PCM is a data format, not a manufacturer. It's not in the same schema as AKM, TI, Cirrus, etc.
Now, regarding "why would I hear a big difference" - it's hard to actually quantify your subjective qualification of "big difference" and it's hard to know that you're hearing a difference between the actual D/A stages; there's far too many variables being changed to make that comparison valid. That's usually the problem in comparing two devices when you don't have lab equipment.
Regarding the immediate jump to the logical fallacy "well a more expensive one exists, it MUST have a purpose" - dismiss that, it doesn't help. More expensive options will always exist, and that doesn't make them better or worse, they're just more expensive. It's a completely disjoint conversation. It's like if I tell you that any car that can do 75 mph in a safe manner is suitable for on-road use (if you're not an American, 75 mph is the speed most of our Interstate Highways are graded for, it's like 120kph), and you respond "nonsense, if that was true there'd be no reason for anyone to own a Maybach, we could all use Honda Insights" - of course we COULD all use Honda Insights! Over-building and over-specing a device doesn't matter when it comes to serving a specific need; just because it *can* go 150 mph or measures 3x lower distortion or whatever else, it ceases to matter when you can't audibly perceive that difference (or legally go that speed), or when there's too much other noise/interference to make that distinction possible (both are true).
Regarding the "sound signature" argument - that's a bit more contentious. It depends on what's going on inside the device; some components apply a "house curve" (noise shaping, EQ) to their signals in order to produce an effect that the designer liked. Some devices simply have a hotter output which makes them identifiable in non-controlled tests (and usually they "win" those comparisons).
Regarding why TI and Wolfson and so on produce so many different models - different applications. For example the chips that go into your car are different than the chips in your CD player, and the chips that can talk USB are different than the chips that can talk I2S. There's also different feature-sets; does it have to decode 24-bit signals, or 20-bit signals, or 16-bit signals? How physically big or small can it be? How much power does it consume? Things like that.
When you get into "hi-fi" though, there's a relatively small pool of chips that are picked from, and they're used very frequently. Like TI's PCM1704. It does not change because it is in a Yamaha or Denon or Krell device; it's the same 1704. Now the device it gets put into may have some sort of "house curve" or some DSP adjustment/correction suite going on, or it may have better shielding, or it may have a hotter output, or it may have a better volume control scheme, or whatever - those things can make a difference in the right context. That still doesn't change the 1704.
Finally, when you look at the actual specs for these converters, most of them (and this is a "modern times" kind of thing) exceed the requirements placed on them by the material they're playing back (like CD audio), or hit the limits of physics (that 120 dB SNR line) - so sure you might gain a dB or two of SNR or you might drop a millionth of a percent of distortion, but it's already so far below the overall noisefloor and below being audible that it stops mattering. That doesn't mean some people don't want to pay for the piece of mind associated with a fancier component. Finally, some of those fancier components do a lot more than just DtoA - like the Benchmark DAC, which is also a headphone amplifier, input selector, and so on. It's not just a raw DtoA box.
So what I'm saying is, approach things logically. Know what you're getting into. Any (modern) DtoA chip by itself is going to be "good enough," so as long as it's in a device that's competently put together (meaning it won't fall apart in a month, and it doesn't have grounding issues, and it won't damage your other equipment, and so on), it should still be "good enough" - there's no point in spending a million dollars when $100 can solve the problem. I don't think any of the components suggested to you in this thread fit into that "bad" category - they will all be perfectly transparent as a DtoA, but many of them have different features (for example the X-Fi has the phono preamp; that isn't a DtoA function, but it may be useful to you, the HeadRoom devices usually have crossfeed, the Audio-GD devices are supposed to have extremely robust amplifiers, etc).
Quote:
But, why would I then hear a big difference in clarity between my Digigram vxPocket (around 800$ back in the old days) and the internal soundblaster in my laptop?
Only the headphone amp?
If that is true, nobody would need Lavry converters, a soundblaster would suffice...
There are also DAC-s like Burl (AD/DA) that are described like creamy sounding, is that only because the analog stages after the converter (transformers?)?
So, are you then saying that there is no difference between Wolfson and PCM or any other DA chip?
Why would Texas instruments then produce so many different ones?