Sign magnitude R2R ladder DAC has better THD at low level, since MSB switch only if required, so it's acts as a lower resolution network.
So a 24 bit resolution with resistors matched to 0.01% has perfect accuracy when only 12-13 bit switch.
But starting from 13/14th bit the problem of accuracy is the same (very similar), only you save 1 bit accuracy since you need 2 sections of 23 bit resolution.
So it's not I2S format.no ...really
You have to use S/PDIFbad idea
you can get I2S signal directly from the decoder chip TC94A70FG inside the devicewhat is the difference between directly from the TCA94A070 and on the actual connector ( to the DAC ) ?
Serge
I repeat, it's not I2S format, I2S is 2's complement MSB first.
Please, read I2S specification (attached from Soekris).
Never said S/PDIF is a good idea, I'm not the designer of your CD player.
I don't understand your question, please explain better.
Gentlemen (and Ladies), please remember that this thread is in the Vendor section of the forum, and all questions should be addressed to the Vendor, in this case Søren, directly for answers.
It is IMHO, that if you wish to discuss the minutia of the design amongst yourselves, perhaps you should start a new thread in the main discussion forum.
We are all waiting anxiously for Søren to finish and test the final design, and then hopefully offer it to us for sale as soon as possible thereafter.
It is IMHO, that if you wish to discuss the minutia of the design amongst yourselves, perhaps you should start a new thread in the main discussion forum.
We are all waiting anxiously for Søren to finish and test the final design, and then hopefully offer it to us for sale as soon as possible thereafter.
Very interesting product Søren!
Is it possible to add, by means of external components, some kind of analog gain control / attenuation between the DAC and the balanced output buffer or do such variable gain control directly on the output buffer itself?
The reason for asking is that I would like to have this feature in a multi-channel DSP/preamp.
Is it possible to add, by means of external components, some kind of analog gain control / attenuation between the DAC and the balanced output buffer or do such variable gain control directly on the output buffer itself?
The reason for asking is that I would like to have this feature in a multi-channel DSP/preamp.
I assume it's for tracking when multichannel ? My plan is actually to have an option of connecting serial out to serial in between multiple DAC boards, to enable tracking digital volume control when doing digital crossovers....
Otherwise you would have to use the direct R-2R output.
Thanks Søren. The serial in/out for tracking volume control is a very nice feature. My question was related to optimizing max output level on each channel without loosing anything in the digital domain. Depending of speaker drivers and amplifier combinations the difference in needed gain can differ conciderably, especially when combining high efficiency compression drivers and direct radiators.
An easy fix would be the option of having a jumpers between the R2R outputs and the buffers. This would enable any signal tweaking on the outputs. But off cource, I understand if this option will not be implemented.
Right.. but sometimes someone loves to answer in place of the vendor.
Just a question for the Vendor, to understand the product philosophy:
how did you implement the sign magnitude in each channel
- with single ladder network and single voltage reference (multiplexed)
- with single ladder network and dual voltage reference (switching the ladder reference from positive to negative and vice versa)
- with dual ladder network and dual voltage reference?
Thanks
Andrea
Maybe some LVDS Transceiver for I2S on board, it would easier implementation of multi-channel DAC in each active box.
This is just a suggestion.
+1
This could be very convenient. Like Tytte71 I need a multichannel DAC for use with a DSP crossover solution. Both my PC-based software (Audiolense) and standalone solution (MiniDSP) need at better DAC.
Andrea,
Søren said 3:with dual ladder network and dual voltage reference in post#34.
by the way
I used to planned sign-magnitude DAC with single DAC-IC,
but gave up why it seems impossible to keep the gains of upper and lower output accurately since add them (other than adjust with trimmer...?)and it must be necessary sample and hold circuit to switch upper and lower.
Do you have any idea to overcome these problem, Andrea?
Søren said 3:with dual ladder network and dual voltage reference in post#34.
by the way
I used to planned sign-magnitude DAC with single DAC-IC,
but gave up why it seems impossible to keep the gains of upper and lower output accurately since add them (other than adjust with trimmer...?)and it must be necessary sample and hold circuit to switch upper and lower.
Do you have any idea to overcome these problem, Andrea?
Discussion is free here, as long as it's nice and on subject....
As Shinja already noted, see post #34.
I actually first tried to find a trilevel driver, but couldn't find anything good, closest was a A3357 SP3T switch. So I ended to with two complete driver/resistors sets as the way to do it.
If there is enough business then my long term plan is to make a custom chip
If there is enough business then my long term plan is to make a custom chip
Excellent idea. Go on then. Interest and sales will surely follow.
Shinja, Soeren
I have never built a discrete DAC, I thought to build one several time, but every time I abandoned the idea because of the complexity of the job, and every time I said to myself that was better to take a TDA1541A from the shelf.
BTW is enough obvious that the option 1 should be the ideal, while the 3rd option should be the worst, since there are 2 ladder network and 2 voltage reference that will be not equal, causing a dissymmetry in the sinusoid.
Reading the datasheet seems that BB implements the 2nd option "The two DACs share a common reference, and a common R-2R ladder ...".
In the PCM1704 they used 2 voltage rails, positive and negative, but one reference only with a servo to ensure the voltage tracking, and a single ladder network. So I presume they switch the R2R ladder between positive and negative depending on the sign bit.
I have no idea how they did implement the switch, but I presume that a sample and hold is required.
In the early 90's I designed with a friend an integrated amplifier remote controlled only. Since the budget was limited, we decided to use the crappy CD4053 switch to act the source selection and the CD4051 as the volume control.
While these switch are surely crappy device, the amplifier didn't sound so bad (it did win a "Best buy" from HiFi Choice or HiFi News, I don't remember exactly). We solved the problem of switch linearity placing them in the feedback loop of an opamp and the result was very good.
But I don't think this tecnique could be adopted to switch the voltage reference of the ladder network.
Thinking to other solution to implement the 3rd option, the easiest way could seem to use something similar to a DC motor controller with four mosfet drived by the sign bit, but the gate capacitance will limit the speed of the circuit, and the transition between positive and negative will generate high glitch.
Another way could be to use a single ladder network with only positive voltage reference, followed by a pair of buffer and a pair of amplifier, the first inverting and the second non-inverting. A pair of mosfet drived by the sign bit should switch to ground the input of the unneeded amplifier. But maybe it will be the same glitching issue.
Finally, I have no idea if is possible to implement the 3rd option, but I hope Soeren will investigate this opportunity to improve his DAC.
I have never built a discrete DAC, I thought to build one several time, but every time I abandoned the idea because of the complexity of the job, and every time I said to myself that was better to take a TDA1541A from the shelf.
BTW is enough obvious that the option 1 should be the ideal, while the 3rd option should be the worst, since there are 2 ladder network and 2 voltage reference that will be not equal, causing a dissymmetry in the sinusoid.
Reading the datasheet seems that BB implements the 2nd option "The two DACs share a common reference, and a common R-2R ladder ...".
In the PCM1704 they used 2 voltage rails, positive and negative, but one reference only with a servo to ensure the voltage tracking, and a single ladder network. So I presume they switch the R2R ladder between positive and negative depending on the sign bit.
I have no idea how they did implement the switch, but I presume that a sample and hold is required.
In the early 90's I designed with a friend an integrated amplifier remote controlled only. Since the budget was limited, we decided to use the crappy CD4053 switch to act the source selection and the CD4051 as the volume control.
While these switch are surely crappy device, the amplifier didn't sound so bad (it did win a "Best buy" from HiFi Choice or HiFi News, I don't remember exactly). We solved the problem of switch linearity placing them in the feedback loop of an opamp and the result was very good.
But I don't think this tecnique could be adopted to switch the voltage reference of the ladder network.
Thinking to other solution to implement the 3rd option, the easiest way could seem to use something similar to a DC motor controller with four mosfet drived by the sign bit, but the gate capacitance will limit the speed of the circuit, and the transition between positive and negative will generate high glitch.
Another way could be to use a single ladder network with only positive voltage reference, followed by a pair of buffer and a pair of amplifier, the first inverting and the second non-inverting. A pair of mosfet drived by the sign bit should switch to ground the input of the unneeded amplifier. But maybe it will be the same glitching issue.
Finally, I have no idea if is possible to implement the 3rd option, but I hope Soeren will investigate this opportunity to improve his DAC.
Solution on Chip is mega bucks...
Hey Kazap....I like your confidence....So will you be providing the $100,000 or more required to fund a Solution on Chip product...???!!!
Soekris has probably spent $20,000 on his time alone to get to this point...
I'd guess another $10,000 worth of lab equipment, prototyping and general office expense on top of that....
At the very competitive ( low!) price points he is aiming to sell the boards at I reckon he needs to sell at least 1,000 DAC boards to make a decent return on his total investment....
If you were a " Dragon" would you invest $100,000 of your cash at this stage into a Solution on a Chip future venture...??
We should all support Soekris and vote with our wallets order one of his DACs when they become available....If he does well he might just reward us all by taking the huge risk with his own ( or borrowed) money and develop a killer DAC SOC ( Solution On Chip) at a $25.00 price point rather than a still very competitive and great value $250.00 ( approx.) ....That's smart DIY audio crowd funding!
Cheers
Derek.
Hey Kazap....I like your confidence....So will you be providing the $100,000 or more required to fund a Solution on Chip product...???!!!
Soekris has probably spent $20,000 on his time alone to get to this point...
I'd guess another $10,000 worth of lab equipment, prototyping and general office expense on top of that....
At the very competitive ( low!) price points he is aiming to sell the boards at I reckon he needs to sell at least 1,000 DAC boards to make a decent return on his total investment....
If you were a " Dragon" would you invest $100,000 of your cash at this stage into a Solution on a Chip future venture...??
We should all support Soekris and vote with our wallets order one of his DACs when they become available....If he does well he might just reward us all by taking the huge risk with his own ( or borrowed) money and develop a killer DAC SOC ( Solution On Chip) at a $25.00 price point rather than a still very competitive and great value $250.00 ( approx.) ....That's smart DIY audio crowd funding!
Cheers
Derek.
Actually the CMOS 405X switches are not crappy devices, as long as you don't load the output....
T.ex. QUAD used them in 80's preamplifiers with great success.
But they're too slow and too high impedance for a R-2R DAC use.
See my post #223
ooops.... I wrote wrong, I meant the 1st option, single ladder network and single positive voltage reference.