acquired an ET team. I think it came out of TDK (prior to their acquisition of Epcos). That team had a good product for the time, but it got drowned in the organizational dysfunction. So then Q had to go out and buy Nujira. No idea which one of those made it into today’s product.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s