• 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!

Matching pre-/amp, input sensitivities, output voltage ...

andrew

Active Member
Joined
May 3, 2018
Messages
192
Likes
42
I’m confused with matching gain across my (DIY) system and after advice. The heart of the system is an RME UCX which feeds power amps with an input sensitivity of 1.4VRMS (or 0.775VRMS). This audio interface has +4BU and -10dBV settings on the analog outputs with manual stating that:
  • High is +19dBU (0dBFSC) with 15dB headroom – calculated to deliver 16.7 VRMS
  • +4dBU is +13dBU (0dBFSC) with 9dB headroom – calculated to deliver 3.45VRMS
  • -10dBV is +2dBV (0dBFSC) with 12dB headroom – calculated to deliver 1.27VRMS
using http://www.sengpielaudio.com/calculator-db-volt.htm for the VRMS calculation. Questions:
  1. So, assuming all things are equal, is it correct that the output of the RME should be set at -10dBV as the higher level +4dBU is too hot for the input of the power amp? (This seems to make sense although I thought that +4dBU was pro-audio set at 1.228VRMS so I might be missing something).
  1. Applying DSP (using Audiolense) sees a circa 20dB decrease in output across the various channels. What is the right path to addressing this issue given the above context? One option is to set the outputs to +4dBU but I’ve got no idea as to what this means for matching the input to the power amp(s). Another option is to insert a Topping Pre90 as a gain stage but this appears to have an input sensitivity of 9.2V at 0dB and outputs up to 50V P to P / 18V RMS which doesn’t seem to work.
Any and all advice would be welcome
 
OP
A

andrew

Active Member
Joined
May 3, 2018
Messages
192
Likes
42
At present, I’m using the digital input of the UCX with volume control in the digital domain at source. My plan to solve the second problem has been to add an analogue pre-amp(s) that provides a gain stage and can also do volume control but I wanted to understand the various issues with ensuring compatibility before moving on this idea
 
Last edited:

mdsimon2

Major Contributor
Forum Donor
Joined
Oct 20, 2020
Messages
2,477
Likes
3,315
Location
Detroit, MI
OK, to answer your first question about output voltage you want to use the lowest setting that gives you an acceptable output level. The lower level you use the lower residual noise and less hiss you will have. The output voltages you have listed look correct with the exception of 19 dBu which is 6.9 V not 16.7 V. It also important to note that those voltages are maximum levels if you are sending a 0 dBFS signal to the DAC, if your signal is lower than 0 dBFS either through volume control, DSP attenuation or just the source being lower level you will not see full output voltage.

Gain staging with DSP can get complicated when you have a lot of degrees of freedom. In terms of the 20 dB level decrease in DSP do you know if this is across all frequencies? A common complaint from DSP users is that the overall level decreases but you need to be careful because some frequencies may have been boosted, these boosted frequencies can still result in clipping (either in DSP or an amplifier) even if your perceived output level is lower. If you truly had 20 dB attenuation in DSP at all frequencies you would never be able to drive your amplifiers to full power regardless of the UCX gain setting, 6.9 V with 20 dB attenuation is only 0.69 V. I like to setup my DSP such that maximum output level will never exceed 0 dBFS (will never clip), typically this involves attenuating boosts but could also involve boosting cuts.

In your case I would do the following. Measure (or understand) DSP being applied and normalize output level digitally such that maximum output level is 0 dBFS, this will ensure you have maximum output level from DSP and no clipping within DSP. Then start with -10 dBV setting and see if this gives you an acceptable output level. If it doesn't you can move to +4 dBu but this comes with the possibility of clipping your amplifiers depending on volume level and source level. For example if you have the amplifier at the 1.4 V input sensitivity setting and you are at a volume control position of -8 dB or less you will never be able to clip the amplifier. It is not uncommon to use a slightly higher nominal output voltage from your DAC to allow you to accommodate lower level recordings.

If you are happy with the noise performance I wouldn't worry about an analog preamp. If you did use the pre90 I would probably use the +19 dBu setting to send maximum output voltage and then attenuate in the pre90, this has the potential to give you better noise performance but if you do not hear any noise with nothing playing when using digital volume control there really is no benefit to analog volume control.

Michael
 
OP
A

andrew

Active Member
Joined
May 3, 2018
Messages
192
Likes
42
@mdsimon2 - First and foremost thanks for the detailed response. I've read a couple of times and have a few questions comments:

OK, to answer your first question about output voltage you want to use the lowest setting that gives you an acceptable output level. The lower level you use the lower residual noise and less hiss you will have. The output voltages you have listed look correct with the exception of 19 dBu which is 6.9 V not 16.7 V. It also important to note that those voltages are maximum levels if you are sending a 0 dBFS signal to the DAC, if your signal is lower than 0 dBFS either through volume control, DSP attenuation or just the source being lower level you will not see full output voltage.

The speakers are active 3-way (bass + mid-high stereo with dual subs) and the amps for the bass modules have gain control as do the subs. I'm thinking that, this being the case, it'd be best to use the highest possible output setting from the RME UFX with amp gain control used to adjust volume. The thought is that this approach sees the RME UFX noise (which I'm assuming is fixed) being a small fraction of the overall signal. Your suggestion above is the opposite. What am I missing?
Gain staging with DSP can get complicated when you have a lot of degrees of freedom. In terms of the 20 dB level decrease in DSP do you know if this is across all frequencies? A common complaint from DSP users is that the overall level decreases but you need to be careful because some frequencies may have been boosted, these boosted frequencies can still result in clipping (either in DSP or an amplifier) even if your perceived output level is lower. If you truly had 20 dB attenuation in DSP at all frequencies you would never be able to drive your amplifiers to full power regardless of the UCX gain setting, 6.9 V with 20 dB attenuation is only 0.69 V. I like to setup my DSP such that maximum output level will never exceed 0 dBFS (will never clip), typically this involves attenuating boosts but could also involve boosting cuts.

In your case I would do the following. Measure (or understand) DSP being applied and normalize output level digitally such that maximum output level is 0 dBFS, this will ensure you have maximum output level from DSP and no clipping within DSP. Then start with -10 dBV setting and see if this gives you an acceptable output level. If it doesn't you can move to +4 dBu but this comes with the possibility of clipping your amplifiers depending on volume level and source level. For example if you have the amplifier at the 1.4 V input sensitivity setting and you are at a volume control position of -8 dB or less you will never be able to clip the amplifier. It is not uncommon to use a slightly higher nominal output voltage from your DAC to allow you to accommodate lower level recordings.

Michae
I've gone back to compare the measurements and correction filters, and had this a bit wrong. The rough average attenuation is about 10dB and at no point do the filters boost. The manual states "We often get questions about why the filters attenuate the output so much. The short answer is that in order to avoid digital clipping, the average sound pressure level has to be attenuated by, typically 6-10 dB to give enough headroom for the single frequency that needs the most amplification. The frequency response of the correction filters – as they appear in the main form will give a 100% accurate picture of how the filters will attenuate the output."

 

mdsimon2

Major Contributor
Forum Donor
Joined
Oct 20, 2020
Messages
2,477
Likes
3,315
Location
Detroit, MI
I should say I am basing this off my experience with a Fireface 800 which also has the same -10 dBV, +4 dBU, high gain settings and assuming that the UFX performs in a similar manner.

As far as I can tell RME uses analog attenuation for the different gain settings so the dynamic range is preserved. My original response was based on the understanding that you have an amplifier with an input sensitivity of 1.4 V / 0.775 V. To explain this a little more let's look at the effective dynamic range using my Fireface 800 as an example and an amplifier with 1.4 V input sensitivity.

I've attached some measurements using a Cosmos ADC in mono mode with XLR splitter on one channel for best possible SNR at the various gain settings. Output voltage is 2 mV in all cases, the idea being we want to make sure there is no low level muting going on while also not polluting the measurement with THD from higher output signals.

-10 dBV, FS output = 1.233 V, noise = 3.27 uV, dynamic range = 20 x log (1.233 x 10^6 / 3.27) = 112 dB
+4 dBu, FS output = 3.404 V, noise = 5.98 uV, dynamic range = 20 x log (3.404 x 10^6 / 5.98) = 115 dB
high gain, FS output = 6.800 V, noise = 11.3 uV, dynamic range = 20 x log(6.800 x 10^6 / 11.3) = 116 dB

I think the reason that -10 dBV has a bit lower dynamic range is because I am bumping up against the ADC noise floor. You can clearly see that at the same output level the lower gain setting always has lower noise from an absolute noise perspective.

At high gain you can't use all of the 6.8 V output because you will clip your amplifier so you need to attenuate the signal by 20 x log(1.4 / 6.8) = -14 dB which will cost you the same amount in dynamic range, so effective dynamic range becomes 116 dB - 114 dB = 102 dB.

At +4 dBu you don't need to attenuate as much but you still need to attenuate by 20 x log (1.4 / 3.4) = -8 dB so your effective dynamic range is 115 dB - 8 dB = 107 dB.

Finally at -10 dBV you won't quite be able to drive the amplifier to full power but you do retain the full dynamic range of 112 dB. So even if -10 dBV measurement is a bit conservative due to ADC noise you are still getting the best dynamic range at -10 dBV.

Amplifier gain is of course another variable. Any noise from your DAC will be multiplied by your amplifier and in general lower gain amplifiers are lower noise. So if you have the option of using a low gain amplifier and the high gain RME setting that may end up better for you. It all depends on how much gain and how much residual noise the amplifier has. Of course it would certainly make more sense to optimize your gain staging around your tweeters rather than your bass / subs as they are likely more sensitive from both a speaker sensitivity and human hearing standpoint.

Michael
 

Attachments

  • fireface800_-10dbv_2mvout_1.7vadc_splitter.png
    fireface800_-10dbv_2mvout_1.7vadc_splitter.png
    167.8 KB · Views: 77
  • fireface800_+4dbu_2mvout_1.7vadc_splitter.png
    fireface800_+4dbu_2mvout_1.7vadc_splitter.png
    170.3 KB · Views: 72
  • fireface800_high_2mvout_1.7vadc_splitter.png
    fireface800_high_2mvout_1.7vadc_splitter.png
    171.1 KB · Views: 55
OP
A

andrew

Active Member
Joined
May 3, 2018
Messages
192
Likes
42
@mdsimon2

Much appreciated. I’ll read, digest and remeasure over the weekend. Out of interest, where in your chain are you doing volume control?
 

mdsimon2

Major Contributor
Forum Donor
Joined
Oct 20, 2020
Messages
2,477
Likes
3,315
Location
Detroit, MI
You are welcome!

I have two setups, both are DIY active stereo utilizing CamillaDSP on RPi4s.

My main system has an Okto DAC8 pro and I use that for volume control (downstream of DSP).

The other system has had a variety of DACs (MOTU Ultralite Mk5, Apogee Ensemble, MOTU Traveler Mk1, RME Fireface 800, etc) and is always in a state of flux as it is my test bed so to speak. In this system I do volume control in CamillaDSP but have added a display and IR receiver so that I have volume indication and remotely control without using the CamillaDSP web interface, see the link in my signature for more details.

Michael
 
OP
A

andrew

Active Member
Joined
May 3, 2018
Messages
192
Likes
42
You are welcome!

I have two setups, both are DIY active stereo utilizing CamillaDSP on RPi4s.

My main system has an Okto DAC8 pro and I use that for volume control (downstream of DSP).

The other system has had a variety of DACs (MOTU Ultralite Mk5, Apogee Ensemble, MOTU Traveler Mk1, RME Fireface 800, etc) and is always in a state of flux as it is my test bed so to speak. In this system I do volume control in CamillaDSP but have added a display and IR receiver so that I have volume indication and remotely control without using the CamillaDSP web interface, see the link in my signature for more details.

Michael

Thanks. My DIY set-up is similar except that I'm using Audiolense's AudioConvolver (running on a stripped Windows 10 machine) for executing the 6-way x/o and overall EQ before using the RME Fireface UCX for D/A conversion. The volume control is done in a miniDSP DDRC-22D which I use to switch between multiple digital sources and ensure re-sampling to 24/96 ahead of the DSP. A slightly out-of-date diagram is below and it was, from memory, informed by reading your posts on CamillaDSP.

I was hoping that AL would ultimately provide volume control but this remains to be committed. The Okto DAC8 Pro looks fabulous but has been out of stock forever and second-hand units are non-existent here in Australia.

Screenshot 2022-02-12 081547.jpg
 

Kijanki

Active Member
Joined
Dec 16, 2021
Messages
106
Likes
69
When I was young -10dBV (0.316Vrms) standard was very popular. AFAIK it was cost saving measure. You need only one item (amplifier) more expensive (more gain), while everything connected required less gain stages. The problem is that it moves most gain to less clean environment (power amp) and increases relative electrical noise pickup by interconnects (lower signal level). Today 4dBu (1.23Vrms) standard is more common, but Benchmark recommends using AHB2 amp at the highest selectable input level +22dBu (9.8Vrms) to improve signal to noise ratio. I believe recording studios use 4dBu standard, but 24dBu is also common. My Benchmark DAC3 outputs 27.5dBu with 30dBu clipping point. Extra 5.5dBu of gain between DAC3 and AHB2 (27.5-22) allows for softer recordings.
 
Top Bottom