Optical heart rate of the Garmin FR 935: How accurate is it for cross-country skiing?

Hiking and cross-country skiing were the main reasons for me to get a smart sports watch. During hiking, I just don’t want to wear a chest strap, and I don’t want to exhaust my phone’s battery by recording with GPS, but I’d still be happy to track both my walk and my heart rate. While cross-country skiing, I’d like to have a device monitor at hand for which I would not need to stop, take the glove off, and reach for the side pocket. To that end, I got a Garmin FR 935. Because, why not? I figured that I’d pay 300,- EUR anyways for a well-performing simpler watch, so I could pay a bit more for a more feature rich watch while it was on discount.

Marketing worked on me. Or, to be correct, indirectly via my instructions on my Mum, who sponsored this amazing Christmas gift.

As it is winter, I didn’t go hiking yet. But I had been cross-country skiing for a week. That was a good occasion to look at how good the optical heart rate sensor of the Garmin FR 935 could handle cross-country skiing. In fact, you don’t find much coverage on cross-country skiing with sports watches in reviews of the common sources. They bike. They run. Maybe they swim. They won’t typically delve into winter sports, because it’s not often winter when they test. I come to suspect that Garmin’s development suffers from the same problem.

Some background

Optical heart rate measures the blood flow through your arm. The sensor should be placed with a finger or two distance to the wrist, and while it shouldn’t squeeze your arm like a device to measure blood pressure, it should fit snug. Lots of dark hair can be a problem. I have lots of hair (keep that in mind), but it’s light: I’m a blond German.

Lots of flex and vibrations in the wrist can impose much larger problems, however. This is the reason that optical heart rate sensors at the wrist continue to struggle with cycling outdoors (and, for this and other reasons, with swimming. Wrist movements while running are apparently much more predictable. In cycling, the hand touches the handlebar. With direct contact, it becomes part of the system. The system picks up random noise from the street.

In cross-country skiing, movement seems predictable from one perspective. You move the arm back and forth in a predictable manner, as long as you move. However, it’s not those big movements, but the micro-movements in the wrist. Those depend on the strength of the stroke, the softness of the snow, the exact point where the stick hits the ground… also in cross-country skiing, the wrist is part of the moving system, and I expect there to be more noise than in swimming. I just wouldn’t know how well the Garmin FR 935 handles it. After all, optical heart rate sensors have improved and now do okay in road cycling.

On the other hand, heart-rate measurement on a chest-strap still is the standard for most consumer applications. It requires moisture in the skin. On dry winter days, that can be a problem at the beginning of a session. Obviously, you can manually add some moisture (read: rub the chest with some water from the tab). The data itself is amazing; if it’s not, there’s something wrong with your strap. Chest straps also have a massive advantage in battery endurance. I usually need to switch my battery once a season. Optical heart rate devices won’t last for more than a week. In fact, the Garmin FR 935 supposedly could go for 24 hours of continuous GPS/heart rate tracking. That’s a lot for what it is (a smartwatch), but obviously nothing compared to a chest strap. However, this argument is also a bit pointless: if I were to use a chest strap to record my heart rate, and then send it to my watch, battery life limitations are the same. It’s just that I won’t need to charge the extra device that I’m using for more accurate heart-rate measurement, but I still need to charge the device that stores the data.

Some testing

In a separate test before, I found that the optical heart rate from the Garmin FR 935 was fairly accurate in non-active settings when I compared it with the heart rate from my chest strap. To that end, I connected the chest strap via ANT+ to my mobile phone. I use a chest strap from o_synce, which is a German company, and have been using the sensor for five years already. I only replaced the strap itself last summer.

I also did an indoor cycling workout, in which the Garmin FR 935 did okay. I was mostly within 2-5% of my chest strap, often closer, and only occasionally farther off. That performance was within expectations that I had from an extensive benchmark test by DC Rainmaker. I am therefore confident that any accuracy issue that I encountered is related to the activity that I am doing.

On my first activity in Deutschnofen, I only tracked heart-rate and GPS with the Garmin FR 935. There is no benchmark. I simply had forgotten to put on the chest strap. I would not need to have a benchmark to understand that my heart rate was likely higher than what I measured with the watch. My heart was pumping on two or three occasions. I was cruising at tempo – in the sense of what tempo is for me – and breathing with an open mouth. My average heart rate was only 121bpm, and the maximum reached 167bpm. Things looked not too shabby for the first half, when I was in hilly terrain, but still seemed a bit low. As I reached easier terrain, I found it to be substantially under-reading my heart-rate.

For comparison, there’s data from another cross-country skiing trip (Oberhof) with very comparable intensity. I measured heart rate on my mobile phone with a chest strap. In both visuals, heart-rate is the third plot. In the Oberhof activity, my heart rate was 141bpm on average, and 175bpm max. And that reflects my impression: the Garmin FR 935 seemed to underestimate my heart rate, particularly at lower intensities.

Obviously, numerous factors could explain these differences. It’s really worth nothing if it’s not a direct comparison (same activity, same day). So, that’s what I did. The next day, I did remember my chest strap, paired it with my smartphone, and then started to record both sessions at the same time. I had a look at my watch occasionally and observed the same low readings as the day before. At some point (i.e. after 30min), I switched the heart rate source on the Garmin FR 935 from its own optical sensor to the chest strap.

This is pretty much a tell-it-all picture. (I used DC Rainmaker’s analyzer tool to generate this graph. The watch is in red, and my mobile phone is in blue.) While I was actually doing some rather intense activity on the Marcialonga, the Garmin indicated that I was totally in recovery mode. I mean, 100bpm is a heart rate that I would see while walking casually around the house. Furthermore, while the pattern has some resemblance across devices, there’s obviously some large deviations as well.

My day on the Marcialonga was a pretty consistent activity. After I switched the heart rate source on the watch, there is this second area in which my heart rate is much lower than before and after, and where the mobile phone mostly did not even really record heart rate (hence the straight blue line). I was actually not really moving there (only twice to slightly change position), as I was having a phone call. Take-away: during the activity, the optical sensor of the watch showed readings that were very identical to those that I had with the chest strap while taking a break. But the watch accurately tracks heart rate during the pause, because it doesn’t auto-pause like the app (OruxMaps) that I use on the phone. Note that in its proprietary calculation of average speed, Strava excludes movements below a certain speed. Hence, auto-pause is not really necessary.

Anyways, to put it simply: you can throw away the heart rate data from the watch.

Some conclusions

I’m not the first one to notice this problem, and it’s not limited to Garmin. As one Fitbit user puts it.

Name a sport outside of running and walking that has consistently rhythmic movement of your hand.  How can I return this piece of junk?

I’m not going to be that harsh. He has a point, though. It’s just that these devices for many activities might not be the right source of heart rate information. Heart rate instead is still best measured with a chest strap. In fact, chest straps first came into being as a tool to help cross-country skiers train. (The Finnish equipped their athletes with a wireless heart rate strap already back in 1977.)

The problem is: Garmin itself claims that the heart-rate measurement is one of the selling points of its watch products. This is a statement from their own product website on the Vivofit HR tracker.

With Elevate™ wrist heart rate technology³, you can ditch the chest strap and measure heart rate from the wrist for all your sports except swimming, where optical heart rate isn’t available.

But, and that’s the point, no, you can’t. And in the foot notes, Garmin well acknowledges that. It’s a bit hidden. There’s a link to a disclaimer on activity tracking and fitness metric accuracy.

While our wrist HR monitor technology is state of the art, there are inherent limitations with the technology that may cause some of the heart rate readings to be inaccurate under certain circumstances. These circumstances include the user’s physical characteristics, the fit of the device and the type and intensity of the activity as outlined above. The HR monitor data is not intended to be used for medical purposes, nor is it intended to diagnose, treat, cure or prevent any disease or condition.

It’s certainly some sort of deceptive advertising. I personally think that they are more than over-selling it. Most users won’t complain. (Actually, they do and sue.) They will use it for running, where there’s no problem, and they will use it for cycling, where they know before they buy that there’s a problem. Triathletes, who use it while swimming, also know in advance it can’t accurately measure heart rate during swimming, so they won’t care or use a dedicated source of heart rate information. Other sports are a niche.

Just, it’s no drama. That’s not really the role of these devices. (You should click here. Amazing article on Polar’s own blog.) Take this final quote from Garmin’s website for the Garmin FR 935.

Wrist-based heart rate might be an easy choice when you want to move without restrictions on race day, but Forerunner 935 gives you the option to tap into much more data with the addition of a compatible heart rate strap.

You get it.

This watch is my monitor. And with it, I have all the advanced training functions that I could not implement with a simple watch. It’s all the information on my hand. It’s a wonderful tool that does exactly what I want from it. As long as I feed the watch with the right information, that is.

During hiking, I can easily rely on the optical heart rate. For daily stress and activity monitoring (in order to observe steps, sleep patterns, resting heart rate etc.), the watch does as good as any cheap tracker. While cross-country skiing, I should better use the chest strap. Other people could place an optical heart-rate tracker on the upper arm, where they are apparently much more accurate and comparable in performance to chest straps, regardless of the activity type. I still have that monitor at reach on my wrist, unlike the phone, which I now only carry for photos and for macro navigation.