Measuring is fun - and you learn stuff.

[antcollinet]

So I want to digitise at least one of my Vinyl LPs (Reason unimportant).

I have two devices with an anlogue input.
Soundblaster omni 5.1 USB "sound card"
Lexicon alpha Audio Interface.

I wanted to work out which was best - I assumed it would be the lexicon since it is aimed at musicians. So I could have just tried digitising with each, and listen to the result, but I've been playing with REW for my first attempt at room correction...

Hmm - could I use REW to measure the two devices?

They both have analogue outputs and inputs, I decided to loop back on each, and see what gave the best measurement. Doing that, the Soundblaster gave dramatically better results. I'm sure the main reason is the lexicon has twiddly pots on the front that control the analogue gain. They don't seem that good, and I suspect the analogue circuits for that aren't the best. The pots are noisy also.

To confirm, I decided to measure the Soundblaster again, but then measure the input of the lexicon using the analogue output of the Soundblaster.

Here is the comparison - red is the Lexicon (THD+N 0.36%), and Green is the Soundblaster (THD+N 0.025% - 144 times or 40+dB better).
Output fedback to input was a 1kHz sinewave from REW generator, at -12DBFS.

 

[amirm]

After measuring almost 1000 devices, I still can't read the book by its cover and tell what is going on without measuring.

 

[antcollinet]

As an aside, while I was experimenting I took a measurement of the soundblaster and got the line below. What the heck - noise at 50Hz and odd harmonics.

Decided it was mains interference. Sure enough for this measurement I'd set the soundblaster on top of the amp, directly above the power supply transformer.

For this measurement, THD+N was 0.055% - 22 times (27dB) worse than placed away from the amp.

So not only have a learned which of my two analogue input devices is the best (and by a very clear margin), I've also learned where not to put it.

 

[antcollinet]

After measuring almost 1000 devices, I still can't read the book by its cover and tell what is going on without measuring.

My test setup isn't quite as amazing as yours . But I'm blown away by what I can do with a £70 sound card and some free software.

 

[AnalogSteph]

So not only have a learned which of my two analogue input devices is the best (and by a very clear margin), I've also learned where not to put it.

Well, actually, it's your measurement setup that is b0rked. If you're going to use an unbalanced connection, you must have galvanic isolation between output and input or else all hell is going to break loose in terms of ground loops. The whole setup is forming a big loop antenna, so no wonder you're going to pick up the mains transformer's magnetic field. The other direction is no good either.

If you have a spare laptop with something resembling a working battery kicking around, this would be a good time to drag it out.

The Lexicon Alpha should be based on a CS4270, the Omni 5.1 uses CS4361 (DAC) + CS5345 (ADC). Pretty equivalent converter wise (note: recording at 96 kHz very much recommended, the anti-alias filters aren't that hot), everything else is down to the analog part. If memory serves, the Omni will manage 99.4 dB(A) worth of dynamic range in loopback according to RMAA, fairly close to what you would expect from converter specs (which would be ~101.5 dB(A)).

I imagine if you're just recording consumer line level sources like a phono preamp, there is little point in using the Alpha, its line-in can take up to +12 dBu, and I imagine you would be giving away 4-6 dB worth of dynamic range over the Omni on that basis alone. At least it's a balanced input, so if you want you could have some fun making adapter cables that actually make use of that fact and largely eliminate ground loops that way. (I couldn't tell you what line output impedance on the Omni is, but the headphone output with its MAX97220A should sport 22 ohms or thereabouts so should need no special consideration for a cable build.)

BTW, the Alpha must have the most pathetic specs for mic input performance I have ever seen on any USB audio interface (which means they're probably legit) - EIN -115 dBu(A), max input level -10 dBu. Then again, let's not be too harsh with a cheap entry-level interface from 2006, they just weren't that good back then. My favorite for a small USB interface like that from those days would have to be the Digidesign Mbox 2 Mini (which, mind you, cost 2 or 3 times as much at the time). Not too far from a Scarlett Solo gen2, really. They can be picked up for the better part of a pittance these days, but beware, Windows drivers are ASIO only (no WDM) and it may not like USB 3.0 ports.

 

[antcollinet]

Well, actually, it's your measurement setup that is b0rked. If you're going to use an unbalanced connection, you must have galvanic isolation between output and input or else all hell is going to break loose in terms of ground loops. The whole setup is forming a big loop antenna, so no wonder you're going to pick up the mains transformer's magnetic field.

That is a really good point - and when I made a minimally inductive bundle of the interconnect, the mains pickup was more or less eliminated. Good shout.

Then just for fun, I maximised the loop area.

Comparison of non inductive bundle, and maximum loop looks like this (Soundcard in the same place for both). I shouldn't need to state which trace is which...

See, still learning - just like I said.