I am trying to convert this design from passive to active crossover (miniDSP).
My plan was to trace the transfer function shown on the website, convert to biquad and insert in DSP. However im not sure which software allows me to import the filter function as an FRD file and use it as a target to finally export the filter as biquad.
Ive tried x-over pro which seems to have potential but is buggy. Any other suggestions?
My plan was to trace the transfer function shown on the website, convert to biquad and insert in DSP. However im not sure which software allows me to import the filter function as an FRD file and use it as a target to finally export the filter as biquad.
Ive tried x-over pro which seems to have potential but is buggy. Any other suggestions?
Last edited:
The passive crossover relies on the impedances of the drivers to determine it's EQ and filtering effects.
If you reverse engineer the passive filter, you will be excluding all the driver/crossover interactions.
I tried to use the dcx2496 to mimic the passive crossover, with the intention of calling up all the dcx settings and implementing them in a discrete active filter/eq setup.
But I could not get the dcx to sound like the original passive filter. I lost patience after a couple of months, but I will try again.
If you reverse engineer the passive filter, you will be excluding all the driver/crossover interactions.
I tried to use the dcx2496 to mimic the passive crossover, with the intention of calling up all the dcx settings and implementing them in a discrete active filter/eq setup.
But I could not get the dcx to sound like the original passive filter. I lost patience after a couple of months, but I will try again.
Last edited:
Andrew, as far as I understand a typical filter transfer function graph plots the final filter effect on the drivers AFTER taking impedence into account.
For example, this graph here from a zaph design: http://www.zaphaudio.com/ZD5-modeled-transferfunction.gif
Is this not the actual electrical function applied to the driver including impedence, one that can be transposed directly to an active crossover?
For example, this graph here from a zaph design: http://www.zaphaudio.com/ZD5-modeled-transferfunction.gif
Is this not the actual electrical function applied to the driver including impedence, one that can be transposed directly to an active crossover?
Well, a typical electrical filter transfer function plot shows the transfer function of the filter working into its intended load.
No
The zaphaudio transfer function you link to appears to be a plot of SPL, for given input, against frequency, of two drivers.
In order to calculate the electrical transfer function of the crossover in the speaker you're interested in, you need to know the electrical impedances of the drive units used, plotted against frequency, in those enclosures.
If you want to duplicate the response of the speakers with passive crossovers, using dsp, it might be a better starting point to plot the transfer function of the crossovers working into the drive units, in the enclosures, then work out how to obtain that.
If you can get an electrical model of those drivers, in those boxes, then possibly. I've no experience of the programs you mention so I can't say.
If you have these speakers, or are building them, you could test the transfer function from the speaker terminals, through the passive crossover to the drive unit terminals, with the drive units in the enclosure, using something like spectra lab or spectra plus. Then implement that transfer function in dsp.
But it seems a lot of effort to go to, implementing this in dsp and bi-amping, just to duplicate the passive crossover - surely you want some benefit or improvement from it? You could use the given passive crossover for a while to see if there's something about it that you'd like to alter, eg make adjustments for listening room conditions.
As I said, I've no experience of the software you mention, you'd have to immerse yourself in their manuals to find out what they will and will not do.
If you have these speakers, or are building them, you could test the transfer function from the speaker terminals, through the passive crossover to the drive unit terminals, with the drive units in the enclosure, using something like spectra lab or spectra plus. Then implement that transfer function in dsp.
But it seems a lot of effort to go to, implementing this in dsp and bi-amping, just to duplicate the passive crossover - surely you want some benefit or improvement from it? You could use the given passive crossover for a while to see if there's something about it that you'd like to alter, eg make adjustments for listening room conditions.
As I said, I've no experience of the software you mention, you'd have to immerse yourself in their manuals to find out what they will and will not do.
Have you tried modelling or even measuring the acoustic response with and without the crossover?
samadhi,
LspCAD will allow you to optimize to a "target" response. The target response can be an electrical transfer function of a passive network, or other "generic" responses you might pick. (.FRD files (ascii text) are certainly usable for this optimization.)
I believe the transfer function links to Zaph's ZD5 project are indeed electrical transfer function equivalents (that include driver parameters.) If you performed a transfer function measurement right at the speaker terminals relative to the input jacks, that's exactly what you'd measure. In that case, those would be the curves you would want to match with a line-level crossover and you'd create an exact duplicate of the (now removed) passive crossover.
The easiest way to do a passive-to-active conversion is to measure an actual speaker right at the driver terminals and save the measurement so it can become the target. If you don't have access to an actual speaker you can simulate the transfer function but you need to do some estimation on driver impedances, inductor DCR, etc, etc.
Cheers,
Dave.
LspCAD will allow you to optimize to a "target" response. The target response can be an electrical transfer function of a passive network, or other "generic" responses you might pick. (.FRD files (ascii text) are certainly usable for this optimization.)
I believe the transfer function links to Zaph's ZD5 project are indeed electrical transfer function equivalents (that include driver parameters.) If you performed a transfer function measurement right at the speaker terminals relative to the input jacks, that's exactly what you'd measure. In that case, those would be the curves you would want to match with a line-level crossover and you'd create an exact duplicate of the (now removed) passive crossover.
The easiest way to do a passive-to-active conversion is to measure an actual speaker right at the driver terminals and save the measurement so it can become the target. If you don't have access to an actual speaker you can simulate the transfer function but you need to do some estimation on driver impedances, inductor DCR, etc, etc.
Cheers,
Dave.
That's labelled "SPL" in the top right hand corner. It's not an electrical transfer function. It appears to be a plot of SPL output of the two driver units used against frequency, given a constant voltage electrical input. They won't let the OP do what he was wanting to.
It's a SoundEasy plot of an electrical transfer function. The "SPL" is just labeling from that program since it uses the same plotting graph for all measurements, either acoustic or electrical. In the SoundEasy measuring suite the 90db line is referencing to zero db difference between reference and measurement channels on the soundcard input....ie, if you performed a loopback connection it would be a straight line at 90db on the graph.
It's also much too smooth to be a real, acoustic measurement.
If I understood the OP's original query correctly, I believe it's exactly what he's interested in.
Cheers,
Dave.
It's also much too smooth to be a real, acoustic measurement.
If I understood the OP's original query correctly, I believe it's exactly what he's interested in.
Cheers,
Dave.
.....It's also much too smooth to be a real, acoustic measurement.
You're quite right about that, though I was thinking that it's much too smooth to be a real electrical measurement.
Hope I haven't been unnecessarily baffling the OP
(Covering my back now) "Properly labelled graphs would have been nice"
It's all good Simon, we humans do that from time to time
Here's the available before and after plot. I appears to agree with the transfer function (ignore the red line on the transfer function plot, the 0dB line appears to be 10dB higher).
Samadhi, I wouldn't normally settle for a crossover based on a single axis when more are available. There's a polar plot or two on the ZD5 page which Zaph appears to have given some consideration.
Based on the limited available information you will need to take your measurements at zero degrees for creating the transfer functions, then when you are finished, to listen to the speakers toed in to meet in front of your listening position.
Here's the available before and after plot. I appears to agree with the transfer function (ignore the red line on the transfer function plot, the 0dB line appears to be 10dB higher).
Samadhi, I wouldn't normally settle for a crossover based on a single axis when more are available. There's a polar plot or two on the ZD5 page which Zaph appears to have given some consideration.
Based on the limited available information you will need to take your measurements at zero degrees for creating the transfer functions, then when you are finished, to listen to the speakers toed in to meet in front of your listening position.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.