Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Updated DSP Module
#1
Here is a sneak preview of our updated DSP module (Revision-C) used in the up coming Ultimate-Preamplifier Plus, which replaces the B2 revision board currently being shipped in the Ultimate-Preamplifer. Updates include the all new ES9038PRO Sabre DAC which replaces the existing ES9028PRO Sabre DAC along with a redesigned IV stage to handle the increased current requirements and new regulators to take advantage of the lower noise performance offered by this DAC.

[Image: Ultimate_Processor(Rev-C).png]
Reply
#2
(04-14-2018, 02:33 PM)Thanks David - looks like SOTA. Is it possible to upgrade the DSP board in the current units with this upgrade? Cheers Wrote: Here is a sneak preview of our updated DSP module (Revision-C) used in the up coming Ultimate-Preamplifier Plus, which replaces the B2 revision board currently being shipped in the Ultimate-Preamplifer. Updates include the all new ES9038PRO Sabre DAC which replaces the existing ES9028PRO Sabre DAC along with a redesigned IV stage to handle the increased current requirements and new regulators to take advantage of the lower noise performance offered by this DAC.

[Image: Ultimate_Processor(Rev-C).png]
Reply
#3
(04-14-2018, 09:51 PM)MattB Wrote:
(04-14-2018, 02:33 PM)Thanks David - looks like SOTA. Is it possible to upgrade the DSP board in the current units with this upgrade? Cheers Wrote: Here is a sneak preview of our updated DSP module (Revision-C) used in the up coming Ultimate-Preamplifier Plus, which replaces the B2 revision board currently being shipped in the Ultimate-Preamplifer. Updates include the all new ES9038PRO Sabre DAC which replaces the existing ES9028PRO Sabre DAC along with a redesigned IV stage to handle the increased current requirements and new regulators to take advantage of the lower noise performance offered by this DAC.

Hi Matt

Should be upgradable because the new board will be the same dimensions as the old one but will have to test first because the current requirements are higher with the new board.

cheers
david
Reply
#4
Hello,
other useful features that would be appreciated by many audiophiles:

1. With no significant cost increase:
- output/preamp stage in a dedicated board; main dsp/dac board with direct differential dac output, so to give chance to any diyer to use his preferred output/preamp stage
- DIP8 sockets for all opamps, allowing users to use IC/discrete opamps of choice; PSU capable of supplying at least 20mA each (there should be no significant compensation issues, as many of them are direct replacement for LM4562). LM4562 are good opamps, but not quite comparable with today's better discretes...
- optional fully sync/fs128 mode for all sources, to get the best from 9038

2. With probable cost increase:
- 32/384 and DSD512 modes to fully exploit Amanero (and other I2S sources) capabilities
- hardware PCM upsampling and conversion to DSD function, hqplayer-quality, for each channel after dsp and before entering dac
- cape for BBB (I2S connection)
- option for Crystek clocks
Reply
#5
(05-28-2018, 09:47 AM)luca72c Wrote: Hello,
other useful features that would be appreciated by many audiophiles:

1. With no significant cost increase:
- output/preamp stage in a dedicated board; main dsp/dac board with direct differential dac output, so to give chance to any diyer to use his preferred output/preamp stage
- DIP8 sockets for all opamps, allowing users to use IC/discrete opamps of choice; PSU capable of supplying at least 20mA each (there should be no significant compensation issues, as many of them are direct replacement for LM4562). LM4562 are good opamps, but not quite comparable with today's better discretes...
- optional fully sync/fs128 mode for all sources, to get the best from 9038

2. With probable cost increase:
- 32/384 and DSD512 modes to fully exploit Amanero (and other I2S sources) capabilities
- hardware PCM upsampling and conversion to DSD function, hqplayer-quality, for each channel after dsp and before entering dac
- cape for BBB (I2S connection)
- option for Crystek clocks

Hello luca

The ES9038 has 4 times the output current of the ES9028 so requires an I/V converter with high current output capability. This limits the selection of suitable opamps in the I/V stage. For this reason we have gone for a fully balanced differential opamp configuration which also has excellent common mode rejection whereas the dual opamp scheme as commonly used or discrete opamp provides little or no common rejection and passes all of the common mode noise through to the balanced outputs. Also with the limited board space available and very limited candidates for the balanced opamp means that through hole option is not useful for this design.

The 384K option is not available in direct dac or preamp modes because the 49.152MHz master clock limits it to 192KHz and the Sample Rate Converter is also limited to 192KHz. Remember the primary purpose of this device is for a high quality active crossover and not a normal standalone DAC. Also the higher the sample rate the more the DSP has to work and use up its resources particularly for FIR filters where 384K is a complete overkill considering the very limited amount of genuine source material available.

There is also option on the board for other clock devices.

Regards
David
Reply
#6
Hello David,
i don't want to force you as you already made a magnificent work and your project is quite unique; just trying to give you some advices, as i see you're really clever and open minded.
Obviously i'm talking about hiend discrete opamps, not about 95% general use IC opamps that are of little use when building a true hiend system. Hi end opamps should have higher current output and better performances than 4562 or the like; but it would be the choice of every user to invest on them or to just stay with good IC opamps: anyone could choose by his needs/possibilities/tastes.
About board space, i'm considering this project from a diyer perspective, so i see no limitations about space or layout: 2 or 3 or even 10 cms more would be no problem in a custom project. But i guess they can be in a fully assembled project...
384 khz option surely will force an upgrade in hardware components, that's why i classified it as a "cost increase" update.
I know this project was started as a top-notch active crossover, but today the borderlines between components and functionalities are fading: to avoid multiple conversion and guarantee top end sound quality, while still maintaining processing functionalities, we're forced to move the dsp/crossover towards the source, in the digital domain; so using a separate dac would be redundant and useless. This means that in a true hiend multichannel system the dsp/crossover has to incorporate (and become) a true hiend dac by itself. You were very farsighted to utilise a true hiend dac chip like 9038 in your project, that really makes it stand out and opens to the market of audiophile active multiamp setups. But future moves towards hd files, hifrequency dsd, and so on...
More, there is a discretely growing community of aficionados of dsd conversion/upsampling solutions (hqplayer, roon and so on) that is stuck by the high computing demands for a normal PC player; they would highly benefit from a dsp/dac component that could ease the load on the PC side by permitting hardware dsd conversion after dsp calculations.

Then i think that things like allowing anyone to use his proper output stages/preamps (or even just allowing the use of better, hiend opamps), increasing frequencies and resolutions limits and providing an hardware processing platform for dsd conversion/upsampling, would make this project a true reference milestone for the audiophile community, particularly if costs can be kept comparable to Revision A and B productions.
Reply
#7
Hi luca

There is much more to be gained from implementing a properly designed active crossover rather than fiddling around the edges with 768K sample rates and convolution filters done in DSD. I know someone who went down this track and spent around 16K on an 8 channel 9018 networked DAC along with a fanless PC just so he could do all of his convolution filters in DSD. In the end his conclusion was that there was no more advantage than having done it using PCM which happens to be the native format for about 99% of the program material out there ! And why did he do this ? Because he read about someone on an audiohpile forum spreading misinformation and unsubstantiated claims about chip solution sigma-delta DAC's and why DSD is infinitely better !

The differential opamp we have chosen for good reason. There is only a handful of possible contenders and I very much doubt that just because it is discrete means it is better is actually going to make a difference in the scheme of things. Originally I was going to install DIP sockets for the opamps but I just did not have enough board space to do it and there is no DIP variety for the differential opamp we are going to use anyway. Also our design is highly integrated with the DSP, ADC, DAC, IV converter and low jitter master clock all in close proximity to each other on the same multi-layer board ! There are no clock fly-leads anywhere running off to different boards. It is no coincidence that it has been designed like this

With this project you need to get your priorities right. Is your requirement about speaker design or is it about DAC design because there are plenty of standalone stereo DAC's with everything you need, but they can't be used for anything else except to drive headphones or passive crossover speakers !!

Regards
David
Reply


Forum Jump:


Users browsing this thread: 2 Guest(s)