there's a lot of different di's available and they have different ways of "getting there". tube-mains powered /solid state-fet-phantom powered / transformer- passive. all sound different and have advantages.

the main plus to a di over the direct input on a pre is you can run the signal long distances through a snake and you can split the signal to direct / amp head.. maybe the best way to go about it would to rent or borrow a few different ones to see if anything is an improvement over the isa (i'm doubtful about that).

Kurt has identified a couple of the factors in choosing whether to use the built-in DI of a pre-amp or an external DI box feeding into the microphone input. I have 5 or 6 different DI boxes, from which I would pick two or possibly three to try on a given sound source (usually guitar), and amongst those, I would also have the option of the pre-amp's built-in DI as another sonic variant.

If it's live sound, I would normally prefer DI boxes on the stage in order to use low-impedance microphone cables back to mix position. However, I have come up against problems when changing acts (as in a milti-performer folk evening, for example), since you may be swapping from a piezo pickup output to a high-level pre-amp output on the guitar, and it's very difficult to set DI box attenuations to suit all comers. In these instances I tend to go for Radial transformer-based (passive) DI boxes to minimise the chance of overload. However, any transformer input does not give the very high impedances that piezos really need.

There are two other points about using external boxes versus DI inputs to pre-amps. The first is that the pre-amp DIs may themselves not have a very high input impedance, and so not give a good sound with passive pickups, with the DI inputs on API pre-amps being an example. Secondly, using an external DI box via the XLR input of a pre-amp usually uses the pre-amp's microphone input circuit. If this is transformer-coupled, it can impart an extra degree of sonic colour that is not available via the same pre-amp's DI input. Again, the API pre-amps are a case in point. An active DI box feeding a pre-amp's transformer input gives a very different sound from a transformer DI box feeding a pre-amp that has an electronic (non-transformer) input.

Apart from not running long cables from a guitar, there are no hard rules about DI inputs or DI boxes; you select whatever route gives the sound you want.

Interesting info about transfo and impedance.. I know that my LA-610 has 2 Di impedance to choose from but haven't really test it..
I Wonder if the ISA di goes through the transfo..
I'm gonna test the difference with my actual di.
It's only for studio work.
If it makes a good difference I'll invest in a radial. they seems like the most thrustable maker...

I'll add that in the case of other preamps, you need to be sure that the 1/4" DI on them will actually tap into the same path of circuitry as the XLR input does. For example, on the ADK AP1, the 1/4" instrument input on the front doesn't route to the input transformers, it routes to the OpAmp only... so if I want to record something like bass through the pre, and I'm looking for the sound of a Lundhal 1538 or a Sowter 920C XFO, I don't use the 1/4" instrument input, I use a direct box and then send it to the XLR input on the back of the pre, which does route the signal to the input XFO stage.

FWIW

My initial tests shows that there is very little difference between the Di and a from DI to mic input sound on the Focusrite ISA. Other than being out of phase. (might have been my Di fault which was a cheap one. Testing the result with a null test showed very small difference in the HF (0 to 1 db) but not a strait line, more like random saw shape of equal and not equal peaks here and there.

But this is not a proof that a good external DI can't make a great difference. I'll need to test it also I guess.
At least I know that the DI on ISAs are not cheap.. they sound very good !!
Dawm ! I just sounded like a sells rep... where's my commission ?? Common Focusrite, add me to your payroll !!! ;)

I guess for me, I consider the quality of DI boxes in the same way that I do with cable; in that as long as they do what they are intended to do, and I don't hear any problems, then I'm happy.

If it doesn't mess with the gain, or add noise or hum, ( and is preferably in-phase ;) LOL ) then I'm satisfied.

I'm not implying that there aren't garbage models out there; like anything else I suppose, there's always a chance that you can come across the occasional lemon...

I currently have 3 DI boxes; a Countryman, which I purchased around 15 years ago for my acoustic electric for live performance; I have an ancient ProCo that I think was made when Lincoln was President, that I carried with me for years on the road back in the 80's when I played bass - and while I never ran it over with a lawnmower - it has been stepped on, thrown around, tossed in and out of milk crates and equipment trucks, has dents, dings, and a permanent duct tape rash; and the third is a Samson model that I bought a few years ago at a yard sale for $5.

Two of them are passive, and one, the Countryman, is active ( requiring a 9volt battery), yet all three sound just fine; and honestly, I don't hear any sonic differences between them. I suppose it's possible that they would show differences if they were each put on a scope, but as far as what I am able to hear in terms of tone, they all sound exactly the same to me, they are all quiet, and they all get signal from point A to point B without any problems.

IMO