How to get the best from the Zoom H2

Discuss the Zoom H6, H5, H4, H4n, H2, H2n, and H1. Please don't "post and run". Participate in the discussion. Thanks.
Post Reply
mhelin
member in good standing
member in good standing
Posts: 30
Joined: Thu Dec 13, 2007 2:22 am

Post by mhelin » Thu Dec 13, 2007 11:26 am

trevmar wrote:H2 HARDWARE DETAILS
The CPU is a TI320 VC509A. There is no clock Xtal trimmer, so your recorded data will always be a little bit deviant from an accurate timeline.
The DSP (CPU) is propably TMS320VC5409:
http://focus.ti.com/lit/ds/symlink/tms320vc5409.pdf

Too bad it's serial port doesn't support I2S, if it did support it then we could put a S/PDIF receiver chip there to bypass the A/D.

I think the noise comes from DC/DC (up-)converter mostly (in addition to EMI from CPU and various signals on PCB). Don't know if it is bypassed when you use the 9VDC input or not, but it would be nice to test the noise level with an external 9V battery vs. internal 2x AA batteries.
0 x

diwo
been here a while
been here a while
Posts: 87
Joined: Sun Nov 04, 2007 3:44 am
Location: Neumarkt, Germany

Benefits of 24 Bit over 16 Bit !?!?

Post by diwo » Wed Dec 19, 2007 3:10 am

Setting the H2 format to 44.1/16, mic gain to "M" and record level to "100" seems to be a good all-purpose setting. In most cases you have not to change the record level.

Record level is working in the digital domain and a ~ 50% level is stored in the WAV file, using actual only 8 Bit of 16 Bit.

Normalizing such a file in post processing will rise the overall level, but there will be a lack of dynamic information, due to the reduce used Bits.

Recording with such a all-purpose setting, it would be a improvement, to record with a quantization of 24 Bit.

what's your opinion?

Dieter
0 x

trevmar
been here a while
been here a while
Posts: 67
Joined: Fri Oct 05, 2007 11:40 pm

Post by trevmar » Wed Dec 19, 2007 3:16 am

I use 48/24 unless I am short of recording time. There is a measurable HF improvement with 48 rather than 44.1
0 x

dosdan
been here a while
been here a while
Posts: 58
Joined: Mon Aug 06, 2007 9:08 pm
Location: Brisbane, Australia

Re: Benefits of 24 Bit over 16 Bit !?!?

Post by dosdan » Wed Dec 19, 2007 4:14 am

diwo wrote:a ~ 50% level is stored in the WAV file, using actual only 8 Bit of 16 Bit.
No, Dieter, we're dealing with a log. response. It depends if, by the term "50% level", you mean power or voltage.

50% of max. voltage = -12dB which is 14 out of 16 bits used.

I presume you meant 50% of max. power = -6dB which is 15 out of 16 bits used.

Or perhaps, you just meant half-way on the level meter. On my H4 that's -24dB or 12 out of 16 bits used.

8 out of 16 bits would be -48 dB, which, on my H4, would only be the lowest meter mark illuminated (assuming the meter response is that accurate).

(Close approximations shown, since 1 bit = 6.02dB)
0 x

diwo
been here a while
been here a while
Posts: 87
Joined: Sun Nov 04, 2007 3:44 am
Location: Neumarkt, Germany

Post by diwo » Wed Dec 19, 2007 2:18 pm

@dosdan

thanks for correcting the figures ...

I want to say, that using only 50 or 60% of the max. recording level and normalize in post processing is anyhow a loss of dynamic information. It would be better to record with 24 Bit and normalize and transform to 16 Bit in post processing.

@trevmar
If the result should be a CD, I guess 48 kHz isn't an advantage - or?

Dieter
0 x

tweakfreak
member in good standing
member in good standing
Posts: 47
Joined: Sat Dec 15, 2007 7:18 am

Re: Benefits of 24 Bit over 16 Bit !?!?

Post by tweakfreak » Wed Dec 19, 2007 3:28 pm

dosdan wrote:
diwo wrote:a ~ 50% level is stored in the WAV file, using actual only 8 Bit of 16 Bit.
No, Dieter, we're dealing with a log. response. It depends if, by the term "50% level", you mean power or voltage.

50% of max. voltage = -12dB which is 14 out of 16 bits used.

I presume you meant 50% of max. power = -6dB which is 15 out of 16 bits used.

Or perhaps, you just meant half-way on the level meter. On my H4 that's -24dB or 12 out of 16 bits used.

8 out of 16 bits would be -48 dB, which, on my H4, would only be the lowest meter mark illuminated (assuming the meter response is that accurate).

(Close approximations shown, since 1 bit = 6.02dB)
Oops! The A/D output is the digital representation of the voltage. Therefore half the voltage means one bit less. Driving the input meters "only" to -6dB means sacrificing only one bit.

And voltage to dB conversion is 20log(V2/V1), meaning that half the amplitude is -6dB, not -12dB.

Power does not come into this.
0 x

dosdan
been here a while
been here a while
Posts: 58
Joined: Mon Aug 06, 2007 9:08 pm
Location: Brisbane, Australia

Post by dosdan » Wed Dec 19, 2007 4:15 pm

tweakfreak, your right. The figures should show:

50% of max. voltage = -6dB which is 15 out of 16 bits used.

50% of max. power = -3dB which is 15.5 (I think) out of 16 bits used.

Decibels are a power ratio, that's why we multiply the ratios of the log of the two compared powers by 10 (to get the "deci" part) and the voltage ratios log by 20 (to square the voltages as well, as part of P=V2/R)

As regards "50% level", are we talking "half power", "half amplitude" or "half scale"? It would then come down to what type of level scale we use. The H4 has a log scale. -24 db is about half-way between -48dB & 0dB. A PPM (Peak Programme Meter) has a log response, so PPM 4 is halfway between PPM 1 & PPM 7. A voltmeter has a linear response, which is why a VU meter does not show -10dB halfway between -20dB & 0dB. -10dB is way over to the left.

I think it's better to avoid percentages in audio and just say a "level of -6dBFS" which is both 1/2 max. amplitude and 1/4 max. power.
0 x

Ozpeter
dues paid
dues paid
Posts: 327
Joined: Sat Sep 22, 2007 7:41 pm

Post by Ozpeter » Wed Dec 19, 2007 5:24 pm

There is a measurable HF improvement with 48 rather than 44.1
Very easy to measure whether the HF is recorded - less easy to objectively test whether it's audible!
0 x

trevmar
been here a while
been here a while
Posts: 67
Joined: Fri Oct 05, 2007 11:40 pm

Post by trevmar » Thu Dec 20, 2007 5:31 pm

Ozpeter, the issue is not frequency response, it is inter-modulation distortion due to inadequate aliasing of the sampling rate. This will make sibilants harsh, and affect instruments in a manner which is hard to quantify. Think of the difference between vinyl and digital and many of those HF artifacts are due to inadequate HF inter-modulation. The ADC chip is designed for 48KHz, and it has less IM at that sampling rate.
0 x

greenmachine
member in good standing
member in good standing
Posts: 43
Joined: Wed Sep 05, 2007 5:05 pm

Post by greenmachine » Fri Dec 21, 2007 4:09 am

trevmar, I remember reading in one of your posts that you suspect that recording in 48kHz and down-sampling via software might be preferable sound quality-wise to recording in 44.1kHz in the first place. I would be interested in any more insights you or others may have on this matter.
0 x

Ozpeter
dues paid
dues paid
Posts: 327
Joined: Sat Sep 22, 2007 7:41 pm

Post by Ozpeter » Fri Dec 21, 2007 6:46 am

The ADC chip is designed for 48KHz, and it has less IM at that sampling rate.
Ah yes, I'd forgotten your previous mention of that. Sounds a bit like Creative soundcards, many of which are locked at 48kHz and resample for other rates - or so I've been told.
0 x

diwo
been here a while
been here a while
Posts: 87
Joined: Sun Nov 04, 2007 3:44 am
Location: Neumarkt, Germany

Post by diwo » Fri Dec 21, 2007 3:39 pm

Sorry for causing confusion about my 50% level post :wink:

I'm meaning half the range of the H2 level meter - so a max. recording level up to - 24 dB.

There is a measurable HF improvement with 48 rather than 44.1
Any idea where, apart from a PC, the majority can play the 48 kHz recorded stuff?
I only see a CD for distribution. A Video-DVD is running with 48 kHz, but it's anyway the wrong media playing music.

So a 48 kHz record has to be resampled to 44.1 kHz - but I guess the resampled result is inferior than a pure 44.1 kHz record, because of necessary interpolations - or?

Resampling a 96 kHz record to 48 kHz needs no interpolations and should be as good, as a pure 48 kHz record.

On the other hand, I think a transcoding from 24 to 16 Bit is without loss - the interpolation will not be audible.
... of which are locked at 48kHz and resample for other rates - or so I've been told
I can't believe this - all record will be sampled with 48 kHz and on the fly resampled to 44.1 kHz or to 96 kHz. In this case the H2 should have the smallest processor load with 48 kHz.

This should be visible in the manual "dropout risk list", but it isn't.

Dieter
0 x

trevmar
been here a while
been here a while
Posts: 67
Joined: Fri Oct 05, 2007 11:40 pm

Post by trevmar » Fri Dec 21, 2007 5:32 pm

Dieter,
48KHz is a common sampling rate - it is used in DV and HDV video, for example, as well as DVDs. I use it as standard for everything except YouTube videos, which I transcode to 22.050 KHz

Generally, transcoding on a computer, with decent post-prod software, should not produce the artifacts which are evident in the spectra of an H2 running at 44.1KHz

The H2 has similar Nyquist aliasing problems at a 96KHz sampling rate to those which it exhibits at 44.1 - indeed, a 48KHz rate deliver fewer intermodulation components in the baseband than does 96KHz. There is no point in using 96KHz sampling with an H2.
0 x

diwo
been here a while
been here a while
Posts: 87
Joined: Sun Nov 04, 2007 3:44 am
Location: Neumarkt, Germany

Post by diwo » Sat Dec 22, 2007 1:07 am

@trevmar,

thanks ... so it would be interesting to have, or produce, some audible H2 44.1/48 kHz examples.

I guess, with many instruments it will not be audible - what instruments would be adequate for test records?

Recording with 48 kHz / 24 Bit in 4ch mode is running the H2 to the performance limit. I've done some test with my 4GB card - in 2 out of 6 tests I had a dropout ~ 1 minute before exeeding the card space. So it would be better to record without risk on a lower level. 44.1 kHz/ 24 Bit, or 48 kHz/16 Bit, what would be the better (audible) choice? More dynamic, or less inter-modulations?

Dieter
0 x

trevmar
been here a while
been here a while
Posts: 67
Joined: Fri Oct 05, 2007 11:40 pm

Post by trevmar » Sat Dec 22, 2007 10:03 am

Dieter,
Although the ear can hear subtle differences between sound samples, it is almost impossible for the ear to figure out why it hears that difference. IMO, test equipment is the only way to figure that out.

One of the best tests is to apply two very pure tones (distortion <0.1%) simultaneously to the input and look at the output spectrum. If you do that with the H2 you will see significant IM (intermodulation) components. You can also see some components with just one pure tone applied, especially at higher frequencies. Spectral analysis can be done, for example, as I described in another post, with Sound Forge (and similar packages).

Another good test is transient intermodulation (TIM). That is much harder to apply, but IMO it does simulate the real world of changing dynamics somewhat better.

I choose 48KHz/24bit and haven't noticed any dropout problems, but I haven't gone to the limits of the SD cards, where I imagine that the CPU is crunching numbers to figure out remaining space in the FAT tables.

I would choose 48/16 for typical recordings (where noise floor is not a problem) and 44.1/24 if the noise floor might become the limiting factor (wide dynamic range orchestral or choral recordings, for example). There is not a lot of difference between the Line In 24bit and 16 bit noise floors, but there is some.
0 x

diwo
been here a while
been here a while
Posts: 87
Joined: Sun Nov 04, 2007 3:44 am
Location: Neumarkt, Germany

Post by diwo » Mon Dec 24, 2007 12:14 am

@trevmar,

thanks ... maybe I own a slow SD-card.
Are you recording in 4ch mode, have you checked the marklist for dropouts?


Happy christmas to all :D :D :D
Dieter
0 x

trevmar
been here a while
been here a while
Posts: 67
Joined: Fri Oct 05, 2007 11:40 pm

Post by trevmar » Mon Dec 24, 2007 12:25 am

Dieter,
No, I haven't been using 4 channel yet. I am still getting used to my new Mid-Side microphone, recording directly from the cardioid and figure-8 capsules and using Voxengo software to get the stereo image. I will be stuck optimizing that for another week or two, I fear. But I am happy with the results so far.

I used the H2 as a USB stereo ADC front-end to Audacity earlier this evening, recording direct to the XP laptop's hard disk through Audacity. That seems an interesting way to do things. It worked fine :):)

Merry Christmas to you and yours :):)
0 x

Post Reply

Who is online

Users browsing this forum: No registered users and 1 guest