5G WAN failover gateway check

I have a failover WAN setup on an RB5009 pretty much identical to the example here Failover (WAN Backup) - RouterOS - MikroTik Documentation

On one ethernet port sits the main FWA antenna, with an external IP, via PPPoE.

On another ethernet port sits the failover 5G modem, with its own local address. The RB5009 port is 192.168.178.2 and the 5G router is at 192.168.178.1.

In one direction the failover works: when I disconnect the main FWA antenna, its external IP is no longer reachable so the router switches to the 5G.

But when 5G (sim) is down, the attached router is still up and still has its own IP 192.168.178.1, thus leading to a failure in detecting a loss of ping to the address.

I have to manually disconnect the ethernet cable in this case for the failover to happen.

Is there a way out of this?

Instead of saying "pretty much identical", post your actual configuration, instructions here:

why would two different WAN connections from two diff providers both be on the 192.168.178 subnet, that raises red flags for me.

Concur need to see the config and also a network diagram may be helpful.

Why?
Both the FWA and 5G devices are routers and the 192.168.178.x is their LAN side address, it looks pretty much normal to me.

I know how to export the config. There's simply no need for it. My question is rather broad. Let me rephrase:

For a WAN failover interface whose check method is via ping, in the case of a 5G router losing the 5G connection (external) the router's OWN (internal) IP address obviously stays up.

Therefore the ping method fails because the router's internal address is still up.

This is valid for any WAN device with both an internal and external IP.

Not what I said.

Well, you should be pinging a known reliable host on the internet, not the local device, usually 1.1.1.1, 9.9.9.9, 8.8.4.4. and 8.8.8.8 are used for this pinging.

If you don't want to share your full configuration (no matter if for - inexisting -privacy reasons or for whatever other reasons you may fancy) post at least your current:
/ip route print
and
/ip route export
otherwise it will be impossible to attempt giving you some advice.

In case of need for recursive routes, check this:

I was missing or did not understand completely this part. Now it makes sense. I thought we were able only to detect the whole gateway being down.

Thanks.