Check it out. I still live on Skyline Drive, the beginning of the track! The road is in terrible shape now and way to much growth of brush around the road to see what is coming anymore. I still push it a bit but don't sweep past people in the turns like I did in the day. Never went over the side but bottomed the suspension in the middle of a turn once and spun it around and hit the edge. That Camaro in the picture, the orange one is about 6" narrower than mine was, full panels from front to back, not added on flares. 302 Z'28 that could turn 8,000 rpm and was not even close to stock. Waited six month for the custom sway bars front and rear to fit that car. And I would question that guy's king of the hill status! My friend would bring his 1967 Can-Am aluminum motored 427 Vette up to the hill once in awhile, everyone thought they had the fastest car.
http://jalopnik.com/5513595/the-good-ol-days-of-mulholland-racing
http://jalopnik.com/5513595/the-good-ol-days-of-mulholland-racing
Thank you Scott
Have you-or anyone else here- experimented with inverse digital filtering (DSP convolution) with the impulse response of loudspeakers?
Re Barcelona food options:
I still remember the joy of tasting the various 'Jamón serrano' and 'jamón ibérico' there.
George
Do you mean de-convolution? Normally that's considered the inverse filter. That process has a large sensitivity to noise and there is of course a huge literature base on the subject.
Yes, but clarification needed. I tried two things:
1) room response lf eq, impulse response, convolver
2) current drive, convolution with impedance impulse response.
That’s the problem Tom. I haven’t got it yet.
That’s correct Scott. De-convolution.
From what I am able to understand, the sensitivity to noise you mention, is due to the environmental noise present in the acoustic recording of the loudspeaker’s IR.
The literature is huge but this is written for normal people. I am looking for an ‘Idiot’s Application Guide to De-convolution’
Pavel
For No1 you applied a sound impulse in the room, you recorded with a mic the response of the room to this impulse and then what did you do with this recording?
For No2 did you record the electrical current (Iim)at the speaker terminals while applying an impulse voltage Vim at the driving amplifier (with gain A) input? Did you then use the Iim to alter the flat gain A? How?
George
Sorry, that would be tough especially since acoustical noise has patterns. I'm more familiar with optics where the noise is mostly just noise. There are some amazing demos around on removing blur from moving objects, like police reading license plates off of photos.
Do you have any street racers, now, yourself? I spend more time and money on cars than audio by far. Started with my first car - a 57 Chevy which i put a Sprint Car engine in it. At the moment working on engines (same in each car) and suspensions for CTS-V and ZR1. Recently picked up a W12 twin turbo Bentley which could use another 100HP/TRQ. Looking for sources.
-RM
Last edited:
That’s the problem Tom. I haven’t got it yet.
George
Hi Gapag
Well the fast and easy way is the Mini-dsp drc unit (a stand alone convolver).
OpenDRC-AN | MiniDSP
Use a program like ARTA to measure the magnitude and phase (you should average several measurements), take that measurement and put it in Re-phase.
Use Re-phase to eq the magnitude and phase to what you want and save the correction impulse.
Put the impulse into the miniDSP software and send it to the dsp and your done (but read the tutorial haha)
Re-phase;
rePhase | SourceForge.net
The problem is the temptation is to try to fix too much and unlike electronics where you can stick your probe and look at anything you want and “that is the signal”, what you measure with a loudspeaker with a microphone changes if you move it.
It takes a rare and VERY uniformly radiating loudspeaker and an absence of reflections to have a loudspeaker even start to resemble a ‘wire” with A signal as opposed to a myriad of paths lengths and radiation sources that is a normal loudspeaker.
Since you are limited to applying a correction to the wires only, not the physical arrangement of sources, the hard part is knowing what part of what you see is universally addressable electrically and what “ measurement wart” goes away if you move to the right a foot. The safe path is only addressing lower frequencies at first.
You will have the best results if you take the loudspeaker measurements for this outdoors or in a big room, well off the ground all to minimized reflected sound even if you have a gated system.
You can go on to "fix"t the room but the fix only applies in that one spot. As a result, if your listening area is three seats wide, you want to measure across that span and average and not try to fix the high warts.
An "automatic" software which might be fun is REW.
REW - Room EQ Wizard Room Acoustics Software
Hope that helps
Tom
Richard,
I have a 1967 Firebird first generation sitting in my garage just waiting for me to do my magic on it. I have 6 large boxes of full body panels sitting for me to cut up and create another monster car. I still have three DZ 302 engine blocks and cranks sitting in storage. Today I'd used Airflow research heads and other modern parts if not just drop in one of the newer crate engines like in the new Corvette, Camaro and Cadillac. Nothing like a modern fuel injected engine with all the electronic controls to have both power and fuel mileage. I added a link to a car similar to what I am going to do but it will be wider than the car shown and I won't copy the front nose. My Camaro makes this look tame, it was really wide. I still have a mold I took off one side of that car which I was going to turn into a full carbon fiber body before the car disappeared! It had running boards on the side to change the aerodynamics as those early bodies had way to much lift at speed. It felt like you were floating at anything over 120 and by 150 was really on the edge without the body modifications. Those bodies had way to much air spilling under the chassis and with spoilers front and rear and the running boards it was a different car completely.
hotrodscustomstuff.com | 1968 Pontiac Firebird
I have a 1967 Firebird first generation sitting in my garage just waiting for me to do my magic on it. I have 6 large boxes of full body panels sitting for me to cut up and create another monster car. I still have three DZ 302 engine blocks and cranks sitting in storage. Today I'd used Airflow research heads and other modern parts if not just drop in one of the newer crate engines like in the new Corvette, Camaro and Cadillac. Nothing like a modern fuel injected engine with all the electronic controls to have both power and fuel mileage. I added a link to a car similar to what I am going to do but it will be wider than the car shown and I won't copy the front nose. My Camaro makes this look tame, it was really wide. I still have a mold I took off one side of that car which I was going to turn into a full carbon fiber body before the car disappeared! It had running boards on the side to change the aerodynamics as those early bodies had way to much lift at speed. It felt like you were floating at anything over 120 and by 150 was really on the edge without the body modifications. Those bodies had way to much air spilling under the chassis and with spoilers front and rear and the running boards it was a different car completely.
hotrodscustomstuff.com | 1968 Pontiac Firebird
That’s the problem Tom. I haven’t got it yet.
That’s correct Scott. De
The literature is huge but this is written for normal people. I am looking for an ‘Idiot’s Application Guide to De-convolution’
George
Apple Inc has a nice Help file on the subject --- Google "Deconvolving Your impulse response --- Apple"
If that fails you... Might look at DADiSP software. Might find info there. I have used it... still have old copy ver. And MLSSA from DRA labs might help as well. Or Linear X soft/hard ware info. These are all T & M stuff and a good mix of practical and theory.
THx-RNMarsh
Last edited:
Richard, you are not alone.
I went perhaps a step further than you by getting together my maginificent 7 and we built a car from the ground up. The only thing we didn't make and really only modified is the Yugo coachwork, i bougt the chassis, the version intended for FIA Group 4 racing. Stiffened with dual plating wherever possible. Engine a FIAT 1.6 litre, orginally fed by a twin carb, now tih a fuel injection Ford reccomends for racing when their factory installed unit cannot be tuned any more. In brief, 142 bhp at 6,800 rpm, soft limiting (no further increase of fuel) 8,200 rpm. hard limiting at 8,500 rpm.
Nothing exceptional, the point was to make a street leagl sleeper/small GT vehicle. Yugo is light, just 820 kg (1,800 lb), and the power to weight ratio allows it to do 0-60 mph in 6.5 sec, not half bad, with a maximum speed of 211 km/h (131 mph), and a breaking distance 60-0 mph in 35 m (app. 77 feet), using Pireli P6000 tyres.
An oddity is that for technical reasons, we had to construct a cascode fuel injection system, standard Bosh 435435 pump to get the fuel from the tank to the engine space, and there an Audi RS8 fuel pump to get the pressure up. It goes like stink, but better yet, its roadholding is literally incredible, for a Yugo, the first next thing after impossible. Anyone with a good ear for engines would love the magnificent sound of the engine revving at 8,000 rpm, that's just a symphony to my ears.
Imagine the poor bugger who looks down on me, only to find that I can slingshot from standstill way faster than his diesel SUV. Priceless! I bought a panoramic rear view mirror just to watch their faces for another 3 seconds extra.
I went perhaps a step further than you by getting together my maginificent 7 and we built a car from the ground up. The only thing we didn't make and really only modified is the Yugo coachwork, i bougt the chassis, the version intended for FIA Group 4 racing. Stiffened with dual plating wherever possible. Engine a FIAT 1.6 litre, orginally fed by a twin carb, now tih a fuel injection Ford reccomends for racing when their factory installed unit cannot be tuned any more. In brief, 142 bhp at 6,800 rpm, soft limiting (no further increase of fuel) 8,200 rpm. hard limiting at 8,500 rpm.
Nothing exceptional, the point was to make a street leagl sleeper/small GT vehicle. Yugo is light, just 820 kg (1,800 lb), and the power to weight ratio allows it to do 0-60 mph in 6.5 sec, not half bad, with a maximum speed of 211 km/h (131 mph), and a breaking distance 60-0 mph in 35 m (app. 77 feet), using Pireli P6000 tyres.
An oddity is that for technical reasons, we had to construct a cascode fuel injection system, standard Bosh 435435 pump to get the fuel from the tank to the engine space, and there an Audi RS8 fuel pump to get the pressure up. It goes like stink, but better yet, its roadholding is literally incredible, for a Yugo, the first next thing after impossible. Anyone with a good ear for engines would love the magnificent sound of the engine revving at 8,000 rpm, that's just a symphony to my ears.
Imagine the poor bugger who looks down on me, only to find that I can slingshot from standstill way faster than his diesel SUV. Priceless! I bought a panoramic rear view mirror just to watch their faces for another 3 seconds extra.
Last edited:
Sorry George I think I put words in your mouth. De-convolution would be to record a signal and recover the original knowing the impulse response of the speaker. This amounts to division in the Fourier domain and small errors can get large. In optics the FFT of the aperture has zeros so noise there blows up.
The opposite is the case when pre-applying a correction a .5dB error in the FFT of your measured impulse response is just that a .5dB error in the output. As others said you don't want reflections, etc. when you measure it in the first place.
Last edited:
Dynamic range has been well studied but do we know what the standards say? The AES and International standards should apply as a minimum for performance. Note that the standards are measured results and not mfr specmanship nor theoretical performance as we fall far short of that.
View attachment 1 dynamic range requirments.pdf
Noise is measured with a specific tone applied and then tone only removed at dut output. This way all noise + artifacts are measured.
Here are some International standards (IASA) ---
Dynamic range of not less than -115dB UNweighted (20-20KHz).
THD+N measured at -1dB FS.... less than -105dB UNweighted.
THx-RNMarsh
View attachment 1 dynamic range requirments.pdf
Noise is measured with a specific tone applied and then tone only removed at dut output. This way all noise + artifacts are measured.
Here are some International standards (IASA) ---
Dynamic range of not less than -115dB UNweighted (20-20KHz).
THD+N measured at -1dB FS.... less than -105dB UNweighted.
THx-RNMarsh
Last edited:
Could you please explain where some of these measures come from, if I play a 1 16bit LSB random noise signal with the volume set for loud but still sensible listening, I can't hear anything. 118dB is silly (depending on your definition of DNR).
So I wish you had included the precise definitions from these articles there is a big difference between spot noise and integrated noise over the whole BW.
Last edited:
I just finished reading the Ampex paper cited in the Wikipedia entry you cited.
And I have the same criticism of their figure as I had of yours. It's just a numbers game and has absolutely nothing to do with the context of ACTUALLY LISTENING TO MUSIC.
They simply take the just-detectable level of noise in a quiet environment (4dB SPL) and then the peak level of unamplified music (122 dB SPL), subtract 4dB from that and come up with a dynamic range of 118dB.
But as I've already said, you're never going to be able to hear that 4dB of noise AND the music with the 122dB peaks AT THE SAME TIME.
se
- Status
- Not open for further replies.