FIR And FIlter Stuff

Subjects that don't have their own home
chris661
Shed dweller
Posts: 2556
Joined: Mon Feb 22, 2010 7:29 am
Location: Sheffield

#16 Re: FIR And FIlter Stuff

Post by chris661 »

vinylnvalves wrote: Wed May 27, 2020 7:31 pm What I am suggesting is I don’t think the DSP will add anything if done right. The source will add, I have a rule for vinyl, if it was recorded on an analogue system/mixing desk - I would buy it. If it was recorded and mixed in a modern studio then it’s digital, so keep it digital.
I like this approach - it makes a lot of sense.


Mark, fair enough regarding re-jigging a large portion of the system. I'm hoping that dropping the Powersoft amps in there (they could feasibly be fed from the outputs of a favourite valve amp w/dummy load - 10V RMS is the maximum input level) will give you an idea of what can be done.

Chris
vinylnvalves
Old Hand
Posts: 1074
Joined: Sat Jan 26, 2013 12:08 pm

#17 Re: FIR And FIlter Stuff

Post by vinylnvalves »

Finally I have confidence in REW, I can use any of the 3 or 4 measurements tools in my arsenal that give the same answer, for SPL at least. The phase however is confusing me, here is the SPL and phase plots for the midrange AMT. This is the raw driver with its protection capacitor. The waveguide is given a bit of unwanted boost at the bottom end, which is normally removed by the LR4 @300 Hz.
2180BA4B-1EC0-4ED5-800B-C6F09501B818.jpeg
. I don’t quite understand why the phase is looping like it is ?
vinylnvalves
Old Hand
Posts: 1074
Joined: Sat Jan 26, 2013 12:08 pm

#18 Re: FIR And FIlter Stuff

Post by vinylnvalves »

Right I now understand what is causing all the rapid phase changes in the response. It’s the open baffle.... obviously the reflections screw up the measurements.... but do they screw up the sound? I have about 1m behind the midrange driver. I have old pillows - I thought to sort out the back wave. So I fashioned up an enclosure with just a cardboard box and lightly stuffed it - ducktaping it to the back of the horn.
47AF9035-A181-488E-B080-500EE6C00A6C.jpeg
. If it wasn’t for the protection cap - the phase would be almost level across the useable freq range. I have measured the little satellite AMT dipole super tweeter which kick in around 5 k. They have the combed freq response too, even when I dragged them into the middle of the room. So the big question for me, Is the dipole screwing up the phase - or do you average it smooth and level the mean. As if the reverb is as bad as the measurements suggest - open baffle shouldn’t sound coherent.
Never had any of these problems, with the horns using CD’s... everydays a school day... or more like a work day, where we chase standing waves an HOM’s around jet engine bypasses, to ensure we don’t get wave speed coincidence- which breaks things.
vinylnvalves
Old Hand
Posts: 1074
Joined: Sat Jan 26, 2013 12:08 pm

#19 Re: FIR And FIlter Stuff

Post by vinylnvalves »

I have started to play with FIR’s I am currently not thinking about the maths... just using Rephase :D . I can unstuck yesterday evening when I realised the Nadja needs an equation on the input .. otherwise no signal gets through to the channels. Currently I have written simple LR linear phase crossovers, for trial. For the input equation which is basically a pass through equation I generated an equation that does nothing, giving it an arbitrary 100 nodes. There must be a better way of defining pass through, but my brain currently cannot cope with generating the maths. Any ideas/experience? The Nadja was hitting DSP overload on the simple crossovers I defined, I may have added too many nodes, as no ringing was detected on measurement. Sound not bad - something’s better something worse from a SQ perspective, it’s the 48k sampling probably. I know I prefer 96k over 48k and 192k with IIR filters.

The fun may begin next week... I have a second Nadja I have acquired - primarily as a spare, but I need to work out how I can synchronise the clocks on each board, so I can run one for each channel, doubling my fir capability. Still woeful though. I have seen that you can get programmable effects pedals with 68000 nodes - reasonably cheap, so that could be an approach for those without a Lake or DEQX who don’t want a PC in their music chain.
chris661
Shed dweller
Posts: 2556
Joined: Mon Feb 22, 2010 7:29 am
Location: Sheffield

#20 Re: FIR And FIlter Stuff

Post by chris661 »

Cool, someone's trying it.

Well, you can tell RePhase to make a linear impulse. It'll just be an impulse signal.

Here's a rather good article on digital signal theory: https://web.archive.org/web/20200426194 ... young.html
Worth watching the video linked at the start, too.

For FIR processing with small time delays, high sample rates make sense. The rest of the time, I can't see any problem with 44.1kHz/16-bit for home HiFi.
Remastered/mastered-for-hi-def notwithstanding - if the mastering engineer has done something different, then of course there'll be differences in the sound. YMMV.

Chris
vinylnvalves
Old Hand
Posts: 1074
Joined: Sat Jan 26, 2013 12:08 pm

#21 Re: FIR And FIlter Stuff

Post by vinylnvalves »

The only thing I have noticed is that with the lower sampling rates a stone wall filter in Rephase for HF is about 22Khz, so no signal above that. I know I cannot hear that frequency band, however I do feel that super tweeters add something. I assume this stone wall filter cannot be removed in Rephase. Looking at the maths you can make a low pass filter with a lot less taps than a high pass one, as you don’t care about the ripple at -80db.
I will hopefully get the second Nadja connected in parallel this weekend, so will be able to have mixed IIR and FIR, or FIR’s at different rates for HF and LF.
chris661
Shed dweller
Posts: 2556
Joined: Mon Feb 22, 2010 7:29 am
Location: Sheffield

#22 Re: FIR And FIlter Stuff

Post by chris661 »

The HF filter is what happens when you get towards Nyquist - the article I linked goes into detail on that, and more.

Super-tweeters are often high-efficiency devices implemented with a single cap - they'll be adding non-trivial SPLs well into the audible frequency range.

Chris
Post Reply