If you have a problem with me agreeing with you let's sort it out in PM. No need to clutter up the thread beyond this point.
check my post, its already edited before I saw yours, because I realized how it read. you disagree we are arguing about minutiae? (sp) I thought thats what we do here mostly… the original post had an edge I didnt intend, my apologies
my problem is with arguing ideals, but without actually supplying any alternative. we have illustrated on a single page that the posters all want a different dac, how do you suppose there is a viable and top shelf niche product within a niche within a niche? and besides it doesnt exist and i'm not about to develop it; are you?
someone is much more likely to develop a board that can be used by many (or so you would think), even for DIY, once you integrate the dac onboard you limit your market. it certainly limits its appeal to me
my problem is with arguing ideals, but without actually supplying any alternative. we have illustrated on a single page that the posters all want a different dac, how do you suppose there is a viable and top shelf niche product within a niche within a niche? and besides it doesnt exist and i'm not about to develop it; are you?
someone is much more likely to develop a board that can be used by many (or so you would think), even for DIY, once you integrate the dac onboard you limit your market. it certainly limits its appeal to me
Last edited:
No, no, if you use an ESS pin you can fit nine angels on its head. Gives a sound reminiscent of noval tube sockets but with enhanced sparkle.
I'd say the edited version of the post probably doesn't come across as intended either, but no matter; let's move on.
I'm developing mine because its the product I'd like to buy myself if it were already available - the secondary aim is to make it easily available and as affordable as possible to others. I disagree that a device with on-board DACs limits the market - I'd contend that there are plenty of people out there who trust the designer's choice of DAC, probably more than those who want to faff around with choosing different DACs. Integrating the DAC I've chosen for several reasons - the main one being cost-effectiveness and a secondary one being the one Steve Jobs often favoured - so that others can't screw it up so easily.
Your requirements don't appear to me to be in the mainstream.
Ok, so it's possible but maybe not technically the best way to do it.
And there is no way that we can settle for Ess-dac's
"twest820" Thank's for the post, it was interresting. We have been guessing
about a multichannel cs-dac for "Chaparks" project but he still want to keep it
to himself (fair enough). Exiting that there are more dsp xo's coming as well.
Qusp want's an all digital because he already has the dac's (likes them to of course).
The rest would rather see at least the dac's as an onbord function, right ?
And there is no way that we can settle for Ess-dac's
"twest820" Thank's for the post, it was interresting. We have been guessing
about a multichannel cs-dac for "Chaparks" project but he still want to keep it
to himself (fair enough). Exiting that there are more dsp xo's coming as well.
Qusp want's an all digital because he already has the dac's (likes them to of course).
The rest would rather see at least the dac's as an onbord function, right ?
An idea, spdif or lvds i2s from DSP system to each speaker. Then, as close as possible to the speaker have DAC+[reclocking if that's how you roll, synchronising L/R speakers, is an issue here]+IV+amps all intelligently integrated. For 2 way this would be feasible today I expect, for 3-way speakers it is becoming more cumbersome.
Well that is my utopia anyway.
Well that is my utopia anyway.
no, I suspect your requirements are no different from mine, would you accept an ESS dac for example? no, no you wouldnt and have stated as much. mainstream? who said anything about mainstream? most people dont like to choose what goes in the food they eat either and look at the obesity rates.
i'm endorsing something that means nobody HAS to choose. for instance, if you make a reasonably priced and quality DSP unit like this, with built in dacs, a number of people will just hack up and bypass your dacs, still sound like good signal integrity?
what you WILL find however is a number of people who already have spent quite some time building high spec dacs that would prefer not to throw all that away in favor of using 'something cost effective that cant be screwed up easily' sounds real attractive, better get a marketing team; something Steve knew very well. seems like you cherry picked and morphed an idea of what he might say. cost effective was never really part of his MO
twest lol tubes, angels and magic? dont use them, bad aim with your generalization. I prefer things I build are modeled and tested on high end agilent and AP analyzers, do I have one that can reliably measure sub ps close in phase noise? no and i'm guessing you dont personally either and would need to rely on your models and the work of others for much of what you do too.
yes I prefer ESS, I just do, just as you just dont, but I dont expect others to fall in line with that
thats what you call moving on is it?
Last edited:
I already guessed what the dac was and said as much, wasnt difficult to figure out. his dacs can never work for me, a number of problems are dealbreakers for the analogue side, the largest being it only has single ended output.Ok, so it's possible but maybe not technically the best way to do it.
And there is no way that we can settle for Ess-dac's
"twest820" Thank's for the post, it was interresting. We have been guessing
about a multichannel cs-dac for "Chaparks" project but he still want to keep it
to himself (fair enough). Exiting that there are more dsp xo's coming as well.
Qusp want's an all digital because he already has the dac's (likes them to of course).
The rest would rather see at least the dac's as an onbord function, right ?
OK i'll just leave you guys to talk about replicating existing products but with a different dac.... sounds exciting!
The difference is I design to my requirements, you (to my limited knowledge, do correct me here if its mistaken) adopt other's designs for your ends.
I don't follow - I don't design in ESS for a number of reasons - price being the main one, difficulty of implementing a high quality analog stage is another, relative secrecy surrounding the device a third.
That was my interpretation of your words '...can be used by many...'.
That would be fine by me, one very good reason for not selling people ESS DACs that they're then going to bypass is it not?
Yep - they're not my intended market, too niche for me
<edit> Referring to your later amendment - who's Steve? Ah OK you meant Jobs, I'm a bit slow
Last edited:
not if people want a finished product that they can deploy straight away. It is your requirement that is in the minority. Just ask Behringer
You're welcome. You can find my mutterings about my WM8805+LPC4300+CS4365 DSP+DAC project on Abraxilito's open source DSP thread for the moment. I plan to make the firmware open source and should have a limited quantity of PCBs available once the boards are back from fab. But, well, don't hold your breath; it's been in the works for a year and remains vapourware.
For SPDIF it's generally handy to have a microcontroller to program settings and manage enable pins to avoid pops and so on. So you might as well put a DSP capable microcontroller on the board and use it for XO and EQ while you're at it. I'm planning to deploy my WM8805+LPC4300+CS4365 boards this way so that the listening room has less clutter. Also doesn't hurt that plugging in one Toslink cable is cheaper and way less hassle than all the chassis work and soldering to set up three XLR interconnects. A six channel DAC is overkill for this but it's cheaper and easier to lay out than two two channel DACs of comparable performance.
For LVDS a handy default is Ethernet hardware. That provides four signal lines. MCLK, BCLK, LRCK, and an IxS data line. If you stick to I2S yeah, it's limited to two way, but if you control both ends of the link I3S or beyond can be implemented without much fuss---look at NXP's SGPIO peripherals on the LPC4300 parts and National's DS90LV LVDS transcievers. Otherwise it's nominally one ethernet cable per I2S which isn't too bad as two Ethernet cables to two stereo DACs is a pretty natural setup for three or four way. I didn't go with LVDS I have a requirement to support SPDIF sources, at which point it's simpler and cheaper to use a WM8805 as a transciever than to put down LVDS hardware. Headline jitter in Wolfson's PLL is comparable to headline jitter on the DS90LV804 though it's unclear how baseband jitter compares (both are about an order of magnitude higher than an XO's output---a few tens of picoseconds rather than a few picoseconds).
Main thing is configure the analog gain structure so there's good bit depth utilization in the DAC for the desired listening levels. With that in place it's no big deal to do volume at the source or the DSP in a DAC direct to power topology using 24 bit samples downstream from the volume. Personally I use my Squeezebox for this as it saves DIYing a remote control but it's not hard to hang an IR receiver off a microcontroller or use the micro's ADC peripheral to read a volume level off a pot. For folks who like to occasionally blast it there's probably value in implementing a loud mode which boosts power amp gain (or uses higher wattage amps), increases crossover frequencies to optimize for SPL rather than directivity, and so on.
Correct, hence the check for data back in post 13. I'm hoping I may evenually be able to get some WM8805 phase noise measurements done. But, without having hardware, that's kind of tricky.
Holy Grail DSP...worth a shot?
Hi Abraxalito & Twest,
I am interested in developing a high end DSP to equal or better the DSP from DEQX that I have used since 2005.
The on board DAC's, power supply, analog out stage, general build quality has always been average PC grade ie not what we are all looking for.
But the core DSP is simply the best, at the moment...!
If you guys are interested in a DIY DSP board to beat the DEQX then I would like to help in any way I can. Through my " day job" contacts I have access to OEM pricing on a huge range of electronic components and services. Even in small DIY group buy volumes ( say 100 boards) I can get pricing that would normally be in the 1,000 plus MOQ bracket.
I can get LiPo batteries and chargers at prices make the mains transformers & shunt reg supplies look crazy!
Also casework, really cool stuff.
My personal holy grail is fully active three way loudspeakers with alloy rear baffle (or side panel if open baffle) acting as the heatsink for the on board power amps and DAC's, being fed a fully balanced digital signal from the one central DSP / solid state music storage source which controls all volume, switching and Eq / crossover ( linear phase) functions. Not too much to ask...?!
PS Happy to PM more on this if you prefer.
What do you think?
Thanks
Derek.
Hi Abraxalito & Twest,
I am interested in developing a high end DSP to equal or better the DSP from DEQX that I have used since 2005.
The on board DAC's, power supply, analog out stage, general build quality has always been average PC grade ie not what we are all looking for.
But the core DSP is simply the best, at the moment...!
If you guys are interested in a DIY DSP board to beat the DEQX then I would like to help in any way I can. Through my " day job" contacts I have access to OEM pricing on a huge range of electronic components and services. Even in small DIY group buy volumes ( say 100 boards) I can get pricing that would normally be in the 1,000 plus MOQ bracket.
I can get LiPo batteries and chargers at prices make the mains transformers & shunt reg supplies look crazy!
Also casework, really cool stuff.
My personal holy grail is fully active three way loudspeakers with alloy rear baffle (or side panel if open baffle) acting as the heatsink for the on board power amps and DAC's, being fed a fully balanced digital signal from the one central DSP / solid state music storage source which controls all volume, switching and Eq / crossover ( linear phase) functions. Not too much to ask...?!
PS Happy to PM more on this if you prefer.
What do you think?
Thanks
Derek.
Having reflected myself on this for what seems an inordinate length of time, I'd not put all the electronics inside the speaker. Rather I'm planning a DIY DAC-amp arrangement which has individual driver outputs. I'm not settled on the digital interface to this yet - in the first instance it'll be SPDIF though, most likely and probably with a loop out to allow building an any-number-of-ways speaker.
That's essentially what I'm building, though I fall into the 40% of users who need 2V or less out of the power amps and hence the only heatsinks I need are thermal planes on the PCBs (running one's power amp off a 6V trafo has its advantages). Given sufficient programming effort the LPC4300 can deliver most DEQX HDP features, though it lacks the processing power for long FIRs. I (like Bruno Putzeys) am not a fan of the measure and automatically synthesize an FIR approach to room correction---with the effort it takes make the measurement and synthesis smart enough to do a good job you can dial in a lot of biquads. It's also my preference to implement linear phase crossovers via offline time reversed IIR. This means I've no need of FIR and so an LPC4300 is more than sufficient.
Delivering a headline compete against the HDPs requires audio SHARCs or Blackfins for the necessary FIR power. Unfortunately last I checked entry level licensing for Analog Device's Visual DSP enviroment for the parts started at USD 3600. Not what I would consider DIY friendly pricing. There's also the problem of DAC choice, probably the best solution to which is to offer a selection of antialiasing filters for an ES9016 or ES9018 so that one can combine good headline specs (THD and DNR) and good time domain behavior. Unfortunately synthesizing and optimizing minimum phase coefficients for the DACs is rather a hassle. Nothing that sufficient time and effort can't address but, from a pragmatic perspective, one can get a turnkey solution with well chosen tradeoffs and good optimization simply by buying a $4 Cirrus DAC. (ESS could trivially address this by listing filter coefficients in their filter synthesis app note but, at least as of when I got a copy if it, they hadn't done this.)
That said, the current schematic and layout I have for the WM8805+LPC4300+CS4365 board support one I2S input and up to 5 I2S outputs via an expansion connector. So if one wanted to hang a stereo ADC and an ES901x off it to extend the on board IO capabilities that's relatively straightforward. The LPC4300 is also friendly to USB Audio Device Class 2.0's asynchronous mode and supports high speed USB; the hardware wire up is simple, so most of the effort to support USB streaming is in implementing the device class (alternatively one can buy libraries for this, but the ones I've found are far from cheap).
From a DIY perspective it's easiest to use something like a Squeezebox for music storage and user interface tasks. If one's interested in DIYing a DSP XO + Squeezbox Touch kind of solution the LPC4350's worth a look; it has the display driver and ethernet interface in addition the DSP and audio IO capabilities of the 4320 and 4330.
DSP options
Hi Twest,
I really am just a speaker designer, not an electronics / DSP / software guy!
I have tried a lot of active crossovers, Pro, consumer and DIY.
So far the DEQX is easily the best of the bunch, but you do have to use your own DAC's and power supply to get the best out of it.
It would be great to have a lower cost DIY option to the DEQX Digital board.
Count me in if I can help in any way.
Cheers
Derek.
Hi Twest,
I really am just a speaker designer, not an electronics / DSP / software guy!
I have tried a lot of active crossovers, Pro, consumer and DIY.
So far the DEQX is easily the best of the bunch, but you do have to use your own DAC's and power supply to get the best out of it.
It would be great to have a lower cost DIY option to the DEQX Digital board.
Count me in if I can help in any way.
Cheers
Derek.
Hmm, so if you don't have a hard requirement for FIR what are you looking for that the miniDSP 2x4 or 2x8 don't do? miniDSP is oddly quiet about their hardware choices but I know the 2x4 uses an ADAU1701 and there's one mention here on DIYA that the 2x8 uses the CS42528 codec. If the latter is correct it's using the same DACs as I am---Cirrus reuses their -100dB THD/114dB DNR block across several different parts---though the codec's SPDIF baseband jitter rejection likely isn't as robust as that of the Wolfson receivers.
With respect to cost my prelimary BOM is around USD 135 for a fully populated board. That's potentially attractive for folks who want to buy a board and solder it up but not many people are comfortable with soldering 0.5mm pitch parts along with, oh, about 180 0603s. By the time one pays for board assembly it's not really any cheaper than the miniDSP 2x8, though it wouldn't surprise me if the analog side of the hardware is higher spec on my board. However, I seem to recall the 2x8 is ADAU1445 based, in which case it has about twice the DSP capability of an LPC4300. There's also the matter of developing a software stack to the level miniDSP's done.
As an aside, it's possible the Hypex DLCP may actually be available soon. So far as I know all the AKM DACs use linear phase antialiasing with fairly standard brickwall and slow rolloff options. If one likes the ESS parts that shouldn't be much of an obstacle as the ESS filters are similar (though ESS's slow roll is about the slowest roll around) and ESS based DACs tend to be on the pricey side, rather like where the DLCP is likely to come in.
Because Cobranet uses one central clock and synchronises all parts. Latency is not a big problem if all parts in a complex system are kept within the exact timeframe. When digital processing and distribution arrived in the studio and life sound, keeping everything in sync was a big problem with phasing effects as a result.
I mentioned Cobranet as a example how something is done right if you want to distribute digital audio in a more complex system, for example a seperate dsp with DAC's in the loudspeakers.
If you want to use seperate dac's in the same box with the dsp, interconnection with flatcable and boxed headers will be fine if you use a groundwire between every signal wire and keep it as short as possible.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.