I can report that both my devices are working as expected. I don't remember anymore which version of the OEM firmware I was using before. At lease one of the devices was at a point where I needed to install from the rescue-uboot service.
The LED over the wifi symbol also works (as soon as I de-disable the devices in /dev/config/wireless). Adding the config shown above did not change anything.
The AC calibration data reads FF FF on devices with not working AC wifi, indicating a corrupted art partition, however, they have AC working in the stock firmware, so the only conclusion I can have is D-Link have the calibration data elsewhere on these devices.
@jackcolentern that is exactly what I am trying t determine. But comparing a C1 to C3 is probably not going to help us.
So far we have two C3 series I and one C3 series P which work out-of-the-box with the current firmware. On the other hard, there is at lease one series P which doesn't work out of the box. So far it's a poor sample size, but the theory that there is a hardware change in the C3 series P devices seems to be invalid.
It's unfortunate, because I find these devices to be quite decent in price/features (especially for a mesh network, which is my main use). It would be really great to find a solution.
I'm definitely not an expert... Is it possible to integrate @blabla's solution in post 33 into the firmware and at the same time support what's working so far?
If calibration data is stored in another place on some devices, the solution will be to try loading from both places. I saw that approach on some other targets somewhere.
The BMI id is contained in the calibration data, It can be extracted using a bit mask (31bit '1', or 0xFFFFFFFE), the driver loads the correct board file according to the bmi id.
If the calibration data we got is wrong(in this case it is all FF FF FF FF) , the bmi id becomes 31 ( 0xFFFFFFFF & 0xFFFFFFFE = 0xFFFFFFFE or 31).
The workaround will make AC wifi "working" but without the correct calibration data, the signal strength, range and speed will not be as good as the original software ( the effect might not be too noticible in everyday usage)
I've downloaded 3.10b02 and 3.11 EU firmware for DIR-842 C3 from https://eu.dlink.com/ website (PL and UK versions both led to the same files). Unfortunately, after a brief look I don't think they use similar encryption scheme to the one used in DIR-850L as described in the blog post you've linked - entropy suggests D-Link have used something better for DIR-842.
BTW. I've recently bought DIR-842 rev. C3, with S/N SY6A2IA00xxxx. Originally with 3.10EU firmware, now running 3.12EU after OTA upgrade. I haven't tried installing OpenWrt yet, but I'm preparing to do so. I just need to make sure I'll be able to return to stock if anything goes south (I've recently bricked my mir3g, after regular snapshot upgrade - probably only soldering serial pins can help me).
I just bought a couple.
P/N: EIR842MEU....C3E
H/W Ver.: C3 F/W Ver.: 3.10EU
S/N: SY6A2IB0XXXXX
And I cannot flash the firmware image, I select the file, press upload, a windows appears, followed by Firmware Upgrade failed!.
Too bad as I am urgently looking for a replacement of the TL-WR1043N which seems to have been EOL'd.
After twiddling with the board file as mentioned, I do get the wireless 5GHz.
I also opened up the router and soldered a serial connection to it.
I searched through the GPL source and found the login/password to use to acces the console. If anyone is interested in the bootlog or such?