then quickly recover to continue recording the softer sounds of interest as before.
In summary, the likely usefulness of the LIMITER is itself very limited and may be considered for: 1) Elimination of clipping distortion caused by long duration (>.3 sec) non-repetitive overloads that exceed the normal signal by 6 dB VU or more and . . . 2) For audio signal dynamic compensation by hard-gain-limiting 50%-100% of the total signal for usable non-distorted voice intelligibility in an unpredictable noisy ambient.
Best NOT to use this setting if you want consistent live sound recordings. Stick to Manual (or even Limiter) for better results. While this newly designed AGC system works extremely well for music recordings in not producing even a hint of the usual audible pumping with bass rhythms, it has an 'elephant's type' of peak signal memory with not forgetting that last loudest peak and will virtually turn down the volume accordingly and permanently. NOT changing this recording level until the recording is stopped or at least put in pause mode.
In other words, having (even if inaudible) loud peak signal, the recording level in AUTO(AGC) mode will be adjusted to accommodate recording this particular sound without clipping, but often maintaining a VU recording level only good for one-time or very infrequent peaks. This same now remembered AUTO/AGC setting is likely way too low for best quality recording the primary sounds.
Another reason to not use the AUTO(AGC) setting is the same 'programming' fault found in the D7/8 AGC circuitry that consistently places the average peaks too low at -14 dB VU when the ATTENUATOR is set (correctly) at 20 dB. When the same Attenuation switch is set WRONGLY at 0 dB, the AUTO(AGC) records at a more desirable -6 to -2 dB average VU level, but it is now slightly and audibly clipped/distorted sounding.
4th Switch Option: 32KHz LP, 44.1K / 48KHz SP Sample Mode
Sampling rate is directly proportional to recording bandwidth. The faster (48K per second sample rate) is a better choice if 2,000 cycles of higher 22,000 cycle bandwidth quality is more important than later making a highest quality CD with no sampling conversion liabilities. I'd suggest always using the 44.1KHz switch setting instead of the 48K; especially when there's any chance of wanting to more easily make a quality CD later.
The lowest available 32K rate selection is best chosen when maximum tape recording time is most important and 12 bit non-linear A/D with 15,000 cycle bandwidth is more than adequate. At the 32K rate, tape time is doubled to 4 hours from the 2 hour running time at the highest quality 44.1K & 48K sampling rate. Be advised that 32K sample rate is not well supported for playback or recording on all DAT tape recorders, many decks will not play or record 32K tape.
Some Things Could've Been Better
Although these two virtually identical portables are nearly perfect in most respects for amateur and professional purposes, a few minor details may be improved on future models.
First, as previously mentioned, the Manual Recording Level knob is a even smaller and turns fewer degrees than on the previous D7/8 model. This makes critical smooth level adjustments more difficult, taking a fair amount of practice for most recordists.
Second, the hatch mechanism that opens the clam-shell lid doesn't open much to even notice it's open at all! This is a sharp contrast to the D8's ability to visibly and audibly indicate the latch has released and lid has popped opened. The photo at the right shows the lid latch popped open(!) and it's nearly impossible to tell its really open until some very careful prying opens the hinged lid.