Its was so much easier in the old days. It all sounded much better too.
Ha finally all your card on the table, thanks for the more elaborate explanation. I hear you. Really.
But I have to walk this path. I want to hear and compare for myself. It's a hobby, I have the time. And it's also fine to go back in time for me and experience and see the different ways to go about building/buying/tweaking an amplifier. You even have me considering working with two or even three amps now to compare, the icepower's are on their way, rotel is cheap, and hypex (2 hours by train from here) plenty second hand modules.
You see I'm typing this from my workstation inside my darkroom (photography lab). I'm a professional photographer and 30% from my work is shot on film, printed on paper even before I scan it. Both black and white and color. I love these old technologies. It even pays the rent and my hifi projects!
I'm curious to see how an active crossover compares to a dsp path. I will do both and report back to you😉 I trust your experienced, but other experienced people swear by analog active crossovers (with eq etc, no worries). I wanna hear for myself.
And in order to do both the dsp can not be inside the speakers, but inside my pc. So no plate amp. But I promise to use one before 15 years have passed.
Same stubborn character trait which makes me never print a negative scan on a digital printer, but with chemistry and paper. I also know 99% of the people won't see the difference, let alone decide which is best. I just love it.
My work is already creative, so for my hobbies I just want the extra puzzle to entertain my brain.
I'm 52, I'm okay with old technology, and compare it to new ones, or let them coexist. I bought old AKG K340 electrostatic/dynamic and Sennheiser HD 560 Ovation 2 headphones next to new Stax and Audeze. To get a sense of history. I think @Guerilla gets that;-) I switch them, I like both sounds. I want both.
If you ever get your hands on an AKG K1000 ... that's a piece of history 🤓
Much fun with your journey!
Much fun with your journey!
XLR (2 channel) mics are preferable for sure, as long as you have a pre-amp with +48v phantom power for the mic or happy to pay for a good-un.
USB mics are deficient in reliable and accurate impulse timing. This is required to get consistent relative phase between drivers measured at the same point.
However, you can work around this relatively easily by:
1. Measuring each driver individually
2. Measure each adjacent driver pair (running in parallel)
Use extracted / minimum phase in your crossover software (not measured phase as not reliable due to USB timing volatility) and adjust the Z offset of the backset driver (one with voice coil furthest away in the pair) so that the FR curve of both drivers in #1 = the combined curve of #2.
So for a 3 way
Measure T, M, W individually
Then measure T + M
and M + W
Put T + M in parallel in your simulation software and increase Z (offset) on M until the T, M individual curves = the combined T+M curve
Put M + W in parallel in your simulation software and increase Z (offset) on W until the M, W individual curves = the combined M+W curve
Finally add the M offset to the W offset and store in woofer as offset.
I've had good success with the above method. I usually reach for my USB mic because for me it is more convenient.
USB mics are deficient in reliable and accurate impulse timing. This is required to get consistent relative phase between drivers measured at the same point.
However, you can work around this relatively easily by:
1. Measuring each driver individually
2. Measure each adjacent driver pair (running in parallel)
Use extracted / minimum phase in your crossover software (not measured phase as not reliable due to USB timing volatility) and adjust the Z offset of the backset driver (one with voice coil furthest away in the pair) so that the FR curve of both drivers in #1 = the combined curve of #2.
So for a 3 way
Measure T, M, W individually
Then measure T + M
and M + W
Put T + M in parallel in your simulation software and increase Z (offset) on M until the T, M individual curves = the combined T+M curve
Put M + W in parallel in your simulation software and increase Z (offset) on W until the M, W individual curves = the combined M+W curve
Finally add the M offset to the W offset and store in woofer as offset.
I've had good success with the above method. I usually reach for my USB mic because for me it is more convenient.
I also manage with a USB mic (UMIK-1) & REW software.
REW generates a timing tone to solve for USB timing, (which is not needed for crossover & PEQ work.)
My workflow for impulse timing is different from Dave's.
Better? Worse? Don't know. Different.
In MiniDSP add 1ms to M.
Measure a sweep for M & T covering the M/T crossover.
Look at the Impulse Response (IR) and locate the M & T spikes, and calculate their relative offsets, remembering to remove the 1ms that was added to artificially separate them.
Record this value as the T->M delay to use later.
Reset M to 0ms.
Set T to the T->M value a couple of lines above.
Remeasure the sweep to check the IR has a single sharp spike.
Reset T & M to 0ms.
In MiniDSP add 1ms to W and repeat everything above for W & M. instead of M & T.
Record this value as the M->W delay to use on the next line.
Calculate the total W delay by adding the T->M and M->W delays. Set this as the T delay in MiniDSP (yeah, not to W)
Set M->W as the M delay in MiniDSP
Done.
I then use RePhase software to unwrap the crossovers, too.
That requires that your DSP is capable of performing FIR transforms.
[I tried using RePhase for EQ, too, but didn't like the result. Possibly I made too many, too detailed FIRs. From memory, it only made sense to calculate from one room location, too, whereas I can use REW to make PEQs from an average response at many room locations.]
REW generates a timing tone to solve for USB timing, (which is not needed for crossover & PEQ work.)
My workflow for impulse timing is different from Dave's.
Better? Worse? Don't know. Different.
In MiniDSP add 1ms to M.
Measure a sweep for M & T covering the M/T crossover.
Look at the Impulse Response (IR) and locate the M & T spikes, and calculate their relative offsets, remembering to remove the 1ms that was added to artificially separate them.
Record this value as the T->M delay to use later.
Reset M to 0ms.
Set T to the T->M value a couple of lines above.
Remeasure the sweep to check the IR has a single sharp spike.
Reset T & M to 0ms.
In MiniDSP add 1ms to W and repeat everything above for W & M. instead of M & T.
Record this value as the M->W delay to use on the next line.
Calculate the total W delay by adding the T->M and M->W delays. Set this as the T delay in MiniDSP (yeah, not to W)
Set M->W as the M delay in MiniDSP
Done.
I then use RePhase software to unwrap the crossovers, too.
That requires that your DSP is capable of performing FIR transforms.
[I tried using RePhase for EQ, too, but didn't like the result. Possibly I made too many, too detailed FIRs. From memory, it only made sense to calculate from one room location, too, whereas I can use REW to make PEQs from an average response at many room locations.]
Maybe this can be interesting for you:
X-Over with UbuntuStudio, Pipewire and Calf
Next time I will do a test with my setup similar to the Description of CharlieM
BR
Guenter
X-Over with UbuntuStudio, Pipewire and Calf
Next time I will do a test with my setup similar to the Description of CharlieM
BR
Guenter