Unless I'm mistaken, it seems the limiter in the Tascam DR-05 recorder is digital rather than analog, and therefore is not any help in avoiding clipping. E.g. If a 1dbFS signal causes clipping without limiting, it will also cause clipping when limiting is turned on, because the clipping happens in the A->D conversion before the digital limiting. (Edit: Read on to discover that I later concluded there can be some use to digital limiting)

I did a quick test to verify this... I played a song from my ipod into the dr-05 through an 1/8" cable. I had the volume adjusted to a level that just barely caused the Peak LED to flash (Aside: Beware that the input on the DR-05 does not accept true line level signals; its input clips below consumer-line level). Turning on the limiting function did not seem to reduce the amount of time the Peak LED flashed.

So is Tascam's marketing message:

TASCAM handheld recorders have a limiter during recording,
so that sudden loud sounds won’t be distorted.

totally incorrect? I could see advertising the limiter as a compressor to smooth out vocals or something, but unless I'm mistaken a digital limiter cannot have that effect, unless implemented with a higher-dynamic-range A->D converter before the limiting which the DR-05 clearly doesn't have.

At any rate if they wanted this to be a legitimate feature they should publish the limiting ratio (e.g. 10:1) and the level at which it engages. They don't so I'll have to assume this is a fantasy feature.

I don't mean to be downbeat on the DR-05; I think it's the best $100 that could be spent and quite competitive with the H4N.

I did some more testing, feeding in a continuous 1k tone. Without limiting engaged, clipping happened at -14dbu on the meter and I clearly heard the clipping at -12dbu. With limiting engaged, the peak light on the meter never engaged even with 8dbu of input, but I started to hear clipping through the headphones at -1dbu with limiting engaged.

Wow, this is cool... seems to be working. What would happen if I turned the input level setting to something other than the 22 which it was for the above test?

With input level setting at 0, the peak light goes on at -3dbu, and I hear clipping at 0dbu. Turning limiting on, and the peak light never goes on even at 8dbu, but I hear clipping at 0db. Just the same as without clipping.

With input level setting at 90 (presumably highest digital gain), the peak light goes on at -47dbu, and I hear clipping at -44dbu. Turning limiting on, and the peak light never goes on even at 8dbu, but I hear clipping at -33db.

These results encouraged me to check out input level 10: the peak light goes on at -9dbu, and I hear clipping at -5dbu. Turning limiting on, and the peak light never goes on even at 8dbu, but I hear clipping at 0db.

To summarize:

Input Level db of limiting
--------------- -------------------
90 11
21 11
10 5
0 0 (this seems to be digital gain unity, so no digital headroom for limiting)

So what it looks like is that this recorder's input level control is a digital gain, and when gained up digitally that digital gain can be limited away effectively. But if you are already maxing out the recorder input level with a hot signal, then you don't have room so no or little limiting will occur even if you turn on limiting.

So good news is that limiting might be useful, bad news is that's only if the signal is degraded by digital gain from the input level control. This experiment is making me think maybe instead of turning on limiting at input level 22, I should adjust my gain structure so as to feed in a hotter signal at input level 0, to maximize the recorder's dynamic range without unneccessarily adding the slight distortion that limiting adds.

I don't like that the peak LED never goes on with limiting engaged... that peak LED sometimes alerts me that I need to turn down gains and I would miss it.

But I wonder if engaging limiting would be a workaround for my laziness around making a cable to pad my mixer output to the input level 0 (unity or no digital gain), instead of the current cable I have which pads it to input level 22.

To explore that (indulge my cable-making-laziness :-), I wanted to see just how much the limiter changes the recorded signal level. So I fed in different signal levels (with input level set to 90) and looked at the recorder meter. Basically the limiting seems tuned well in that it doesn't change the signal much until you get within a few db of the clipping point. Then the limiter gives an extra 14db of headroom:

Input signal Meter (limiting off) Meter (limiting On)
---------------- ----------------------- -------------------------
-33dbu clipping heard clipping.
-34 clipping -2dbfs
-46 clipping -2
-47 clipping -2
-48 -2dbfs -4
-50 -4 -4
-52 -6 -6
-54 -8 -8
-56 -10 -10
-58 -12 -12

So this seems like a reasonable "soft clipping" form of limiting with a ratio somewhere in the neighborhood of 6:1 or 7:1, probably a bit less harsh than the 10:1 limiter in my SD mixpre, so should compliment it well. I think I will engage the limiting function on my DR-05.