• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Impulse Response

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
7,063
Likes
3,374
Location
Riverview, FL
#1
Presented for your amusement:


Single Full Scale Bit sent through speakers, playback recorded in Audacity:

upload_2017-7-4_3-9-35.png


Impulse response calculated from a ten second 10-24kHz sweep tone (three weeks ago) in REW:

upload_2017-7-4_3-7-56.png


Well, I am amused.
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
7,063
Likes
3,374
Location
Riverview, FL
#4
Looks like you have a one second delay through the system? Is that due to fir filters/convolution you are using?
The absolute time in the recording has no meaning, though the duration should be accurate.

I just moved the peak to one-second on the display above.

Here are 4 recordings showing what I get to work with.

First track is source. One second silence (record settling time) then the full-scale bit then more silence. Not perfect, but free, and useful.
The recorded data comes "before" the source data in this instance. Sometimes it comes later.

upload_2017-7-4_13-12-1.png

The MiniDSP has 6144 taps, and AcourateDRC centers its calculated impulse around tap 3072, so, 1/48000 * 3072 = 0.064s filter delay (I guess. Or is it 0.128s?). It may be noticeable with video, but just barely.

Here's 400 samples of the filter data:

upload_2017-7-4_13-31-0.png
 
Last edited:

Blumlein 88

Major Contributor
Joined
Feb 23, 2016
Messages
6,482
Likes
6,204
#5
You act as if there is some connection between math and physics and this really works.
 

DonH56

Technical Expert
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
2,982
Likes
3,539
Location
Monument, CO
#6
I hate reading threads like this. Makes me realize how inadequate my impulse responses are. I have solved that problem, however; I no longer measure them. :)

The other interesting thing, given these are really outstanding responses, is that little tiny response blip at about 1 ms. It is about 5% if I read it rightly, and looks totally negligible. However, 5% is -26 dB, and yet we worry about DAC artifacts 100+ dB down. Hmph.

Oh, the delay (latency) is normally from initial input to first output at the other end, so 128 us. "Centering" the response at the center tap allows the filter to manipulate the output using pre- and post-emphasis for correction (acting on samples before and after the main cursor/signal).
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
7,063
Likes
3,374
Location
Riverview, FL
#7
You act as if there is some connection between math and physics and this really works.
I'm somewhat surprised it works as well as it seems to.

Somewhere else, I posted a similar comparison between a calculated Step Response and a recorded 10Hz signal. All but identical...

---

My later work history included plenty of taking the output of the Telco Software/Firmware/Hardware Design and Development Engineers, thinking to myself "Oh yeah? Well, we'll see about that!", pushing the button, and figuring out what really happened, and whether or not it met the design spec or if we could otherwise slide it past the customer during their acceptance testing.

That mind-set has carried over into my Golden Years.
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
7,063
Likes
3,374
Location
Riverview, FL
#8
Last edited:

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
7,063
Likes
3,374
Location
Riverview, FL
#9

tomelex

Addicted to Fun and Learning
Patreon Donor
Joined
Feb 29, 2016
Messages
759
Likes
276
Location
So called Midwest, USA
#10
thanks man, interesting stuff.
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
7,063
Likes
3,374
Location
Riverview, FL
#11

restorer-john

Major Contributor
Joined
Mar 1, 2018
Messages
2,350
Likes
3,734
Location
Gold Coast, Queensland, Australia
#12
Basically, your single impulse was generated digitally and sent through what type of D/A?

If it has an FIR oversampling filter (likely), it has become something very interesting by the time it has been through the amplifier, loudspeaker, microphone and A/D...

Here is a single impulse, 1 sample 44/16 at 0dBFS on the world's first CD player, the Sony CDP-101: (500us/div)

RIGOL Print Screen23-02-2018 12_15_20 PM.842.jpeg


This is what a non-oversampled, analogue 9th order Tchebyscheff (brickwall) filter looks like on one of my actual CDP-101s (tested the other day actually). Looks rather familiar?

Just for fun, here's the 100Hz square wave response:

RIGOL Print Screen23-02-2018 12_12_47 PM.672.jpeg


Here's a typical impulse from a OS player with excellent D/As and a superb measured performance (Marantz PMD-325 professional)- note scales are not the same so it looks better than it is. (1ms/div)

RIGOL Print Screen23-02-2018 10_49_37 AM.948.jpeg


Zoomed just a little:

RIGOL Print Screen23-02-2018 10_49_58 AM.403.jpeg


So, sending an impulse that looks like the one above through amp/speakers/microphone/A-D results in the top trace.

I put it to you all, what have we gained in 35 years? :)
 

Wombat

Major Contributor
Joined
Nov 5, 2017
Messages
3,861
Likes
2,159
Location
Australia
#13
Focus too much on minutiae and its relevance to the bigger picture can be overlooked.
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
7,063
Likes
3,374
Location
Riverview, FL
#14
Basically, your single impulse was generated digitally and sent through what type of D/A?
One full-scale bit created in Audacity sent through a Benchmark DAC2.

The DAC has a symmetrical impulse response (what is that, linear filter?)

If it has an FIR oversampling filter (likely), it has become something very interesting by the time it has been through the amplifier, loudspeaker, microphone and A/D...
I don't measure much electrical here.

Most of what I present as "measurements" is what a mic at the listening position picks up. I want the "in air" to correspond to the "recorded signal". What happens in the middle - Room Correction IIR and FIR filters modifying the digits thereby modifying the electrical signal sent to the speakers is of less interest to me.

Reasons:
Can't hear electrons
Don't have a digital scope or high-high-quality ADC
Do have a 25 foot cable to preamp output to PC but it is noisy
DSP Room Correction (niniDSP OpenDRC-DI and AcourateDRC) whacks out the electrical frequency response and phase from pristine textbook flat, so electrical signals become "don't care".

Preamp Output Frequency Response examples (electrical) with and without DRC in the path (with a 25 foot cable to PC) for MartinLogan:

upload_2018-3-14_13-11-2.png


Signal Phase:

upload_2018-3-14_13-13-4.png


For JBL LSR 308:

upload_2018-3-14_13-14-45.png


Phase:

upload_2018-3-14_13-14-20.png



I put it to you all, what have we gained in 35 years?
The ability (at home) to modify music on-the-fly noiselessly to reduce audible problems presented by the speakers and the room at low to no incremental cost with all the digital toys created for us to play with.

I couldn't do that at home in 1983.
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
7,063
Likes
3,374
Location
Riverview, FL
#15
Here's some electrical square wave - around 200hz in the sweep:

25 foot noisy cable to PC, some amplification applied to enlarge the received signal (makes it look more noisy)

upload_2018-3-14_13-56-37.png
 
Top Bottom