Just to clarify -I am not getting clipping on the Input but on the Output. I now have the inputs set at -3db and the outputs set to -6db and it is still clipping. With my 104db sensitivity speakers it is pretty nasty!
I have also tried using coax fron the Touch to the Najda, same results.
Bass/Treble is Off.
As far as I am aware the Touch does not change the signal, I have it set at 100% output.
DSP cores are 48% and 35%, there are no clip warnings in NUC.
Since I prefer using the Touch I would really like to sort this problem out.
I have also tried using coax fron the Touch to the Najda, same results.
Bass/Treble is Off.
As far as I am aware the Touch does not change the signal, I have it set at 100% output.
DSP cores are 48% and 35%, there are no clip warnings in NUC.
Since I prefer using the Touch I would really like to sort this problem out.
Hi Mike, thanks for letting me know that you got the parcel! Looking forward to your build!
Hi Alan. I start with a few remarks:
- Assuming you're playing the same content, there should be no difference between the digital outputs of 2 devices. If there's a difference, then at least one device is scaling audio data.
- There's no level difference between SPDIF coax and optical.
You write there's no clip warning in NUC. So how do you know the outputs are clipping?
The amount of headroom that's required depends on your processing parameters. It's impossible for any of us to tell you if you need -6 or rather -12 dB without knowing how you've set the things. Can you please post your setup file?
Nick
Nick,
I know I am getting clipping in two ways, firstly through the speakers and secondly by watching the LEDs, the Output Clip led is lighting.
I am about to try some other settings so I will send you my NSF file at the end of the session.
The weird thing is that I never had this problem the previous times I have used theNajda board, but that may be that I did not notice it with much more inefficient speakers.
Back to NUC for a while.
I know I am getting clipping in two ways, firstly through the speakers and secondly by watching the LEDs, the Output Clip led is lighting.
I am about to try some other settings so I will send you my NSF file at the end of the session.
The weird thing is that I never had this problem the previous times I have used theNajda board, but that may be that I did not notice it with much more inefficient speakers.
Back to NUC for a while.
Probably slightly off topic, but it´s related to my Najda build, so here goes:
I´m using a 1U chassis from Modushop, anodized black. The anodization is non-conductive, and the screws that holds it together does not seem to form a very good connection between the panels. Is it sufficient to have one good connection between each touching panel, or should it be continuous for shielding purposes? I could achieve the first by milling off a small section from each panel, and connecting them with wires/eye connectors. If not, I would have to remove the coating right at the edges where the panels meet, risking some "bleed" onto the outside the chassis, which won´t look very nice.
Modushop should really mask off some areas when they anodize. Also, the chassis is built so that you need to put the back panel on last, which is a pain. So i was forced to "mod" it slightly to be able to put the top on last (which i suppose is what most people want to do..)
Otherwise my experience with modushop has been good. When sent a properly formatted drawing of the front panel, they milled it exactly like I wanted it.
I´m using a 1U chassis from Modushop, anodized black. The anodization is non-conductive, and the screws that holds it together does not seem to form a very good connection between the panels. Is it sufficient to have one good connection between each touching panel, or should it be continuous for shielding purposes? I could achieve the first by milling off a small section from each panel, and connecting them with wires/eye connectors. If not, I would have to remove the coating right at the edges where the panels meet, risking some "bleed" onto the outside the chassis, which won´t look very nice.
Modushop should really mask off some areas when they anodize. Also, the chassis is built so that you need to put the back panel on last, which is a pain. So i was forced to "mod" it slightly to be able to put the top on last (which i suppose is what most people want to do..)
Otherwise my experience with modushop has been good. When sent a properly formatted drawing of the front panel, they milled it exactly like I wanted it.
Hi Alan,
I got your setup file.
So you've set a global headroom of 6 dB in the input processing section.
In the channel processing section, you're using only 4 channels and have defined a 2-way LR4. You're also attenuating all outputs by 3dB after the processing (i.e. in the post-gain section).
It's not clear to me why you're having post-gains. If your intent is to add more headroom and prevent clipping, then you should move these gains to the pre-gain stage. Indeed, post-gain is applied to the signal after the filtering. So if it's clipping, the post-gains will not help as they're coming behind.
Anyway, you end with a global processing that's rather simple and looks like this:
Why this is clipping, I don't really know.
The first thing you should do, that's monitor the vu-meters in NUC. The top segment (labelled OVF) will light on when the corresponding channel is clipping.
The 'Output Clip' led is the ORed combination of all OVF segments, i.e. the led is lighting on when any of the OVF segments is on. It tells you that output is clipping somewhere, but the OVF segment tells you precisely for which output this is happening.
For these clipping channels that you've identified, you should add a bit more headroom. Do that in the pre-gain section.
Still I'm surprised: this configuration should not clip.
I got your setup file.
So you've set a global headroom of 6 dB in the input processing section.
In the channel processing section, you're using only 4 channels and have defined a 2-way LR4. You're also attenuating all outputs by 3dB after the processing (i.e. in the post-gain section).
It's not clear to me why you're having post-gains. If your intent is to add more headroom and prevent clipping, then you should move these gains to the pre-gain stage. Indeed, post-gain is applied to the signal after the filtering. So if it's clipping, the post-gains will not help as they're coming behind.
Anyway, you end with a global processing that's rather simple and looks like this:
Why this is clipping, I don't really know.
The first thing you should do, that's monitor the vu-meters in NUC. The top segment (labelled OVF) will light on when the corresponding channel is clipping.
The 'Output Clip' led is the ORed combination of all OVF segments, i.e. the led is lighting on when any of the OVF segments is on. It tells you that output is clipping somewhere, but the OVF segment tells you precisely for which output this is happening.
For these clipping channels that you've identified, you should add a bit more headroom. Do that in the pre-gain section.
Still I'm surprised: this configuration should not clip.
Hi Nick,
Things going from bad to worse at my end. Just disconnected my display to test the cables for problems (my display issue you recall). In the mean time I turned on the Najda but controls stopped working, so I connected to NUC on the laptop. No sound whatever I did. Plugged the display back in and realised the on/off button also would not work. Also, cannot connect to NUC at all now. Noticed the above post and tried updating the software, but since I cannot connect I am therefore unable to update the firmware. Tried unplugging the unit at the mains but of course the on/off button (all buttons) does not work so I cannot even turn it on.
Help!
Nick,
I have now changed my setup completely:-
Inputs down by 3db.
Ch1/2 LR4 at 20 and 55 hz - runs the tapped horn
Ch3/4 no crossover at all.
No gain changes at all.
Analogue Outs set to global 3.0Vrms.
This has very little processing of the signal and for the full range Lowthers the signal should pass through with little or no change.
This setup does not seem to clip, the bass is a little high but that I will reduce slightly.
So I will leave it at that for the time being and build some bigger tapped horns.
Thanks
Alan
I have now changed my setup completely:-
Inputs down by 3db.
Ch1/2 LR4 at 20 and 55 hz - runs the tapped horn
Ch3/4 no crossover at all.
No gain changes at all.
Analogue Outs set to global 3.0Vrms.
This has very little processing of the signal and for the full range Lowthers the signal should pass through with little or no change.
This setup does not seem to clip, the bass is a little high but that I will reduce slightly.
So I will leave it at that for the time being and build some bigger tapped horns.
Thanks
Alan
Same, got a very good experience from Modushop, although delivery quiete a bit longer than planned. Can you explain the mod you made ?
Tks
Nobody knows the answer to my question regarding the connection between anodized chassis panels?
Also, one question that surely fits in this thread: Nick: What kind of plug did you have in mind for the remote connector? (I did search the thread, but I suppose the search term is too generic).
Also, one question that surely fits in this thread: Nick: What kind of plug did you have in mind for the remote connector? (I did search the thread, but I suppose the search term is too generic).
Thanks Nick,
I can post a pic, but I´m afraid it´s not much help. It´s just a regular chassis, but with anodized panels (anodized surfaces are not conductive). My question is: How important is continuity between each panel that touches, so that the "shield" is complete? Is it enough with one solid connection between each chassis part, or is it necessary with a "continuous connection" along the edges of each panel?
Anyway, here are a few photos of the build so far. I have yet to populate the S/E to BAL boards (detailed here), install power supplies properly, integrate the mux for multiple SPDIF inputs etc. Moving very slowly, as I have a working nanoDigi setup right now, and I want to get everything right.
The display on the right shows the led status pins - it works fine with polling, but I agree with you that interrupts are more proper. Using multiple pins with interrupts on the Arduino requires a bit more effort on the coding side, so I´ll see if it´s necessary to do so. However, I don´t think it will be a problem that the LEDs won´t show exactly the Arduino is processing something else, as the Arduino won´t process anything that takes more than a a split second anyway - but we´ll see.
I can post a pic, but I´m afraid it´s not much help. It´s just a regular chassis, but with anodized panels (anodized surfaces are not conductive). My question is: How important is continuity between each panel that touches, so that the "shield" is complete? Is it enough with one solid connection between each chassis part, or is it necessary with a "continuous connection" along the edges of each panel?
Anyway, here are a few photos of the build so far. I have yet to populate the S/E to BAL boards (detailed here), install power supplies properly, integrate the mux for multiple SPDIF inputs etc. Moving very slowly, as I have a working nanoDigi setup right now, and I want to get everything right.
The display on the right shows the led status pins - it works fine with polling, but I agree with you that interrupts are more proper. Using multiple pins with interrupts on the Arduino requires a bit more effort on the coding side, so I´ll see if it´s necessary to do so. However, I don´t think it will be a problem that the LEDs won´t show exactly the Arduino is processing something else, as the Arduino won´t process anything that takes more than a a split second anyway - but we´ll see.
Attachments
