@BobC:
Have a look at this rather long post of a few years ago, contains a lot of very good information relating to your current situation.
https://forum.kitz.co.uk/index.php/topic,22086.msg379336.html#msg379336As previously stated in this post, from personal experience on my line and as reported by many forum patrons, I don't believe that DS ES rate is the main criteria DLM considers with respect to DS G.Inp re-tx profile selection.
Have never seen any definitive explanation as to exactly what modem factors DLM monitors for re-tx profile selection but have found that significantly decreasing DS FEC error rate by modem DS speed capping the only way to achieve the desired result of low DS re-tx profile selection by DLM.
Of course, modem speed capping also reduces other factors such as G.Inp LEFTRS which is also believed to possibly be a DLM re-tx profile contributory factor.
My line would certainly sit for weeks/months with zero DS ES/SES rate and DS G.Inp re-tx profile high with DLM making no attempt to revert to low profile.
Did find that after a long period on re-tx high profile that quite a severe DS speed cap was required and DLM action could take a considerable time to revert to re-tx low profile.
Once DS re-tx low profile was attained found that DLM was more tolerant to higher DS FEC rates without re-application of re-tx high profile.
Regarding the modem speed clamping process:
I have always manually dropped the PPP session, powered down the modem and unplugged the data cable.
Re-powered the modem and entered the desired clamp rates from my Win 10 PC via Telnet.
Plugged in the data cable ensuring at least 30 mins. had elapsed from initial modem power down.
As I'am sure you know, the modem speed clamp won't survive a modem re-boot, will have to be re-entered.
I believe, entering the modem speed clamp command on a synch'ed line will cause an immediate re-synch, have never tried.
As advised, would still wait until DS 3dB SNRM is achieved, assuming your line can support this, before starting any speed capping operations.