Topic: Tascam DR-100 MKII / a limiter bug?

[jimbee]

In testing the DR-100 MKII, I've noticed that in MONO mode, with a mic in the left XLR and no mic in the right XLR, with the limiter switch on the back set to OFF, the limiter still seems to be engaged.  And it is very ugly, with overmodulation at -6 db, the wave forms hitting a brick wall.  With the limiter switch set to ON, the limiter works like I would expect, the wave form can approach 0 db and it sounds much better.

In STEREO mode, the limiter is off when the switch is off, and on when the switch is on.

Maybe I should just use stereo mode as the workaround, but for recording film dialogue with a shotgun mic on a boom, I would prefer to have the two duplicate channels of audio that you get in mono mode, rather than dialogue only in the left channel and an empty right channel that you get in stereo mode.  And I would like to be able to truly turn off the limiter in mono mode if I want to. 

If someone else has noticed this, or could reproduce the results, maybe Tascam could fix it with a firmware update.

Any advice appreciated,
jimbee

[DSatz]

A belated reply, I know, but I'm just now reading up on the Tascam recorder. Are you certain that you weren't overloading the input of the recorder during the test you described? This would depend on your test signal voltage and the setting of the mike sensitivity switch.

Pardon me if this is something you're thoroughly familiar with, but with most recorders, there is a range of signal voltages high enough to cause distortion even when you have turned the rotary control down such that the meters never even get close to 0 dB (full scale). This is because the first thing the signal encounters within the recorder is nearly always an active circuit stage with a fixed level of voltage gain (set by the sensitivity switch if one exists in the recorder)--and like all active circuits, that stage has a maximum voltage that it can put out without distortion.

The rotary level control is usually placed after that stage. If you reduce it to keep the meters from reaching full scale, you are merely reducing the level at which an already distorted signal is being recorded ...

--best regards

[runonce]

The description of the mono limiter "off" - does not sound like the behavior of a limiter.  Sounds more like overload as described in the previous post.

If the mic input has low/high switch, might try the low setting to reduce chance of overload - and try the test again...

[sk-1]

Here's my speculation on what's happening. Mono mode takes the L and R inputs and sums them to create a mono stream which is then written in duplicate to both channels. The -6dBFS wall that you hit is probably the result of 6dB attenuation applied to both channels before summing to ensure that the summed signal doesn't go over 0dBFS. The point at which the channels clip is probably the same in mono mode as in stereo mode, except that the 0dB clip point is shifted by attenuation to -6dB creating the impression of an artificial 'brick wall'.

This could be tested by setting the unit to stereo mode and feeding into one channel a constant tone signal that meters at for example -12dB. If you then switch it to mono mode, and the above explanation is correct, then that -12 signal will drop down to -18 (and -6dBFS will be the new clip point). 

With the "limiter" (such as it is, presumably performed in DSP and therefore ultimately ineffective) engaged, I'd guess that this -6dB attenuation isn't applied, with level control instead left to the limiter algorithm, hence a single channel's ability to reach 0dBFS in the absence of a right channel signal.

That's my take on the situation.

edit: clarity
 

[ale]

true! the speculation is (almost) right!  A limiter implemented by DSP is useless because a loud signal has already been clipped by the AD converter.
The channels sum happen in the digital domain, where the signal is already "limited" by the 0dBFS issue. This is very ODD!
Another problem is that the metering happen AFTER this process, inhibiting the meter to show the real input level of the converters and detecting the overs.
This second part is the only one that can be solved (maybe) via firmware.

Basically in the single mic scenario, the only way to deal with this problem is to keep the signal below -6dBFS, while on that tiny meter, is really hard! My solution will be the use of an Y XLR cable to feed both inputs with the same signal and setting the DR to stereo mode. Another advantage will be the ability to set different levels on the two chs in order to better manage some occasional clips.

Much better could be the ability to save a single ch to a simple mono file halving the needed storage. One must buy the 702T from sound devices to have this incredibly simple feature...