I took delivery of 3 R5H cards to test against the beloved XR5 and found that they all reported a noise floor on the bench of approx -83 (+-2dBm) across the whole 5GHz band. This was on an RB411 running V3.16 in a 19dBi Rootenna in the office. I tried the 3 cards in sequence and saw the same thing. I then changed the pigtail and saw the same thing. I then put in an XR5 and saw a noise floor of -94 to -106 across the 5GHz band. I went ahead and deployed the XR5 bundle and packed up the R5Hs for RMA.

Unfortunately I am now at home so I can't tinker with them anymore but has anyone else seen this?

I have several R5H running and i got no less than -90/-89.

Same problem here, R5H with 3.17 give me a -83 noise floor, but work with no problem even with a -6 dB S/N so maybe it's only a cosmetic bug?

Regards,
Giorgio

Maybe, but I couldn't trust putting it into a backhaul.

I have already some 20 pcs of R5H at production.
Noise level of R5H cards needs some explanation indeed. "Frequancy usage" tab gives usually range from -87dBm to -83dBm.
However, getting decent connection even at RSSI of -85dBm is still possible.
This indicates that the noise reading is faulty. The card is reporting it from wrong register, it seems.

There is more - you can choose at software either A or B connector, with no difference of course.

My conclusion is that R5H needs some driver tweaking for RouterOS - choice of antenna connector should not be available, and noise figure corrected.

IMHO, using R5H for BH is safe and most beneficial, considering the ESD protection and adjacent channel filtering improvements.

Confirmed so.

Mikrotik engineers - can you acknowledge it?

I went ahead and deployed the XR5 bundle and packed up the R5Hs for RMA.

For me, the only numbers which matter are speed, reliability, and cost.
For my tests, I check how fast the link can transfer data, not just the noise floor and signal level.
It's not unusual to get better performance out of links which may appear weaker.

I hear you, but these numers are key performance indicators on which descisions might be made!

Little problems like this just smell bad. We lived through years of bad Tranzeo firmwares.

I suppose we should give them time for the product to mature. It could be a great card yet.

Mikrotik, any input?

these numbers should not be used as "key indicators" to anything, this applies to any card you use

There was a software problem with interpreting the data from the card, we made a fix for the next version