Hi Josip,
Glad to read you solved the issue, well done
- and thanks for letting us know how you fixed it!
I have news too. I dropped the expansion board PCBs and parts at the assembly plant yesterday.
Today, I received the front board panels
The 2-euro coin gives you an idea of the size of the PCB.
I'll post later a picture with all parts assembled.
Cheers
Nick
Glad to read you solved the issue, well done
- and thanks for letting us know how you fixed it!I have news too. I dropped the expansion board PCBs and parts at the assembly plant yesterday.
Today, I received the front board panels
The 2-euro coin gives you an idea of the size of the PCB.
I'll post later a picture with all parts assembled.
Cheers
Nick
Hi Please don't flame me on my first post!
I know #$%# all about programming dsp and love what you are doing here.
I have a simple question and do not want to start a debate on this topic
I read and reread the posts about "phase" tried to work out if it was possible to have the phase at the crossover point coherent. All I could find is links to psycho acoustics and flat phase stuff. If my speakers are moving in sync when they are making the same frequency I'll be happy.
Thanks
Jeffrey
I know #$%# all about programming dsp and love what you are doing here.
I have a simple question and do not want to start a debate on this topic
I read and reread the posts about "phase" tried to work out if it was possible to have the phase at the crossover point coherent. All I could find is links to psycho acoustics and flat phase stuff. If my speakers are moving in sync when they are making the same frequency I'll be happy.
Thanks
Jeffrey
Not yet, I'll post the drawing on the website this week.
Below is a picture of an assembled control board. Mounting screws between the display and the control board are missing, they are on order (M2.5). In the foreground, the supplied knob.
No worries, Jeffrey. I don't understand your question though
Maybe another forumer will.
If I understand correctly, you want to know if Najda lets you design phase coherent/integrated crossovers.
Well, yes. But this all depends on the crossovers you design. In this post I will assume you will be working in IIR (Which is the "Standard crossover/filters" mode). There is no individual phase correction or linear phase filtering in IIR mode, so you have to be careful with phase alignment. This is not specific to Najda, this goes for ALL kinds of analogue and IIR crossover design.
For drivers to be phase aligned at crossover, you can go down many paths. Different crossover steepnesses require different kinds of care to be taken.
IIR filtering works with the exact same phase/transfer functions as analogue/passive filters. That means that each 6dB/oct of IIR filtering produces a 45º phase shift at crossover point, exactly like the analogue equivalent. Additional steepnesses (12/18/24/48) are essentially many 6dB filters cascaded on top of eachother, so that a 12dB/Oct filter creates 90º phase shift, just as a 24dB/Oct makes for 180º.
A low pass filter will create a phase shift inverse to what an equivalent high pass filter would do. That means that if you cross two drivers at 12dB/Oct, say a midrange driver and a tweeter, you have a midrange lowpass filter that shifts 90º to the left, and a tweeter highpass that shifts 90º to the right. Comparing these, your drivers are now perfectly OUT of phase at crossover point (180º apart from eachother), meaning they will move inversely to eachother. The easiest way to work around this is by inverting the polarity of one of the channels (Say the highs) by using the Invert Phase function in NUC or by simply inverting the +/- on your drivers as is done with most passive crossovers.
Taking this logic further, for example with a textbook 24db/Oct filter, you will see that each filter (both the midrange lowpass and the tweeter highpass) now have a 180º phase shift each, producing a total of 360º, meaning that their phase shifts line up perfectly. Thus you do not need to invert your driver polarity for a regular 24dB crossover.
Now, it's important to note that text-book crossovers like these are not always (rarely, actually) the answer in the real world, and that there are things that are more important than perfect phase integration. Often you will see people crossing drivers using for example 12dB for midrange lowpass, and 24dB for tweeter highpass. This works well, because even though the drivers are technically out of phase by 90º that's not enough for the drivers to be cancelling each other out, and often has other benefits to the overall sonic performance of the crossover. There are many common variations.
Two other things you need to remember are:
- EQ. IIR EQing creates phase shifts in the same fashion as the crossovers. If you need to do EQing close to the crossover point, you might want to do that in the Input section of the Najda instead of the channel outputs, so that both sides of the crossover gets the exact same phase skews applied.
- Time alignment. When it comes to phase integration, this is perhaps the most important point of them all. No matter how perfect your phase integration is on the DSP side, it won't matter unless your drivers are not aligned perfectly distance wise. For example, a common problem is that when the tweeter and midrange are mounted on the same baffle, the center of the midrange speakers is usually further away from your ear than the center of the tweeter. This is why you often see tweeters recessed or speaker cabinets tilted backwards in high end commercial speakers. With a digital crossover such as the Najda you can compensate for this by simply delaying the tweter's signal by the time equivalent of 1-2cm, relieving you of having to make clumsy speaker design decisions.
so, TLDR: Yes, you can make perfectly phase integrated crossovers with the Najda, but as with everything else it requires careful design and consideration.
(DISCLAIMER: Ramblings above are based on my understanding of the subject and are not guaranteed to be correct. Feel free to correct me.)
Last edited:
It is my understanding that - prior to any crossover implementation - you should also individually eq the drivers flat beyond where they will cross over. This means that when you apply the crossover filtering you know that the phase (and frequency) response in the crossover region is (entirely) down to the crossover response and not a sum of the crossover and driver responses.
IIR filters are minimum phase and naturally compensate for the drive units which are also minimum phase
Disclaimers as above...
IIR filters are minimum phase and naturally compensate for the drive units which are also minimum phase
Disclaimers as above...
Hi Mike,
I've just uploaded the front board documents here. That should be sufficient for a front panel project.
Cheers
Nick
Nick
is it possible to overload I2S inputs ?
I'am using rPi as input and sometimes my (output) bars go to red (bas/mids/highs)...digital clipping...input is always high but don't go to red (I think)
for now I attenuate input for like -2db
It all depends on recordings...some is good but on some I have some stupid problems of peaking to red...
I read somewhere that people have solder something like 22ohm on all three I2S lines (not related to Najda but general I2S connection) ?
is it possible to overload I2S inputs ?
I'am using rPi as input and sometimes my (output) bars go to red (bas/mids/highs)...digital clipping...input is always high but don't go to red (I think)
for now I attenuate input for like -2db
It all depends on recordings...some is good but on some I have some stupid problems of peaking to red...
I read somewhere that people have solder something like 22ohm on all three I2S lines (not related to Najda but general I2S connection) ?
Last edited:
Oh yeah. When you connect, all outputs are muted, that's a safety feature. That's also documented in the first chapter
To unmute, 3 possible ways to do that:
- Press spacebar
- Click the 'Unmute' button
- Menu-> Settings-> Unmute
Never mind. It worked today... Yesterday it didn´t react to the unmute button in NUC, even though I was connected to the board.
No, there's no clipping possible in a digital transmission. Digital material is usually normalized, this means that the level will hit the 0dBFS. What you need is a bit more headroom. You have 2 dB for now, try to slightly increase that figure, that should solve the problem.
As for the termination resistors, leave that out for now. What's more important is that you have short I2S lines and a nice ground return path for each line.
Hm... Could it be that your amps were off?