Shootout: 828mkII w/ BLA Mods & MicroClock mk2 vs 828mkII Stock
 

this is a shootout between a stock MOTU 828mkII and one with the Black Lion Audio analog modifications and the MicroClock mk2.

i recorded each track of this test song individually, and the signal was split after the preamp...with one channel going to the stock 828mkII hooked up to one computer, and the duplicate channel going to the modded 828mkII which was hooked up to a different computer. both computers were running DP7, and both recorded at 44.1/24. not that it matters, but here's the skinny:

snare/cymbal -> C414 -> SCA N72 -> splitter -> both converters
kick -> PR40 -> SCA N72 -> splitter -> both converters
bass -> avenson isodi -> SCA N72 -> splitter -> both converters
acoustic -> C414 -> SCA N72 -> splitter -> both converters
el guitar 1 -> PR40 -> SCA N72 -> splitter -> both converters
el guitar 2 -> PR40 -> SCA N72 -> splitter -> both converters
shaker -> PR40 -> SCA N72 -> splitter -> both converters

so here are the files. want to guess which is which?

aa
bb



EDIT 1: here is a 10x loop of each file through their respective converters:

aa ten loop
bb ten loop



EDIT 2: here are some pictures from both units:

stock one
stock two
stock three
stock four
mod one
mod two
mod three
mod four



EDIT 3: and in the interest of being comprehensive, here is a set of test loopbacks provided by gearslutz member nms. unlike the other files above, these have not been level matched (i thought nms might want them as raw as possible):

original
aa one pass
aa four pass
aa ten pass
bb one pass
bb four pass
bb ten pass



EDIT 4: click here for the results

i like aa

awesome thanks Jon! I decided the best way to match levels is to match it from the 1k sine part.. so I'm in the middle of RMS matching both channels to a hundredth of a db with the original using that tone. This way if there's low end loss on a converter that file won't be artificially boosted to match RMS and mask its low end loss. You'll be able to hear or see on a meter for instance what level the 60 & 80hz tone registers at with nothing masking it. I'll post the processed files in one hour.

Btw, any chance you can run my file through the lynx still?

Here are the level matched files and seamless loops extracted for both units at 1, 4, 10 passes. Try null tests with the diff loops. Not a lot of difference I'm finding.

http://www.bass-skidz.com/Motu_828Mk2.zip

http://www.bass-skidz.com/Motu_828mk2_BLA_Mod.zip

Wow, are you 1000% sure one of those was BLA modded? The analysis stats I got from testing those ended up identical! Looks like separate recordings made from the same unit. crazy. Check how the Motu's rank against the others though. Impressive. Thanks again for supplying the 828 clips.. insightful for sure.

oh yeah, one is definitely modded. check the original post and look at the pictures...you can clearly see the replaced chips.

that is strange...it could be psychosomatic, but i swear i can hear a difference between "aa" and "bb" in the original post (which is just one instance of A->D). anyone else?

I thought I could at first.. But not so sure. 10 loops should exaggerate the results, so it's odd to still match at that point.

any chance you could perform the same analysis on the "aa" and "bb" files in the original post (the little "song" i recorded)? i know there is no "original" to compare it to, but they could be compared to each other to see if there is actually a difference, right?

it's tricky getting worthwhile results that way since not comparing it to an original. You can't use sample rate correction to align the clock drift because it will correct one file to match the other, thus giving the untouched file the advantage. They definitely don't null with or without clock drift corrected though on those pieces. When nulling the loopback file I found them to null to -54db rms without clock drift correction.

gotcha.

nms, can you do me one last favor and perform your analysis technique on the 1-pass stock & bla files? i have a cockamamie theory brewing about all of this, and the numbers on a single pass for both units would tell me something...

Here's another test of stock vs BLA mod that I ran Diffmaker analyses on thanks to the test recording you left in that thread ;).

https://www.gearslutz.com/board/low-e...ml#post7160380


Here you go, your comparison of stock to BLA modded 828mk2 at 1 pass:

Corr Depth: 95.2 dB (L), 103.1 dB (R) // RMS: L=-95.5 R=-92.9
*Sample Rate error is -0.0005 ppm, adjustment unnecessary

I'm not sure what your "cockamamie theory" is but if it's that the unit either wasn't modded or doesn't make a difference apparently then I'd say you are correct!

so are those number for both units at one pass?

i don't get it. i just listened to "aa" and "bb" with some ATH-M50 headphones, and i picked the BLA unit 14 times out of 15. does anyone else hear a difference? am i just insanely lucky? perhaps i should go hit the casino...

Well that's the thing. Don't listen to your files, listen to my copies which are accurately level matched. Then try to pick it out. It takes little volume diff to cause bias.

i leveled each track individually to 1/100th of a dB using FreeG...so i don't think that's at work here.

That's not the most accurate way to match them though. It only takes a little difference to sway. You need a wave editor to properly match them rather than a DAW.

Here's your RMS levels for 1 pass as you have them

aa: L=-25.67 R=-25.89
bb: L=-25.37 R=-25.60

10 pass levels

aa RMS: L=-25.68 R=-25.61 (peak: L=-6.16 R=-6.26
bb RMS: L=-25.29 R=-25.54 (peak: L=-5.21 R=-5.68)

even though it defies logic, i feel like i can hear the difference between the BLA and the stock more clearly with less passes. the 10-pass files sound indistinguishable to me.

if that is true and not in my head, it led me to this thought (which might be ridiculous)...with any interface, there are two different sources of degradation: the analog path, and the actual digital conversion. now BLA claims to improve the analog path, they don't touch the actual converter chip. so let's say the converter chip is actually responsible for more degradation than the analog path (seems possible). if that is the case: the more passes you do, the more the slight benefit from the analog path modification will be clouded by converter degradation...so by 10 passes, any benefit from improved chips in the signal path has been nullified. to take this theory to an extreme: imagine playing a song on a guitar with dirty strings to a friend over the phone. now change the strings on the guitar, and play the song to your friend over the phone again. your friend probably didn't notice the string change because the telephone degradation rendered the slight improvement inaudible.

how's that for cockamamie.

Here.. tell me what you think now:


Here are the exact RMS level matched aa & bb direct recorded files. Can anyone tell them apart with any level of consistency?

Motu_This_one_pass.wav
Motu_That_one_pass.wav


And here are the level matched aa & bb 10 pass recordings.. did I rearrange the "this" & "that" assignments or keep them the same? Which is which? Which one is the nice opened up version from the expensive mods?

This_tenloop.wav
That_tenloop.wav

I'll say this.. there is no accurate testing to be had without exact level matching!

Here's a few more Diffmaker test results for you guys.. I tested the 10th gen recordings with the direct recordings. This is particularly relevant because you can see exactly what kind of degradation is happening from the first recording to the tenth and clock drift from recordings made from the same unit. I didn't enable clock clock rate correction on these ones.

Stock 828 mk2: Correlation Depth: 20.4 dB (L), 20.4 dB (R) !Sample Rate error is 0.1721 ppm // RMS: L=-31.44 R=-31.21
BLA Modded 828 mk2: Correlation Depth: 21.4 dB (L), 21.3 dB (R) !Sample Rate error is 0.2351 ppm // RMS: L=-36.36 R=-36.89 <---interesting that we finally find a difference!

Ha, I can see how so much of selecting the right gear does not come down to selecting which piece is better or worse, but what best suites ones taste. I tend to prefer a bit of a raw natural sound over a smooth sound, so I chose the file that best represented my preference as being the upgrade. I don't know how much of this was psychosomatic but the more I listened to the file I chose the more I preferred it.

As much as I would've liked to be right, I'm glad I was wrong because now I know I can get the sound I prefer without spending $$$ on an upgrade. And you're right on, it is too easy to make money off us musicians when it comes to gear. Most of us are chasing after what everyone else says is the best rather than chasing after what we really like. Sadly this is because at times we don't really know, or don't trust our ears.

Did you confirm with some amount of sureness by blindly picking which one you liked? Without definitive blind test results it's really just luck of the draw or psychological.

If I was testing converters, I wouldn't use a Neve style preamp that adds a layer of sheen (distortion) on top of the signal, I'd use a clean preamp that reveals detail. That's the whole point of the BLA mod - detail and clarity. If you've got a clean preamp, you might give it another go with your 414 on acoustic guitar. You might be pleasantly surprised. diddlydoo

Another thing I noticed with my BLA modded 2408 was an increase in headroom. That's a biggie, considering that a stock 2408 or 828 will break up when the levels get too close to zero. Of course, those of us "in the know" don't record with hot levels, but when those accidental spikes come along, having a little extra headroom to avoid analog distortion can be a lifesaver. (I'm not talking about digital clipping, I'm talking about what happens a few DB before digital clipping.)

You can download the combo folders I posted for both units and compare yourself. There is intentionally hot content in that. 60hz @-0.3db, 11khz sine @-0.7db, etc. A stock unit will not break up when levels get close to zero. You're welcome to check out the files to see if you can find a difference though as there's a lot more content to choose from. There are seamless loops so you can play back one against the other on repeat to try to hear it. I'm quite sure you won't though even after 10 recording passes. At least where the mk2/mk3 units are concerned. I can't vouch for mk1's.

ooh, good idea.

i say this=aa and that=bb. i'm pretty sure i can still hear the difference fairly clearly. again, i picked it about five times in a row...or at least i think i did.

so am i correct or a fool with buyers remorse?

There's probably a psychological element to it, but I really did like certain things better about the file I picked.

so what's the verdict with the this/that files?

You picked right on that one. For the single pass files aa=this bb=that

@nostringsgtr - they both sound great and are very close.

i'm someone who believes that there is an embarrassing amount of psychosomatics in the world of pro audio...but i have to go against logic and the analysis results on this one. perhaps the theory i pulled out of my ass earlier (post #15) is solid, i don't know. but despite what the numbers say, i'm confident that there is a subtle but clearly audible difference between a stock and bla modified 828mkII. i can't even say which one sounds "better", but there is a difference.

@nms, i think your converter analysis thread is a great idea that could potentially uncover some serious truths, but unfortunately it seems that the testing method and analysis results aren't telling the whole story.

As I mentioned in the other thread though no one way of gauging tells the whole story. Our ears can be tricked easier than analysis results. In my 16th post here the analysts showed significant difference in results with the BLA and stock and that's after 10 passes. Do you have foobar? Think you could pass a 10/10 abx test with it?

Here's one last pair for you. I adjusted something with the levesl but I won't say what til after. They should be on a level playing field though still. Which one is the stock 828mk2?

here_one_pass.wav
there_one_pass.wav

this time i'm hearing: here=bla, there=stock.

Alas my friend, this time you picked the reverse!

The other issue that could have been at play was that while the previous ones were RMS matched, the peak level was louder on the BLA modded mk2. This time I averaged the peak and rms level differences between them so they'd both be on pretty level ground. If you have foobar for doing an ABX test between them I'd be impressed if you scored well with it. I couldn't.

The only remaining thing to do here I think would be if you can run a RMAA test (PC only) with them. All you do is download the software, set it to 24bit, 44khz, make sure your computer's default input & outputs are routed thru the motu channels you're recording with, levels at unity like last time, then click the button on lower left for 'Run tests Playback/Recording.

This is a single loopback test that provides all sorts of stuff from freq response to THD, dynamic range, noise, cross talk etc in an easy format.

hold on a minute. i was surprised to hear that i got it wrong because i feel like i've become pretty familiar with the difference after so much listening, so i checked...and "here" nulls with "bb", and "there" nulls with "aa". so that means i guessed correctly, right?

Just rechecked and you're right, those 2 got switched somewhere in the midst of here, there, this & that hahaha. What monitoring are you using for this?