[tomb]

Coming right up!

As a for instance, here's AMB's frequency response graphs of the M-Audio Transit in loopback and the Alien DAC:

Here's the same thing with the M-Audio Firewire:

Now, if one were truly interested in the response of the Alien DAC from 15KHz to 20KHz, then perhaps the M-Audio Firewire may not be enough. However, it's dubious that it would follow the loopback curve exactly if the response were not much better than the recording device.

In any event, there's much more useful info to be had from RMAA as well. I'm sure AMB could make much more authoritative statements about this than I can, but either device seems to me to be quite useful for our purposes.

Here's mine comparing the BantamDAC to the AlienDAC, using the M-Audio Transit. You can see that the FR matches the loopback in AMB's images perfectly:

More than that, here is evidence of the difference in lack of cabling between the typical Alien DAC and the BantamCableDAC (Noise):

EDIT: These are at 16/44.1 or 48KHz. The frequency response is better at higher rates, but the Alien and Bantam are not able to respond at those settings. The fact remains that the Transit has a better response at the lower rates, though. Again - so what? Either can perform these tests adequately to see what's going on with the device under test. The M-Audio Firewire is undoubtedly more convenient, though.

EDIT2: I got the BantamCableDAC/BG Alien FR graph mis-placed with one of AMB's. It should be corrected now - refresh your browser if it doesn't show as BantamCableDAC/BG Alien.

 

[linuxworks]

I'll have to post what I'm seeing.

I recently tried the bantam and amb dacs

they both tested as very low noise (as we'd expect) but the test gear (the m-audio fw box) was STILL better in all specs than either. to me that meant the test gear is still 'good enough' to discriminate minor subtle details between these kinds of 'already very good' boxes under test.

I also tested my old m-audio 'superdac 2496' which was a $300 dac back in the 1998 era (iirc). that seems to be my best dac that I currently have and I think THAT was finally where the dac of the maudio firewire box was not as good as the device-under-test (that pro-level balanced-out superdac box). and so now, I may end up using THAT dac as my output stage (to generate test signals) and the input stage of the firewire box since its still my best a/d stage that I have access to (until I find an 0404 box, which is actually better yet).

as far as the 20k rolloff, that is NOT a factor of the m-audio box when in 'analog in/out' mode. when I test my phones amps, of course there is no digital stuff going on and I'm sending out analog and reading in analog. yes yes - it gets converted - but still its running at 'best mode' which is 32bit/96k. when I test amps in THAT setting, there aint no 20k rolloff my friend!

but when you test usb dacs - uhhh - of course usb can't run faster than 16/48k. and so you find a 20k rolloff, give or take.

I still don't see any 15k rolloffs. not in the testing I've done (admittedly just a day of ownership of that maudio firewire box)

but even testing old 16/44.1 protocols I see flatness up to 20ish and not at all 15ish.

maybe that was a demo or old version of rmaa? I'm using the latest and I swear there is no 15k limit.

 

[linuxworks]

for tom - here is an RMAA test I just did on a bantam dac build.

the setup I used:

here is one such test chart:

see the upper limit on the right? its NOT 15k! and this is a '16bit/48k' usb-driven test. sure looks like things are ok up to 20k-ish, which is what I would expect.

one more:

again, I see a good response up to the limit of what 48k SR can do.

and for a totally wild compare, here is the best that this firewire box can do:

you can SEE its freq measurement goes well above 30k, when doing analog level tests (its analog in connected directly to analog-out via 2 short rca-rca cables).

fwiw..

 

[tomb]

Well, all I can say is that I'm no expert on M-Audio products - all I have is the Transit and my/AMB's posted results.

One thing's for sure - your tests make the BantamDAC look d*mn good.

EDIT: On the other hand, take a look at your vertical scales. You have 6dB in one grid mark on the FR graph. That's more than the entire vertical scale in the positive or negative portion of the graphs I posted.

\

Note that I'm using the same scaling as AMB's graphs.

 

[linuxworks]

Quote:

Originally Posted by tomb /img/forum/go_quote.gif Well, all I can say is that I'm no expert on M-Audio products - all I have is the Transit and my/AMB's posted results. One thing's for sure - your tests make the BantamDAC look d*mn good.

so far, its been testing better in noise (which is one spec I like to watch) than the amb dac that I built. I see some low freq rolloff on MY bantam but its probably due to my output caps. the amb dac has double caps (electro and box) while I have only box style low-value caps on the bantam.

I'm still fine tuning my tests - but yes, the bantam is coming out very well indeed in the tests.

Quote:

EDIT: On the other hand, take a look at your vertical scales. You have 6dB in one grid mark on the FR graph. That's more than the entire vertical scale in the positive or negative portion of the graphs I posted. \ Note that I'm using the same scaling as AMB's graphs.

I didn't pay much attention to scaling the verticals - but I did change the default horiz scaling ONLY to show the full left to right spread - and to remove scrollbars from the x-axis. the default display needs the user to scroll and for a screenshot, that makes no sense

I may have scrolled the vertical a bit but I'm not sure I *scaled* the vertical at all.

but the point you cared about was on the x and not the y axis, so I wanted to be sure to show the upper and lower freq limits in the tests.

when you said 15k, I got scared and thought someone turned back the hands of the clock and we were all listening to FM radio again (lol).

I can test amb's dac up to 24/96 via coax-in. I can only test usb 'standard' devices up to 16/48 - but at least I can test cleanly up to the theoretical 20k-ish limit.

its good to be better-than-FM (lol). and all the dacs (even my junky 'bad' audio alchemy DITB) test pretty clean straight up thru 20k.

 

[linuxworks]

Quote:

Originally Posted by tomb /img/forum/go_quote.gif EDIT: On the other hand, take a look at your vertical scales. You have 6dB in one grid mark on the FR graph. That's more than the entire vertical scale in the positive or negative portion of the graphs I posted.

here's one I didn't post but had uploaded to my flicker site.

that shows the fact that I have a 1uF vishay box cap on the output. I suppose a larger value (or even another one in parallel like AMB's does) would have given an even lower freq-resp. perhaps when I perm mount this test unit I'll up that value or put the electro in there as well.

the upper limits - no problem there! clean to 20k, give or take. not FM radio, no sirree!

but see - this is my point - this testing SHOWED me clearly what I needed to do or where something fell short. and where its just fine, too - it showed that, as well.

would fancy cable dress show up on a test that uses fairly decent and sensitive test gear? that's my point.

 

[digger945]

^Yea but what does it sound like?

(sorry, couldn't help it LOL), j/k j/k

 

[linuxworks]

sound? lol.

hey, by this chart, which would 'sound' better?

you now need to say - in which freq range (!) will this or that sound better. when you get beyond single scalar numbers and go 2d with charts, you can see that single number specs don't tell the whole story.

here's the more realistic view showing that once you zoom out a bit, its not nearly as 'bad' (the diffs between the dacs) as it might seem:

I really can't tell much sound diffs between them. so all I'm seeing are techie diffs but not sure I can really say one SOUNDS better than the other, really.

 

[tomb]

Edit: Never mind. There's no good end to this. Sorry for getting involved in the discussion.

 

[digger945]

I dunno, I probably couldn't hear noise that low anyways.
I know for a fact I can't hear anything above 14k at all.
Fact is I can't hear all that good at all.
You didn't have the Gamma in the freq resp graphs above(I'm not asking you to do it).
If you don't find much difference, then I guess you'll have to wait 'till you acquire a good tube amp LOL

(again, i'm just kidding, i'm a kidder you know

)

Thanks for the results, it's very interesting. I did something like that with the 0404, but I think I deleted the results when I finished posting on a thread on the sound science forum(few months back).

 

[linuxworks]

Quote:

Originally Posted by tomb /img/forum/go_quote.gif I don't want to quibble - but you sort of started this by questioning the Firewire's response. However, your results in this graph clearly show a drop beginning before 15K and dropping below -3dB at 20K (the image itself shows below -4dB):

I think you are reading the chart wrong, tom. yes, there is some drop starting 'early' but look at the y-values! they are fractions of a single dB.

the last right-most vertical line is the 20k line. the graph goes BEYOND 20k, the way they have drawn it.

so, at 20k, its down less than half a db. am I right or am I the one reading the chart wrong?

 

[tomb]

No, I was reading the chart wrong (but every scale has been different). You are correct and I am wrong. Sorry for getting involved in this discussion.

 

[digger945]

Just curious, what kinda IC's are you using?

 

[linuxworks]

Quote:

Originally Posted by digger945 /img/forum/go_quote.gif Just curious, what kinda IC's are you using?

lm741. only the best!

oh wait. you mean cables? lol. some belden shielded wire on (admittedly) cheap rca 'gold' male jacks. the cables are similar to rg58 or 59 in thickness but I can't find a belden # of the jacket. its single conductor center plus shield. each cable is separate (not plastic bonded as a typical L/R pair would be). length is about a foot or so (I made the cables short on purpose). I believe I also intended them to be spdif 'quality' which should certainly be enough for passing analog audio over.

and yes, I was kidding about the lm741. I now use better parts than that (lol).