A fully Digital Volume Control for the DCX2496 users

[Victor]

I while ago there were many discussions regarding the implementation of the volume control for the DCX2496. I use the combination of the DEQ2496 and the DCX2496 to realize most the EQ needs of my full range 3-way dipole system. Until now I have been using two motorized 4-gang pots from Alps. The pots, naturally, were placed between the DCX and the power amps. Well, suffice to say I no longer use that scheme.

I wanted to set-up an all-digital signal processing, whereby the signal stays in the digital domain from the source such as the CD or DVD player to the power amps. That should include all the EQ and, yes, the volume control as well. The D/A conversion is to take place immediately before the Power Amps. I now have this system working and the key to this set-up is a pro-audio box from Roland, the M-1000 model. It is no longer manufactured but there are many floating around on the net and some are new. I bought two of them and it should say that I am highly impressed with its capabilities.

The M-1000 is a digital mixer with 5 digital SPDI/F inputs including one Toslink and internal 24-bit 96 kHz processing. That means that it up-samples everything to those numbers. It also has digital volume control!

Many will say that it is a problem because you would lose resolution to the tune of 1 bit for every 6 dB of attenuation. Well while it is true in principle I do not think it applies here. Here is why, - the CD data is 16 bits and when this Roland box turns it into the 24 bit word, it simple does it by shifting the 16 bits. So, no new data is created and no benefits of S/N either, but now you have 24 bits to deal with. The extra 8 bits represent slightly more then 48 dB of usable attenuation. That means that the Roland M-1000 can give you a 48 dB of attenuation before you start chopping the resolution that really matters, - the 16 bits of original data.

Naturally enthused with this new-found reality I did, what I always do with all my new toys, - I riped the cover off and looked inside to see what makes it tick and if I could make it tick a little better, - you know what I mean... I was pleasantly surprised to find Analog Devices AD1895 Asynchcronouse Sample Rate Converters (ASRC), - 4 of them. What does it mean? It means that we have a near complete immunity from jitter related distortion and also the processing to the tune of 125 dB of S/N. That's 20 bits! Analog circuitry notwithstanding since we do not go there, but it was also well done. By the way I am already thinking about replacing those AD1895 with AD1896! I need to take my pill now...

Well, in the end here is what I have now, - the DVD player outputs AC-3 or DTS SPDI/F signal into Technics AC-500 processor. I modified the AC-500 so that now it has 3 SPDI/F outputs for the 5.1 channel sound. It was really simple and it lends itself nicely to such modification with about $50 in parts. I take both front and rear channel in the digital SPDI/F form and input them into two Rolland M-1000 units, - one for the front and one for the rear. The M-1000 controls the volume and the balance. From M-1000 I take the signals in digital SPDI/F form and input them into the DEQ-DCX combination for front and the DCX in the rear. So far all processing is done in the digital domain with no resolution loss. The DCX does the requisite D/A conversion and the final few dB of level adjustment for the high-mid-woofer set-up and outputs six analog signals to the power amps.

Well, the listening test support the reasoning, - I like what I hear. The Roland M-1000 becomes a 5-input fully digital pre-amp with volume and balance control. It is ready for 24 bit sources when they come to market and works with no resolution loss for 16-bit sources. Since the maximum digital volume level at the time of recording is 0 dB the combination of M-1000 and the DCX cannot be overloaded into clipping, - which also nice.

Also if one were to use an additional resistor divider built into cables at the output of the DCX, then an overall level adjustment done by M-1000 can be constraned to at most 6 db, - which 1 bit of resolution loss. Therefore, if and when the 24-bit sources come to market, the M-1000 will still be a great unit to use. Becouse 1-bit resolution loss cannot be ever heard by humans, let alone passed through the noise of the best analog power supply available.

Al in all I am happy. And now I can address my project du jur, - the subwoofer, - for which I do need help! But that is a topic for another thread.

Regards to all,
Victor

[thylantyr]

If the Roland unit is no longer produced, do you know
what product took it's place?

[Victor]

If the Roland unit is no longer produced, do you know
what product took it's place?

none, as far as I know

[TomK]

Many will say that it is a problem because you would lose resolution to the tune of 1 bit for every 6 dB of attenuation. Well while it is true in principle I do not think it applies here. Here is why, - the CD data is 16 bits and when this Roland box turns it into the 24 bit word, it simple does it by shifting the 16 bits. So, no new data is created and no benefits of S/N either, but now you have 24 bits to deal with. The extra 8 bits represent slightly more then 48 dB of usable attenuation.

I'm pretty sure that the addition of 8 lsb's provide very little additional usable attenuation. Let me try to explain why.

The significance of all of these extra 8 bits summed together is still less than the lsb of the original 16 bit information. They probably represent signal levels comparable to the noise floor of the system. Any "usable" attenuation will still affect the 16 bit data in the same manner as if the Roland box was not used.

[Victor]

I'm pretty sure that the addition of 8 lsb's provide very little additional usable attenuation. Let me try to explain why.

The significance of all of these extra 8 bits summed together is still less than the lsb of the original 16 bit information. They probably represent signal levels comparable to the noise floor of the system. Any "usable" attenuation will still affect the 16 bit data in the same manner as if the Roland box was not used.

Tom,

Well, let me think about this, - in principle if the data were to be upsampled from 16 to 24 bits before it gets to Roland, then we are in good shape because the upsampling process will store the 16 bit data in the 24-bit digital word using positions from the bit 23 down to the bit 8. The last 8 bits will be padded with zeros. This last 8 bits are in fact used by M-1000 to do the attenuation, owing to its internal 24 bit arithmetics, - and so those last 8 bits will amount to about 46 dB.

I am certain that Roland does this bit shift anyway and so even if the data is not upsampled before it hits Roland still the last byte is used for attenuation purposes and that is why we have no resolution loss of the original 16-bit data. I would argue that this bit shift is a must due to the need to normalize the full-scale volume of all inputs.

Does this make sence to you?
Victor

[Davey]

Victor,

It's an interesting solution, but it still doesn't solve the basic problem with the DCX (or similar)...the signal level relative to the "noise" inside the unit is not optimum. This problem exists anytime the volume control is upstream of the DCX, regardless of whether it's digital or analog. Whether a digital control is losing resolution by chopping bits or not doesn't really matter when the rest of the gain structure is not optimized. IMHO.

Users who have highish gain power amps and/or highish efficiency speaker systems will still notice the DCX "hissing" at them during soft music passages. There's no solution to this problem other than reworking the analog output portion of the DCX for lower gain and/or attenuating externally.

Regarding digital volume controls.....My current setup has my Squeezebox 2 feeding the digital input on the DCX with some fixed 10db attenuators on the DCX outputs. That works pretty darn good and gives the user a remote volume adjustment capability and keeps the signal level inside the DCX nearly optimum.

Cheers,

Davey.

[thylantyr]

I think people get too paranoid about issues. It's normal
human behavior to over analyze and worry about audibility.

I have a simple front end chain for speaker development,

dvd player -> 5532 based preamp -> DCX -> home amp & proamp -> 52 driver line array

Line arrays are very revealing to bad sources and I don't
hear anything bad from a stock DCX using analog in.
The line array is setup for maximum array sensitivity within
reason, ie, 0.8 ohm tweeters, 2 ohm midwoofers driven
by ~ 3kw worth of amps. {I did the -6dB output for the home amp driving the tweeters}.

This simple setup is quiet as a dead bird. The music is detailed at any volume level.

People considering this unit get dissuaded because of
these issues when in fact they stand to gain alot more
than loose and they miss the boat.

[TomK]

Tom,

Well, let me think about this, - in principle if the data were to be upsampled from 16 to 24 bits before it gets to Roland, then we are in good shape because the upsampling process will store the 16 bit data in the 24-bit digital word using positions from the bit 23 down to the bit 8. The last 8 bits will be padded with zeros. This last 8 bits are in fact used by M-1000 to do the attenuation, owing to its internal 24 bit arithmetics, - and so those last 8 bits will amount to about 46 dB.

Victor

Hi Victor,

I believe that you will get more dynamic range, but how "usable" it is is what I am concerned about. You will get the additional dynamic range on the bottom end from about -96dB on down to about -140db which the 16b information is incapable of representing. Normal levels of attenuation of -2, -4, -10 dB or so have to impact the same msbs in the 16b word as in the 24b word. Think of the 8 additional bits as a fractional part to the right of the radix point of the 16b data. There may be some benefit to retaining the fractional part of the data in the additional 8b after the multiplication to avoid truncation, but I don't know how significant it is since the original data was only 16b. Maybe someone with a little more horsepower would like to jump in here.

Best Regards,
Tom

[Interious]

M-1000 no longer available? Are you sure? It looks like it's still made and still available -- http://www.samedaymusic.com/product--ROLM1000

What am I missing?

Dave

[Victor]

TomK,

Yes I think I see what you are saying. You are making an interesting point. Yes, the effect you are describing will definitely theoretically take place immediately once you begin to reduce the level down from 0 dB. The question is what does this reduction really mean in terms of real numbers.

There is a way to make a digital volume control that will not result in any resolution loss, - one would have to use a scalar multiply technique, which is a linear operation. Although the dynamic range would decrease, the spectral content will remain the same.

Anyway, I think that you might be right and the 16-bit data is impacted by the volume reduction, but because of the shift to 24 bits and use of 24 bit arithmetics this impact is not a linear function as 1 bit loss for every 6 dB attenuation. I would submit that the losses for the first 46 db of attenuation are so very small that it is nothing to worry about. Naturally a simple THD+N test would put this to rest.

I cannot do the test, but I did compare the use of Roland M-1000 to a high quality Alps blue attenuator. The attenuator was placed after the DCX2496 and before the power amps. I heard no perceptual differences. It is not scientific, but even with -30 dB attenuation I heard nothing to warrant a suspicion of resolution loss.

In the end, I really like an idea of using a fully digital processing chain. Also the convenience of the M-1000 is simply outstanding, - it is a full digital preamp. I am already thinking about remote control mod along with an improved AD1896 ASRC.

Davey,

I do not see how the gain structure might be somehow wrong inside the DCX, if the DCX input is a digital signal. The S/N issues only exist when we are talking about the A/D conversion process. Naturally, in this case if the input analog amplitude is low we will have problems with noise.

However, if the input is digital, then it does not matter how many bits arrived, the S/N remains the same. The noise associated with digital processing is due to things like the truncation and round-off errors. Those things will manifest as a reduction in S/N with the corresponding penalties in THD.

The ASRCs inside the M-1000 function with those errors but according to the specs those errors result in better then -125 dB S/N. The S/N penalties inside the DCX unit are due to the arithmetics associated with IIR filters, providing we are talking about digital in-digital out scenario. Unfortunately Behringer does not publish that data.

The hissing you are talking about might be due to the inadequate ANALOG input amplitude to the DCX. It will not happen with digital input providing that this digital input was generated with proper analog levels to beginning with. If M-1000 is used, those analog levels were established during the recording process and are as good as they could ever be. My power amps have 30 dB of gain and I hear no hiss at all at any volume level when I use M-1000.

Therefore, with respect to Roland M1000-DCX2496 combination I do not see any problems. In my response to TomK (see above) I agreed that there might be an impact of M-1000 processing but I am convinced that impact is infinitesimal for the first 46 db of attenuation. Past that point you may loose resolution. To be sure I will have to do the THD+N test, but I do not have a THD analyzer on hand.

I have no experience with Squeezebox 2. Does it have a volume control capability?

thylantyr,

I cannot agree with you more!

Dave,

Those are close out stocks. When they are gone, it is over.

Regards to all,

Victor

[TomK]

[B]

Anyway, I think that you might be right and the 16-bit data is impacted by the volume reduction, but because of the shift to 24 bits and use of 24 bit arithmetics this impact is not a linear function as 1 bit loss for every 6 dB attenuation. I would submit that the losses for the first 46 db of attenuation are so very small that it is nothing to worry about. Naturally a simple THD+N test would put this to rest.

Victor,

I have assumed that the case we are talking about is the same full scale signal for either 16b or 24b words. Then, the significance of the msb of the 16b data and the msb of the 24b data is the same. Bit for bit, the first 16 msbs of either length data are exactly the same for the same amplitude. Multiplication by a constant (attenuation) will affect the first 16 bits identically except for perhaps bit 16 in the 24b case. It's not so much that the 16 bits are shifted, but that an additional 8 bits are tacked on below the least significant of the 16 original bits. Their significance is on the order of the quantization error of the original data.

Tom

[Victor]

Tom,

Can you clarify you comment for me? It seems that you are saying that the upsampling to 24 bits does not help matters. However, I think that due to a necessary amplitude normalization it must?

You are saying that "...Multiplication by a constant (attenuation) will affect the first 16 bits identically except for perhaps bit 16 in the 24b case...."

Well then, the bits 17 through 23 are not effected in the same way. That would mean that the overall 24 bit word is not effected nearly as much. In fact given the arithmetical significants of that last byte, it would seem that the 24-bit word will not be effected all that much at all. Is that what you are saying?

Victor

[Davey]

TomK,

Davey,

I do not see how the gain structure might be somehow wrong inside the DCX, if the DCX input is a digital signal. The S/N issues only exist when we are talking about the A/D conversion process. Naturally, in this case if the input analog amplitude is low we will have problems with noise.

However, if the input is digital, then it does not matter how many bits arrived, the S/N remains the same. The noise associated with digital processing is due to things like the truncation and round-off errors. Those things will manifest as a reduction in S/N with the corresponding penalties in THD.

The hissing you are talking about might be due to the inadequate ANALOG input amplitude to the DCX. It will not happen with digital input providing that this digital input was generated with proper analog levels to beginning with. If M-1000 is used, those analog levels were established during the recording process and are as good as they could ever be. My power amps have 30 dB of gain and I hear no hiss at all at any volume level when I use M-1000.

I have no experience with Squeezebox 2. Does it have a volume control capability?

Victor

Victor,

There's more to this problem than the resolution of the A/D converters at the DCX input. The "noise" and "hissing" at the outputs is present (at the same level) regardless of whether the input is digital or analog. Trust me...I've been fiddling with the DCX2496 for a few years now....both on my test bench and in my listening system.

When I mentioned "gain structure" I was referring to the setup outside the DCX and not within it. We, as users, don't have any control of the gain structure inside the DCX, except within the constraints of the software implementation of the input and output sliders. Obviously, those were never designed to work as a system volume control.

The signal "level" inside the DCX, whether it be sourced from analog or digital, needs to be "windowed" at as high a level as possible relative to the clipping point to achieve the best theoretical and real-world performance. Since the maximum output levels of the DCX are much greater than we need to drive our power amps the only workable "solution" is to add attenuation between DCX and power amps or reduce the gain of the output op-amps or a couple of other solutions I've seen users try. (Those output indicators need to be flashing green and amber (but never red) with much regularity.)

A digital volume control pre-DCX, such as the Squeezebox 2 or the Roland, provides a volume control and an improvement in useability, but it doesn't do anything to help out the dynamic range and "signal to noise" situation with the DCX.

Hang in there.

Cheers,

Davey.

[TomK]

Tom,

Can you clarify you comment for me? It seems that you are saying that the upsampling to 24 bits does not help matters. However, I think that due to a necessary amplitude normalization it must?

You are saying that "...Multiplication by a constant (attenuation) will affect the first 16 bits identically except for perhaps bit 16 in the 24b case...."

Well then, the bits 17 through 23 are not effected in the same way. That would mean that the overall 24 bit word is not effected nearly as much. In fact given the arithmetical significants of that last byte, it would seem that the 24-bit word will not be effected all that much at all. Is that what you are saying?

Victor

Hi Victor,

After talking to one of my colleagues, I realize the problem is a little more complicated than I originally thought. I still believe that the effect of the additional bits is comparable to the initial quantization error in the 16b data. Being careful not to mix up resolution and dynamic range issues, I think that the "usable" dynamic range is unaffected. I'm still considering resolution...

Thanks for the intellectual stimulation!

Tom

[Victor]

Davey,

I must closely examine my DCX ( and I have 2 of them!) for hiss. So far I hear nothing, but I might be wrong about that. When you say

"...A digital volume control pre-DCX, such as the Squeezebox 2 or the Roland, provides a volume control and an improvement in usability, but it doesn't do anything to help out the dynamic range and "signal to noise" situation with the DCX..."

I agree. Can you tell me more about this Squeezebox 2? How exactly does it control the volume and what input/output structure does it have? What exactly does it do?

I do not think that the signal level inside the DCX plays any significant role in the S/N determination. I guess we will need to do the THD test to show this, but from purely theoretical perspective I do not see this.

I also have a -10 dB resistive attenuator at the output of the DCX built into the cables. As for the red LED's, well, they never light up with the use of M-1000 because the maximum level of the digital signal inside the M-1000 is 0 dB.

Regards,
Victor

[Dennis H]

The Squeezebox 2 is way cool. It's a little box that you can connect to your home LAN (wired or wireless) and play music stored on a hard drive anywhere in the house. You can use a Pocket PC as a remote control for the jukebox and browse nice color screens looking for the music you want.

http://www.slimdevices.com/pi_overview.html

[Davey]

Victor,

Well, for crying out loud! If you have a 10db attenuator built into your cables then you've improved the "hiss" situation considerably. Obviously, you probably wouldn't notice any problems since by doing this you've forced yourself to increase the internal signal level 10db above what it would normally be. This swamps the low-level noise and moves the signal to a much better position in the dynamic range of the unit.

I agree with Dennis......the Squeezebox 2 is way cool.....and I'm still using the tip of the iceberg. I'm not exactly sure how exactly it implements the digital volume control.

I think you need to do some further test bench testing of your DCX unit to better understand how it works and how to optimize its operation. I have mastered a variety of test signals that you can use to gain familiarity if you'd like. PM me.

Also, you can still end up clipping the unit even if the max level from you M1000 is 0dbFS. If you program some sort of EQ that curves above 0db....or when you start programming crossover filters like LR24 it will change the crest factor of the signal and cause the unit to go into clipping. There are some other ways as well. The clipping indicators on the two input channels are not useful and may indicate red from time to time, but this does not necessarily indicate clipping. The indicators on the output channels are quite a bit more useful though.....if those flash red it's probable that the DCX is clipping.

Cheers,

Davey.

[Victor]

Hi Victor,

After talking to one of my colleagues, I realize the problem is a little more complicated than I originally thought. I still believe that the effect of the additional bits is comparable to the initial quantization error in the 16b data. Being careful not to mix up resolution and dynamic range issues, I think that the "usable" dynamic range is unaffected. I'm still considering resolution...

Thanks for the intellectual stimulation!

Tom

Tom,

This digital volume control topic does puzzle me a lot. And I thought that I new enough of DSP to see my way clear, but sadly it is not the case. I can see how the level control may be implemented in the lossless fashion, but truth be told, I do not know for sure what does this Roland M-1000 do with volume control.

On the surface it seemed to me that the upsampling to 24 bits does the trick, but your posts forced me to think that there is more to it then what I know. I shall look up a a DSP guru here in Toronto and see what I can dig up. In the mean time, if your colleague manages to explain to you what is going on, I would like to know too.

I am also going to look for a THD analyzer to actually measure how clean the M-1000 processing is at various volume levels. This will definitively tell us if there is in fact any loss of resolution. Again, my listening tests are quite positive, but I know better then to say that they are definitive. I may be happily listening to 12 bits, right! Gotta measure everything!

Well, in the end, I must confess that the DSP is not my thing really, although I did all the graduate course work in the subject, I used to spend all my time doing analog sections of IC's on the transistor level and now I just do descrete electronics and DSP is not on the menu any longer.

regards,
Victor

[Victor]

The Squeezebox 2 is way cool. It's a little box that you can connect to your home LAN (wired or wireless) and play music stored on a hard drive anywhere in the house. You can use a Pocket PC as a remote control for the jukebox and browse nice color screens looking for the music you want.

http://www.slimdevices.com/pi_overview.html

Dennis,

Thanks for the link, - I shall look it up. It seems like a handy little toy.

[Victor]

Davey,
I guess I must play with my DCX a little more. Thanks for te pointers. I just might look you up for the files of the signals that you mention. Right now I am going to be busy with my new sub that I am going to build out of the existing box and Ave15 driver.

regards,
Victor

[thylantyr]

Is this how you will connect the Roland device to the Behringer?

Roland M1000 {RCA digital output SP/DIF?} --> DCX2496 digital inpu {AES/EBU input} ?

Any issues? Or are you converting SPDIF to AES?

I've read on other forums where people are doing;

1. Cheating and sending SP/DIF directing into the DCX with no issues.

2. People doing the DCX mod by installed an impedance matching transformer to convert 75 ohm to 110 ohm.

What's the real story ? I placed an order for a M1000 today
to check it out, it's low enough in price to mess around with.

[Victor]

thylantyr,

I am very impressed with M-1000. I actually bought 2 of them. I connect it to the DCX through the 75-110 Ohm 1:1.2 ratio transformer.

It is really a fully digital pre-amp that has 5 SPDI/F inputs. I am quite sure that there is no resolution loss. Also, there is a satisfaction of having a fully digital signal processing chain.

I use both M-1000s to control my 3-way full-range front dipole speakers and 2-way full-range rear dipole speakers. The only inconvinience as I see it, - no IR remote capability. However, it is a small price to pay in my view. Remote capability might be possible if I replace the encoders with motorised versions, but I am not sure it is worth the trouble.

The M-1000 motivated me to go ahead and modify my Sony N975 DVD player for 3 SPDI/F outputs. So, I am spitting 1's and 0's all the way....

...reagrds
Victor

[thylantyr]

Roland M1000 + DCX2496 Impressions

Here is my audio madman methodology. I gather data,
digest, then experiment. Can I hear a difference? It's
either yes or no, this is simple. I do nothing complex.

I've been doing auditions with the line array using different front end electronics to see what I
really can hear.

Step 1
I started simple with 20 year old analog active crossovers,
18dB/octave, Nakamichi brand.

The signal chain;
Source -> preamp -> crossovers -> amps -> speakers

All analog paths. It sounds good, but the crossovers lacked
features to allow best optimization of the sound.


Step 2
The signal chain;
Source -> preamp -> DCX -> amps -> speakers

All analog paths. It sounded better because the DCX has
so much flexibility that I was able to find the sweet spots
when tuning the system. :T

I also swapped sources but didn't hear anything better
or worse.

Step 3
The signal chain;
Source -> Roland -> Canare SPDIF_AES transformer -> DCX -> amps -> speakers

All digital paths up to the amps. I don't hear anything
better or worse than the previous setup.

Step 4
The signal chain;
Source analog -> Roland [analog in]->
Roland [digital out] -> Canare SPDIF_AES transformer -> DCX -> amps -> speakers

I wanted to sabotage the front end to see if extra
conversions were audible. The neat thing about the Roland
is that I can connect the analog outputs on the source
to the Roland and also connect the digital output from
the source to the Roland and just pick which input to use
by adjusting the appropriate volume knob, this gives me
a real time ability to compare both paths within a second of
switching time. This helps preserve the human memory
of the sound to see if you really can hear a difference.
I can't, both paths are great.

Note: If both analog and digital volume knobs
are used together, you can mix both signals and get
a bizarre effect in sound. :twisted:

My next attempt to make it sound bad is to test 5 conversions.

Source [D/A] -> Roland [A/D] -> Roland [D/A] -> DCX [A/D] -> DCX [D/A] -> amps.

FYI, the loudspeaker is a 52 driver line array setup for
~104dB sensitivity {calculated, not measured} with 3kw
of amps.

[Dennis H]

Thanks thylantyr. While you're experimenting, you might try what some people claim is the optimum for maintaining digital bit depth and S/N ratio.

Source -> Canare SPDIF_AES transformer -> DCX -> analog preamp or volume control -> amps -> speakers

If the preamp is active, you might need fixed attenuators between it and the DCX so the DCX can run at full output without overdriving the preamp's input stages. If you are going to hear an improvement, I'd expect it to be at low volume where the DCX loses bits with the preamp before it.

If there is no difference, that certainly makes life simpler.

[Victor]

Isn’t it a great hobby or what?

The last time I really thought long and hard about perceived audibility of analog verses digital signal paths was when I witnessed some friends of mine singing praises to a megabuck turntable-driven system while in reality only listening to about 12 bits of resolution courtesy of the turntable S/N capability. Everyone loved the 12 bit sound coming from a $33,000 (Can.) BW Nautilus rig. Makes you wonder!

If we look at the Roland M-1000 analog input stage, we would notice a rather competent signal path with at least 17 bits capability. Its AD1895 Asynchronous Sample Rate Converter (good to about 21 bits of resolution at all frequencies of interest) leaves any and all concerns about jitter in the dust. However, the versatility of M-1000 is truly remarkable for the price it commands and the extremely high performance it offers.

The question you raised, - is the digital path any preferable to the analog path that has inherent A/D conversion? Your answer is, - No, it is not, - the analog sounds just as good. I know from experience that you are quite correct here.

In my view the M-1000 offers the intellectual satisfaction of having a fully digital and for the purposes of 16 bit data lossless signal path and that is all. It does not offer a better sound because it’s S/N numbers render the M-1000 to be completely transparent for the purposes of human hearing capabilities, i.e. it has no sound at all.

...regard,
Victor

[thylantyr]

Thanks thylantyr. While you're experimenting, you might try what some people claim is the optimum for maintaining digital bit depth and S/N ratio.

Source -> Canare SPDIF_AES transformer -> DCX -> analog preamp or volume control -> amps -> speakers

If the preamp is active, you might need fixed attenuators between it and the DCX so the DCX can run at full output without overdriving the preamp's input stages. If you are going to hear an improvement, I'd expect it to be at low volume where the DCX loses bits with the preamp before it.

If there is no difference, that certainly makes life simpler.

I have some ganged pots and plan to do this test someday
when I get unlazy. Branwell did a three tests and here is
his results.

///

Branwell posted this message;

http://groups.yahoo.com/group/DCX2496/message/415

From: Branwell <branwell1@...>
Date: Thu Jun 16, 2005 1:08 pm

Subject: Digital Volume Control for DCX based systems

Hello,

I have an active setup with the Behringer DCX 2496 crossover, and like most
home users, have had to address the volume control issue and level matching
for non pro amps.

The setups I have tried are:

1) CD ( digital out ) – DCX ( analog out ) – DACT 6 channel pot - Amps
2) CD ( digital out ) – DCX ( analog out ) – VCA based 6 channel pre –
Amps

Both of these setups dealt with the level matching and allowed good volume
control.
Of the two, I preferred the sound of the active volume control as opposed
to the passive pot. A little more dynamic and sets up a more 3 dimensional
sound stage.

To try something different, recently I got a Roland M1000 digital mixer and
am using it to act as a volume control.

http://www.rolandus.com/products/details.asp?catid=12&subcatid=49&prodid=M-1000

CD ( digital out ) – M1000 ( digital out ) – DCX – Amps

While this setup does not deal with the level matching issues ( Can add L
pad resistor setups in the interconnects between the Amp and the DCX ), it
does deal with the volume control.

Compared to the two analog volume control setups, the Digital volume setup
sounds more dynamic, more detailed and interestingly, more musical.

While this was fairly close to being apples to apples, the variables include:

Passive Volume Control:
Behringer to Pot – Single ended.
Pot to Amps – Single ended.

Active Volume Control:
Behringer to VCA - Balanced
VCA to Amps - Single ended.

Digital Volume Control:
Behringer to Amps – Balanced.

The amps I have will take single ended inputs or balanced, so there was not
an amp change.

Could the balanced factor account for the perceived sound quality increase?.
Perhaps, but either way, for those that like the idea of a Digital volume
control, the M1000 is a viable solution.

[thylantyr]

My next attempt to make it sound bad is to test 5 conversions.
Source [D/A] -> Roland [A/D] -> Roland [D/A] -> DCX [A/D] -> DCX [D/A] -> amps.

:blah: Auditioning this sabotaged method was comedy.
It sounds great. :lol:

I'm now convinced that anyone auditioning my rig will
fail a blind test; All digital path up to amps vs. 5 conversions up to amps, if you count SRC, 6. :rofl:

DCX running hot with reduce amplifier gain
I haven't done the output attenuator test yet, that is saved
for last, but one of my amplifiers has gain controls, the other not so I did a midrange only test.

I ran the midwoofers on the line array with no low pass
filter and I 'zeroed' out the DCX setting except the high
pass filter.

I turned up the Roland to near clipping and the DCX
is running 'hot', but no red LED's.

I can't say it sounds better or worse than running the
DCX and Roland 'cold', where you have higher gain on the
amplifiers, but operating the front end near zero volume
control. The array is very high in sensitivity so even a watt
generates enough SPL to hear details in music.

I don't hear more or less detail with any configuration so
far. Where is all the bad sound that is claimed all over
cyberspace.

The only way to get bad sound was to run the Roland right
at clipping where the red LED blinks, I can hear distortion.
The clipping circuit seems to be calibrated well, just below
this threshold, I hear no distortion.

The other way to make the setup sound bad is
to -> :stupidpc:

The best way to improve SQ -> :beer:

I think the real issue is noise floor and hiss, you can
optimize this for sure, but the hiss issue is really not
related to the 'SQ issue' where the sound loses resolution
by taking a long path to the amplifiers and canibalization
occurs, this is really what I'm trying to listen for.

It's obvious that lower gain amplifiers and running the
front end 'hotter' will reduce hiss, but are people
associating this hiss with SQ? I think most people do.

[thylantyr]

The Roland seems sold out now.

Is there any pro audio gear in the $500 or less range
that will do what Roland + DCX did ? All digital path
with digital input volume control.

Plan B. is the analog path. This is for a 'customer' who
wants the digital path. I'm not really a pro audio guy,
I'm not familiar with all the offerings out on the market.

I'm using my first life line - poll the audience :rofl:

[Snap]

500 is going to push it. But there is a BOAT load of products that can do what the roland does. Mackie, Yamaha, Apogee, Digi Design.

To give you the best place to look, check out digital recording stuff. Anything dealing with Cake Walk, Pro Tools, that kind of gear should have some extra stuff that would help you out. My lunch is over so I can not look it up for you.

Also did you check out Yahoo Shop? I typed in Roland M1000 and got a ton of sites. Are you sure that they are all sold out?

Check out www.sweetwater.com

They have TONS of pro audio gear at a good prices.

[Victor]

[QUOTE=thylantyr]The Roland seems sold out now.
QUOTE]

they might still have them here...

http://www.podcastingnews.com/items/Roland/ROLM1000.htm

Vic

[thylantyr]

The places that used to sell them for under $250
no longer show stock. There are some hits for the $500
prices. 8O

I will check again for stock.

Question 2: Is the Behringer SRC2496 not able to do the
same thing as Roland M1000 ? Digital volume control?
SRC cost $129. The reviews I find are great but I can't
find information on using it as the master volume in
the digital path.

[Dennis H]

The SRC2496 doesn't have a digital volume control. The gain knob adjusts the analog input.

[thylantyr]

The SRC2496 doesn't have a digital volume control. The gain knob adjusts the analog input.

;x(

Thanks!