Your first problem is the statement of the subfloor being borderline.Backer board might add a miniscule strength, but with tile, if you have bounce and give with the subfloor,it's going to be a constant maintenance issue with your grout joints. Your first consideration would be to get your subfloor structurally sound enough not to have any bounce/flexing/give.

After that, come back and get advise from a flooring pro here regarding the backer,self leveling part of the show.

nothing beats a mortar bed installation.

It's how stone has been installed for thousands of years and still how the majority of the world installs stone and ceramic tile, and considering its historical longevity, it's a no brainer, but in it's use, the cost may be prohibtive when compared to thin-set and backerboard technology. Also it's hard to find mud-bed isntaller in the USA and expensive when you do find them.

As to your subfloor movement, or deflection, the industry standards are as follows:

The Tile Council of North America recommends a deflection of no more than L/360 of the span for ceramic tile, and the Marble Institute of America recommends the deflection of the subfloor/substrate to be no more than L/720 of the span for natural stone, or twice as much as that of ceramic tile.

To calculate deflection, convert the area span from feet into inches then divide by 360 or 720. The end result gives you the maximum amount the floor can move under an "expected load".

To give you a close estimate of your subfloors deflection rate, here is a handy tool at John Bridges website called: Deflectolator subfloor/substrate deflection calculator.

Ok now on to your choices if you dont want the problem of finding a mud-bed isntaller or the cost of mud-bed installations.

There are 4 products I recommend whole heartedly and proved to be suitable solutions when deflection is within the above tolerances:

Schluter Ditra - waterproofing, load-distribution, and ucoupling membrane

Wonderboard - Cementious Backer Unit

Durock - Cementious Backer Unit

hardibacker

It is important to note that no mortar bed, membrane, or backer unit will sturdy up a subfloor/substrate in terms of deflection, and that the substrate needs to be withing deflection paramneters BEFORE any of the tile system is installed to support that system. The subfloor/substrate needs to be within specified deflection prior to the installation of any tile system. It simply adds rigidity between joists to carry the load without it cracking, and in cases of membranes, it adds waterproofing, load-distribution, and uncoupling properties.

A self leveling compound will not add to any framing/joists ability to carry a load. In fact, it adds alot of extra weight to that system possibly limiting the load even more than any mudbed or CBU/FBU system. According to the specifications, the framing needs to be within specified deflection before topical coatings, menbranes, backer units, mudbeds, or thin set is added.

In conclusion, it depends on your framing and deflection rate as to what is best for you. I am not a huge proponent in SLC's for ceramic tile over a wood substrate simply because the way alot of houses are built today the extra weigh of the SLC can cause problems. I believe in fixing the woodworking and subfloor system or patching it to within height tolerances. Over concrete it is fine because concrete is much more sturdy. I just think some use SLC's way to liberally as a blanket easy fix, when they may very well work in one persons house and not well in anothers. I have repaired many tile jobs over SLC's over wood because of this. Every year we have a few customers whom we make cracked grout joint and tile repairs for because someone used an SLC over their wood subfloor as a base for their tile job because the added weight of the slc when combined with the added weight of the tile and thinset made deflection that much more, and they dont want to rip it all out and fix it right for the time, expense, and mess involved.

I am not saying anything in particular is right or wrong for you, or that any will absolutely succeed or fail, but I just want you to consider the information.

It all depends on your framing/joist system and its deflection, so figuring that out is the first step then you can decide from there.

Self leveling mortar (SLM) is certainly going to be more expensive than hardibacker. And, I'm not sure I would agree that the pros tend to use SLM more often. I've layed 6 floors as a DIYer - all of them on hardibacker, and everything has been just fine. As some of the previous responders have written, the subfloor is more important. Neither HB or an SLM are going to provide you significant additional strength. The only way you can get that is (1) adding joists; (2) adding blocking between joists; (3) adding to / replacing plywood.

The only reason I would use an SLM is if (1) the floors are unlevel to begin with and you want to correct this; (2) you are laying a radiant heated floor and need to cover the electrical wires.

In any event, if you still plan to use the Self-leveling underlayment method, I wrote an article on how to get self-leveling mortars to level on my blog.

Fred
One Project Closer