| TobWen |
I forgot to say something:
I am working in a German music-studio.
At the moment we are using ProTools 888-Interfaces with 8/8 ADDAs, but the A/D is ****.
In order to record a perfect sound from our microphones, we need good A/Ds.
I first thought about buying Apogoo-ADs, but they I found this forum where audiophile-profis are chatting about quality and I thougt, it's better to request a self-build A/D.
Can you help me?
Better A/Ds means better sound for you !! |
|
|
| machinow |
Hi, there is a way, of course. I have a pair of CS5394 117dB ADC for 4 years and yet I've done nothing with them. Now there are better, 96/192kHz ones. Check also CS5396. I think it needs a microcontroller. Good layot is very necessary. And some quality opamps or something exotic, - and, of course, good low jitter clock generator.
What is your budget??? And how many channels do you need? In one box or separate stereo ADCs?
regards
marry X-mas |
|
|
| TobWen |
Why does the CS5396 needs a microcontroller?
Of course ... quality OPs ... I thought of OP275 by Analog Devices.
Jitterclock can come in from Wordclock in the studio-beat.
2 Channels needed right now ... for prototyping of course.
Finally we need 8 channels (mono) A/D and 40 channels (mono) D/A.
But I did not find any good pricing D/A in this forum.
The main problem is, you guys build very good D/As and write much about them in this forum.
But nobody can send me a prototyp to test the sound.
I am really interested in paying a prototyp or paying the time to test it :-(
btw: Budget isn't a big problem :-)) |
|
|
| Elso Kwak |
Hello Tobias,
Have a look at this thread. I don't know the Audionote ADC is ready for marketing yet.
http://db.audioasylum.com/cgi/m.pl?...r+elso&session=
I also did receive the CD ROM from Peter Qvortrup and I can assure you it sounds good. The CD ROM contains samples from LP's played back through a conventional AD converter and through Peters new product.
I feel he would make a much stronger case if he used a life music feed to demonstrate the differences.:cool: |
|
|
| peranders |
Have you checked this?
http://www.digitalaudio.dk/
| quote: | | At the moment we are using ProTools 888-Interfaces with 8/8 ADDAs, but the A/D is ****. |
What are they based on, AD chip etc? |
|
|
| peranders |
| quote: | Originally posted by TobWen
Finally we need 8 channels (mono) A/D and 40 channels (mono) D/A.
btw: Budget isn't a big problem :-)) |
Indeed, but what is your budget then?
Have you thought about buying a digital mixer? Yamaha for instance? |
|
|
| TobWen |
Hi peranders,
| quote: | | Have you checked this? |
I know Digitalaudio very well, but many professional mixers (human beings, hehe) don't like it's sound. That#s why I did not choose it.
| quote: | | What are they based on, AD chip etc? |
Unknown ... They changed their chips at every revision.
There was
888-16
888-18
888-20
888-23
888-24
Now they switched to "HD-96" and "HD-192"
I think they only use the flagships of Crystal or AMK.
| quote: | | Indeed, but what is your budget then? |
Budget means ... 8 channels => 2000-3000 EUR
| quote: | | Have you thought about buying a digital mixer? Yamaha for instance? |
I really hope, this was a joke ... If not, I will never write to you again :-) |
|
|
| Guido Tent |
| quote: | Originally posted by TobWen
Why does the CS5396 needs a microcontroller?
Jitterclock can come in from Wordclock in the studio-beat.
|
the 5396 does not need a uC for all applications
If you want serious sound then have the masterclock where it should be, at the AD, and slave the rest
regards |
|
|
| rfbrw |
| quote: | Originally posted by Guido Tent
If you want serious sound then have the masterclock where it should be, at the AD, and slave the rest
regards |
Not really an option in a studio enviroment of any reasonable size. |
|
|
| Guido Tent |
| quote: | Originally posted by rfbrw
Not really an option in a studio enviroment of any reasonable size. |
I know about studio's, but what alternative do you have ? jittery clock at AD conversion will remain in worse recording.
You might use an additional PLL in the AD that is slaved......
regards |
|
|
| rfbrw |
| quote: | Originally posted by Guido Tent
I know about studio's, but what alternative do you have ? jittery clock at AD conversion will remain in worse recording.
You might use an additional PLL in the AD that is slaved......
regards |
You do what the most do. Use a PLL and a VCXO and lock to the house SPG. |
|
|
| Guido Tent |
| quote: | Originally posted by rfbrw
You do what the most do. Use a PLL and a VCXO and lock to the house SPG. |
the loopfilter can be sloooow........
happy Christmas ! |
|
|
| Mark Hathaway |
Howdy Tobwen,
I can say nothing but great things about crystal semiconductors cs5361. A discrete input stage and a kwak clock make it very nice sounding. You could buy their evaluation board and modify it fairly easily.
The stoock eval board is not optimal, with an opamp input stage with electrolytic dc blocking caps, but can be modified.
Its great to have a really low noise adc chip, the hard part is making the preceeding gain stages quiet enough to use the super wide dynamic range it offers. I've reached the point that different brands of to92 transistors make a difference in noise. I've managed about 100db of dynamic range (dc-20khz) with 80db of gain, 24/192khz is very demanding.
The real challenge is if you are going to use multiple adc chips, they need to be clock synced. The crystal semi's have this feature. The next sticking point is the aes3/spdif tx/rx system is less then optimal for multi track recording, adapting an existing adc box could be the answer.
I'm still trying to come to grips on exactly how bad the aes3 multichannel problem is. At the moment I've had some advice from Brian Brown in another thread here, and have read a few aes.org papers covering the topic. I'm still at the very start of this learning curve re: aes3 clock sync/group delay
Mark Hathaway |
|
|
| rfbrw |
| quote: | Originally posted by Mark Hathaway
Howdy Tobwen,
I'm still trying to come to grips on exactly how bad the aes3 multichannel problem is. At the moment I've had some advice from Brian Brown in another thread here, and have read a few aes.org papers covering the topic. I'm still at the very start of this learning curve re: aes3 clock sync/group delay
Mark Hathaway |
Methinks you doth see a problem where none exists,sir. |
|
|
| rfbrw |
| quote: | Originally posted by TobWen
Hi there,
I just found a great A/D on the web: AKM AK5394a
Is there a way to create a stand-alone ADC using the AK5394a?
Greetings
Tobias from Germany |
Is there any particular reason for selecting this A/D as opposed to one from Analog Devices, Crystal or TI ? |
|
|
| TobWen |
Happy X-Mas Mark Hathaway,
| quote: |
I can say nothing but great things about crystal semiconductors cs5361. A discrete input stage and a kwak clock make it very nice sounding. You could buy their evaluation board and modify it fairly easily.
|
The problem is, the cs5361 is pretty old. It was standard for studio-equipment some years ago.
Today standard is: AKM5392a and CS5381
But the AKM is used more and more. Is the CS5381 pin-compatible to the CS5361?
| quote: |
The stoock eval board is not optimal, with an opamp input stage with electrolytic dc blocking caps, but can be modified.
|
A friend of mine developed a nice board for the cs5361.
Which op-amps would you prefer? At the moment he is using OPA2134 by TI / Burr Brown
and Switched-Capacitor Voltage Converter LT1054 by Linear Technology.
The Interface Transmitter is the CS8405A.
| quote: |
Its great to have a really low noise adc chip, the hard part is making the preceeding gain stages quiet enough to use the super wide dynamic range it offers. I've reached the point that different brands of to92 transistors make a difference in noise. I've managed about 100db of dynamic range (dc-20khz) with 80db of gain, 24/192khz is very demanding.
|
Yes ... 100db of dynamic range @ 192 KHz.
But in Europe, music studio produce @ 44.1 KHz & 24 bit.
What about dynamics & noise @ 44.1 Khz?
| quote: |
The real challenge is if you are going to use multiple adc chips, they need to be clock synced. The crystal semi's have this feature. The next sticking point is the aes3/spdif tx/rx system is less then optimal for multi track recording, adapting an existing adc box could be the answer.
|
At the moment I only need a REALLY PERFECT Stereo-A/D for recording vocals or guitars.
Happy X-Mas
Tobias |
|
|
| TobWen |
Happy X-mas rfbrw,
| quote: | | Is there any particular reason for selecting this A/D as opposed to one from Analog Devices, Crystal or TI ? |
Yes ... because you can find the AKM in any A/D > 1000 EUR / $
So there must be something about this chips (argh, my bad English). Perhaps the new Crystal CS5381 is as good ...
I have opened many actual A/D & D/As and inside were:
ADC: AKM5392
ADC: AKM5392a
DAC: AD1852
DAC: CS4396 |
|
|
| TobWen |
ah, I just got informed by a friend of mine, that the "old" Cirrus-Chip only uses 64x oversampling.
the AMK uses 128x - I don't know about the "new" Cirrus :-( |
|
|
| peranders |
| Digital Audio Denmark uses CS5396/97 as ADC and AK4395 as DAC..... |
|
|
| peranders |
| quote: | Originally posted by TobWen
Yes ... because you can find the AKM in any A/D > 1000 EUR / $
|
M-Audio uses AKM in their rather cheap products. |
|
|
| TobWen |
a friend of mine is working @ M-Audio :-))
They are using AKM AK4524VF
I hate these codecs ...
listen: We need a really good A/D :-)))
Watch the attached file ... this is our self-build A/D.
Is it good? Sinus Sweep!
ADC cs5361.
OPA2134 by TI / Burr Brown
Switched-Capacitor Voltage Converter LT1054 |
|
|
| fmak |
| quote: | Originally posted by peranders
Digital Audio Denmark uses CS5396/97 as ADC and AK4395 as DAC..... |
-----------------------------------------------------
I evaluated the unit and didn't think much of the adc/dac sound - rather bland with little dynamic impact; rather inoffensive. The Wadia 17 is better but errs in the wrong direction. A homemade CS5396/BB1704 sounds better than both. |
|
|
| machinow |
| quote: | Originally posted by TobWen
Why does the CS5396 needs a microcontroller?
Of course ... quality OPs ... I thought of OP275 by Analog Devices.
Jitterclock can come in from Wordclock in the studio-beat.
2 Channels needed right now ... for prototyping of course.
Finally we need 8 channels (mono) A/D and 40 channels (mono) D/A.
But I did not find any good pricing D/A in this forum.
The main problem is, you guys build very good D/As and write much about them in this forum.
But nobody can send me a prototyp to test the sound.
I am really interested in paying a prototyp or paying the time to test it :-(
btw: Budget isn't a big problem :-)) |
TobWen,
the bigger problem I think is the synchronizing with low levels of jitter. A good PLL gives about 100pS , which is much for a studio equipment. The design may use an ASRC like AD1896 and master clock and a PLL to generate the output clock from the synchro signal. This is a complex design, and more, maybe the ASRC will have it's own colours of sound. I do not know how the profi-equipment are made, but this will be the solution I would prefer. ASRC are sayd to be used in "digital mixing consoles" mainly
:rolleyes:
There are some good ADCs like AD5396/97, AKM you mention and the new 5381. On my knowledge, AKM are large sales - oriented and are generally difficult to be bought in small quantities /especially in my country /:confused: / Can you buy AKMs there???
I think CS5396 needs some initialization by a micro, but not sure if this is true. The 5394 is more stand-alone friendly, but only 48kHz.
There are many good opamps from AD and TI, but most favourable I think are AD8620/10 and OPA2134 /134 - the first one is pretty expensive at ~$8 at 1000+ quantities.
Again, the good reclocking is the main problem - have you any info for this?? And what is the output format you need?
Regards
:)
PS: If you were a little bit closer I would have nothing against sending you my DIY DAC for a try /AD1852 OPA2134 based - not something special of course/ :) |
|
|
| machinow |
CS5381 seems to be pretty good on paper and is pin-compatible w. 5361 - but this does not mean it is software - compatible or stand-alone compatible, this should be checked out. :rolleyes:
It was at my attention before half an year for one semi-profi design, but only for a while - the design was much more cheaper, say AK4524 ;)
Happy New Year |
|
|
| TobWen |
Hi machinow,
no ... I cannot get AKMs here, too.
I am trying to get some CS5381 ... do you have a good distributor for Crystal / Cirrus?
What do you think about my tests of cs5361?
We changed OPA2134 to the cheaper NE5532 and it got much better:
SNR: 109,5 dB
THD+N is 0,0001%
So I don't like the OPA2134 anymore :-))
Is it compatible to the AD8620/10?
Where can I get it?
Output I need AES3 and S/PDIF.
For clocking we are using a MK1412A at the moment! |
|
|
| fmak |
I think CS5396 needs some initialization by a micro, but not sure if this is true. The 5394 is more stand-alone friendly, but only 48kHz.
--------------------------------------------------
The 5396 can be reinitialised by press switch and I recommend it as reinitialisation during use improves the sound - the chip seems to drift. |
|
|
| peranders |
TobWen, so you like NE5532 better. What are your demands then? If money is no object why can't you hire a consult for the MC programming?
No ADC is good enough for you, have you tested dCS? They are at least expensive.
http://www.dcsltd.co.uk/
Since you live in Germany you can get all normal IC's, including Cirrus and AKM. |
|
|
| TobWen |
Hi peranders,
| quote: | | so you like NE5532 better |
No, the measures are better with the NE5532 :-)
| quote: | | If money is no object why can't you hire a consult for the MC programming? |
Where is the fun then? I don't want to have any guy building my ADC ... I want to do it on my own, to test it on my own and be happy about the results of my diy-project.
The results of these project I am gonna share on the web that anybody can rebuild it and make it better.
Of course I can go to the store and buy a 3000 EUR Apogee-A/D.
That is state-of-the-art-quality.
But I don't feel any fun at spending money in a hifi-store :-)
| quote: | | Since you live in Germany you can get all normal IC's, including Cirrus and AKM |
That is correct for a part.
Of course I can get Cirrus & AKM, but the problem is, you can only get 50-1000 from the distributors.
AKM is _very_ hard to get because the normally send to the manufactors of soundcard (etc.) directly and not to the end-customer.
You can get Cirrus in parcels with 50 pieces ... but 50 x 17 EUR is expensive for tests only :-(( |
|
|
| peranders |
| quote: | Originally posted by TobWen
That is correct for a part.
Of course I can get Cirrus & AKM, but the problem is, you can only get 50-1000 from the distributors.
AKM is _very_ hard to get because the normally send to the manufactors of soundcard (etc.) directly and not to the end-customer.
You can get Cirrus in parcels with 50 pieces ... but 50 x 17 EUR is expensive for tests only :-(( |
Didn't you say that you wanted to build 40 channels? So you need 42-50 IC's at least.
If you want to buy 50 IC's I believe that your distrubutor happily delivers those. |
|
|
| TobWen |
| quote: | | Didn't you say that you wanted to build 40 channels? So you need 42-50 IC's at least. |
Argh, I just have seen that I mixed up 2 things because I hadn't enough sleep.
We need 50 channels D/A in the studio, but only 8 channels A/D.
42-50 ICs isn't much ... 850-900 EUR for ICS.
| quote: | | If you want to buy 50 IC's I believe that your distrubutor happily delivers those. |
In Germany distributors only sell parcels with 50-1000 pieces.
You can't buy Cirrus chips under 50 pieces.
That's why developing at home is so very hard here :-(( |
|
|
| rfbrw |
| quote: | Originally posted by TobWen
That is correct for a part.
Of course I can get Cirrus & AKM, but the problem is, you can only get 50-1000 from the distributors.
AKM is _very_ hard to get because the normally send to the manufactors of soundcard (etc.) directly and not to the end-customer.
You can get Cirrus in parcels with 50 pieces ... but 50 x 17 EUR is expensive for tests only :-(( |
You need to find out who their distributors are. They are listed on the relevant manufacturer websites.
A few years ago I worked on the design of a 4ch ADC using AKM and Crystal parts. For the prototypes I was able to get AKM parts in small numbers (4 off) from A UK based company called DIP International. The CS8402 tx chips came from Sequoia Technology also in the UK and again in small numbers.
http://www.dipinternational.co.uk/
http://www.sequoia.co.uk/
BTW, what does the studio as a whole lock to? |
|
|
| TobWen |
| quote: | | BTW, what does the studio as a whole lock to? |
doest this mean "look like"?
here is a pic |
|
|
| rfbrw |
| What I mean is what Sync pulse generator do you use to synchronize the different pieces of equipment. |
|
|
| TobWen |
| atm Wordclock or S/P-DIF sync from a DAT-Recorder or from the ProTools |
|
|
| Mark Hathaway |
Howdy TobWen,
using opamps in front of a adc like the 5361/81 is a sin, and the ne5532???!!!
I've not played with the cs5381, only the 5361, 5351 and the ad1871. A CLEAN clock, no matter the chip, is crucial.
I'd rather use a cs5351 with a discrete front end using matched transitors and plastic film caps then a 5381 with opamps and non bypassed electrolytic caps, eeeeeeeeeeeeeeeeeeee.
Spose this is why I diy, I think opamps should NEVER ever be used, they suck.
Getting the power supplies clean enough for high end adc work isn't easy, good luck:)
Mark Hathaway |
|
|
| machinow |
| quote: | Originally posted by TobWen
Hi machinow,
no ... I cannot get AKMs here, too.
I am trying to get some CS5381 ... do you have a good distributor for Crystal / Cirrus?
What do you think about my tests of cs5361?
We changed OPA2134 to the cheaper NE5532 and it got much better:
SNR: 109,5 dB
THD+N is 0,0001%
So I don't like the OPA2134 anymore :-))
Is it compatible to the AD8620/10?
Where can I get it?
Output I need AES3 and S/PDIF.
For clocking we are using a MK1412A at the moment! |
Hi TobWen,
Crystal Semi are maybe easier to be bought here in small quantities, but not sure. Anyway, 5..10+ quantities are necessary. The situation with AKM's is like in your country, or worse.
How does your project measure w. OPA2134s??? NE5532 has a good output stage that delivers less THD with low load inpedances at high frequencies. It is a very good opamp for it's price. AD8620 claims to be some state-of-the-art opamp, and very expensive too
:cool: Look for Analog Devices distributor. www.analog.com
You need something like the CS8405/06 for S/PDIF transmitter, if you want sync think about AD1896 from AD, and some PLL for audio... /is MK1412 such, or it is the equipped clock generator in your studio??/
With discrette transistors u can get better results at lower prices, but u need to be advanced analog engeneer to design them, otherway better do with OPAMPS. Good Power Supply and clock is critical.
NE5532 vs OPA2134 - the first can generate much more ID when speed overloaded due to its BJT vs. FET /in OPA2134/ input stage. For best performance use FET OPAMS or Fast Bipolar OPAMPs, some Video OPAMPs with voltage feedback etc. With discrette bjt one can do this easily, at the price of increased noise /1..2uV ref. input/
Regards |
|
|
| machinow |
Also, PCB layout is very important... Not necessary are the n-layer PCB's, but at 2 layers with large planes and SMD components give good enough results. My higher-end DIY projects have double-sided PCB w. good layout, Euroboards only for small cheap experiments :)
btw Analog Devices sell small quantities generally, but seems that they do not have very good audio ADCs, or I would prefer them /here I can buy anything at 10+ quantities from their distributor /
Do you have an opinion for the resampling I suggest you?? Otherway you will have to do with 100pSecs of jitter, which is not very good for studios
Regards |
|
|
| TobWen |
machinow,
| quote: | | How does your project measure w. OPA2134s |
don't have one here :-(
can't say anything about that
| quote: | | You need something like the CS8405/06 for S/PDIF transmitter |
We are using CS8405A
atm the MK1412 is the clock on the A/D and it works for now :-)
At the studio a Fostex D15 acts as a wordclock.
Of course, better analog-equipment could raise quality, but it is hard to get samples and I don't want to change my circuits every day :-((
atm I don't do any resampling. I set the CS5361 to 44.1 KHz and record :-) |
|
|
| machinow |
| quote: | Originally posted by TobWen
atm the MK1412 is the clock on the A/D and it works for now :-)
At the studio a Fostex D15 acts as a wordclock.
Of course, better analog-equipment could raise quality, but it is hard to get samples and I don't want to change my circuits every day :-((
atm I don't do any resampling. I set the CS5361 to 44.1 KHz and record :-) |
Hi again, the MK1412 can have a very high jitter - look in the specs, something like 200pS peak??? This is very much for a studio, only goes for a commersial CD/DVD player in the low/mid price range
:cannotbe:
The jitter problem is greater than the ADC one - whatever new and expensive ADC u use u can not get good /enough/ performance from it, especially at higher bitrates and sampling freq.
Resampling would be necessary just to do with the jitter problem, and synchronize with the wordclock. It is not necessary itself. The other jitter solution is to generate the masterclockl at the adcs and slave ALL THE OTHER equipment :whazzat: with that clock. Here you need one multichannel ADC, can not use n stereo ADCs
These are the only solutions for the jitter I know so far.
Happy New Year |
|
|
| rfbrw |
TobWen
Do you have some idea of the basic features required.
For example, given that the AKM5394 supports 192K sampling would you want to use this or want the option of mono mode AES3. How many different sync options do you need, wordclock AES3 video etc. |
|
|
|
