Understanding Mic pre specs on Audio Interface.

[CFox]

CTRL C'd and V'd from Newbies. ( without the new and lols )

I couldnt decide whether to put it in Recording, or Mics...or Computer.
These are stats from a USB Audio Interface.
What are the important specs here pls ? I wont tell the products name, would just like someone to tell me if its any good. Thanks

They all seem to present their data slightly differently it seems.

First
Mic:
Input Impedance:
4.8 kOhms balanced
Full Scale Level
at maximum Gain
-40 dBu
Maximum Input Level:
+10 dBu
SNR (weighted):
100 dB
Equivalent Input Noise:
-128 dBu (weighted)
THD+N:
0.007%
Frequency Response:
20 - 20000 Hz (+0 / -0.5 dB)

-----------------------------------
Second

Type XLR Female Balanced
Frequency Response (±3.0 dB) 14 Hz to 70 kHz
Input Impedance (Balanced) 1200 Ω
THD+N (unwtd, 1 kHz @ +4 dBu Output, Unity Gain) < 0.008%
EIN (A-weighted, 55dB Gain, 150 Ω Input, 20Hz to 22 kHz) -115 dBu
S/N Ratio (Unity Gain, Ref. = +4 dBu, 20Hz to 22 kHz) > 95 dB
Common Mode Rejection Ratio (1 kHz, 55 dB Gain) > 45 dB
Gain Control Range (± 1dB) 0 dB to 35 dB
Max Input Level (Unity Gain, 1 kHz @ 0.5% THD+N) -3 dBu
Phantom Power (±2 VDC) +48 VDC

 

[CFox]

Hey MS, here is an example of me not knowing something other than recognising signal to noise ratio.

 

[mshilarious]

Generally the specs on the first are better. Notably, the second's equivalent input noise spec is not good; that will limit its application to condenser mics or dynamics on loud sources. Pre #1 is spec'ed 13dB quieter, which is huge. I would probably reject pre #2 on that basis alone, unless there were extenuating circumstances.

(Note that you have to make sure you are comparing apples to apples with EIN, that is V vs. u and A-weighted vs. unweighted. u is 2dB more than V, and A-weighted is about 4dB less than unweighted)

Pre #2's CMRR is terrible although it's probably good enough for government work; I always say that if you need more than 30dB of CMRR then you have other issues . . . but still, that should be more like 60dB or more, ideally. Pre #1 does not spec CMRR which is a serious omission in my book.

Given that pre #2 is really only useful mainly for condenser mics, I'd like to see input impedance higher than 1k2, although that's not a big deal. On the other hand, 4k8 is pretty high for a dynamic mic, a lot of those sound better with some loading (although ribbons tend to enjoy the higher impedance). But that can be fixed with a shunt resistor built into a cable; the opposite direction is not so easily remedied.

Another problem with pre #2 is that minimum gain seems to be 20dB (55dB maximum gain but only 35dB gain range). So if it doesn't have an internal pad, next you'll be shopping for an external pad when you use a condenser mic on a loud source (you can use the mic's pad instead if it has one, that degrades SNR but nobody really cares on a loud source).

On the other hand a couple of its specs claim unity (0) gain, which doesn't jibe with the gain range/max gain stats. My wild guess is that the unit has an internal 20dB pad which accounts for the difference.

But in the final analysis specs don't tell you if a preamp is good, they tell you if/when it would be bad. Both of them could sound great; both could sound like crap, but pre #2 is never gonna work for whisper tracks with an SM7B no matter how good it otherwise sounds.

 

[CFox]

Thanks MS. Your analysis seems to match the anecdotal evidence concerning the second pre amp that the signal is weak on dynamic microphones.

Trying to uncrack a locked PDF manual with another spec run down.

Maybe you could just take a squiz at this page and the tests someone has claimed to have performed.See below.

Fig. 8 shows the A-weighted response, which at better than -120dB beats Mackie's own quoted spec of -114dB. Granted, parts vary and there might be some units out there that hit Mackie's spec, but I'm happy to see that they're conservative about their claims.

Fig. 9: Intermodulation distortion plus noise (swept frequency)

Here's where the rubber meets the road - under worst-case conditions, without A-weighting, and taking distortion and noise into account, we're looking at better than -84dB (Fig. 9). I'm a little fuzzy on my math, but I think that's probably well under -100dB A-weighted.

http://www.harmonycentral.com/docs/DOC-1621

 

[mshilarious]

I don't really have time to peruse somebody else's tests, but I will say there is no standard for intermod, and intermod + noise is an especially dubious spec. For example, intermod is highly dependent on frequencies tested, gain setting, and also together with issues like crosstalk as well (for example, was the second input terminated or floating?) So anybody comparing two units on their own needs to be very careful to compare apples to apples, and should be explicit in their explanation. If they have done so, then they don't need my help in translation.

Given that -84dB (V? u? FS?) unweighted could not be anywhere near -100dB A-weighted unless there was a really freaky spectrum at play (all above 10kHz, for example), I worry about that test without even looking at it. Also, there is no way component variation is going to create a 6dB swing in noise. I don't think I've ever seen 0.5dB variation in an IC opamp's noise. So I again would suspect something is being done incorrectly.

 

[CFox]

Thanks again bro. Sounds like this review "industry" needs a shake up. Until then, I guess we defer to good old word of mouth.

 

[mshilarious]

Oh dear I had to satisfy my curiosity and read that. Anderton should really know better, he's even written a book

First, on the noise chart, he confuses spectral noise vs. integrated noise, and that graph clearly shows unweighted noise and NOT A-weighted noise. He's lucky he got within 6dB Actually his graph shows unweighted noise of about -80dBFS or -84dBFS-A. Integrated noise is about 43dB higher than spectral noise across 20kHz bandwidth, as given by:

20 * log (20kHz ^ 0.5) = 43dB

So I wonder if he had gain on minimum (which Mackie spec'ed noise at -112dBFS-A). Probably not.

Given that maximum gain (60dB?) is more than sufficient to bring analog noise well above minimum converter noise, that's how you derive EIN: you measure integrated noise at max gain with 150 ohm termination, and provided that is at least 10dB or so above converter noise (it should be way more here), you subtract 60dB from that figure and that's your EIN.