I use a dam1021 with AES/EBU in a studio environment.
Using it in phase critical applications, like eg parallel processing, might be problematic since the macro timing seems to fluctuate slightly. I heard it while recording two loops of the same sound file with the dam1021 (into the pcm4222 evaluation board adc).
When reversing polarity on one recording and listen to the leftover of the two files running in parallel I heard a that the two files change phase relationship while running. As if the dam1021 corrects its fifo slightly, but not always the same way. With traditional (FIFOless) DAC converters the PCM4222 ADC does not do that, so I blame the DAM1021.
The jittertest shows a very clean signal, much better than my RME UFX sound card. So micro timing seems to be flawless.
Maybe my aes/ebu arrangement is faulty, but I rather think, that the Dam1021 locks to the clock but corrects the fifo every once in a while so the sound file shifts a little in time.
Please correct me if my assumption is wrong.
The boards will follow clocks to within a few uS, but two dam1021 will never be synchronous. If that's a requirement, just use synced dam1121.
Yes it is! Best software on the market, and it never let me down. Sound quality is good enough for a lot of things, but the the dam1021 is the better Dac and the pcm4222 is the better adc.
Interesting. I don't understand how you can have audio integrity while stile exhibiting changes in macro timing though. For instance, my RME audio cards and my Lynx Aurora16 all have re-clocking (calling it SteadyClock, Syncro Lock etc.), but it all results in perfect sample-accuracy. However, these are jitter-prone and thus acoustically non-transparent.
I currently use the PCM4222EVM ADC as my main AD converter (BTW Soeren, what about that SAR ADC? I would buy that in a heartbeat). For my PCM4222EVM I have build a custom NE5534-based input stage with precision-trimmers for low DC, so you can turn off the acoustically non-transparent high pass filter. It also sounds better than the standard slightly mushy balanced op amp input stage utilizing modern "high end" parts.
As for the RME software, it's rock solid, but I preferred the old mixer to the new "modern" looking one.
If a few uS are below the ca. 23 uS of a single sample at 44,1khz I wonder if there is any practical relevance. The phase shift would effectively occour outside of the audio band, would it not?
Easy, I just use it as a master DAC on the buss. One sample timing deviation might be a problem with parallel processing and in a multi microphone setup, but not there...Interesting. I don't understand how you can have audio integrity while stile exhibiting changes in macro timing though. For instance, my RME audio cards and my Lynx Aurora16 all have re-clocking (calling it SteadyClock, Syncro Lock etc.), but it all results in perfect sample-accuracy. However, these are jitter-prone and thus acoustically non-transparent.
I simply put an high quality 4:1 Pikatron Transformer (terminated with 600ohm) in front of the chip, bypassing the electronics. No noise, no dc, perfect level and impedance adjustment, great transparent sound, that is a very elegant solution to multiple problems.
You have to look at two issue, let me try to explain:
AD converter: First, you need to have the sample clock with low jitter as that affect the sound directly. Second, you need to have the same sample frequency for all your other AD converters as you otherwise can't process them together, like mixing.... But that clock can have jitter as the samples can be buffered as long as the frequency are locked to the wordclock and the jitter are removed again when doing the DA conversion.
The simple way is to take your wordclock and lock your sample clock to it with a fast pll, usually with some jitter and use the same clock for the AD converter... The advanced way would be lock you low jitter slow pll clock to it, and have a small fifo between the AD converter and your audio output which will be locked directly to your wordclock. Kinda the reverse of a dam1021....
DA converter: Since you have analog out you don't have any need to lock it to the wordclock. And it doesn't matter if one channel is a couple of uS away from the other, nobody can hear it as it's like having the speakers a couple of mm from each other....
The RME audio cards and Lynx Aurora16 probably just use the wordclock input for the AD converters....
I don't like to have any trimmers, and I don't want transformers, so when I get to doing my AD converter it will have a fully balanced low noise input amplifier, actually based on my dac1541 output amplifier, and a digital high pass filter, they don't affect the sound at all if the cutoff frequency is low enough.
AD converter: First, you need to have the sample clock with low jitter as that affect the sound directly. Second, you need to have the same sample frequency for all your other AD converters as you otherwise can't process them together, like mixing.... But that clock can have jitter as the samples can be buffered as long as the frequency are locked to the wordclock and the jitter are removed again when doing the DA conversion.
The simple way is to take your wordclock and lock your sample clock to it with a fast pll, usually with some jitter and use the same clock for the AD converter... The advanced way would be lock you low jitter slow pll clock to it, and have a small fifo between the AD converter and your audio output which will be locked directly to your wordclock. Kinda the reverse of a dam1021....
DA converter: Since you have analog out you don't have any need to lock it to the wordclock. And it doesn't matter if one channel is a couple of uS away from the other, nobody can hear it as it's like having the speakers a couple of mm from each other....
The RME audio cards and Lynx Aurora16 probably just use the wordclock input for the AD converters....
I don't like to have any trimmers, and I don't want transformers, so when I get to doing my AD converter it will have a fully balanced low noise input amplifier, actually based on my dac1541 output amplifier, and a digital high pass filter, they don't affect the sound at all if the cutoff frequency is low enough.
If you get the transformer circuit right it can be great. I tried a Beclere trafo on the in place I want to keep as free of any coloration as possible.
There's always DC with the PCM4222EVM boards, the transformer should not change that. A trimmer was the only way I could get it to acceptable levels (below 110 db).
I'm afraid I don't quite understand. In my setup the 2-channel AD converter is the master. Every DAC (currently one 16-channel converter, 3 8-channel converters and two 2-channel converters) slaves to it. Since every DAC gets the same clock they are syncronized (but some of them have a different delay converting from D to A, so one should use the same unit for tracks that need to be phase coherent). If I were to replace these DACs with lot's of DAM1021 all synced to the same clock via the AES/EBU input they should be syncronized in a phase coherent manner within the audio band, shouldn't they?
Any idea when your ADC will be availible? I would buy it, and a lot of other pro audio people would, too.
Seems to work pretty well for Mr Putzeys, Mola Mola.
Lampizator changes DAC's more often than his underpants
and every oneis by far the best LoL.
Implementation is the key....DS or MB.
T
Yeah, you can measure different delays for the different DACs, but unless the delay is large enough (I don't know how large it need to be), it will not affect the audio quality, in your case it's like moving the microphones around, the speed of sound is around 300 m/sec, so t.ex. 100 uS is 30 cm (1 feet for the few non metric).
The dam1021 with its few uS delay difference, it's definite not something you can hear, multiple units will sync to having the same delay, without any jitter.
Note the term delay difference, the absolute delay only matter in live settings, although if using minimum phase filters the dam1021 delay will be a little over 1 mS. And jitter is a different thing that ARE creating artifacts.
Sorry, no time frame, but I'm aware that there aren't any non delta sigma AD converters on the market, pro or consumer.... Maybe I should put priority on it now I'm done with the dac1321, dac1421 and dac1541 series, although I also have other projects I'm working on....
I'm not saying it's not working, I'm saying take a look with a spectrum analyzer, downstream it could create intermodulation effects....
Lampizator changes DAC's more often than his underpants
is by far the best LoL.
Implementation is the key....DS or MB.
T
Implementation is important, but Delta Sigma, and therefore DSD, will always have modulation effects, for the best ones it might be small but it will be there....
Sounds great, multiple units syncing having the same delay is all I need.
Soeren, what are your thoughts on the balanced outputs of the DAM1021? Are they acoustically transparent?
There is but one studio quality ADC i know of that doesn't use sigma delta (I think): The Lavry Gold AD-122mk3. But it's really pricy.
No, additional parts always do something.... But the used opa1602 are pretty good parts, but of course they don't beat my discrete buffers used in my dac1xxx series
I haven't been able to find any info on that anywhere, if you have some facts I would like to know what ADC chip Lavry use nowadays, I believe their really old ones didn't use Delta-Sigma, but those parts are not being manufactured anymore.... If just someone could take a photo of the internals....
Hi Soren
Are you 100% positive that multiple Dams in parallel have exactly the same delay?
As I wrote I did a test with recording multiple takes of the same track through the dam and with reversing polarity I heard that the takes had different phase relationships and even worse, they changed during the course of the track. As if the fifo adjusted...
But of course my testing setup might have been faulty....
Last edited:
Where these changes continuous or like steps? If they where indeed continuous like a warble that would indicate that the buffer don't work kas intended. If distinct steps, how often? Please, if possible, post a short .wav snippet?
//
As I said, you can probably measure it, the 45/49 Mhz master clock is adjusted in one hz steps, so of course sometime one is one hz away from the other.... If two dam1021 are feed the same clock, the delay will be the same down to a few uS. But it's so tiny that it's not something you can hear when playing music.