Reasonably priced good quality 6 or 8 channel USB DAC?

[JustIntonation]

Yes thinking about it / balancing things it seems the MOTU DACs aren't the perfect match for my intended setup..
The Anaview AMS0100 has the following dynamic range (at a gain of 20.6):
Dynamic range 4Ω 15Vrms/idle noise 115 dB
The Anaview AMS1000 has this dynamic range (at a gain of 27.5):
Dynamic range 4Ω 36Vrms/idle noise 115 dB

It makes sense to me that I don't want the DAC to have a noisefloor above the noisefloor of the amps. I can allready guess this will be the case with an unattenuated MOTU, but I'll do the calculations now based on the calculator links you gave.

 

[JustIntonation]

There are tons of good articles here for that sort of DIY. An opamp buffer shouldn't make too much of a difference from a measurement point of view.

A few good articles to start with:

http://sound.whsites.net/dwopa.htm
http://sound.whsites.net/articles/balanced-2.htm

Thanks for that. But I've never done such a thing and to build a 6 channel one.. mm I'd rather look at other options first..

 

[Blumlein 88]

Yes thinking about it / balancing things it seems the MOTU DACs aren't the perfect match for my intended setup..
The Anaview AMS0100 has the following dynamic range (at a gain of 20.6):
Dynamic range 4Ω 15Vrms/idle noise 115 dB
The Anaview AMS1000 has this dynamic range (at a gain of 27.5):
Dynamic range 4Ω 36Vrms/idle noise 115 dB

It makes sense to me that I don't want the DAC to have a noisefloor above the noisefloor of the amps. I can allready guess this will be the case with an unattenuated MOTU, but I'll do the calculations now based on the calculator links you gave.

Notice they are comparing max signal with idle noise. With most power amps, if you could run at high power and notch that frequency out, you'll find the noise floor has risen some. Some DACs are that way though most aren't very much.

Also all circuits with gain, to my knowledge add some noise in the process of supplying gain. In microphone pres they talk about EIN, equivalent input noise. The best of those add 3-4 db onto whatever noise floor is input to them. What that means is suppose you had an amp with 120 db dynamic range. You feed it an input source with 120 db dynamic range. The process of adding gain to that input would probably add 4 db of noise and the result would be 116 db dynamic range at best. And you can't get around that.

I'm not trying to dissuade you from getting as good a result as possible, and gain staging properly to get the quietest widest range result is wise. But don't let perfection become the enemy of the good. Getting better results in multi-channel DACs than the Motu or similar will get rather expensive. Such devices are still improving. Something like the Motu now and something similarly priced in 5 years (which from recent history indicates it will have quieter circuits) will get you there. And I'm not so sure other than pathological signal conditions that anything past 100 db is normally relevant to music listening anyway.

 

[maverickronin]

Thanks for that. But I've never done such a thing and to build a 6 channel one.. mm I'd rather look at other options first..

It is pretty intimidating if you've never done it before.

On the plus side though, you'd only need to build 4. You can go straight from the DAC to the amp input for whichever set of drivers don't need to be padded down.

 

[March Audio]

Notice they are comparing max signal with idle noise. With most power amps, if you could run at high power and notch that frequency out, you'll find the noise floor has risen some. Some DACs are that way though most aren't very much.

Also all circuits with gain, to my knowledge add some noise in the process of supplying gain. In microphone pres they talk about EIN, equivalent input noise. The best of those add 3-4 db onto whatever noise floor is input to them. What that means is suppose you had an amp with 120 db dynamic range. You feed it an input source with 120 db dynamic range. The process of adding gain to that input would probably add 4 db of noise and the result would be 116 db dynamic range at best. And you can't get around that.

I'm not trying to dissuade you from getting as good a result as possible, and gain staging properly to get the quietest widest range result is wise. But don't let perfection become the enemy of the good. Getting better results in multi-channel DACs than the Motu or similar will get rather expensive. Such devices are still improving. Something like the Motu now and something similarly priced in 5 years (which from recent history indicates it will have quieter circuits) will get you there. And I'm not so sure other than pathological signal conditions that anything past 100 db is normally relevant to music listening anyway.

Indeed.

JustinIntonation, I already do what you are attempting. dac noise is not audible, although my drivers are a bit less sensitive.

 

[JustIntonation]

Indeed.

JustinIntonation, I already do what you are attempting. dac noise is not audible, although my drivers are a bit less sensitive.

I'll be listening at 1.3m distance (4 to 4.5 feet) in a completely quiet room with the tweeter at 97dB 4ohm efficiency pointed directly at my ears (in fact it is about 98dB efficiency at the most sensitive part of our hearing 3-4kHz) I have to calculate things.. Other speakers have been too noisy at this distance in even less quiet conditions.

 

[JustIntonation]

Notice they are comparing max signal with idle noise. With most power amps, if you could run at high power and notch that frequency out, you'll find the noise floor has risen some. Some DACs are that way though most aren't very much.

Also all circuits with gain, to my knowledge add some noise in the process of supplying gain. In microphone pres they talk about EIN, equivalent input noise. The best of those add 3-4 db onto whatever noise floor is input to them. What that means is suppose you had an amp with 120 db dynamic range. You feed it an input source with 120 db dynamic range. The process of adding gain to that input would probably add 4 db of noise and the result would be 116 db dynamic range at best. And you can't get around that.

I'm not trying to dissuade you from getting as good a result as possible, and gain staging properly to get the quietest widest range result is wise. But don't let perfection become the enemy of the good. Getting better results in multi-channel DACs than the Motu or similar will get rather expensive. Such devices are still improving. Something like the Motu now and something similarly priced in 5 years (which from recent history indicates it will have quieter circuits) will get you there. And I'm not so sure other than pathological signal conditions that anything past 100 db is normally relevant to music listening anyway.

Wise words and agreed
Yes I saw that, comparing with idle noise. Not true dynamic range ok. Thanks for explaining it in more depth than I knew.
Just trying to balance things for the right decision now. If I do it correct now I'll never have to worry about it again Little work for many years of pleasure.
I'm not going to spend more now than what I have described. But perhaps I can still come to the conclusion that an even less costly DAC will serve me just as well or even slightly better. Or that perhaps 2 Hypex Fusion amps FA253 which will also cost slightly less will do the job slightly better in the end.
Will calculate everything in detail and compare later tonight or tomorrow.

 

[JustIntonation]

It is pretty intimidating if you've never done it before.

On the plus side though, you'd only need to build 4. You can go straight from the DAC to the amp input for whichever set of drivers don't need to be padded down.

Ah but I've never built a multi-driver speaker either.. I just have to like learning about it and it'll work out ok. Just don't know if I like going down that route yet
Btw, it seems to me I'll have to build 6 as +20dBU doesn't work for any of the drivers / amps. They all need to come down to about +4dBU (for mid) perhaps ideally less for the tweeter but this doesn't matter much it seems to me. A 6 channel +4dBU output dac would be a good match without any attenuation it seems. (perhaps I should look into partially building a DAC, that seems like a lot of fun But far harder than active attenuation circuit as wel lol)

 

[March Audio]

I'll be listening at 1.3m distance (4 to 4.5 feet) in a completely quiet room with the tweeter at 97dB 4ohm efficiency pointed directly at my ears (in fact it is about 98dB efficiency at the most sensitive part of our hearing 3-4kHz) I have to calculate things.. Other speakers have been too noisy at this distance in even less quiet conditions.

Ok, fair enough. In that case I would suggest your best solution is still a multichannel professional dac with a simple resistor attenuator (just requires high tolerance resistors to maintain CMR) but change to say the Hypex amps that don't have the unusual input impedance of the Anaview. Or check out the noise level on the Fusion amps and do the XO with that which will be a much simpler solution.

 

[March Audio]

Hypex nc122

Fa122 - appears the dac is not as goodas say Motu.

Going full shill here, the bare Hypex amp modules are OEM only and not for diy, however I am about to release a range of amps utilising them.

 

[maverickronin]

Ah but I've never built a multi-driver speaker either.. I just have to like learning about it and it'll work out ok. Just don't know if I like going down that route yet
Btw, it seems to me I'll have to build 6 as +20dBU doesn't work for any of the drivers / amps. They all need to come down to about +4dBU (for mid) perhaps ideally less for the tweeter but this doesn't matter much it seems to me. A 6 channel +4dBU output dac would be a good match without any attenuation it seems. (perhaps I should look into partially building a DAC, that seems like a lot of fun But far harder than active attenuation circuit as wel lol)

I saw the Okto DAC mentioned in another thread. Don't really know much about it other than that it exists, but it's a bare board with 8 balanced outs.

http://www.oktoresearch.com/
https://www.diyaudio.com/forums/ven...ormance-8-channel-module-sabre-es9028pro.html

You'd loose the analog inputs from the pro audio interface though. Not sure if your PC is your only source or not. If it isn't, it's best to have all the ins and outs running though the same device, or at least the same driver stack, for lower latency and stability.

Besides Hypex and Anaview you could also look into IcePower amp modules.

 

[pos]

My hypex amps have 26dB gain and work fine with my 8A and ultralite. No additional attenuation required and software volume control.

Just, what's the sensitivity of your drivers and the input impedance of the anaview?

2216nd woofer, with a sensitivity exceeding 100dB/2.82V/1m around 2kHz.

 

[pos]

The okto DAC8 might be a good alternative solution, with a much more reasonable 3.4Vrms output, and very good noise specifications.
I hope Amir will be able to measure it soon.
The motu 24Ao I tried was a no go for me. I hope the 8A and ultralite are different...

 

[March Audio]

Okto looks very intetesting

 

[JustIntonation]

I saw the Okto DAC mentioned in another thread. Don't really know much about it other than that it exists, but it's a bare board with 8 balanced outs.

http://www.oktoresearch.com/
https://www.diyaudio.com/forums/ven...ormance-8-channel-module-sabre-es9028pro.html

You'd loose the analog inputs from the pro audio interface though. Not sure if your PC is your only source or not. If it isn't, it's best to have all the ins and outs running though the same device, or at least the same driver stack, for lower latency and stability.

Besides Hypex and Anaview you could also look into IcePower amp modules.

Well doesn't seem like it can get more perfect than the Okto DAC
Very good that you mention it! I did a big search for multi-channel DACs and completely missed this one somehow..
The PC is my only source so no problem there. Only have to figure out how to get 6 channel audio in i2s into the Okto DAC from my computer but there are good options there it seems to me. (perhaps I can even consider an external DSP instead of using my computer..)
So thanks! This one is going to be it!

As for the amp. I've given it a good thought and I'm going to stick with the Anaview AMS amps. I know their sound and love it. They approach neutral transparent sound from the good sounding / musical side. I've seen them compared in ways to a very good tube amp and strangely enough I can understand that, somewhere in the distance it reminds me of a good Manley tube amp I once had (in fact the only good tube amp I ever heard, not into tubes otherwise at all). The sound is smooth and big and transparent and in no way tends to do "harsh" by itself. Also very "involving". But I've never heard Hypex amps, i've seen the NC400 described as sterile, bright and analytical compared to the Anaview AMS1000 by some (including pros). At least people who heard both seem to agree they sound quite different, some find the Hypex NC400 to be even more transparent though. But since I'll already have a very analytical setup/room I don't want to push it potentially too far with an amp I haven't personally heard. And where the NC400 measures slightly better than the AMS1000 / AMS0100, the NC252 / NC122 modules actually measure slightly worse than the AMS0100 amp below a few Watt of output. And when it comes to phase there are no measurements for the Hypex amps while there are extremely good phase measurements for the AMS amps and there's reason to believe the Hypex amps have a much less flatter phase than the AMS amps as most class-D amps do and AMS has a seemingly unique thing going here in its design (could this explain part of the difference in sound?). All in all without being able to directly compare them both the safest bet (and best measuring one) for me are the Anaview AMS amps. Dont know much about the Icepower amps though, but untill recently the comparisons I read stated clearly that they were not yet in the same leage as the Hypex and Anaview AMS amps. Maybe a very new model could have changed this I don't know but not taking a gamble here.

 

[Deleted member 65]

Only have to figure out how to get 6 channel audio in i2s into the Okto DAC from my computer

I've got this, USB in & various outputs incl I2S.

 

[Deleted member 65]

With configurable I2S ..

 

[pos]

The PC is my only source so no problem there. Only have to figure out how to get 6 channel audio in i2s into the Okto DAC from my computer but there are good options there it seems to me. (perhaps I can even consider an external DSP instead of using my computer..)

The miniDSP USBstreamer is supposed to mate with the DAC8. In the diyaudio.com thread they report using it for testing.
This is the configuration they should send to Amir for testing.

 

[andreasmaaan]

I mean, if one of those DACs is off sync by just one sample at 44.1kHz, and this is between for instance the DAC for the mid-driver and the DAC for the tweeter, then the resulting phase shift between those drivers is audible already at certain off-axis locations (around the -3dB point off-axis for LR for instance where it is very sensitive, or at the flat on-axis for a Butterworth crossover).

Hi @JustIntonation, sorry to pull you back to a post from a long way back to something slightly off-topic in terms of your decision about which DAC to buy, but I just wanted to get a better idea what you were talking about with this statement.

I understand your calculation, i.e. I agree that a one-sample timing discrepancy between the two DACs would have such an effect.

But what I'm not sure about is the idea that the DACs could be off by as much as one sample.

In general, the worst performing DACs I know off exhibit jitter in the range of about 10 ns. I might be wrong to make that assumption, but for example in the classic jitter study by Benjamin and Gannon, the lowest quality HDTV receiver they measured exhibited jitter of around 10 ns, while other cheap devices that they measured exhibited maximum jitter of less than 1ns (often significantly less).

An error of one whole sample at 44.1 KHz amounts to a timing error of >20,000 ns.

Is it plausible that such a large timing error between two reasonably high-quality DACs could occur?

Being no expert in digital audio, I'm actually not sure myself. It does sound very unlikely to me, however.

I'm wondering what others here think?

 

[JustIntonation]

Hi @JustIntonation, sorry to pull you back to a post from a long way back to something slightly off-topic in terms of your decision about which DAC to buy, but I just wanted to get a better idea what you were talking about with this statement.

I understand your calculation, i.e. I agree that a one-sample timing discrepancy between the two DACs would have such an effect.

But what I'm not sure about is the idea that the DACs could be off by as much as one sample.

In general, the worst performing DACs I know off exhibit jitter in the range of about 10 ns. I might be wrong to make that assumption, but for example in the classic jitter study by Benjamin and Gannon, the lowest quality HDTV receiver they measured exhibited jitter of around 10 ns, while other cheap devices that they measured exhibited maximum jitter of less than 1ns (often significantly less).

An error of one whole sample at 44.1 KHz amounts to a timing error of >20,000 ns.

Is it plausible that such a large timing error between two reasonably high-quality DACs could occur?

Being no expert in digital audio, I'm actually not sure myself. It does sound very unlikely to me, however.

I'm wondering what others here think?

I can tell you my reasoning as to a DAC being off by 1 sample.
First of all, I don't think slow clock "drift" is usually counted as jitter. But if we look at for instance the many measurements by Amir we see that the frequency of a 1kHz sine output by a DAC is almost never exactly 1kHz, it is often off by a bit. So a clock can have 0 measured jitter but still drift slowly over time relative to another clock as their frequencies don't match exactly. As I understand it, it differs per implementation how far DAC clocks can drift apart when fed the same signal, for instance spdif.
If we take for instance a 44.1kHz spdif signal and 2 DACs which have an asynchronous sample rate converter ASRC in front of them (like two Hypex Fusion plate amps) running at 93kHz (as it is in the Hypex Fusion plate amps) and the DACs also running at this frequency. Then we have the situation that at any one exact moment in time if you look at the value the ASRC of plate amp 1 has of the incoming spdif data it could have one value while the other plate amp already has already sampled the next value. This goes both ways so the max error (not counting many other effects of ASRC) would be twice the resample rate of the ASRCs (as their clocks are not synced). These are my own ideas btw following my simplistic logic of limited understanding of how these things work. But I personally wouldn't trust using 2 plate amps where one is driving the tweeters and another the mids. As for left-right use of these plate amps this seems less sensitive to me, but still with earlier Hypex plate amps there were several people who commented that the center image was doing weird things (don't know what and if those were doing something differently though).
And then there are DACs who sync to the incoming clock / extract it from the SPIF for instance, I don't know what can go wrong with those and which DACs still use this method.
What my common sense does tell me is that things can go wrong enough here to matter it seems me. And that the ideal solution if I don't know exactly how things are done with detail and certainty that a single multichannel DAC will not have these problems, and a single device with multiple DACs all running from the same clock will more likely be OK as well compared to using several stereo DACs for multichannel audio.
This is my thinking at least, hope someone can chime in with more knowledge on this.

edit: btw, in the case of the Fusion amp ASRC example, those max errors will be reached all the time. But the 93kHz is chosen as it doesn't sync with 44.1 or 48kHz or their multiples, so the max error is there just one sample, next one has another error, etc. The real world effect of this goes beyond my understanding though..