Audiodidakt said:
I think the ESS Sabre would be excellent candidate. I guess
the new 32bit Sabre DAC can be set up to do thes x-over
functions internally. I would still prefer the control and display
functions of the DCX, though, and still think that the Sabre's
would be excellent as DACs and still use the control circuitry
of Behringer's
BTW does anybody know if anybody ever came out with mod
to run DCX at 192K? My own personal preference would be
a chain that could keep original sample rates of source material
and do all processing at those freqs to avoid SRC.
You found the link, just read it! It takes time, but you will find what you are looking for.
A number of companies offer upgrades. Here's a couple with a DIY output section.
http://www.pilghamaudio.com/
http://www.selectronic.fr/dcx2496_US.asp
If you want to go with a passive and total DIY check this link.
http://www.dcx2496.fr/en/why.php
I went with the last link only because I use mine in a car. With the road noise I didn't believe I would be able to hear the difference.
Good Luck!
A number of companies offer upgrades. Here's a couple with a DIY output section.
http://www.pilghamaudio.com/
http://www.selectronic.fr/dcx2496_US.asp
If you want to go with a passive and total DIY check this link.
http://www.dcx2496.fr/en/why.php
I went with the last link only because I use mine in a car. With the road noise I didn't believe I would be able to hear the difference.
Good Luck!
AndrewT said:
That one.
Not as good. But I wanted to stay vague and not get into transformer bashing. Who knows, maybe some folks like the sound of Sowter. I didn't - at all. Sure thought I would, tho....EDIT: Just about any 1:1 transformer will work. A 10K:10K might be a bit better. Put about ~3K across the primary of the transfo and you're ready to go. Simple.
I agree. It needs to be thought out first.
Remember the AKM DAC needs to see at least 600 - 1000 Ohm (although this resistance can be on the other side of the transformer depending on what is plugged in to).
Also, you still need some anti-alias filter. This can be provided by the transformer itself (possibly) and also an RC filter. One recommendation is to perform the RC LPF before the transformer so that high frequency junk (not a technical term) does not produce any sort of modulation products in the audio band. At the very least, I would sketch out what the source and load impedance's are in your configuration. When I did this it became very clear that using anything other than a 1-to-1 ratio is going to be a problem (in most cases this rules out trying to use it as a step up transformer).
Remember the AKM DAC needs to see at least 600 - 1000 Ohm (although this resistance can be on the other side of the transformer depending on what is plugged in to).
Also, you still need some anti-alias filter. This can be provided by the transformer itself (possibly) and also an RC filter. One recommendation is to perform the RC LPF before the transformer so that high frequency junk (not a technical term) does not produce any sort of modulation products in the audio band. At the very least, I would sketch out what the source and load impedance's are in your configuration. When I did this it became very clear that using anything other than a 1-to-1 ratio is going to be a problem (in most cases this rules out trying to use it as a step up transformer).
Nothing lackadaisical about it at all. I did tons of measuring, calculating, talking to chip engineers, transformer makers, etc.
I compared a number of output transformers in the identical circuit. They all sounded different, and much better than the stock active output of the Berhinger.*
I just liked the sound of some better than others, that's all. To me, the Sowter was much too forward, too midrangey. Like a little full range speaker with a bright midrange.
Sure, it could well be that the circuit was not optimum for the Sowter, which is why I hestitate to knock it. But all the other transformers did well in the same circuit.
Why the 3K across the primary? To load the DAC output. As Mr. W. Tarragon states, these chips like a fairly low load. This is true of most of the differenitial output chips. 1K ohm works well as a load. With a load of about 20K on the secondary of the transformer (power amp) - 3.3K across the primary gives a very nice, straight impedance curve of near 1K for the DAC chip to see. Yes, I measured it - swept it.
*My tests were done on the DEQ, not DCX. Also on other chips such as Cirrus logic.
I compared a number of output transformers in the identical circuit. They all sounded different, and much better than the stock active output of the Berhinger.*
I just liked the sound of some better than others, that's all. To me, the Sowter was much too forward, too midrangey. Like a little full range speaker with a bright midrange.
Sure, it could well be that the circuit was not optimum for the Sowter, which is why I hestitate to knock it. But all the other transformers did well in the same circuit.
Why the 3K across the primary? To load the DAC output. As Mr. W. Tarragon states, these chips like a fairly low load. This is true of most of the differenitial output chips. 1K ohm works well as a load. With a load of about 20K on the secondary of the transformer (power amp) - 3.3K across the primary gives a very nice, straight impedance curve of near 1K for the DAC chip to see. Yes, I measured it - swept it.
*My tests were done on the DEQ, not DCX. Also on other chips such as Cirrus logic.
Chip and direct out mod make top class
Changing the chip and running direct out through a Dayton foil coupling cap will provide sonics that are much better than any of the current $1000.00 commericial dacs. Changing the chips is hard but the direct out mod is easily done by stealing the signal from the ribbon cable right after it comes out of the dacs.
.
The schematics are found at the Yahoogroup for the DCX in the files section.
.
http://tech.groups.yahoo.com/group/DCX2496/
.
.Audiodidakt said:
Changing the chip and running direct out through a Dayton foil coupling cap will provide sonics that are much better than any of the current $1000.00 commericial dacs. Changing the chips is hard but the direct out mod is easily done by stealing the signal from the ribbon cable right after it comes out of the dacs.
.
The schematics are found at the Yahoogroup for the DCX in the files section.
.
http://tech.groups.yahoo.com/group/DCX2496/
.
Scott,
last I read you were experimenting with that load resistor value and thought 1k 'stole' something from the music. What did you end up with? I know some were using the 1k and a low pass RC but the chip has a built in low pass (75k)? I am only using a 1k into a Mundorf Mcap of 5.8uf.
andy
last I read you were experimenting with that load resistor value and thought 1k 'stole' something from the music. What did you end up with? I know some were using the 1k and a low pass RC but the chip has a built in low pass (75k)? I am only using a 1k into a Mundorf Mcap of 5.8uf.
andy
What is gained by loading the AK4396 (or4393) so heavily?
Why load the chip so heavily? Loading 1k across the balanced legs is like shorting each leg through 500R to ground. That much load is on the edge for most opamps let alone a dac chip. What is to be gained with the increased load versus something more reasonable like 4-5k per leg?
.panomaniac said:
Why load the chip so heavily? Loading 1k across the balanced legs is like shorting each leg through 500R to ground. That much load is on the edge for most opamps let alone a dac chip. What is to be gained with the increased load versus something more reasonable like 4-5k per leg?
It may depend on the chip, but that is usually what they are spec'd for. I've asked tech support if they found a better operating point, but they said no - 1K was it.
That said, I've run them from 600R to 50K. They didn't seem to have a problem at all. But if you are using a transformer, it's not going to be a linear load unless you put a swamping resistor across it.
But please, feel free to experiment with different loads. It's easy enough to do! Somewhere between 1K and 3K has worked best for me. YMMV.
That said, I've run them from 600R to 50K. They didn't seem to have a problem at all. But if you are using a transformer, it's not going to be a linear load unless you put a swamping resistor across it.
But please, feel free to experiment with different loads. It's easy enough to do! Somewhere between 1K and 3K has worked best for me. YMMV.
Help, information needed.
Sorry guys I interfere with current conversation but need a rapid response to make a decision. My question is about what kind of data is vehiculated between two DCX's, daisy chained by RS-485?
To be more specific, I need to know if using digital input on first unit, which would be for example channel Left, there is a mechanism to transmit channel Right digital stream to the second unit using the link?
Or, the single mode to do the job is by splitting digital S/PDIF cable, connect them to input A of each unit and choose S/PDIF --> A in the first unit and S/PDIF --> B in the second unit as the input stream.
I did'n find such a case in manual's examples, maybe someone somewhere try this and can explain to me. I'm thinking such a configuration (daisy chain, digital input) is the best in a more than 3 way stereo setup (mine is 4-way). Thank you!
Regards,
Dorin
Sorry guys I interfere with current conversation but need a rapid response to make a decision. My question is about what kind of data is vehiculated between two DCX's, daisy chained by RS-485?
To be more specific, I need to know if using digital input on first unit, which would be for example channel Left, there is a mechanism to transmit channel Right digital stream to the second unit using the link?
Or, the single mode to do the job is by splitting digital S/PDIF cable, connect them to input A of each unit and choose S/PDIF --> A in the first unit and S/PDIF --> B in the second unit as the input stream.
I did'n find such a case in manual's examples, maybe someone somewhere try this and can explain to me. I'm thinking such a configuration (daisy chain, digital input) is the best in a more than 3 way stereo setup (mine is 4-way). Thank you!
Regards,
Dorin
Never answered question asked once by oehlrich
As a DCX owner, in my perpetual search for the perfect volume control solution I found an extremely interesting question asked once by a forum user named Oehlrich. He wrote:
"I studied the datasheet of the AK4396 D/A converters. For sure you know a D/A converter is a multiplying device. It multiplies the digital input with the reference voltage. The result of this multiplication is the output signal. This means decreasing the reference voltage also scales down the output signal. In the data sheet of the AK4396 I found the following text:
Analog output voltage scales with the voltage of (VREFH-VREFL). AOUT (typ.@0db) = (AOUT+-) - (AOUT-) = +- 2,4Vpp*(VREFH-VREFL)/5
At the DCX2496 VREFL is 0V and VREFH 5V. This means the output is +-2.4V. Decreasing VREFH down to 0V also will decrease the output signal swing down to 0V! So my idea is inserting a pot in the supply for VREFH for all three D/A converters. That’s all. It’s so simple!
But there is another value in the datasheet. It says that VREFH shold not be lass than 0.5V below AVDD which means minimum 4.5V. The question is what will happen below 4.5V? Will the chip die with a cloud of smoke or will it work despite of the limitation in the datasheet? So maybee increasing of VREFL is the better way. There is no maximum voltage for VREFL in the datasheet so it may be inceased up to 5V which also gives an output signal of 0V!
I have not tested this yet and it is not without risk. So think about it carefully before you try. But if it will work it
really would be a very cool patch!"
Nobody seems to trigger at Oehlrich's ideea, so I put it again to your attention. If this:
"There is no maximum voltage for VREFL in the datasheet so it may be inceased up to 5V which also gives an output signal of 0V!"
works then we just solve in an elegant manner such an hard big problem of DCX. But same time we're leaving without object a pretty good kit industry here.
As a DCX owner, in my perpetual search for the perfect volume control solution I found an extremely interesting question asked once by a forum user named Oehlrich. He wrote:
"I studied the datasheet of the AK4396 D/A converters. For sure you know a D/A converter is a multiplying device. It multiplies the digital input with the reference voltage. The result of this multiplication is the output signal. This means decreasing the reference voltage also scales down the output signal. In the data sheet of the AK4396 I found the following text:
Analog output voltage scales with the voltage of (VREFH-VREFL). AOUT (typ.@0db) = (AOUT+-) - (AOUT-) = +- 2,4Vpp*(VREFH-VREFL)/5
At the DCX2496 VREFL is 0V and VREFH 5V. This means the output is +-2.4V. Decreasing VREFH down to 0V also will decrease the output signal swing down to 0V! So my idea is inserting a pot in the supply for VREFH for all three D/A converters. That’s all. It’s so simple!
But there is another value in the datasheet. It says that VREFH shold not be lass than 0.5V below AVDD which means minimum 4.5V. The question is what will happen below 4.5V? Will the chip die with a cloud of smoke or will it work despite of the limitation in the datasheet? So maybee increasing of VREFL is the better way. There is no maximum voltage for VREFL in the datasheet so it may be inceased up to 5V which also gives an output signal of 0V!
I have not tested this yet and it is not without risk. So think about it carefully before you try. But if it will work it
really would be a very cool patch!"
Nobody seems to trigger at Oehlrich's ideea, so I put it again to your attention. If this:
"There is no maximum voltage for VREFL in the datasheet so it may be inceased up to 5V which also gives an output signal of 0V!"
works then we just solve in an elegant manner such an hard big problem of DCX. But same time we're leaving without object a pretty good kit industry here.
I've answered myself reading AKM4396 datasheat. There is defined a deltaVref which is VREFH-VREFL and this deltaVref cannot be less than 3Volt. Lowering deltaVref from 5V to 3V is about 3dB decrease, not so big deal. So, I'm keep searching on for other volume control solution... sorry for digression!
I did the same for years, tried
* 6-gang volume after DAC and before custom 1 opamp stage
* 6-gang volume forming a balanced shunt regulator in each DAC output
etc. etc.
Ended up buying Jan's active output upgrade and it is defenetly the most convenient and best suited for purpose + it sounds good.
With any other solution it is not easily possible to make the needed level offset between drivers in analog domain.
Ergo
* 6-gang volume after DAC and before custom 1 opamp stage
* 6-gang volume forming a balanced shunt regulator in each DAC output
etc. etc.
Ended up buying Jan's active output upgrade and it is defenetly the most convenient and best suited for purpose + it sounds good.
With any other solution it is not easily possible to make the needed level offset between drivers in analog domain.
Ergo