Was there any changes in the readings when you test the supply voltage in AC mode
at the cluster? If the previous 28V reading was lowered or became zero, then there was a grounding
issue too, but it was not related to the problem...
It still might be a wiring fault between the firewall connector and the back of the
instrument cluster. Disconnect the speed sensor and the connector with two wires
at the back of the instrument cluster. With a piece of wire short the disconnected
connector pins at one end. At the other end of the cable put the multimeter leads
into the connector and check continuity. If you got low readings (0.0something KOhms)
then the wiring harness can be ruled out.
If the power supply is free from voltage spikes, then the fault most probably inside
the instrument cluster.