Overshooting and pre-ringing artifact

May 11th, 2009

One of the problems of digital crossover based on FFT FIR filters is overshooting and pre/post-ringing artifacts (a.k.a pre-ringing effect). this is strongly connected to the Gibbs phenomenon (see below). the effect can be minimized by using different methods to create the filters, by using longer filters or by not using FIR filters :) you can try IIR filters or wavelet FIR filters instead.

Fig 1. Illustration of the Gibbs effect. image taken from mathworld

The sharp peaks depicted in fig 1. are the overshoot, the oscillation before and after are the corresponding pre/post ringing.

Gibbs phenomenon, From Wikipedia, the free encyclopedia

In mathematics, the Gibbs phenomenon, named after the American physicist J. Willard Gibbs, is the peculiar manner in which the Fourier series of a piecewise continuously differentiable periodic function f behaves at a jump discontinuity: the nth partial sum of the Fourier series has large oscillations near the jump, which might increase the maximum of the partial sum above that of the function itself. The overshoot does not die out as the frequency increases, but approaches a finite limit.[1]

Good morning!

May 10th, 2009

What a great morning! I woke up, turned the system. first the squeeze box, then the amplifiers, and during the spare second before music started playing I heard something dirty… paused the music and stepped closer to the speakers: tweeter sounds fine, high-mid and low-mid are also fine, the woofer is not fine! it’s hissing terribly. the woofer units are powered by the old Kenwood KA-6100 vintage amp. the volume knob is turned around ~65% of the amps power to match the gain with the other amps. lowering the gain to mid noon ~50% reduced hiss back to normal levels. I also lowered the gain of the Sherwood (high-mid) and the Harman-Kardon (low-mid and tweeter) via BruteFIR’s digital attenuation (in it’s filter coefficient file), and the result is astonishing! sound is much more precise, bass has more punch, more accuracy, more speed. played Beethoven’s 5′th symphony and Avishay Cohen. this is wesome!

My next project should defiantly include amps, Thinking about LM3875 Gainclones from audio sector powered by PowerRegs (designed by Teddy Prado).

Madisound pissed me off

May 8th, 2009

I wanted to order 3 pairs of the RB-Kit from madisound. for 59$ you get a great value for money kit. I mailed them on the 28/04/2009, asking about shipping cost. they quickly replied 1 day later. mailed them again on the 1/05/2009 asking a few more questions, it took them 5 days to answer… and it was too late because they run out of stock.

I planned one set for the kitchen in-walls and 2 pairs for the front/rear speakers of my HT system. now I need to find something else.

too bad.

Kitchen in-wall loudspeakers

May 1st, 2009

Thinking about placing in-wall/ceiling loudspeakers in the kitchen. Music have always been a part of coocking for me. I need to decide where to place the whole system. I think that the Squeeze box will be the source, perhaps a T-Amp as an amp… havn’t decided on the speakers. probably somehing brandless. before I continue I need to consider all placement posibilities.

Using Matlab to calculate FIR filters coefficients

April 29th, 2009

I wrote a very simple Matlab function for calculation FIR filter coefficients.

I hope it can be of use to somebody.

good news and (not so) bad news

April 29th, 2009

Good new first: I’m finally working with all 4 drivers (until now I worked w/o high mid) this thanks to Anatoli that gave me an extra amp (Sansui RZ-9500).

(not so) Bad news: (1) the system now sounds horrible(*). I need to adjust the gain, gain adjusting for 4 channels is a pain. updates soon. (2) one of the high mids is humming. this can be from any of the following reasons: my previous cat dropped it (while it was still in the box) from a 2m shelf; my son pressed the soft dome really hard (twice); perhaps the amp pops finally made some damages?; all of the above.

(*1) horrible: stage not clear, no micro details and noise ground is too high. whats wrong? I can think of several reasons: (1) fix gain and time alignment; (2) use better regulators for DAC’s; (3) use better FIR filters. currently i’m using Gaussian windowed 16K taps filters. perhaps there is a pre-ringing artifact in the high-mid and the tweeter.

2 days later:
(*1) horrible? no, not bad actually: I replaced the Fc’s from 150-1200-3500 to 200-1500-4500 and everything is much better now, the image is much wider, the high mid is not humming any more (perhaps it needed break-in?). it’s not perfect, but it’s also not as horrible as i suspected. :). no doubt about it, further exploration of parameter space should be conducted.
Also, I’ve tested two filter tap sizes: 1k VS 16k. could not hear the difference in a short test I did. more on this later-on.

Reference earphones for loudspeaker builders

April 28th, 2009

Linkwitz is talking about reference earphones for loudspeakers evaluation.

My new HTPC Case: Antec Veris

April 28th, 2009

I previously owned a 7″ touchscreen HTPC case, after it was sold I bought the Antec Veris case (a.k.a. FUSION 430*)

The case is very quiet and built nicely. it comes with the famous anti-vibrations-rubber-nipples for the hard drive, it has 2 12cm fans pushing air out on it’s right side. the psu is isolated in it’s onw chamber. detailed review can be found on silent pc review.

Whats inside?

AMD Atnlon 64 5000+ cooled by a zalman 7500.
Gigabyte mobo, 780G chipset.
80G sata hard-drive.

Performance:

Plays 1080p with not special configration (via VLC).

Why did I select this case over the others?

Demands: I looked for a mid-price case with silence on my mind. I do not case for more buttons on the front panel. I do care that it will be as small as possible (especially in the depth dimension) and could accommodate full size expression cards. I also don’t mind m.atx form factor limitation, since all good HTPC mobos have a m.atx version Candidates: my first candidate was the Antec nsk-2400, but it looks too cheap for a living room. Second option was the Themaltake DH-101. it’s fairly large,  it can accommodate a full atx board, but for the price of 6″/15.3cm tall. it’s also 17″/43.5cm deep, which is not much but still 1″ deeper than the Antec. it also doesn’t come with a PSU, which make it less of a bargain. the case comes in two flavors, one with VFD and buttons and one without. I prefer the naked version. Another candidate was the NXZT Duet.in short: too deep. There are other, but these are the ones that i was interested in that time (~6 months ago). Recently I’ve seen a new case from Silver-Stone, the GD02. the thing that cough my ey was the fact that this case depth is merely 14.1″/36cm! it will be a good candidate next time I shop for an HTPC case.
Note: I took into considerations some more issues didn’t talk about, like: exhaust fan size. position of PSU, anti-vibration mechanism etc. when selecting your own HTPC case all of these and more should be carefully examined to find the best suitable case for you. 

(*) Note: the Antec Veris is the 3rd generation of the fusion. the 1st came with a 2×16 vfd-display on the silver case and 2×16 lcd-display on the black case. The 2nd generation came with an iMON lcd-display that got several people very angry (read here and here). Then the 3rd generation came renaming the Fusion to Veris and returning the good old 2×16 vfd/lcd display (silver/black).

 

 

DIY interconnects cables

April 28th, 2009

Decided it’s time to make me a new set of interconnect cables. before I make the realy expensive cables, I decided to try and shave on some low-end connectors + an RG59 cable. I need an RCA-Phono cable, since my RME dac’s have PHONO outputs.

 First we skin the cable about 10 mm, it depends on your plug:

 

Now, use shrink or tape to cover 60% of the skinned wire:

And we cover the bare wire with solder

soldring paste allways help people like me with no soldring skill what so ever

Strip off the non-soldered part and attach the RCA plug:

And do about the same with the PHONO plug:

Note: the ground of the two cables is not shortened, only one ground is connected to the plugs chasis

And this is the result:

one Phono-RCA interconnect cable, 1.5m long

Great, now it’s time for coffee :)

The PHONO plugs are extremely lousy and I’m very sorry I bought them. one of them got broken inside the DAC’s female plug. The RCA plugs are not that bad considering their low cost (2.3$ pair). The RG59 cable was found in a dumpster a few years ago. This method of making is based on Assaf’s manual (Hebrew) and an advice from a fried (regarding the ground trick, see note on Phono plug picture). I made 3 cables and it took me ~2 hours including cleaning and coffee.

bad news and good news

April 27th, 2009

Lets begin with the bad news, I’m an idiot. I had (as in past, meaning I no longer have) a working digital filter based on a via-c3 processor (that i traded with Alon for another PC) and an RME Hammerfall light sound card. It all got toasted. why? because I dreamed about a DIY low profile case, so I tested a riser card on the actual system, it was faulty and it burned the mobo+RME card altogether.
Now lets move on to the good new, I bought a new RME card on ebay (for half the price i payed for the old one), i got me a new file server and assigned the old one for the digital filter spot (AMD 3200+, ideal for low power apps) and the setup is working again.
Hooray!

This news is fashiobly late by ~2 moths, but who counts?