lordpil's blog


NO HOMEWORK THIS WEEK
solidworks teacher <3 me

Posted by renesis at 17:06 | permalink | 0 comments


thats might not be a horrible idea honestly
since this isnt for reproduction
its for audiophiles so litz wire is easilly +200% sales

Posted by renesis at 10:52 | permalink | 0 comments


you can 20k and 20k to gnd and the rail
but yeah, i think first thing try and get the best performance out of the input section
i would prob depop c3 and work from there until happy with the input section
they were talking about doing headphone amp design on a solderless breadboard at work
i just stayed quiet
im getting better at that with age

Posted by renesis at 10:47 | permalink | 0 comments


ive seen those kind of values is low power stuff, not like it wont work
but typically you try and push as much current as you can get away with
itea1: his problems are before that
and my guess is it has something built in
yeah so maybe tack 10k on that input to see if it changes things
but just for experiments sake, change R1 and R2 to 2.2k, R5 and maybe C1 so it still works
cant just stack resistors on top?
that prob needs to go up, not down?
anyway, i think you have a barrel of paths to choose from at this point
maybe same thing in this context

Posted by renesis at 10:42 | permalink | 0 comments


shotgun-fuckit approach, replace RL and the Rin to the preamp to 2.2k
youre cap coupled, no?
or your preamp has its own internal loading? or what
yeah would use spec RL, and very low Rin, see how that changes noise
youre splitting the rail for opamp reference node, no?
its active or just resistors and a cap?
yeah so theres the half supply cap you can tweak
and yeah, running that to the MSGeq7 cap pin might help keep them in sync
other hacker advice: never trust a resistor above 100k for audio signals
i really like that advice

Posted by renesis at 10:37 | permalink | 0 comments


what are you loading it with?
macegr: can you FFT the noise?
their SNR spec is A weighted
assholes
try dropping that way down
has a 2.2k Zout
ha, measurement circuit doesnt spec a load
this is with a 5V supply?

Posted by renesis at 10:32 | permalink | 0 comments


try lower, shrug
i think thats backwards tho
signal and noise dont scale the same
shrug, diode or something
honestly dont think its a great suggestion, but you might want to try other voltage electrets
well ask them nice first
try and get samples of the 9.7mm mics, too
oh
thats their big one
heh they make a 4mm with wire tails
that says 3v on their website
looks like they spec min voltage on their site listings
but the noise might scale differently

Posted by renesis at 10:27 | permalink | 0 comments


in parallel or switched somehow?
pretty sure that will take more space than just doing a half inch mic
dude i think youre the only person ever to say a polycom thing worked well
if you keyed on like ~500hz, used some mic horns, prob not to hard to get working actually
wasnt the shogun shit from the 3rd movie?
macegr: what voltage you using on the mic?

Posted by renesis at 10:22 | permalink | 0 comments


macegr: also consider moving to a larger mic, half inch or even larger
you dont need the high frequency response so much
ya
those mems things are pretty cool
made a bunch of little heatshrunk remote mics with just the mems hole sticking out
performance seems pretty comparable to reg capsules
prototypes
of things
ha, naw sensitivity was pretty typical on stuff we got
reflections are an issue outside of a lab

Posted by renesis at 10:17 | permalink | 0 comments


macegr: you have half supply bias held up with a cap?
heh
maybe try reducing instead of increasing
i dont think itll work because of phase lag issues
well, maybe because in the linear region of the transducer
but yeah, the half supply voltage scaling with your power supply voltage might actually help things
your circuit probably doesnt have enough reactive shit to make it impossible
do you have a cap there?
see if you can sc ale noise with the cap value, shrug

Posted by renesis at 10:10 | permalink | 0 comments


yeah ferrites suck like that
macegr: try putting the mic on short wires?
theyre not audio people is my guess
they just sell some audio things
they will be like I KNOW ITS FUCKED UP WTF
macegr: at sonos, QC guys were like WTF ACOUSTIC GUYS WHY CANT YOU KEEP YOUR SPEAKERS WITHIN A +/- 2dB ENVELOPE
then they toured the factory, attitude changed to JEZUS FUCK I DUNNO HOW THESE ACOUSTIC GUYS KEEP THEIR DRIVERS WITHIN QC

Posted by renesis at 10:05 | permalink | 0 comments


np, miss audio geek shit
i get 3 weeks in the shit early january, than back up here
prob not
you can just use a resistor, but if the noise is from the actual mic input, itll just come back with gain unless it drops below noise floor
in which case it doesnt matter because you will have boosted the entire noise floor
bit a regular foam mic condom reduces air noise
yeah if its electric noise then youre dealing with typical grounding and coupling issues
which admittedly in your case is maybe a random factor because open air
a lot of this shit gets sorted out during compliance testing, heh
its not always directly related, but you have to think the same way
ferrites?

Posted by renesis at 10:00 | permalink | 0 comments


whats your max volts?
so youre getting like 4 usable bits?
macegr: maybe try a mic condom
macegr: bigger rails?
depends where the noise is coming from
i dont know what you meant by 04:57:55 < macegr> i was thinking that would be a better solution than a switch
no i mean what is the better solution
software controlled?
mic condom might just reduce air noise
whats ambient like in your space?
fuck i gotta study and shit

Posted by renesis at 09:54 | permalink | 0 comments


macegr: so in this light, you can see why a DSP genelec speaker plot at +/- 0.25 dB from like 100-20khz was impressive
both in terms of the speakers performance, and the rooms ability to display that
itea1: hes not selling speakers or amps, so notrly
macegr: whats your bottom?
abs max mic spl or through your system?
so youre just noise limited?
thermal switch

Posted by renesis at 09:49 | permalink | 0 comments


so imagine trying to deal with this in a professional environment, and being the dude that does the tests that fuck up peoples schedules and budgets
and thats a day in the life of me at work
macegr: yeah dB scale kind of stretches things out
its our ears fault
so the really fucked up thing?
heat dissipation isnt in dB
watts are fuckin watts
meaning +/- 3dB is a 4x variation in power
so maybe 250w, maybe 1000w
deal with that
6dB in power is huge
youd kind of notice
not like, hard to notice, but doesnt seem like a huge change or anything

Posted by renesis at 09:44 | permalink | 0 comments


you can do it you just wont have the bounce you want in normal environments
well honestly dude, youre like IM DOING TEH MATH but youre like using an easy bake oven as measurement equipment
your input is bullshit
yeah but youre trusting a silly iphone app
instead of testing
go into the real world, set the shit up for 70 to 110, compare bouncy light output to SPL meter, tweak
i have no sympathy
testing is pretty much everything in audio
math gets you in the ballpark
math verifies your end result isnt orders of magnitudes stupid

Posted by renesis at 09:39 | permalink | 0 comments


10x is 10dB with power itea1
with voltage and sound pressure its 20dB for 10x, 6dB for 2x
3dB is like ~30%, 1dB is like ~10%
pretty much everything digital audio is voltage referenced because thats what the DACs and ADCs deal in
yes and the log base 10 of 10 is 1
so 20*1 for voltage and 10*1 for power
i think the math is the same for bels and decibels, honestly
and youre still wrong
okay now youre just groveling
do you not want it to clip?
then everything i says applies

Posted by renesis at 09:33 | permalink | 0 comments


users dont know shit about audio
but i mean basically, your usable bits will dictate dynamic range, 6dB a bit, and you just subtract that from whatever you want your max to be
20dB is a typical value, 40 probably works
10dB is not much
26dB, no?
20dB is 10x, 6dB is 2x, volts not power
you have to bin them if you think it matters
shit is as much about QC as engineering

Posted by renesis at 09:28 | permalink | 0 comments


okay so they got 2dB os flop there too
yeah but how many bits
can you use 8?
lets say 8 usable bits
thats 48dB
lets call a concert 130dB worst case
so make the shit start moving at like 85dB
79 for using 9 bits, 73 if using all 10
unlog amp
its big its heavy its wood

Posted by renesis at 09:23 | permalink | 0 comments


so right now, youre tony rodriguez
not completely, but youre in the same bind
the joke about tony was 'tony wants more knob'
he always wanted the knob to be at 12 o clock
ipod, dj mixer, home a/v, pro audio interface DACs
he didnt understand what the knob was for
knob is a big fuckin deal
are you doing FFT shit to this to get binned spectrum values?
whats the linearity on that?
well, hopefully not a dB
macegr: how many useful bits?

Posted by renesis at 09:18 | permalink | 0 comments


and iphone on the desk with mic as close to desk as possible, opening towards speaker
ground plane measurement, biggest reflection is now your friend
gives you a 6dB boost above noise
so i mean best case, assuming the gov guys are right about iphones being +/- 2dB, your 2dB off
which is prob well within the spec of the electret
audio is messy at your budget
like 10dB messy, probably
you want +/- 1dB, prepare to devote your entire warehouse to acoustic measurement, and drop maybe $20k
how is your office way louder than a concert?
oh sorry
well for reference, QSC K12 do like 132 dBSPL
music program material, measured on SPL meters
itea1: actually a lot of it is for the rigid fiberglass panels
and about half is for test gear and infrastructure stuff

Posted by renesis at 09:13 | permalink | 0 comments


at what distance from the speakers and stage with guitar amps and speakers at what dB-SPL?
the numbers for europe are going to be diff for the numbers in the us
pro gear has a way of dealing with this. its called knob
i dunno man most people dont get really moved by the bass until its about 100dB
20dB boost at bass music shows isnt super uncommon, it also doesnt show up on A weighted and not much in C weighted osha type measurements
why not a clip LED and a gain/pad button?
do a 20dB or 30dB pad
youre going to want about 40dB of dynamic range
so i would measure that like, on desk with mic pointing at source, side of mic touching desk

Posted by renesis at 09:08 | permalink | 0 comments


so maybe the electret is 3dB low and your spl meter is 3dB high
youre just using absolutel values on the iphone with an app?
you need a measurement mic and a pistonphone
rent one
or a credible measurement mic
im going to laugh if your iphone is also in a case
but i mean who knows maybe they calibrated the random golden iphone while it was in a case
measuring audio for real is $$$

Posted by renesis at 09:03 | permalink | 0 comments


additionally, or alternatively, move the mic position vs the source several inches
make many measurements, i suspect you will see huge variation in amplitude
been there, kind of like it
um, wed just go outside and smoke weed and cigs in the alley

Posted by renesis at 08:54 | permalink | 0 comments


would just take longer
mostly use log chirps, sometimes MLS noise but not often
only reason to use stead sine is if you need thd+n
you can get straight thd from log sweeps and noise bursts
so are audiomatica, audio precision, and listen inc
no listen inc
they make soundcheck data acq software
macegr: do a sweep
at an amplitude youre seeing 6dB boost
my guess is you will see comb filtering

Posted by renesis at 08:49 | permalink | 0 comments


how do you know the phone is flat?
(its not, look at papers, its like +/-3dB at least)
how do you know youre getting the same signal? youre not, at 1/2" youre seeing out of phase signal at 5khz, and you cant ever trust shit above 10khz
so just at 1khz?
phase always counts if there are reflections
well
a perfect reflection is exactly 6dB
thats why ground plane measurements are fuckin awesome
so ya thats what im saying
youre not going to get equiv spl in a randomly shaped environment
macegr: mic in the air, get a mic stand or use string
get as far from reflections as possible
put the source an inch away
basically your direct will be huge compared to any reflections
except baffle step reflections
so eventually you get like a 6dB bounce
few hundreds hz, then i believe it combs but we never use that shit

Posted by renesis at 08:43 | permalink | 0 comments


and youre going to have tp explain measurement
because if youre in a room, ground and cieling bounce could easily account for the boost
what do you mean by that
one measurement is between 70 and 90
and youre +/- 12 dB
of when you do measurements between 70 and 94 they come out with +- a few dB
so youre seeing up to 6dB deviation
so what are you doing about reflections
in audio twice the voltage is noticable but not a big deal, honestly, unless youre worried about smashing into rails
how are you gating?

Posted by renesis at 08:38 | permalink | 0 comments


theyre usually referenced at 1khz
so dB per dB would be deviation from that
if you mean compression, you have to test that
itea1: http://www.parts-express.com/audiomatica-clio-pocket-personal-acoustic-measurement-system--390-900
CLIO with a mic is significantly under $1k
usually when you get an acoustic engineer something nicer than CLIO, they want CLIO too because its fast, easy
sonos had like 8 of them
they would stick them in closets, forget about them, order another, was kind of sad
ya is one of the big issues with the place, why i didnt mind being gone
ya most places do
you can get that cheap USB measurement mic from parts express, dayton branded, has a curve file you can download based on serial
ac-130h: they arent mics, theyre data acq systems
we mostly had b&k mics, they were much more closely watched
industry standard
itea1: you dont like earthworks gear?
i kind of dig their 1/4" measurement mic
doesnt feel like itll freak out and die if you drop it
is it flat?

Posted by renesis at 08:33 | permalink | 0 comments


which i believe are the same ones they ship with their test your own damn ears system
protip: dont step on your earbuds, there not always the same after
smaller diaphragms have higher noise
slash lower sensitivity, but they have higher frequency response ranges
sure and it amplifies more noise

Posted by renesis at 08:27 | permalink | 0 comments


macegr: are you measuring in a room?
with what source?
right im saying even b&k shit isnt close to that now
and you can get CLIO from audiomatica for bit over 1/10 that
get gated time impulse measurements
ya it is
a lot of their shit even says B&K
ya
itea1: b&k is a sales rep that looks like jeremy clarkson who tells you not to buy GRAS because GRAS just uses the same 80s designs that theyve long ago developed into something better
we bought the GRAS shit at last job =\
ya but hearing aids dont sell well at guitar center
i did really like my cheap etymotics tho

Posted by renesis at 08:22 | permalink | 0 comments


heh
$35k for gated acquisition?
i think you can get a single channel CLIO fir under $1k
the normal systems are probably about $3k, you can get away with an earthworks mic for under $1k, and B&K or GRAS equivs will be bit more but not that much more
02:21:44 < macegr> it's a -44dB electret mic. the thing is, these are only specified at 1kHz at 94dB. so everyone out there assumes that the sensitivity is flat across the dB range
thats not a good assumption, and i dont think its supported by the plots in most electret datasheets (a lot of which have crazy 10dB divs on the plots)

Posted by renesis at 08:17 | permalink | 0 comments


Top | Add to Technorati Favorites


© 2007 lordpil.   XHTML 1.0! CSS! Site design by GNAA  Blog Engine by pbx | MULTI2 | ian hanschen | lolwat