OP
Keith_W
Master Contributor
- Thread Starter
- #401
I spent the whole weekend redoing the DSP from the ground up. I have a lot of interesting measurements for your wonderment (or mirth) as well as some new DSP methods I have developed for your admiration (or ridicule).
Let's start by answering the question asked by everybody on ASR (or maybe not): "should I buy a plasma tweeter?"
In pink, we have a measurement taken from 2022. In red, taken over the weekend. Look at that, in a few years I have lost up to 10dB at 1kHz and a little bit at the top end. Fortunately the tweeters are crossed over at 4.5kHz (where the two curves start to deviate). Why does this happen? It is because the plasma tweeter has two consumables - the electrode, and the PL519 valve. A replacement electrode/valve set from Acapella cost me $1000 the last time I did the replacement, which was about 8 years ago.
Answer: unless you view your speakers as requiring maintenance like a car, the answer is "no". We are reminded that loudspeakers are imperfect LTI systems. Time invariance means that one set of inputs applied at one moment in time will result in exactly the same output as another set of the same inputs from another moment in time. Your speakers are not time invariant - as the voice coil heats up, the output may change. As suspension components lose their elasticity, the output may change. Plasma tweeters wear out in a different way. Fortunately it can be serviced and made as good as new.
The next question - "if I want to take a measurement to linearise a driver, at what distance should I place the microphone?" Let us look at a few drivers.
Woofers. The woofers are mounted on a 30cm wide baffle. According to theory, the baffle step is a volume loss of about 6dB that occurs over four octaves. Its centre frequency f3 can be calculated with f3 = 115824/W where W is the width of the baffle in millimetres. In my case, the f3 should be 386Hz, or roughly where I placed the marker. I intend to use the driver from 50Hz to 500Hz, which means that the microphone should be spaced 2x baffle width away from the speaker to account for the baffle step.
However, if you do this, your measurement becomes contaminated by reflections, as you can clearly see. The solution would be to lug this massive speaker outside, lift it onto a table, elevate the microphone, and get out a tape measure to ensure proper microphone separation from any reflective surface. I couldn't be bothered! So I spaced the mic 10cm from the cone and did a sweep.
Midrange horns. In red, the microphone is placed at the mouth of the horn. In black from 100cm away. Theory says that you should not measure at the mouth because out-of-phase reflections from inside the horn can arrive at the mic and contaminate the measurement. The usual advice is measure from 2x horn diameter away - in my case, 100cm. You can see that the shape of the curve changes. Since I was able to gate out reflections, I chose to measure from 100cm away.
Subwoofers. Question - is it worth taking a measurement of subwoofers for the purpose of driver amplitude and phase linearisation? Is it even possible to take a meaningful in-room measurement of a subwoofer? I think the answer is "yes".
In red, the mic was placed 20cm from the cone. In green, the mic was at the main listening position (MLP). I took the minimum phase response and set a very long gate (15 cycles). For context, a 50Hz sound has a wavelength of 6.9m ... about the same as the length of my listening room. This means that a gate of 15 cycles means that the sound has bounced back and forth my listening room 15 times. Surely the measurement would be meaningless, I thought.
To my surprise, it turns out that the two measurements look very similar. The phase curves are almost the same, the only difference is the MLP measurement is time shifted (here, I time aligned the MLP and nearfield measurement so that you can compare them). Since it appears that straightening out the amplitude and phase of the sub would still be beneficial at the MLP, I decided to proceed with Acourate's new Sinc pulse driver linearisation feature for all the drivers.
Let's start by answering the question asked by everybody on ASR (or maybe not): "should I buy a plasma tweeter?"
In pink, we have a measurement taken from 2022. In red, taken over the weekend. Look at that, in a few years I have lost up to 10dB at 1kHz and a little bit at the top end. Fortunately the tweeters are crossed over at 4.5kHz (where the two curves start to deviate). Why does this happen? It is because the plasma tweeter has two consumables - the electrode, and the PL519 valve. A replacement electrode/valve set from Acapella cost me $1000 the last time I did the replacement, which was about 8 years ago.
Answer: unless you view your speakers as requiring maintenance like a car, the answer is "no". We are reminded that loudspeakers are imperfect LTI systems. Time invariance means that one set of inputs applied at one moment in time will result in exactly the same output as another set of the same inputs from another moment in time. Your speakers are not time invariant - as the voice coil heats up, the output may change. As suspension components lose their elasticity, the output may change. Plasma tweeters wear out in a different way. Fortunately it can be serviced and made as good as new.
The next question - "if I want to take a measurement to linearise a driver, at what distance should I place the microphone?" Let us look at a few drivers.
Woofers. The woofers are mounted on a 30cm wide baffle. According to theory, the baffle step is a volume loss of about 6dB that occurs over four octaves. Its centre frequency f3 can be calculated with f3 = 115824/W where W is the width of the baffle in millimetres. In my case, the f3 should be 386Hz, or roughly where I placed the marker. I intend to use the driver from 50Hz to 500Hz, which means that the microphone should be spaced 2x baffle width away from the speaker to account for the baffle step.
However, if you do this, your measurement becomes contaminated by reflections, as you can clearly see. The solution would be to lug this massive speaker outside, lift it onto a table, elevate the microphone, and get out a tape measure to ensure proper microphone separation from any reflective surface. I couldn't be bothered! So I spaced the mic 10cm from the cone and did a sweep.
Midrange horns. In red, the microphone is placed at the mouth of the horn. In black from 100cm away. Theory says that you should not measure at the mouth because out-of-phase reflections from inside the horn can arrive at the mic and contaminate the measurement. The usual advice is measure from 2x horn diameter away - in my case, 100cm. You can see that the shape of the curve changes. Since I was able to gate out reflections, I chose to measure from 100cm away.
Subwoofers. Question - is it worth taking a measurement of subwoofers for the purpose of driver amplitude and phase linearisation? Is it even possible to take a meaningful in-room measurement of a subwoofer? I think the answer is "yes".
In red, the mic was placed 20cm from the cone. In green, the mic was at the main listening position (MLP). I took the minimum phase response and set a very long gate (15 cycles). For context, a 50Hz sound has a wavelength of 6.9m ... about the same as the length of my listening room. This means that a gate of 15 cycles means that the sound has bounced back and forth my listening room 15 times. Surely the measurement would be meaningless, I thought.
To my surprise, it turns out that the two measurements look very similar. The phase curves are almost the same, the only difference is the MLP measurement is time shifted (here, I time aligned the MLP and nearfield measurement so that you can compare them). Since it appears that straightening out the amplitude and phase of the sub would still be beneficial at the MLP, I decided to proceed with Acourate's new Sinc pulse driver linearisation feature for all the drivers.
