(Unofficial) RME UCX II Interface Review and Measurements

[PhilActiveAudio]

Hi to all, I am a new ASR forum member even if I am an old reader.

I have a 3 way active system based on exaSound e68 DAC, 3 amps and 3 drivers (box and horn). I use a RME Fireface UCX II to measure drivers with Acourate. Measures are use to build filters that include active crossovers and SPL + phase + room slight corrections. Resulting filters are implemented in Roon (convolution).

While measuring drivers, I found that the impulse is having a post little "ghost". I first thought that it was because of the room, but a loopback measurement of the UCX II is also giving me this "ghost" (see picture below). The problem is that this "ghost" is taken into account by Acourate through the whole filter construction process and have a bad impact on the resulting filter impulse... The only way to delete this "gost" is to limit the SPL range to be measured: i.e. 10 Hz to 20 kHz @ 44.1 kHz instead of 10 Hz to 24 kHz, or 10 Hz to 40 kHz @ 96 kHz instead of 10 Hz to 48 kHz.

Of course ,no Fireface effects are ON or I can find where it would be set to ON in TotalMix. I also tried to change various parameters (volume input and output, output dBu level...) but I am stuck with this problem.

Do you have any idea how to solve my problem?

 

[jhenderson0107]

... The only way to delete this "gost" is to limit the SPL range to be measured: i.e. 10 Hz to 20 kHz @ 44.1 kHz instead of 10 Hz to 24 kHz, or 10 Hz to 40 kHz @ 96 kHz instead of 10 Hz to 48 kHz.

Based on this symptom, I suspect that the impulse measurement "ghost" is an artifact of the UCX II's anti-aliasing filter. This is an analog filter which cannot be perfect; It attempts to implement a very sharp stopband filter at Fs/2. I speculate it creates the observed (undesired) time-domain artifact within the passband.

 

[Trell]

Based on this symptom, I suspect that the impulse measurement "ghost" is an artifact of the UCX II's anti-aliasing filter. This is an analog filter which cannot be perfect; It attempts to implement a very sharp stopband filter at Fs/2. I speculate it creates the observed (undesired) time-domain artifact within the passband.

You messed up the quote in your post as I did not write that but @PhilActiveAudio

 

[Rja4000]

While measuring drivers, I found that the impulse is having a post little "ghost".
(...)
The only way to delete this "gost" is to limit the SPL range to be measured: i.e. 10 Hz to 20 kHz @ 44.1 kHz instead of 10 Hz to 24 kHz, or 10 Hz to 40 kHz @ 96 kHz instead of 10 Hz to 48 kHz.

It's probably indeed the DA reconstruction filter, which is fixed for the UCX II, if I'm not mistaken (and that makes a lot of sense for an audio interface).

But I have a hard time understanding how limiting bandwidth to 40kHz can be an issue ?

 

[MC_RME]

Acourate has certainly not been developed to 'measure drivers'. That 'ghost' does not exist, it is a display error due to not enough sample points. Get an app that uses proper Sinc interpolation, and with enough sample points you might see the real impulse response.

 

[Rja4000]

That 'ghost' does not exist

So, if I understand correctly, this is just a side effect of the bandwidth limitation, linked to the sampling frequency, which is not managed correctly by the Acourate software ?

 

[MC_RME]

Correct. You can see this in MI DSO. Activate Sinc and use 10 ms horizontal. From about 4 ms on it will look correctly. Deactivate Sinc and check the sample points available...

The whole point in measuring impulse response is to use a real DSO , which has a very high sample rate. When using a normal ADC you need to use 384 kHz or better 768 kHz sample rate, otherwise the representation will be inaccurate, especially and most easy to notice with NOS.

 

[PhilActiveAudio]

Correct. You can see this in MI DSO. Activate Sinc and use 10 ms horizontal. From about 4 ms on it will look correctly. Deactivate Sinc and check the sample points available...

The whole point in measuring impulse response is to use a real DSO , which has a very high sample rate. When using a normal ADC you need to use 384 kHz or better 768 kHz sample rate, otherwise the representation will be inaccurate, especially and most easy to notice with NOS.

Thanks for all your answers that clarify my situation. I will contact Mr Brueggemann and share with him your feedbacks.

For my understanding, what is MI DSO (Multi Instrument ?)?

In regards to 40 kHz, limiting the bandwidth is indeed not an issue since anyway my mic stop at 30 kHz and my ears far before... It was to mention that I have the same issue with 48 kHz (and not 44.1 as I mentioned before) and 96 kHz so quite a reproductible effect.

 

[PhilActiveAudio]

Initial measurements were done using a cable to link Output 1 (TRS) to Mic input 1 (XLR) so as to be close as possible to driver / room measurement conditions. Thus, making use of DAC and ADC. Acourate is being developped to perform such measurements (please refer to the various articles about Acourate and Mitch Barnett's book).

Following Mr Brüggemann advice, I tried a new measurement using the hardware loopback mode of the Fireface UCX II. Can't find any "ghost":

So, the origin may be the Fireface DA reconstruction filter.

Since there is no way to modify parameters of the DA reconstruction filter (if I inderstand well previous answers), I will continue working at high samplerate (e.g. 96 kHz or above) with a bandwidth from 10 Hz to 30 kHz.

 

[morillon]

small aside
in other circumstances but on speakers... I could observe that the threshold was 88k... so of course 96k will do it very well..
just you have a microphone with a precise calibration file up to 30k (m30 or m50 ?)..? (otherwise it doesn't seem to make much sense to me...)
;-)
like 45hz 22khz modest "8000" microphone

 

[PhilActiveAudio]

I only use the RME Fireface UCX II to perform speakers (3 voices) raw measurements with target crossovers. Acourate (and other equivalent softwares) then manage everything to build target impulses for each drivers including crossovers, SPL linearization, groupe delay and room corrections. The 6 generated impulses are used with Roon (convolution). My listening DAC is an exaSound e68. But if the measurement process includes errors (it is already quite a difficult process), target impulses will also include errors and this is what I want to avoid.

I am using an Earthworks DM23 that goes from 10 Hz to 48 kHz (I just opened again the file) so above 30 kHz that I had in mind. I will push the Fireface to "it's limits" to see exactly where it becomes really problematic and adjust the range accordingly.

 

[morillon]

the m23 is announced at ...23khz certified
(for this use.. the "new" version "r" very adapted)
but for an "30k"...we have the m30 -50
as much to work in the "reasonable" characteristics of our microphones etc.
;-)

 

[PhilActiveAudio]

Indeed, just checked. Thanks Morillon.
Then I will limit at 23 kHz.

 

[morillon]

22khz...88k-96k (and like 44k or less in distortion mesureaments ;-) )
that s good for this use ( and this mic)
;-)

 

[PhilActiveAudio]

Yes, as far as the interface is concerned, it is good.
I now need to perform a new speakers measurement.

 

[bequietjk]

What is the dOO dAD doomaflooper on the back for?
View attachment 154831

In case you want to sport this on your hip or on a necklace.

 

[Trell]

In case you want to sport this on your hip or on a necklace.

Add a 12V battery along with an USB stick for recording using DUREC on the UCX II, and you’ll have a real on-the-go system. Don’t forget a nice microphone or two.

 

[Tuan Huynh]

I prepared some kind of "Dashboard"

Cheking all Balanced Output

SINAD
View attachment 154826

SNR
View attachment 154828

I would like to know the SINAD of + 24Dbu from Main XLR output! Can you share?

 

[Rja4000]

I would like to know the SINAD of + 24Dbu from Main XLR output! Can you share?

The UCX II maximum output level is +19dBu

 

[Tuan Huynh]

The UCX II maximum output level is +19dBu

OMG Sorry I misread it with UFX 2