I agree with Davey and would add that these days it is way too easy to have too many volume controls in the signal chain. You need to make sure the level out of the miniDSP is as close to max as you can and don't use it to control the volume.
I don't see how the speaker sensitivity affects this I would think the comfortable listening level to noise floor would scale.
Clearly we have a difference of opinion on what constitutes "hi-fi". What is the approximate sensitivity of your speakers that you're running miniDSP on?
If you had to listen to that noise, you'd be looking for a solution, too, and one that doesn't resort to adding resistors in series to cut noise.
Chris
Noise performance would be one part of "hi-fi." I will stipulate the noise of the 2x4 unit is somewhat higher than decent analog circuitry would create. (I already mentioned this earlier in the thread.)
However, I don't reject the unit out of hand just because of the noise performance. There are many features and capabilities to consider here.
Adding a post-miniDSP attenuator is certainly a band-aid approach....I will stipulate to that. But hey, I'm just trying to offer some options that might do the job well enough to make it workable....in this case.
Clearly, it's not a device that you're interested in. I think we have the gist of that now. 🙂
Dave.
A low-pass filter can be created by simply shunting an appropriate value capacitor across the miniDSP outputs. This will attenuate out-of-band frequencies above the audio range that might be causing an audible issue in the audio range. This might not work, but I have seen a few cases where it did.
Attenuating the outputs can be implemented with a two-resistor divider network. A 6db (one half) reduction might be about right. So, two resistors of the same value in series.....probably 5k each would be okay.)
Dave.
^ Much better to look at your gain structure from end-to-end and make sure that isn't what's really killing you noise-wise.
As usual, Scott packs about 6 posts worth of insight in a single sentence. I think you misread him.
"I don't see how the speaker sensitivity affects this I would think the comfortable listening level to noise floor would scale."
Jan
Consider it a public service announcement: a defect that miniDSP knows about and has done nothing about...for a long, long time.
What is the approximate sensitivity of the loudspeakers that you are running your miniDSP on, if I might ask?
Chris
I went and found a couple of folks elsewhere whose problems were resolved by the end of their thread by paying attention to gain structure issues. Yes you will have problems by turning the PA up full and running a low level through the DSP. Think about CD's for that matter and any DAC the digital is run at full scale.
The OP could at least look at the actual input level to his PA at normal listening levels. I don't think white noise at -86dB FS is particularly audible. I realize that 99dB/W speakers and a 100W amp without a volume control could be a problem.
I used headphones in my case, Sennheiser HD650 and I was using the miniDSP to do RIAA. There was no audible noise with needle up.
Last edited:
They're Linkwitz LXmini's, so approximately 84db. My amplifier is Av=29db with low idling noise. I have to put my ear to within about 6" of the drivers to hear any audible noise. That's acceptable...in my book.
I think much of the noise of the miniDSP 2x4 unit is inherent in the ADAU1701 DSP chip. They use a simple, passive output filter per Analog's recommendation.
A different/better layout of the circuit board 'might' yield some improved performance. But, I think that's water-under-the-bridge for them now. They're probably not going to redesign it. It is what it is. 🙂
Cheers,
Dave
I think part of the problem is folks can't abide by setting things up so the DSP can't clip their amplifier i.e. in the 100W amp and 99dB/W scenario the amp has to be turned down (I assume) to run the DSP anywhere near .9V rms.
EDIT - I did notice several of the folks having problems were trying to set everything up with no measurement instruments at all, I admit the gain structure issues can be hard to sort out.
Last edited:
No one that I know of (about 1-2 dozen cases: which is probably statistically significant...) has resolved the issue using gain restructuring. In fact, the attempts at using resistors in series have not been very successful to my knowledge. That's why I spoke up.
The bastion of knowledge also may not entirely reside here on this forum: you're assuming that you're proposing a fix that hasn't been tried elsewhere, in a rather forward manner.
The main problem is in the DSP and/or analog circuitry, as has been described above, which is too noisy for all applications that I care about. Another issue is PS noise, as it has also been described here by the OP. There is no way that I'd call this a hi-fi device. You might.
This also isn't the first time I've run into DSP noise in digital crossovers. For instance, Crown XTi DSP amplifiers also exhibit irritating levels of DSP noise, among others.
It would be constructive to talk about something that works with higher efficiency loudspeakers without resorting to resistors or added output filters to cut the inherent DSP noise? There are many examples of digital loudspeaker controllers (crossovers) that have extremely low noise-- out of the box.
By the way, any of you guys own stock in, or work for miniDSP? Perhaps we could get the word to them to upgrade their hardware.
Chris
You could have searched the forum, as I said your DAC's run at full scale not 10bits off of it. ANY DSP board that does not have some adjustment for the user to set the gain structure has the same issue. The AD and DAC need to run within a couple of dB of full scale. If your amp wants 25mV FS at your listening level no DAC is going to be quiet straight in.
Last edited:
Hmm, interesting quote there, Scott.
I think my HK3490 amplifier attenuates the signal before sending it to the MiniDSP. There is no way for me to go around that if I still want to use the DAC in the HK. However, I do have an O2+ODAC headphone amplifier/DAC combo. Maybe I could connect my computer to the O2+ODAC, the MiniDSP to the O2 headphone amplifier via a 3.5mm to 2xRCA and then connect from the MiniDSP out to a normal input on the HK amp?
But how should I leave the volume on the O2 amplifier? I have the gain set to 1x (no gain resistors installed) so should I turn the volume all the way up or should I leave it somewhere in between?
I think my HK3490 amplifier attenuates the signal before sending it to the MiniDSP. There is no way for me to go around that if I still want to use the DAC in the HK. However, I do have an O2+ODAC headphone amplifier/DAC combo. Maybe I could connect my computer to the O2+ODAC, the MiniDSP to the O2 headphone amplifier via a 3.5mm to 2xRCA and then connect from the MiniDSP out to a normal input on the HK amp?
But how should I leave the volume on the O2 amplifier? I have the gain set to 1x (no gain resistors installed) so should I turn the volume all the way up or should I leave it somewhere in between?
Sorry, not sure but experimentation is easy. Just try a setup where both the input and output of the DSP are a couple of dB below full scale on peaks. That quote is from an earlier thread here on the same issue in 2012.
I don't think that will improve the situation since your setup is still the same on the output side of the miniDSP unit. However, it would isolate/eliminate any noise coming from your front-end gear. But you essentially already did that with the first testing configuration you mentioned.
If you can easily separate your ODAC from the O2 you could install the O2 between your miniDSP outputs and the amplifier inputs. You then have an adjustable attenuator/driver. I suspect that would do the job but you can't dial in too much attenuation. Every db of attenuation you apply post-miniDSP forces the internal signal higher by that same amount. That's a good thing.....until you reach 0dbFS within the DSP.
Dave.
Hmm... Seems like I was thinking backwards then. But yes, I could install an O2 after the MiniDSP. I am not sure that I have enough cables for it since I need a 2xRCA-3.5mm between the MiniDSP and the O2 and then another one of those from the O2 to the HK amp, but I will give it a try.
The objective is to attenuate the noise. But you will also attenuate the signal as a result. Attenuating the signal is not necessarily a bad thing...in this case.
Give that a try and report back.
Cheers,
Dave.
This noise issue has been a bit of an ongoing behavior for some users for years, along with the absense of power on-off muting. These obvious problems still persist today and need to be addressed in a revision.
Here is how I would configure my own system with a DSP like this:
The input may not be the best place for attenuation (other than lowering the incoming signal to prevent clipping), since this would reduce the available resolution and what SNR is available. The analog output may not be the best place to permanently attenuated if the user desires full amplifier drive capability. A suitable work around may be to place the preamp, or volume attenuator, after the DSP. This would reduce idle noise and scale it according to playback level. It's not a cure, though.
Here is how I would configure my own system with a DSP like this:
The input may not be the best place for attenuation (other than lowering the incoming signal to prevent clipping), since this would reduce the available resolution and what SNR is available. The analog output may not be the best place to permanently attenuated if the user desires full amplifier drive capability. A suitable work around may be to place the preamp, or volume attenuator, after the DSP. This would reduce idle noise and scale it according to playback level. It's not a cure, though.
Well, you edited your post, but I think you might be confused by my previous post (#14) regarding the noise level.
Here's another measurement of an output of my miniDSP 2x4 unbalanced unit with no programming (green) and with my soundcard input shorted (yellow.)
ARTA in this mode displays voltages relative to one volt RMS in a 1Hz bandwidth.
This is pretty darn good noise performance.....relatively speaking. This 2x4 unit connected to my 29db gain power amplifier attached to an average sensitivity speaker driver yields audible noise only if my ear is within a few inches of the driver. I have some DSP gadgets from other manufacturers on my shelf that exhibit worse performance than this.
Cheers,
Dave.
Here's another measurement of an output of my miniDSP 2x4 unbalanced unit with no programming (green) and with my soundcard input shorted (yellow.)
ARTA in this mode displays voltages relative to one volt RMS in a 1Hz bandwidth.
This is pretty darn good noise performance.....relatively speaking. This 2x4 unit connected to my 29db gain power amplifier attached to an average sensitivity speaker driver yields audible noise only if my ear is within a few inches of the driver. I have some DSP gadgets from other manufacturers on my shelf that exhibit worse performance than this.
Cheers,
Dave.
Attachments
I would disagree on the noise there is nothing to address as Davey shows. Consider a CD player with DAC at line level out and a PA with no volume control what do folks do? What I do is use a Goldpoint passive attenuator at the output and of course the amp now may not be able to be driven to full power. I don't edit the music files to use fewer lsb's so I can run the amp full on. There are plenty of CD players and pre-amps with ~90dB SNR's and people do enjoy LP's on their high sensitivity speakers it's all in the gain structure.
EDIT - Davey could you clarify a previous comment, what is the SNR 20-20kHz. I got the impression there was some out of band noise included in the number quoted.
Last edited:
- Status
- Not open for further replies.