[Davey]

Dave - is the result the same with the analog inputs muted?

Yep.  There doesn't seem to be any combination of plug-in settings....low-pass filtering, muting, attenuation, etc, etc, that changes the output noise.  It appears to be strictly hardware related.

I haven't integrated a miniDIGI yet so can't judge the contribution of the ADC stage to the overall situation.

I returned Paul's board to him but I still have another one on my bench.  Anybody have other specific testing you'd like me to attempt?

Cheers,

Dave.

[pelliott321]

I heard back from miniDSP
They still insist that there is nothing wrong with the design or my board and that this noise we are seeing is out of band noise and cannot be heard.  I just do not buy it.
I still think this is a flawed product and should not be accepted by anyone that cares about quality sound

[JohnR]

They still insist that ... this noise we are seeing is out of band noise and cannot be heard.

Well... this is correct, so...

I still think this is a flawed product and should not be accepted by anyone that cares about quality sound

Um... I think you may have lost perspective here. We're talking about a $300 board, which can be used to deliver a lot of mileage on the cheap, if you put a bit of effort into it. I see people around here spending that much on a USB cable for goodness sake. Anyway, good luck with whatever you do end up doing.

[JohnR]

Yep.  There doesn't seem to be any combination of plug-in settings....low-pass filtering, muting, attenuation, etc, etc, that changes the output noise.  It appears to be strictly hardware related.

Thanks Dave. Later I realized that the ADC question wasn't too bright as of course the 100kHz noise won't get passed directly through the ADC.

Anybody have other specific testing you'd like me to attempt?

I don't know how much you'd like to do, and I feel like we're about to get to the point where you should tell me to invest in some equipment and do it myself   However, any chance you could stick the probe on the supply rails (specifically the +/-)?

[JohnR]

Anyone know if schematics are available?

Hi Mike, I rather doubt it, although I guess you could always ask for that section. Perhaps they'd be willing on the basis of encouraging "mods" in the community.

[jtwrace]

I see people around here spending that much on a USB cable for goodness sake.

 

[Davey]

I guess it brings the question whether any DSP gadget like this (even one that works well) is acceptable in a "quality" sound system.  My own feeling is probably not, but I realize I might be in the minority on that. 

I find myself in agreement with Paul that this type of performance is not acceptable.  I don't really think the monetary aspect or the relative value of the product factors in.

As an example, consider the situation a user with a wide bandwidth power amp might find themselves in.  100khz noise of approximately 200mV RMS amplified by a power amp gain of 20 means your tweeter would see 4 volts RMS worth of noise......all the time.  It might not be that much of an issue since tweeter inductance might be rather high at 100khz....but even still. 

The unit supports a 96khz sampling rate so I don't see the logic in not bandwidth limiting the analog stages to 48khz.  The CS42528 DAC data sheet even shows a sample output filter DAC output filter/buffer for this purpose.  I see a variety of op-amps and associated cap/resistors in front of the unbalanced outputs so there is some circuitry.....but without a schematic I'm uncertain of the exact implementation.

Cheers,

Dave.

[JohnR]

From your scope shots, it's 200 mV peak-to-peak, right? (not RMS).

It looks like the output stages have filtering in them - hard to be sure without a schematic though, as you say. I was assuming the noise is coming from the output stage itself rather than the DAC. (Hence my question about the +/- rails )

[Davey]

No, it's 200mV per division.  (Legends at the bottom.)

Yes, I suppose there's signal filtering in the output sections....but if the noise is contaminating the output section via the power supply rails and/or another path then the filtering doesn't have the desired effect. 

I did some crude outboard filtering this evening and this cleaned up the noise quite a bit.....but that is not a solution. 

It is what it is......and the miniDSP folks have declared this normal operation.  There's not much point in discussing further......unless modding is on the agenda. 

Cheers,

Dave.

[JohnR]

I think modding would be an interesting topic... although it's a fair way down the list for me. I was hoping you'd be interested in figuring out the cause, but no worries, thanks for the measurements

Just for clarity though - your scale is 200 mV per division - the sine wave is a bit under 3 divisions peak to peak, and the noise riding on the sine wave is less than a division peak to peak. No?

[pelliott321]

I think the measurments were only important to prove what I felt what this unit sounds like.  At first I got sucked in on the wow factor and the neat software set up, which was a great design.  But after listening for a while I was getting bored with the sound.  It was not anywhere as compelling as what I was use to on my maggys.  Thats what started me questioning and looking into whats going on.  I did my measurements and found the noise.  There is a lot of good things about this board but great sound from a high res system is not one of them. 
The board in now on the junk pile.  If I ever see a discussion about this board on a forum I will add my 2 cents worth.   I am dissatisfied with this board and I feel miniDSP should refund my money, which they will not.  Its buyer beware
I goes back to the old saying "If it is too good to be true; its probably not"

[Tyson]

I just got my board, I don't have any measurement capabilities, but once I got the hiss from my tweeter dealt with (used an isomax to lower it, same as I have to do w/the DCX249), I can say that the digital input on the 2x8 sounds better than the digital input on the DCX. 

Once I figure out how to lower the gain on the analog input I'll be able to compare how it sounds to the DCX with analog in.  If it's better, then it's well worth the $$, IMO.  Plus it seems to be a better platform to do mods with than the DCX. 

I guess for me, the question is whether the 2x8 is better than the stock DCX since they are similar in price.  And I really like the DCX - it has some weaknesses, but to me the advantages of being fully active with speakers outweighs them.  If the 2x8 can get me a better experience and better sound for the same $$, then it's a good buy.

Too bad you are kicking it to the curb.  I am curious - when you set up the maggies with the miniDSP, what crossover points and slopes did you use?  I'm asking because I've had consistently bad experience with steep crossovers.  Anything over 6db sounds sterile and lifeless.  12db is not tooooo bad, but 18, 24, and 48 db slopes are just terrible in my experience, with multiple platforms now (both the DCX and miniDSP exhibit this life-sucking behavior with steep slopes - I think it's the slopes, not the equipment, since they both sound great with 6db or even 12db slopes).  Anyway, my $.02

[Davey]

Tyson,

Just for grins, what you should do is configure your DCX, or miniDSP, or whatever to an appropriate (constant voltage) crossover with greater than 6db slopes and then sum the two outputs together and listen to the result via headphones or some other full-range system.

This allows to evaluate the phase shift of a particular crossover alignment....and ONLY the phase shift....without all the other variables possibly confusing the issue.

Also, if possible, you should rig this test so you can do it blind and compare to a reference (non-phase-shifted) signal.

Cheers,

Dave.

[Tyson]

Sure, I can give it a shot.  But I think one of the reasons 1st order sounds good is because of the greater driver overlap, and that would be negated in a single driver (headphone) setup.

[Davey]

Driver "overlap" is one of the variables I was talking about possibly confusing the (phase) issue.

My point is that phase performance is just one of the factors involved in determining the audibility of a particular crossover alignment.  When you say "sterile" and "lifeless" are subjective evaluations regarding steep slope crossovers, you just can't automatically attribute all of that to phase performance.

Do you follow?

Cheers,

Dave.

[Tyson]

Well, I did have a DEQX for a while and it did phase perfect steep crossovers, and ultimately I sold it and went back to a DCX.  It's "phase perfect" steep slopes sounded better (to me) than the DCX's non-phase-perfect steep slopes, but in the end I prefer the sound of the shallower slopes.

What would be cool is if miniDSP came out with the ability to do FIR filters on the 2x8 and you could switch between them in real time.  They did mention something about this being a possibility in the future.

[pelliott321]

well I am not really "kicking it to the curb" just yet, but that project is on hold for the time being, too many others I need to move on.
I did try many different slopes.  The last one I tried (and listened to the most) was very steep on the sub but I copied the the slopes on the passive I am using (12db/oct).

I just think this was a rush to market item that needs some design work.
 

[wushuliu]

This is not unlike the noise spectrum I've seen from various other wide-band gadgets on my test bench.  In fact, I was looking at the output noise from my E-MU Tracker-Pre just the other day and it's very similar.

I guess it just highlights the difficulty in designing these sorts of gadgets at a price point with the almost ubiquitous use of switcher supplies nowadays.

It's difficult to judge how any of this translates to subjective audio performance because we don't know the characteristics of the downstream amplifier, the signal levels that will be used in the 2x8, etc, etc.  However, it sure doesn't look good to an analog design guy like me. 

Cheers,

Dave.

So is the noise present regardless of power supply used?

[JohnR]

I expect it would be, the board has an onboard switching regulator to generate the negative rail for the opamps, I forget the number but I believe I posted it earlier in the thread.

[pelliott321]

I havent had time to revisit this but I have a few ideas.  The small caps across the out puts make since and certainly will do no harm. But I would try shielding the chips by cutting small copper squares with ground wires on all the chips.  This is an old mod done on dacs.  Its too bad that the audio section cannot be moved away but maybe but something between the audio and digital sections to shield