The MiniDSP cards use an ADAU1701 SigmaDSP chip, analog specs are on page 5. No issues with processing precision, however the onboard A/D/A converters aren't really high end.
http://www.analog.com/media/en/technical-documentation/data-sheets/ADAU1701.pdf
Temperature and build are closely related, I see no reason to ignor the effects.
Well, I mean by using the simple modelling options of components and existing models. The more precise you need the simulations to be, the more effort needs to be put into modelling of a device, but still there is a limit. It really depends how much effort you feel is worth the precision you are looking for. Normally it is a tradeoff of risk in determining the invested effort. For audio, the risk is really low.
I wish I could share with you the models we use in my day job. They are not simple. They also provide incredibly good match between simulation and reality. This level of precision makes modern circuit design possible.
Those who believe computer simulation is no better than a guess are really showing their lack of experience with simulation (and perhaps lack of experience in general).
Tom
No they're not, actually. A temperature increase is the result of the dissipation of power. An LM3886 amp running on ±28 V supplies delivering Pout into Zload will dissipate X amount of power regardless of the build. The absolute temperature depends on the heat sink and airflow more than anything else.
Tom
If the DSP handles the volume control, you're probably right. If the DSP lets the DAC handle the volume control, I'd think you'd get pretty darn close to full precision even at low volume. Whether that's MiniDSP's implementation is unknown. They're rather tight-lipped about their implementation.
Tom
Tom, I'm not sure what you mean by letting the DAC handle the volume control. If you're talking about the digital volume control built into many DAC chips, then you're still losing resolution as you turn down the volume. If you mean an analog (stepped or continuous) attenuator after the DAC, that is different. However, if you're using your miniDSP as the crossover in a multi-amped system, you then have to make multiple analog volume controls track each other.
Not wanting to go too off topic, but since Mod-86 is perfect (IMO) for active conversions I guess we are ok. It all depends whether you think that a 24 bit signal reduced down to 16 bits is audible. If you need more than 8 bits of attenuation you possibly have the gain setup wrong. But requires you put the gain at the end of the chain. For me I find that I only use about 15dB of the gain range for my listening but I have no idea if this is normal or unusual!
Thanks much for the reply. I was thinking the same thing, that while the miniDSP 2x4 isn't terrible, it's noise is not nearly as low as the amazing Modulus-86. I've been trying to think of alternatives to achieve high quality without spending thousands of dollars. Finding a way to take advantage of Modulus-86 still seems like a good path though.
Kind Regards
Given the DSPs available with high-precision math (32- or 64-bit) there's really no technical reason a DSP filter can't outperform an analog filter. However, as with any precision design, one has to pay attention to the architecture and design. I'm not familiar with the architecture or code base of the MiniDSP, so I can't say specifically if has the capability to outperform analog filters. Above plots make me wonder, though...
Even analog filters have their limitations. You really have to pay attention during the design phase to ensure that you get the stop band attenuation you need. I ran into that during the design phase of the DC servo in Modulus-86 Rev. 2.0 (there! back on topic...) I suggest reading Bonnie Baker's note on the topic: http://www.ti.com/lit/an/slyt306/slyt306.pdf
Tom
I believe that's right. I think the noise that I hear people complain about with the miniDSP 2x4 is primarily in the analog output. Proper gain staging of the whole system is key.
At any rate, I'm hoping that a miniDSP NanoDigi 2x8 B (all digital with coaxial SPDI/F in and out, would be a way to get around that. I feel that sound quality would then be more dependent on which DAC is used after the DSP, which brings me to my question.
I spent a weekend reading through this entire thread, and your website. How does one control Volume on your kit. I could see simply using a potentiometer on the input, but from what I understand, it's balanced, not single-ended. I'm not sure where to go from there.
Another question if I may. It seems like a decent DAC should output 1.5 or a little over 2V. Is that enough to drive Modulus-86 without a preamp to add gain? My speakers will be around 95dB 1m/1w by the way. I ask because I've heard of balanced sources outputting 4V, are they counting both phases to get that number? I'll be single-ended for now, and leave that as an upgrade path.
Thanks for your time,
AlexQS
Limitation there is the A/D/A converters used in the SigmaDSP chip that's the core of the MiniDSP card.
You could add a MiniDIGI card to the MiniDSP, use a SPDIF signal source and buy an external DAC. This might cost a fair bit, but probably not a thousand dollars.
Alternatively, you've got me thinking... If anyone's interested, I could bang out a high end I/O card for the MiniDSP that has high end DAC/ADC parts, digital input, a clean local clock, differential I/O, etc. (If this is something you might be interested in, PM me and I'll go from there, don't pollute this thread)
gmarsh,
What about the nanoDIGI solution AlexQS has mentioned:
nanoDIGI 2x8 K | MiniDSP
To me also looks this like a better solution vs MiniDSP, as the DAC is coming into play after the DSP... more expensive though as more DACs will be needed.
What about the nanoDIGI solution AlexQS has mentioned:
nanoDIGI 2x8 K | MiniDSP
To me also looks this like a better solution vs MiniDSP, as the DAC is coming into play after the DSP... more expensive though as more DACs will be needed.
So you are saying that heatsink selection is not related to built? It would affect time to stable temperature and temperature variation during operation.
Yes, the more investment you spend on modelling, the more accurate you can get. But you cannot get this level with models available to the general public. Even chip makers have to tune their modelling to more up-to-date technology to reduce risk. I do not believe that anyone is doing this level of modelling in audio.
There are different limitation in computer simulation depending one what area of simulation you get into. How much you spend on modelling and research also depends on how much customers are willing to pay to get the quality. You get into investment and return considerations as well when you start amortising the cost.
Please re-read my previous post. You'll find the answer there. You'll note that we agree on this.
Tom