-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Topping D10s noise via Mac mini
- Thread starter Thomas_A
- Start date
That HDMI cable could be causing the problem. There have been numerous measurements in Amirs reviews where HDMI connections introduce noise into his components even when they aren't turned on. I am assuming your Macbook is not hooked up to the HDMI cable? Try measuring you Mac mini with the HDMI cable unplugged and see if there is any difference.
OP
- Thread Starter
- #44
Will do that.
So the implication then is there is some other source of noise creating ground difference between the two macs (or dac and adc), such as the HDMI to AVR mentioned above. Bear in mind you are talking tiny amounts of noise - of the order of 20uV (20 millionths of a volt). It is really difficult to eliminate noise in a system down at this level.
Remove ALL connections to anything not needed for the measurement. For example, can you remove any monitors from the mac mini and remote into it via wifi to control it?
After that, I would try (to start) optical connection from mac mini to d10 s with the dac being separately powered. You may need to get a cheap usb to toslink device.
What length analogue interconnect to you have between DAC and ADC - if you can, shorten this down to the absolute bare minimum (2 or 3 cm). If the spacing between left and right is the same on both devices you could even use couplers such as these.
Also worth trying to bond the ground of the dac and adc together using a very low impedance connection, such as a braided earth bond:
Also worth trying grounding both macs to mains ground - or individually one at a time. It may create a ground loop, but will also give a drain path for ground currents. It is not always clear what effect lack of a mains ground will have.
Another option would be use of a transformer based ground isolator between DAC and ADC - though these tend to be expensive. You could also try unbalanced to balanced converters at each end of the analogue link to connect via a balanced connection - with the ground connection broken in the analogue connection at one end.
Sokel
Master Contributor
- Joined
- Sep 8, 2021
- Messages
- 6,254
- Likes
- 6,383
Agreed on everything measurement wise.
The debate here though is how to achieve the performance he payed for in normal circumstances and not watching at a picture that shows a 20 yo noisy DAC.
If Amir showed that perfomance in any dac here the result would not only be decapitated panther but also tar and feathers on top of it.
It's not about audibility of course,if it was just that he could just stay with the Trident and be done with it.
True. But Amir - of necessity - is measuring the performance of a component. It would not make any sense for him to measure in the presence of external noise - which would just mask the component performance.
The fact that is hard to achieve that performance in real world conditions is irrelevant. We want to be sure that any component we buy doesn't *add* to any problems that exist in our system. But lets be clear that those problems are in our system, not a problem with the component.
In fact the best way to minimise those problems is to go for balanced interconnect (and optical) wherever possible. A point that @amirm makes frequently.
And we should also understand that buying new components - even if they measure much better than those we already have - are probably not going to improve on our real world system performance - unless we have set that up really really well.
Sokel
Master Contributor
- Joined
- Sep 8, 2021
- Messages
- 6,254
- Likes
- 6,383
I totally agree,specially about the not adding problems.
BUT,we are often told that good DACs are immune to noise or other PC problems,something that clearly holds lots of water.
Making such a statement as we often read needs to be backed-up by facts,for example measuring it with a proved noisy front end and such.
That's my argue,cause people think that they may buy a 120 SINAD DAC,connect it without any care and that's it.
Nope.
I went through the same scrutiny,Multitone thread is full of my (newbie) trial-error effort (backed with measurements) to get were I wanted (specially with noise as after the DAC my active system is all analog) .
The mostly are. It probably is in this case.
But if you have a ground difference (noise) at two ends of an analogue unbalance interconnect, then there is nothing the DAC can do.
The output of the dac can be absolutely perfect with zero noise on it's output. The device at the other end of the interconnect will add the ground voltage difference (noise) to the signal it receives. This has nothing whatsoever to do with the DAC being immune to noise or "PC problems" The noise is generated along the ground of the analogue interconnect - not at the output of the DAC.
Sokel
Master Contributor
- Joined
- Sep 8, 2021
- Messages
- 6,254
- Likes
- 6,383
Here's a measurement that NO interconnect involves (other than the 20cm loop cable made for this reason).
It's only connected to PC,powered by the PC.
Look at the mess in the second measurement,almost 5db noise hit.
And all I did is I changed the USB ports,the good one is USB2 and the bad is USB3 (does the same with all ports of the same kind).
Ok,you can argue that E-MU is not a good DAC,after all is 15 years old but it's noise is decent otherwise when plugged properly.
Would I know that without measurements?No.
The same goes for d10s,at -98db noise is a horror story,something will amplify that tens of times down the way,someone must start from the beginning to see what hurts it.
Well said!
The AP test system that Amir uses has been engineered to present as close to a perfect environment as possible to the device under test. The CMRR is excellent, the power system is isolated from the AC mains as well as possible, and so on. That allows AP users to really test a device with minimal external influences, revealing its fundamental performance. Heisenberg go home... It's the best that anybody can do.
But, who has a home audio system that is as pristine and isolated as that?
Unless a tester goes to great lengths to simulate various external influences, the behavior of the device under test as might be found in a real system isn't measured. There's few or no standard tests for this, anyway, and the number of permutations is probably overwhelming. It's no surprise that nobody does this testing, at least to my knowledge. (Simple example: Power amplifiers are tested into resistive loads for distortion, not the complex load impedances that speakers present, especially at the ends of speaker cables.)
I've always thought that system set-up may be as important as the basic individual component performance. It's not as sexy or exciting as opening a shiny new box holding a new component. It's painful, because the whole idea hasn't been quantified well, and, in fact, is often not understood and therefore dismissed. Many approach the whole idea as a game of mixing and matching various components, which might be a solution but not a great one.
This is a great opportunity for somebody. Perform system testing and publish your methodology and results. Be prepared, though - wear a suit of armor...
A mess? You are talking about 5dB at -140dB. Assuming 2V 0dBFS - then we have a difference between 0.2uV and 0.35uV. A 0.15uV change. And you still have a 20cm interconnect which could be coupling that noise. But even if it is on the output of the dac - it is still absolutely amazing performance.
I think it is safe to say that it fails to indicate a significant failure in noise immunity.
And for the D10 s - so what? If the noise is added after the signal has left the DAC - it says nothing about the DAC performance.
I think it is safe to say that it fails to indicate a significant failure in noise immunity.
And for the D10 s - so what? If the noise is added after the signal has left the DAC - it says nothing about the DAC performance.
Sokel
Master Contributor
- Joined
- Sep 8, 2021
- Messages
- 6,254
- Likes
- 6,383
No,no,I'm talking the noise itself in my measurement,-110db against -105db,you can see it in the window in the chart,up right corner.or in the SNR change (also 5db).
The cable is not a factor here,I have made literally thousands of measurements with it with the same results,port is the problem there,I have verified that lots of times.
Inaudible of course but...
OK - so now you are talking about the difference between 6 and 11uV. Still microscopic in terms of system noise.
Not sure what you mean by "port" or what your actual test setup is in this case. Or where you have measured the noise (is it common mode noise on the input power for example?)
In any case, this is sort of irrelevant for the OP
Last edited:
Sokel
Master Contributor
- Joined
- Sep 8, 2021
- Messages
- 6,254
- Likes
- 6,383
Ok,my bad,I should describe the rig.
So:
1st measurement. PC > USB2 port > E-MU0204 DAC > 20cm cable >E-MU0204 ADC > (same) USB2 > PC.
2nd measurement.PC > USB3 port > E-MU0204 DAC > 20cm cable >E-MU0204 ADC > (same) USB3 > >PC.
(a loopback really,E-MU powered by the PC,only change is the USB port,USB2 vs USB3)
This is apparently what you can get in real life. It seems that some people report the same issue with the SMSL su1 also .
This is enough for me to not buy those dacs. As Vintageflankers review of a very cheap and new smsl dac show, there might be quality problems on some units were the product might be faulty . As a customer you will never now If the gear is measuring correct or not. And those dacs dont sell in normal stores so you cant listen or get any guaranty for a faulty unit. Do they have any quality control at all ? How do we know ?
As Audiosciencereview grow bigger, this site is probably also targeted from those companys as a commercial sellplace. They send a unit to Amirm that measures perfect , it gets a golfing pink panther and there is a sell success of the product.
Good work Thomas for showing us this ( possible ) issue.
Last edited:
So in this case the noise is either coupled in the 20cm cable, or is coupled internally in the E-MU0204, between DAC and ADC circuits. Most likely the latter, as it is difficult to think how the USB connection could change noise coupling in the interconnect.
Either way - your system is not OP's system. Noise coupling mechanism is almost certainly completely different - not least since DAC and ADC are in different boxes, powered from different supplies.
What problem? Noise spikes at -120dB? It is incredibly difficult to get system noise down at this level. What on earth makes you assume this is a problem with the device being tested, rather than the much more likely test setup?
Folks - there is a reason an audio precision costs 10's of thousands. Don't for one minute think you are going to get results as good as it using a couple of hundred £ ADC (No matter how good that ADC may be), combined with consumer grade PC, power systems and interconnects.
Last edited:
Sokel
Master Contributor
- Joined
- Sep 8, 2021
- Messages
- 6,254
- Likes
- 6,383
You can see my #32 post for a similar (and much worst in terms of conditions) test with different devices.
I'm trying to point out that everything is specific to the measured system,essentially we're saying the same thing.
The problem is that Thomas bought a dac with a reallity SINAD of 98 when he expected 110.
Even if you cant hear such a difference, the selling points of this cheap dacs are a low price combined with spectacular measurement results. People expect to get highend results for almost no money.
A lesson learned from this might be that If one connects a computer thats not running on batterys, one can not expect a SINAD higher than 100 with a dac.
Last edited:
