Maybe we can try to figure out the reason step by step:
1. To use the same belt for testing between GH-625 and GH-625XT to check the problem comes from belt or device;
2. The measure HR with two conditions: statistic and dynamic to check it is algorithm or timing issue;
3. Starting timer of GH-625XT without wearing belt to check the peak comes from noise or not;
Another example below. Comparing a common ANT HR belt with the 625XT. F-GGH-2P-1111022 firmware.
A very slow run, quite cold so not much sweat. The data is clipped some for the first 8 minutes, both ANT and 625XT displayed high readings, even if I had moistened the bands and wore them for 30 min (with contact with clock) prior to running. The belts are much the same in this aspect.
When using the belt with zingos 505, the displayed value has been similar in both ANT devices.
The ANT readings are much more what I expect.
I have been out of battery for the ANT belt for a couple of weeks and have no other good comparisons.
Since the 11-11-02 update, the occasionally high readings have been improved. The high readings at maximum effort is much reduced (but I have only my perception to compare with). There are a few unexpected peaks in the data but no constant high values.
I have used the 625XT without HR (just to get GPS data to compare with the other device). The belt has then been in my backpack. Quite often there has been data from the belt, maybe one sample every minute about 70-100bpm. There has been moist training clothes in the backpack too, but I do not think they have been in contact with the belt.
No belt close to the 625XT, I get no data.
For me, the still a little high distance reading is a bigger issue. As the GPS data is more accurate to the position, this should be doable.
With the latest update, the data at faster runs are better than previously, less change at slower runs.http://hem.bredband.net/gerhardnospam/tmp/st3/delsj%C3%B6ar-hr.jpg
(Link will likely be removed in a couple of months.)