Hi Paulo,
Hobbyists always assume that they are following the advice correctly. They do the wrong things, the plants fail and then they automatically blame the advice. I know without a doubt what symptoms are associated with what fault. Falling and melting leaves are only ever associated with poor CO2 uptake. The reasons are discussed here+>
http://www.ukaps.org/forum/threads/what-causes-leaves-to-melt-and-what-to-do-now.20421/
With that as the basis of our troubleshooting methods we must then determine WHY there is a poor CO2 uptake.
If the diffusion method is at fault then there is a likelihood that more plants in the tank would suffer CO2 efficiency. The exception of course is if all or most of the other plants are exceptional CO2 feeders and if they can make due with lower CO2 concentration levels.
Carpet plants typically grow semi-submersed or in the shallows, not in deep areas, meaning that they are sometimes exposed to air and can access CO2 directly. They are therefore very poor CO2 feeders because they have not had to develop better strategies to accumulate CO2. Therefore, in a tank, they are susceptible to CO2 deficiency more so than other plants which have a better developed CO2 mechanisms.
CO2 concentration levels are at their worst at the bottom of the tank because the gas is always trying to escape the tank. There is a pressure gradient which is lower at atmosphere than it is in the water column, so movement of the gas is constantly up and away. The concentration level can be 10X lower at the bottom, near the plant beds than at the top of the tank.
The critical problem therefore is to do a better job of distributing the gas to the lower reaches. It's not very clear how the water is being distributed.
The timing of the gas is also in question. We do not know what time the gas comes on versus when the light comes on.
I have not eliminated the dissolution method as a possible flaw. I've only surmised that distribution to the lower reaches is a higher probability than dissolution method. If the timing is correct and if the pH drop is correct then the flow rate and distribution to the substrate should be considered.
The easiest way to check the diffusing effectiveness is to port the gas directly into the filter mistake tube and to measure the pH profile. Compare it with the profile using the current configuration.
CO2 is very difficult and I don't think the right questions are being asked to the OP. These are the critical paths to investigate: dissolution, timing, flow rate + distribution method, injection rate.
Cheers,