Keith_W DSP system

[Keith_W]

New amplifier time. I decided to sell the Vitus solid state amp that was driving the woofers. For a few months I was using a borrowed SGR EL30S amplifier (apologies for the 6moons link) and it was continually going into clip protection. That amp is 150W/ch at 8 Ohms, and 300W/ch at 4 Ohms. So I bought:

A Yamaha PX3. On paper, it looks really good: 300W/ch at 8 Ohms, 500W/ch at 4 Ohms. So approximately double the power of the amp it's replacing.

BUT I did notice rather suspect looking specifications on Yamaha's webpage:

And:

Meaning, they test the amplifier with a 1kHz 20ms burst to obtain a "power output" of 300W, rather than the standard test of driving the amplifier to distortion and reading the power output at the "knee". Meaning, this amplifier has 300 marketing Watts. And 0.1% THD = SINAD 60; 0.3% THD = SINAD 50. More on this later.

The non-standard specifications suggest that Yamaha is trying to pull the wool over my eyes, hoping that i'm not smart enough to notice. That would normally be enough for me to say "screw you, i'll go buy something else". But this amp was available second-hand for very cheap, and ... it's only for the woofer. So I will likely never hear the bad SINAD and it SHOULD have enough power for my woofers, right?

After I got it home and started playing with it, I realised there were a few very nice features:

- it has built-in DSP, but it is rather rudimentary. You can dial in LPF's, HPF's, and a couple of PEQ's. I left the DSP features bypassed, since I will BYO DSP.
- individual input sensitivity controls and gain controls - I LOVE this feature!
- dual analog inputs (1/4" TRS jack and XLR), the amplifier auto-selects the input based on whether it is getting signal or not.
- Routing controls. It can send one analog input to both outputs. Or it can sum two analog inputs into mono and send it to one output. Or both. Or you can bridge both channels into one without having to re-cable. Really cool.

And here are some not-so-great features:

- the fan. It is always on. I put it in my entertainment cabinet and closed the door, but I can still hear it. It's not too noisy, but it's definitely still audible. When the music comes on, the fan noise is masked. There is a Noctua fan mod which I may perform if I decide to keep the amp.
- the speaker binding posts. They suck. They only accept bare wire of fairly thin gauge. No banana plugs, no spade lugs. So I bought some high end audiophile speaker cable for $5/meter from a respectable audio boutique. I tinned one end with solder and fixed spade lugs to the other. It is reasonably secure.

I hooked it up and within an hour, I had dialled it in with the help of some sweeps and I sat down to listen.

Oh dear. My first impression was - yeah, it really is 300 marketing Watts. I don't need a blind test to see the clip indicator come on and the clip protection cut out the music. Fortunately it only does that at very loud volume, so it is actually subjectively putting out more power than the amp it replaced. But there was also a bit more, there was a bit of wooliness as if the bass wasn't right.

I checked to make sure that I did not accidentally leave any DSP settings on, then I pulled out the measurements and took a closer look.

This is the step response (red) and pulled up an old step response of the same speaker for comparison (green). You can easily see that the red measurement has a double peak and the green measurement looks like it should. So this amp adds an extra 1.5ms of latency compared to the old amp.

 

[Keith_W]

After that I had no choice but to redo the time alignment.

Red = new DSP filter with the amplifier time aligned, green = old DSP filter designed for the previous amplifier. It looks reasonably okay now.

While I had my measurement rig set up, I decided to make a few new filters and have a listen.

The graph on the left is full range, the one on the right is zoomed to 10-200Hz. Here are 3 mono measurements (L+R speakers with L+R subs all playing together) with 3 DSP settings:

Blue: subwoofers low-passed at 50Hz, main speakers high-passed at 50Hz.
Yellow: subwoofers low-passed at 80Hz, main speakers equalised to run down to 20Hz with no high-pass filter.
Magenta: no subwoofers, main speakers equalised to run down to 20Hz.

The purpose of running the speakers full range with the sub going up to 80Hz is to have more bass sources in the room so that I can make the FR as flat as possible. Normally, dips in the frequency response in the bass region is due to the direct sound interacting with its reflection (the Allison effect). But if the dip is due to two speakers interacting with each other, then it is potentially possible to adjust the phase of specific frequency bands to get rid of the cancellation. This is the power of FIR filters! I haven't done that yet. I generated all the filters in one sitting and it took me hours.

Despite the apparent similarity of the measured frequency response, there is a huge difference when listening.

The 3-way system (no subwoofers) sounded notably anaemic with no bottom end. The two dips in close proximity to each other is "only" 8dB deep but it spans a frequency range of one octave (35Hz - 70Hz). I haven't calculated how many ERB's wide it is but it should theoretically be easily audible, and listening confirms that. It is a good demonstration of how much good bass contributes to the sensation of spaciousness. After all I did deliberately set up the subwoofers in the two front corners to try to take advantage of Griesinger's stereo bass effect. But what the measurement shows is that the speakers are definitely capable of going down to 20Hz with no subwoofers.

The big surprise was the comparison between speakers with 50Hz high-pass and with no high-pass. As you can see, the blue and yellow measurements look almost the same, in fact you might think that the yellow measurement (with no high-pass) might have a little less bass. Not so! The bass is overwhelming and powerful when the speakers play full range and augmented by subwoofers! I was blown away and shocked by the difference. After all, the FR is almost the same.

The spectros don't really say much about why there is such a large difference between main speakers with/without HPF. As you can see, there is a difference, but it's not huge. The spectro does look noticeably cleaner without the subwoofers though.

Anyway, right now I am not sure whether to keep the amp. I have two options: (1) sell the amp and replace it with its big brother the PX10 or something like a Crown, or (2) keep the amp and buy another PX3. Run both in bridged mode. I will have to think about it.

 

[Holmz]

20ms is a bit short for sustained bass.

 

[Music1969]

Blue: subwoofers low-passed at 50Hz, main speakers high-passed at 50Hz.

your subs struggle with 20Hz output? Relative to other frequencies

 

[Keith_W]

your subs struggle with 20Hz output? Relative to other frequencies

I have a 25Hz subsonic filter applied. Not because the subs can't produce anything below, but because there is an annoying resonance in my house structure somewhere below that. Cutting bass at 25Hz reduces it, but does not eliminate it. I can still hear the walls resonating if I play the subs loud enough.

(EDIT) the other reason for the subsonic filter is that it substantially reduces pre-ringing.

 

[dualazmak]

At least for my preference and in my setup, I do need nice reproduction of around16 Hz.
Please refer to my typical posts:

- Excellent Recording Quality Music Albums/Tracks for Subjective (and Possibly Objective) Test/Check/Tuning of Multichannel Multi-Driver Multi-Way Multi-Amplifier Time-Aligned Active Stereo Audio System and Room Acoustics; at least a Portion and/or One Track being Analyzed by Color Spectrum of Adobe Audition in Common Parameters: [Part-09] Organ Music: #641

- Reproduction and listening/hearing/feeling sensations to 16 Hz (organ) sound with my DSP-based multichannel multi-SP-driver multi-amplifier fully active stereo audio system having big-heavy active L&R sub-woofers: #782

- A nice smooth-jazz album for bass (low Fq) and higher Fq tonality check and tuning: #63(remote thread)

 

[3ll3d00d]

Oh dear. My first impression was - yeah, it really is 300 marketing Watts. I don't need a blind test to see the clip indicator come on and the clip protection cut out the music. Fortunately it only does that at very loud volume, so it is actually subjectively putting out more power than the amp it replaced. But there was also a bit more, there was a bit of wooliness as if the bass wasn't right.

Fwiw this indicates a gain management problem as the clip light coming on is based on the input voltage (if we ignore DSP as it sounds like that's not in use) so clip light on means it's not set to accept your max output (from the dac) voltage cleanly. If I got your DAC model correct, i seems like your DAC can do +18.0 dBu so I would think you want the +14 dBu setting on the amp with a small gain reduction on the frontend of the amp. This assumes you do send a full scale signal (which you might not, e.g. your DSP for the woofer might not peak at full scale or your volume control may be clamped at a lower level) and obviously you still need to get the level right across all the amps (which likely means further adjustments somewhere)

 

[Keith_W]

Fwiw this indicates a gain management problem as the clip light coming on is based on the input voltage (if we ignore DSP as it sounds like that's not in use) so clip light on means it's not set to accept your max output (from the dac) voltage cleanly. If I got your DAC model correct, i seems like your DAC can do +18.0 dBu so I would think you want the +14 dBu setting on the amp with a small gain reduction on the frontend of the amp. This assumes you do send a full scale signal (which you might not, e.g. your DSP for the woofer might not peak at full scale or your volume control may be clamped at a lower level) and obviously you still need to get the level right across all the amps (which likely means further adjustments somewhere)

Hmmmm. You may have a point. I'll go and adjust the settings later and see if this fixes it.

 

[Vuki]

After that I had no choice but to redo the time alignment.

View attachment 465441

Red = new DSP filter with the amplifier time aligned, green = old DSP filter designed for the previous amplifier. It looks reasonably okay now.

While I had my measurement rig set up, I decided to make a few new filters and have a listen.

View attachment 465442View attachment 465446

The graph on the left is full range, the one on the right is zoomed to 10-200Hz. Here are 3 mono measurements (L+R speakers with L+R subs all playing together) with 3 DSP settings:

Blue: subwoofers low-passed at 50Hz, main speakers high-passed at 50Hz.
Yellow: subwoofers low-passed at 80Hz, main speakers equalised to run down to 20Hz with no high-pass filter.
Magenta: no subwoofers, main speakers equalised to run down to 20Hz.

The purpose of running the speakers full range with the sub going up to 80Hz is to have more bass sources in the room so that I can make the FR as flat as possible. Normally, dips in the frequency response in the bass region is due to the direct sound interacting with its reflection (the Allison effect). But if the dip is due to two speakers interacting with each other, then it is potentially possible to adjust the phase of specific frequency bands to get rid of the cancellation. This is the power of FIR filters! I haven't done that yet. I generated all the filters in one sitting and it took me hours.

Despite the apparent similarity of the measured frequency response, there is a huge difference when listening.

The 3-way system (no subwoofers) sounded notably anaemic with no bottom end. The two dips in close proximity to each other is "only" 8dB deep but it spans a frequency range of one octave (35Hz - 70Hz). I haven't calculated how many ERB's wide it is but it should theoretically be easily audible, and listening confirms that. It is a good demonstration of how much good bass contributes to the sensation of spaciousness. After all I did deliberately set up the subwoofers in the two front corners to try to take advantage of Griesinger's stereo bass effect. But what the measurement shows is that the speakers are definitely capable of going down to 20Hz with no subwoofers.

The big surprise was the comparison between speakers with 50Hz high-pass and with no high-pass. As you can see, the blue and yellow measurements look almost the same, in fact you might think that the yellow measurement (with no high-pass) might have a little less bass. Not so! The bass is overwhelming and powerful when the speakers play full range and augmented by subwoofers! I was blown away and shocked by the difference. After all, the FR is almost the same.

View attachment 465453

The spectros don't really say much about why there is such a large difference between main speakers with/without HPF. As you can see, there is a difference, but it's not huge. The spectro does look noticeably cleaner without the subwoofers though.

Anyway, right now I am not sure whether to keep the amp. I have two options: (1) sell the amp and replace it with its big brother the PX10 or something like a Crown, or (2) keep the amp and buy another PX3. Run both in bridged mode. I will have to think about it.

How do decay graphs look? Does the one with subwoofers have longer decay in bass?

 

[Keith_W]

How do decay graphs look? Does the one with subwoofers have longer decay in bass?

I just realized that the animated spectro GIF I put in my post does not work. I'll try again ... hopefully it works this time.

It explains why the 3 way system sounds more bass shy. Bass simply decays faster. But there's not much between the two 4 ways.

 

[Vuki]

I just realized that the animated spectro GIF I put in my post does not work. I'll try again ... hopefully it works this time.

View attachment 465767

It explains why the 3 way system sounds more bass shy. Bass simply decays faster. But there's not much between the two 4 ways.

Look how much more bass energy is there with no highpass between 50-100hz! And that is IMO "fullness" frequency range.

 

[Keith_W]

I spent some time today making some high end speaker cable. Well, it was actually the same cable I was using, but I decided to go premium and buy some Speakon connectors for one end, and spades for the other end. After a couple of hours in my workshop (my soldering skills are extremely poor!) I was done. I plugged them in and had a listen.

Well, it was obvious that one speaker was louder than another. I swapped the cables left to right, and now the other speaker was louder. So obviously the quieter cable was not constructed properly.

I pulled it apart and to my amazement, it looked fine. I was utterly baffled. So I put it back and it sounded OK for a while. Then the problem came back. So I got REW out and did some sweeps:

Son of a bitch!!!!!!

(the reason it looks so funny is because this is a bare sweep of the woofer, with the mic sitting in the corner of the room. I didn't bother positioning the mic properly since all I wanted to do was check the SPL)

This is why I don't DIY, I can't even construct speaker cables properly!!!! I am 100% sure the problem is on the Speakon side since jiggling it produces all those SPL variations that you can see. I snipped those damned things off and i'm back to using bare wire.

 

[Keith_W]

So this happened last night. I had a small gathering and a few friends who are all into classical music came over. One of them is a professional musician, and I asked him to bring his violin along. He obliged. He did not want me to share his picture, so I have blanked his face out. This is the first time I have heard a live instrument in my listening room.

Immediately after this, I played a recording of a violin sonata by Bach / Julia Fischer. There really was no comparison, the sound of a live instrument only a meter in front of you, and speakers is vastly different. I think my system sounds pretty clear, but it was no match for my friend's playing.

 

[Robocop]

I've done the same but with piano and guitar. What we do listen to is unlikely to ever be performed in your lounge!!! All the more reason to support live artists.

I've just been to Melbourne I only realised you lived there after returning to NZ.

Robert

 

[Holmz]

…
I've just been to Melbourne I only realised you lived there after returning to NZ.

Robert

Head to NZ soon… Christ ️ and Nelson.

 

[Hayabusa]

View attachment 482024

So this happened last night. I had a small gathering and a few friends who are all into classical music came over. One of them is a professional musician, and I asked him to bring his violin along. He obliged. He did not want me to share his picture, so I have blanked his face out. This is the first time I have heard a live instrument in my listening room.

Immediately after this, I played a recording of a violin sonata by Bach / Julia Fischer. There really was no comparison, the sound of a live instrument only a meter in front of you, and speakers is vastly different. I think my system sounds pretty clear, but it was no match for my friend's playing.

if you want to really compare you have to record the same violin in an-echoic room.

 

[Keith_W]

I have wondered for some time whether reverberant fields actually exist in listening rooms. Dr. Toole says that the RT60 should be renamed "early reflection decay", because small rooms have specular reflections and not reverberant fields. But ... a 20kHz wavelength is 17.2mm, surely that is short enough to form a reverberant field, as well as specular reflections? And given that specular reflections become more diffuse as you get further from the wall, at what distance do they not become "specular"? We are talking short wavelengths here, no need to tell me that long wavelengths form room modes.

A reverberant field is supposed to measure the same no matter where in the room it is measured. I have made this theoretical point to others. And given that I am writing a REW guide, I decided to test this theory.

So I got the microphone out and took measurements from 6 positions. At MLP, 1m behind MLP, and 2m to the left and right ... as shown in the diagram.

I then overlaid all the measurements and selected Topt. Now we look to see where the measurements diverge. As you can see, they are all fairly tightly bunched down to 600Hz, which is where they start to diverge. In my room, the Schroder frequency is 133Hz, which means the transition zone ends at about 532Hz. Nice to see the measurement following theory!

As you can see from my photo, I did not bother moving furniture out of the way when taking these measurements. This is why we have some specular reflections at 1.6khz and 2kHz.

What is surprising is how "dry" the bass is. The room has five smaller windows, one almost full-height picture window, and large openings to an adjoining staircase and the dining room. Plenty of opportunity for bass to escape. I guess this is why the bass is so "dry". As for that huge spike at 50Hz, that's the noise floor. I thought I could avoid it by measuring loud and long, but it looks like I didn't. It also wasn't helped that I measured in the middle of the day on a weekend, and the neighbours were having a party.

So there you go ... "proof" that reverberant fields exist in small rooms. At least down to 600Hz in my room.

 

[Duke]

I have wondered for some time whether reverberant fields actually exist in listening rooms. Dr. Toole says that the RT60 should be renamed "early reflection decay", because small rooms have specular reflections and not reverberant fields. But ... a 20kHz wavelength is 17.2mm, surely that is short enough to form a reverberant field, as well as specular reflections? And given that specular reflections become more diffuse as you get further from the wall, at what distance do they not become "specular"? We are talking short wavelengths here, no need to tell me that long wavelengths form room modes.

A reverberant field is supposed to measure the same no matter where in the room it is measured. I have made this theoretical point to others. And given that I am writing a REW guide, I decided to test this theory.

View attachment 484027

So I got the microphone out and took measurements from 6 positions. At MLP, 1m behind MLP, and 2m to the left and right ... as shown in the diagram.

View attachment 484030

I then overlaid all the measurements and selected Topt. Now we look to see where the measurements diverge. As you can see, they are all fairly tightly bunched down to 600Hz, which is where they start to diverge. In my room, the Schroder frequency is 133Hz, which means the transition zone ends at about 532Hz. Nice to see the measurement following theory!

As you can see from my photo, I did not bother moving furniture out of the way when taking these measurements. This is why we have some specular reflections at 1.6khz and 2kHz.

What is surprising is how "dry" the bass is. The room has five smaller windows, one almost full-height picture window, and large openings to an adjoining staircase and the dining room. Plenty of opportunity for bass to escape. I guess this is why the bass is so "dry". As for that huge spike at 50Hz, that's the noise floor. I thought I could avoid it by measuring loud and long, but it looks like I didn't. It also wasn't helped that I measured in the middle of the day on a weekend, and the neighbours were having a party.

So there you go ... "proof" that reverberant fields exist in small rooms. At least down to 600Hz in my room.

Very enlightening! Thank you for doing this, and for directing me here from another thread. The real world is more complicated than my assumptions about it, and more interesting as well.

 

[DWPress]

So there you go ... "proof" that reverberant fields exist in small rooms. At least down to 600Hz in my room.

I can correlate your theory! Here's my room and the sum of 14 measurements like yours but with additional elevations. Down to about 500Hz in my room and I have multiple diffusers on my sidewalls along with treated front and back walls.

 

[Flaesh]

He did not want me to share his picture

It would have been possible to do without this scary rectangle, something like this:

reverberant fields

Old, sloppy, low-level measurements. Greens are 5" cone in box, top-firing, in the middle of the room. Oranges are XT1464 in normal corner position..

I'm glad to find a thread about your system, I'll read it later.