Reply to thread

Message: <blockquote data-quote="Al Walker" data-source="post: 100954" data-attributes="member: 5264">Re: Uli Behringer of The Music Group Q&A    Question and suggestion Re  ,X32Dear all,My name is Al Walker and I am a senior systems engineer working for the MUSIC Group. I have been working on understanding the causes for the noise issues reported with our ADA8000, and have recently posted on the SAWStudio forum to reach out to the members there to learn about the set-ups that they are using in order to improve our understanding of the mechanisms responsible for generating the noise.  I would like to do the same here, and invite the members of this forum to share with us the details of the audio interfaces, computers and software packages they are using to allow us to fully capture a list of equipment that people are experiencing problems with.Previously we had received reports about noise generated when ADA8000s were used in conjunction with the MOTU 2408 unit, and to investigate the cause we obtained a MOTU 2408 Mk3 unit and PCI-424 card. It was during this testing that we identified the difference in performance between the original Alesis AL1401A version of the ADAT Encoder (transmitter) IC and the current Wavefront AL1401AG version when the ADA8000s were slaved to the ADAT output from the MOTU 2408.  We have now had a report of this same problem occurring with the RME Digiface, which we will be investigating.We discovered that some ADA8000/AD8200 units do work just fine with the MOTU 2408, and to this end we conducted 100% screening of the Wavefront Optical Encoder ICs used in our ADA8200 production as we found that the noise generation varied with temperature.  We tested sample units with our MOTU 2408 Mk3 and did not see any difference in performance between 44.1 kHz and 48 kHz operation – if we had, we would not have made the decision to ship the ADA8200.  As all our testing results confirmed correct operation at both 44.1 kHz and 48kHz, we still need to better understand under what circumstances the noise that has been reported with ADA8200 appears.The fact that an audio interface always operates with other devices means that careful examination is needed to find out on which side the problem occurs. Our investigation has shown that the MOTU 2408 clocks are subject to clock jitter (noise) which leads to syncing problems, which the Alesis version of the ADAT Encoder IC is better able to reject.  The noise generated in this condition varies, but is usually acombination of high amplitude digital noise and clicks and pops indicative of a clocking problem. This is why we felt that we should at least advise users of the known issue when used with MOTU 2408, hence our disclaimer notice which you can find on the product pages of our website. We decided that our only choice moving forwards was to be able to control the manufacturing of these ICs ourselves and as has been previously posted we are now ramping up production. We expect the first lot to arrive in our factory in September and from then onwards we will be switching to the new silicon.As an aside, most digital audio ICs use three separate clock inputs, the master clock frequency, the bit clock (the data rate for the individual data bits) and the word clock which runs at the sampling frequency.  In typical designs operating at 48 kHz and 44.1 kHz sample rates, these clock frequencies are 12.288 MHz, 3.072 MHz and 48 kHz, and 11.025 MHz, 2.822 MHz and 44.1 kHz respectively. The most typical way of generating these clocks is to successively divide down the output of a very accurate crystal oscillator operating at the master clock frequency to obtain the bit and word clocks, this ensures the phases of the clocks are maintained and that any (very small) error in the crystal oscillator output is not magnified when generating the other clock frequencies.The design of the ADAT Encoder IC is unusual in that it only has a word clock input and internally multiplies up this clock frequency to generate the much higher bit clock and master clock frequencies. This multiplication process, done using a phase-locked loop (PLL), needs to be able to reject any jitter noise present on the incoming word clock signal, as obviously otherwise when the sample clock frequency is multiplied up first 64 times and then 256 times, so is the effect of any incoming noise. This has been the issue with the Wavefront version of the ADAT Encoder IC compared with the original Alesis version. Whilst Wavefront themselves have not been at all forthcoming about the changes to the IC design that we have seen to have occurred, we have worked with our supplier to obtain die revisions of the IC that we were advised had improved jitter rejection, and then screened these ICs ourselves for correct operation at multiple sample rates and temperatures.With regard to some of the comments about power supply-generated heat on ADA8000, one of the key improvements we made with ADA8200 was the change to a high-efficiency switch mode power supply to eliminate the concerns with heat generation, so the power supply design is completely different compared with ADA8000.I would appreciate it if forum members could help provide answers to the following, and also any other information that they think may be relevant to help us fully resolve this issue.<ul> <li data-xf-list-type="ul">What type of noise (clocking noise, low level noise) is generated?</li> </ul><ul> <li data-xf-list-type="ul">Does noise occur with both PCs and Macs?</li> </ul><ul> <li data-xf-list-type="ul">What audio interfaces have been tried that result in noise being generated?</li> </ul><ul> <li data-xf-list-type="ul">Does the problem only occur if the ADA8000/ADA8200 is slaved to the audio interface which is in turn slaved to the computer or PCI internal clock?</li> </ul><ul> <li data-xf-list-type="ul">What DAW software packages have been tried?</li> </ul>Thanks for your support.Kind regards,Al WalkerSenior Engineer, SystemsMUSIC Group</blockquote>

Verification

Top Bottom