Detailed Topic Index

Saturday, January 8, 2022

Pitfalls in DFA a1 - Polar belt position

 See also this for another example

Although already discussed in a previous post, a recent series of examples have come to my attention illustrating the importance and need for ECG wave inspection in cases where DFA a1 does not behave as it should.  In the cited post, I spoke about the importance of sharp, high signal to noise ratio R peaks.   Since measuring "correlation properties" of HRV, AKA, DFA a1 requires exact recognition of beat to beat patterns, sub optimal precision of the RR interval will be a problem in recognition.  Please read though that post first.

Let's look at a few new examples then some recommendations.

First is from a participant in one of our studies who was wearing both an H10 and Movesense ECG during a session, cycling near the AeT.  We were analyzing the HRV and noted that the a1 appeared way too low for the intensity performed.  I then took a look at the ECG waveform since we had the Movesense data.  Here it is:

  • The bulk of the signal voltage was negative!  You would think that Kubios would auto invert this (especially since that was chosen in the preferences) but it does not.

I re imported the data with "reversed polarity" and here is what it looked like:

  • Does this make a difference with DFA a1?

Here is a plot of the identical ECG (standard and inverted) and the simultaneous Polar H10 RR series:

  • Both the standard ECG and H10 derived a1 data run low, but the inverted ECG (which is identical to the standard ECG, just reversed polarity) runs higher, near where it should be.
  • This illustrates several points:
  • Kubios auto polarity selection does not work as it should.
  • The Polar H10 does not have the ability to auto sense optimal polarity - we will need to do this ourselves!
  • Simple signal inversion (polarity reversal) can lead to substantial a1 changes.

How often does this situation occur?  

I simply don't know.  However, here are another couple of cases I've seen.

One of my blog readers noted low a1 at low cycling intensity.  Here is the Polar H10 waveform recorded from Fatmaxxer:

  • Both samples show very low relative R peaks with most of the signal in the negative range.
  • The second tracing has the R peak on the wrong segment of the QRS.

The H10 was inverted (Polar logo upside down) with the module centered (at the sternal notch). An additional test of 1 inch left placement was also compared:

  • We now have a dominant positive voltage with better R peaks (either is ok).  
  • Repeat ramp and a1 measurements are pending for this individual.

Next Example - a friend and cycling coach who noted many artifacts in the RR series during high intensity.  When H10 was recorded with the Polar logger app, the R peaks were low and Kubios was having trouble deciphering signal to noise:

  • During high intensity, the chest/muscle motion related noise is overwhelming the mediocre ECG signal, producing artifact:

  • Moving the module about 1 inch left wasn't enough for best signal (good but not great):

  • However, The module placed at the level of the left nipple gave us twice the original voltage:
  • End result - Best voltage so far and now we see visible P waves (red) - this is where I recommended he keep it.


Takeaways so far:  

  • DFA a1 calculation can alter depending on signal strength, waveform shape.
  • Module/belt position can affect a1 measurement greatly.
  • It's possible to do this yourself (next part).

How to optimize signal and waveform:

You will either need Fatmaxxer (free, android) or this app for iOS.  There may be other options for the PC as well.

Use cream or gel on the H10 sensor pads

Try module centered with Polar logo right side up (default position).

Analyze the wave - if it looks good (R peak sharp with minimal negative waveform) you are probably good to go.  

If the negative part of the wave is greater than the positive - try just inverting the module.

Repeat analysis (after inversion).

Optional - move the module 1 inch at a time to the left, potentially as far as the nipple and measure the waveform at each site.

Compare all measurements and find the best choice for you.


A walk through example:

An individual with a mediocre looking "standard" Polar belt waveform.


  • Not bad, but negative signal>positive signal, can we do better?

Inverted module/centered:

  • As expected, the mirror opposite as above.  Slightly better?


Standard module (not inverted)/1 inch to the left of center:

  • Looking better than center.

Standard module/2 inch to the left of center:

  • Success!  Normal looking QRS, best positive voltage!


As expected, the inverted module/2 inch left has poor positive signal (since it's now inverted):


Conclusion - moving the module 2 inch Left yields the best QRS waveform and signal strength

What does Polar have to say?

Yes, they recognize it also.


The developer of Fatmaxxer has graciously added a feature to help us decide on optimum signal strength without the need to graph out the ECG.  The peak R voltage is now displayed (see below in red circle) beat to beat.  This is accurate as long as noise is not present.  My recommendation would be to try various belt positions at rest, wait several seconds for the signal to stabilize, then note the voltage.  If you are seeing artifacts detected, that could be a sign of more noise or signal distortion (so wait for no artifacts).  

Find the best voltage and then monitor it as you exercise.  It should not change very much.  If you are interested, a visual approach is still the best IMO, but this gives us a quick, accurate option.  

What are good voltages?

In my experience try to aim for above 1000 uV.  If you are seeing <500, chances are that the HRV and even HR may not be captured properly.

Further comments and examples will be added and this post amended.


  1. I've been observing "low a1" in all my ramp treadmill tests with the H10. Unfortunately, seems like Fatmaxxer hasn't implemented ECG into the Android app yet, so I think best option is to change position during live exercise and see if a1 goes up or down during steady state effort. Not very accurate.

  2. The current version of Fatmaxxer on Github is what you need - the playstore version is very out of date. Even so, better belt position may not fix the "low a1 while running", so try doing a ramp or equivalent on the bike instead

    1. Thanks! Found and installed the github beta with ECG. Will test further.

      Until I found this post, I was on the verge of completely upending my Z2 running training for triathlon to drop my target HR ~10-15bpm. I'm glad I didn't yet! Besides belt position, what other causes may there be for "low a1 while running" that I could try to debunk?

      I've run a number of successful bike ramps for AeT. Is it fair to assume that bikeAeT = runAeT?

    2. The low a1 while running continues to baffle us. It's somehow "artifactual" in that it does not occur in the same individual with biking or stair climbing (no impact). I don't think it's belt position but optimizing voltage will give you a more precise a1 (so it's always good to do). Yes, bike thresholds should equate to running thresholds within the accuracy of the metric (a few bpm). Good luck with your training.

    3. As a follow up to my "low a1 on treadmill" reports, I'll note that I'm seeing much more realistic and consistent a1 result running on the road. Baffling.

  3. Hello, have you done any test if the garmin hrm-pro HRV is accurate ?

    1. Sorry, no testing or experience with that one.

    2. I tested both and H10 was noticeably better. So much so I won't trust my HRM Pro for HRV anymore since I purchased and tested H10.