maybe with the eevblog 121 meter some compare would be possible it use the same ic ? no ?
AFAIK the EEVblog 121 meter uses the HY3131, but in combination with an analog RMS chip. And thus a different AC solution.
The XDM1241 seems to use the internal digital RMS solution that is lower bandwidth but ideally with less issues at a small fraction of the range.
There is a chance that the exactly 500 counts could be some kind of default setting and a cal step could have been skipped.
I see no reason to cripple the AC readings that much. Normally the standard circuit with the correct "offset" for the meters own noise should work well. The weak point is the limited bandwidth (the specs are only for 1 kHz). The limited BW also makes noise less of an issue than with the analog RMS. So even without cheating one should get a rather stable zero and OK readings even below 0.1 % of FS. The "cheating" is more from the low bandwidth, that can make it look better than a higher BW solution. So I see absolutely no need for cheating with a fixed zero - it's more like a bug or missed/failed calibration step.
A solid zero is more a solution that analog RMS solutions may use / need, as these by design work poorly at low amplitude. These often requite more than 1% or 5% of the FS to work. So there 500 count dead zone may be possible, though already a pretty bad case.