Nord Lead 3 - Digital Aliasing Issues

[madtheory]

It would be better to record at 44.1kHz, the resampling introduces another variable. Can you record it using a really short cable, Nord directly to RME, no other audio devices at all (including speakers) connected, to see if we can isolate the source of ground loop. Also hold the note for a lot longer than you feel is necessary.

[aeon]

It will really help if you gain stage correctly.

The newest sample has a peak level that is -4.77db below FS. At a sample value of 18901 out of 32768, or 57 percent value, you are not making the best use of your dynamic range.

Also, when I have isolated and pushed gain on the tail of the sample, it is clear there is some degree of hum.

So yeah, a 44.1KHz sample so as to avoid SRC artifacts, done as close to FS as you can would be most excellent. As has already been said, do sustain your sound for a bit. Also, are you sure you don't have envelopes set to decay upon key release?

If you are wondering if there is a hardware fault, it may be better to go with a simple saw wave at C5 or C6 - that will push harmonics well past Nyquist so as to show issues with aliasing, if any, and it would be nice to have a few seconds of that followed by an octave-down pitch bend via the stick. This will help show/animate any inharmonic artifacts.


cheers,
Ian

[madtheory]

IMO making an issue of a peak level of -4.77 dBFS is a red herring. Using a percentage is misleading because it is a logarithmic scale, most of the resolution is at the bottom, so to speak. It's roughly 6dB per bit. The AES/ EBU standard is to aim to peak at -18dBFS (or -18dBFS= 0VU= +4dBu typically), anything above that is a headroom margin to avoid clipping. At -18dBFS it's in the 21 bit range. True 24 bit is difficult to achieve, most of the time the noise of the analogue circuits swamps it. Causing clipping for the sake of getting closer to zero is totally not worth it.

[aeon]

That's fair. I simply think that optimization in that regard will help to isolate any and all variables. Also, if overall gain is increased, we will be able to better hear the tail issues, if any.

I agree that clipping is totally not worth it. That said, when the source is an instrument where you have total dynamics control via envelopes, I think the overall signal level can be effectively managed so as to avoid it.

I also hope the OP has the volume knob on the NL3 set to full-scale. Needless to say, if not, more of the signal will fall below the threshold of noise, and we can't really know what the instrument is capable of.


cheers,
Ian

[madtheory]

I agree that clipping is totally not worth it. That said, when the source is an instrument where you have total dynamics control via envelopes, I think the overall signal level can be effectively managed so as to avoid it.

Very true! -18dBFS is designed for recording unpredictable musicians I personally prefer the convenience of headroom. When tracking a MIDI'd synth I'd leave 5 to 3dBFS.

I also hope the OP has the volume knob on the NL3 set to full-scale.

Ya, I asked that already...

[Erotic Cabaret]

I also hope the OP has the volume knob on the NL3 set to full-scale. Needless to say, if not, more of the signal will fall below the threshold of noise, and we can't really know what the instrument is capable of.

Are you talking about the main volume knob on the upper left part of the unit? No, I didn't turn it all the way up for the recording...even at half-way it nearly clips. It's a very "hot" instrument in terms of volume output. Am I supposed to be turning it up all the way?

This isn't an issue of recording, guys, it's an issue with the synth. What you're hearing in the recording is exactly what you hear from the synth line outs. The Fireface 400 is a good interface and I'm using short Mogami cables (which I've heard are the best) and I'm recording at 24 bit. Do you guys normally track at 44.1 khz instead of 88.2? There shouldn't be ANY issues mixing something recorded at 88.2 to 44.1...it's the simplest form of division.

Also, this is not an issue of envelopes set to decay upon release. I could take the simplest sound and, once again, as long as it has high frequency content and a long release time, the issue will be apparent. After listening to my cousin's NL3, I'm pretty sure the problem is universal...though it's less pronounced in his.

[aeon]

I also hope the OP has the volume knob on the NL3 set to full-scale. Needless to say, if not, more of the signal will fall below the threshold of noise, and we can't really know what the instrument is capable of.

Are you talking about the main volume knob on the upper left part of the unit? No, I didn't turn it all the way up for the recording...even at half-way it nearly clips. It's a very "hot" instrument in terms of volume output. Am I supposed to be turning it up all the way?

Many digital instruments implement the volume knob as a digital scaling before the D/A conversion, so using those with the volume knob down will decrease the signal-to-noise ratio, and increase quantization error in low level signals.

Best practice is to run the synth's volume at max and attenuate afterward (in analog) as needed to match your level-calibrated interface. I've never used a Fireface, but my sense is it should be able to handle the output of a line-level synth without issue or other equipment.

This isn't an issue of recording, guys, it's an issue with the synth. What you're hearing in the recording is exactly what you hear from the synth line outs. The Fireface 400 is a good interface and I'm using short Mogami cables (which I've heard are the best) and I'm recording at 24 bit. Do you guys normally track at 44.1 khz instead of 88.2? There shouldn't be ANY issues mixing something recorded at 88.2 to 44.1...it's the simplest form of division.

I track at the rate of the final delivery medium. That's not so relevant here in any case.

Mogami is fine cable, yes, but as with any cable, issues of shield and grounding come into play, as well as the power to the system itself.

Also, this is not an issue of envelopes set to decay upon release. I could take the simplest sound and, once again, as long as it has high frequency content and a long release time, the issue will be apparent. After listening to my cousin's NL3, I'm pretty sure the problem is universal...though it's less pronounced in his.

Did you listen to his NL3 patched into your setup, or in a different setup?

I know the constant asking of this and that can be tedious and irritating after a time, but it is in the interest of getting your problem sorted, so I apologize in advance if my responses seem a bother.


thanks,
Ian

[madtheory]

Many digital instruments implement the volume knob as a digital scaling before the D/A conversion, so using those with the volume knob down will decrease the signal-to-noise ratio, and increase quantization error in low level signals.

Properly implemented dither solves that problem surprisingly well. I'll test this on the Nord Lead 2x I have here. It has a fairly hot output, as does my Novation KS Rack. When connected directly to my audio interface the Nord is very clean with output level at 12 o'clock for big unison patches. Unfortunately I can't post any examples, the Mac is in hospital.

[Erotic Cabaret]

Aeon -

Thanks for the tip on turning the volume to the max on digital synths. I was unaware that that can improve the signal to noise ratio.

I checked his Nord Lead 3 at his house, so I suppose the power could have been different. I just got a JDI, so perhaps that will improve things. I've heard it can eliminate power issues and hum, though I still don't think that's the issue here.

So, then, Aeon, you must track everything at 44.1khz, yes?

[Erotic Cabaret]

Dang, I just tested the audio again with my Radial JDI and the noise is issue is much improved.

[b3groover]

So, then, Aeon, you must track everything at 44.1khz, yes?

I can't speak for Ian, but I track everything at 44.1kHz. Bit depth is more important than sample rate, imo. 24bit / 44.1kHz sounds great.

[Erotic Cabaret]

For sure, but I hadn't heard of people tracking at 44.1 in a long time. It makes sense, though, to track/mix at the same sample rate the tune will eventually be heard at.

[aeon]

Many digital instruments implement the volume knob as a digital scaling before the D/A conversion, so using those with the volume knob down will decrease the signal-to-noise ratio, and increase quantization error in low level signals.

Properly implemented dither solves that problem surprisingly well.

Agreed, but I have no idea who has or hasn't implemented dither properly, and I am too lazy to test.

Thanks for the tip on turning the volume to the max on digital synths. I was unaware that that can improve the signal to noise ratio.

It can improve signal-to-noise for many analog synths too.

I checked his Nord Lead 3 at his house, so I suppose the power could have been different. I just got a JDI, so perhaps that will improve things. I've heard it can eliminate power issues and hum, though I still don't think that's the issue here.

Radial makes good stuff.

So, then, Aeon, you must track everything at 44.1khz, yes?

My stuff, yes, since I only do audio for myself.
Clients get what they want.

Dang, I just tested the audio again with my Radial JDI and the noise is issue is much improved.

Glad to hear. Part of the mystery solved!


cheers,
Ian

[Erotic Cabaret]

The interesting thing about the Nord Lead 3 is that when you turn up the master volume all the way, even if you've lowered the preamp volume so nothing's clipping and you're getting a normal level, the synth begins to output some very off-putting frequencies. If I turn the Nord's master volume to half way and increase the volume at the preamp stage (and get the exact same overall level) the sound is much better and doesn't have that rumbly frequency issue.

I've always had issues with the Nord's frequency output. I have to EQ the heck out of most sounds, especially sounds I've programmed using the low pass filter that still have a decent amount of high frequency content. The problem is that these sounds are often very "rumbly," especially if you listen to them on laptop speakers. And I always record mono as per Aeon's advice, so you'd think I wouldn't have an issue.

[madtheory]

What do you mean by "rumbling"? I'm guessing it's clipping distortion, but an audio example would help! It would be interesting to hear what difference the JDI makes. I can see how it would fix the ground loop problem, but not the hiss. They have quite a low input impedance so should be quite clean for synth use.