Happy to report that my theory seems to bear out.
I am closer to concluding that my real problem was indeed the very poor S/N from the DAC at nominal listening level (94db drivers and high power hypex amp modules)
I added a -10db attenuator on amp side of the XLR cable and increased the DAC volume to get back the same SPL from the speaker. With this, the distortions are almost inaudible compared to without the attenuator.
In any case, there is something about my setup where the noise floor is elevated. The designers measurement of the DAC showed pretty low noise floor. Need a decent soundcard to make the measurements on my end.
So the options i have now are
1. build some really quite balanced passive attenuators on the DAC to amp XLR interconnect.
2. Lower the gain on the UCD400 and NC400 modules
I can definitely recommend the second alternative. You will have to unsolder a SMD resistor, but it is not as tricky as it seems.
Lowering the gain in the amps has the added possible benefit of increasing the amount of feedback, and thus (at least in theory) increasing linearity (albeit if it already is way beyond audible limits, increasing it won't really be audible).
Reducing the gain on the modules is hard to reverse so i'm reluctant.
For now i switched to using the unbalanced output of the DAC.
This allows higher signal level compared to the balanced output.
the distortion is less audible but is still present.
Any idea what the typical line level rms voltage is for nominal listening?
For now i switched to using the unbalanced output of the DAC.
This allows higher signal level compared to the balanced output.
the distortion is less audible but is still present.
Any idea what the typical line level rms voltage is for nominal listening?
That is true for 400 Wrms in a 4 Ohm load with a gain of 25.8. Take a look in the datasheet, its explained there.
Thanks for verifying - just about to dash off to a trade fair (Integrated Systems Europe), so didn't have time to check the data sheet.
BTW, Hypex are exhibiting at ISE.
Some questions i need to investigate
1. Whats the real noise floor on my DAC outputs
2. Why does the noise appear to modulate with the music
3. What nominal rms voltage should be on the DAC output for good S/N on the amp input, and to operate the AMP at its lowest distortion wattage
4. I want to make sure for nominal home listening level, the DAC output level is where it performs the best and drives the AMP according to #3 above
I did some measurements and i notice that for nominal listening volume setting on the DAC, i measure 16-80mv while playing common songs.
I think across a wide range of source, the signal level from the DAC would be around 100mv rms range.
I am using unbalanced output from the DAC now (offers slightly better S/N at DAC output, as confirmed by slightly reduced audibility of the distortion i started about)
@100mv i am operating the AMP (@20x gain) around 1W range into 4ohm, this is already pretty loud for my listening room.
I think if i drop the gain of the modules down to 4.7 (by removing the resistor), i might be safely able to raise the DAC output level for better S/N at the AMP input while still getting around 1W nominally from the AMP.
Btw, the graph seems to suggest that the AMP provide lowest distortion around 5-10W range.
Also the DAC seems to perform best when its setting is -12dB, my current nominal listening setting is -20dB
1. Whats the real noise floor on my DAC outputs
2. Why does the noise appear to modulate with the music
3. What nominal rms voltage should be on the DAC output for good S/N on the amp input, and to operate the AMP at its lowest distortion wattage
4. I want to make sure for nominal home listening level, the DAC output level is where it performs the best and drives the AMP according to #3 above
I did some measurements and i notice that for nominal listening volume setting on the DAC, i measure 16-80mv while playing common songs.
I think across a wide range of source, the signal level from the DAC would be around 100mv rms range.
I am using unbalanced output from the DAC now (offers slightly better S/N at DAC output, as confirmed by slightly reduced audibility of the distortion i started about)
@100mv i am operating the AMP (@20x gain) around 1W range into 4ohm, this is already pretty loud for my listening room.
I think if i drop the gain of the modules down to 4.7 (by removing the resistor), i might be safely able to raise the DAC output level for better S/N at the AMP input while still getting around 1W nominally from the AMP.
Btw, the graph seems to suggest that the AMP provide lowest distortion around 5-10W range.
Also the DAC seems to perform best when its setting is -12dB, my current nominal listening setting is -20dB
I really suggest putting a potentiometer on the amp input to find out the optimal gain structure - you can remove it again as soon as you have found out exactly what gain you want.
You can't really measure with actual music. Set the volume setting using music, and then measure with a full-amplitude 1 kHz test tone. Easy to generate with audacity or sox.
100 mV amplified with a gain of 20 results in 2 V on the output - that is 1 W into 4 ohm.
No use to worry about that - harmonic distortion won't be an issue below 100W.
Strange - a DAC should perform best at 100% volume.
Yes that is often the case. The easiest way is to connect the signal generator output (test tone) to another scope channel and then set the scope to trigger on that channel. Or to connect the test signal to a trigger input if that is available on the scope.
Jan
Thanks Jan.
My generator is software based through usb interface of a PC.
But as you suggest i could try to use the sine output of the DAC output as trigger input.
I took some measurements of sine test tone on the DAC output
test tone was generated in Jriver, this is a -20db, 100Hz sine tone.
With the DAC volume setting at -20 (my nominal listening level), it produced only around 12mV rms and i see a large amount of crud on the waveform. What is that? Can that explain the distortion i hear from the bass driver at nominal listening levels?
Second picture is the same test tone with DAC volume at 0dB. DAC output is around 112mV
Also tried a 0db sine test tone generated through audacity.
at -20dB voulme, DAC produces about 110mV rms
at 0dB volume, output is close to 1.1 V
test tone was generated in Jriver, this is a -20db, 100Hz sine tone.
With the DAC volume setting at -20 (my nominal listening level), it produced only around 12mV rms and i see a large amount of crud on the waveform. What is that? Can that explain the distortion i hear from the bass driver at nominal listening levels?
Second picture is the same test tone with DAC volume at 0dB. DAC output is around 112mV
Also tried a 0db sine test tone generated through audacity.
at -20dB voulme, DAC produces about 110mV rms
at 0dB volume, output is close to 1.1 V
Good question. ~5 kHz tone leaking through from somewhere.
It is a sign that something is wrong - perhaps oscillating.
Hi Julf,
Is there a possibility of significant power line/ground noise?
I see a lot of high freq crud when I observe on the oscilloscope its ground output stub.
Further i doubt if the low amplitude measurements from the scope can be considered reliable due to 8-bit resolution and limited sensitivity. I'm not sure, may be i should read up on that.
For well designed mordern DACs the noise floor should be in microvolt range right? Hence a high resolution soundcard would do a better job. Unfortunately my focusrite doesnt appear up to the task.
as i had mentioned, -20db resistive attenuation pad on the DAC output make the audible distortions go away. May be the resistive loading makes the marginal instability disappear. Its a bit frustrating that i cant seem to see on the oscilloscope, what i'm able to hear from the drivers and even record on the reasonable microphones on newer smartphones.
Is there a possibility of significant power line/ground noise?
I see a lot of high freq crud when I observe on the oscilloscope its ground output stub.
Further i doubt if the low amplitude measurements from the scope can be considered reliable due to 8-bit resolution and limited sensitivity. I'm not sure, may be i should read up on that.
For well designed mordern DACs the noise floor should be in microvolt range right? Hence a high resolution soundcard would do a better job. Unfortunately my focusrite doesnt appear up to the task.
as i had mentioned, -20db resistive attenuation pad on the DAC output make the audible distortions go away. May be the resistive loading makes the marginal instability disappear. Its a bit frustrating that i cant seem to see on the oscilloscope, what i'm able to hear from the drivers and even record on the reasonable microphones on newer smartphones.
Sure, but a 5 kHz tone is not really typical of power line or ground-related noise.
When you short the scope probe to ground?
8 or 10 bit resolution is fine, as long as the prescaler works properly - you don't have more resolution with your eyes on the screen anyway.
It is important not to confuse range and resolution. Resolution tells you what the smallest *difference* you can measure is, range tells you the smallest (or largest) signal you can measure. The focusrite should be more than good enough if you get the gain structure right.
Didn't the scope pictures clearly show a ~5 kHz noise signal?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.