Going Active - Cura CA10s rise again
Now playing - Raising Sand through Cura CA10 speakers and some Cyrus Power amps I bought a very long time ago for an active loudspeaker project. These are not ordinary CA10s - or in fact real ones.
For reference, real ones looked like this.
And here's mine, about to be measured with REW:
The original project - which I had intended to complete with a crossover built from opamps, and perhaps using a crossover kit - never happened. It was a plan to provide speakers as part of our selfbuild - and that has not been a happy project. The amps stayed in boxes.
But now - I'm listening to music, and the crossover took me 10 minutes. Its not tuned yet of course, but it sounds like music already. Its going to be tough to turn them off, and also tough to stop fiddling with them.
The difference - a shiny new MiniDSP 2x4HD, which I bought from Purite Audio.
First - a little about the loudspeakers. The Cura 10s are a modestly sized ported loudspeaker that was sold by a Britsh company that gave up a good few years ago.
Wilmslow Audio bought some remaining cabinets and sold them on at an attractive price from their 'Clearence Corner'.
- The cabinets were sold without grills or badges but they had:
- routed driver and port holes
- t-nut fastenings already installed, and drilled holes for the bolts
- decent veneer
- vibration damping pads to the inside of the cabinet walls
- a separate plate to cover a small compartment on the back: I assume it was for the crossover
All the comments I have seen suggested that they were fine performers and they retailed for about £700 a pair. I don't have badges - or the grills. But then, the cabinets were only £60 a pair. Delivery was extra but was not exhorbitant and I was buying some MDF cabinets for some other drivers anyway.
I was not able to track down a schematic for the original crossover, but I did find out that the bass driver was a Vifa PL18 and the tweeter a Seas 25TFFC.
Both of these driver models have been obsolete for some time but Falcon Acoustics had some that had been manufactured recently - presumably as spares since they were popular designs with a number of manufacturers, as well as well regarded for DIY designs.
So - I could use original drivers but would have to have a crack at a crossover. And I'm a complete beginner at this, although I have read quite widely (and dreamed a lot). My plan is to finally try active - and if I can get a transfer function that sounds OK and is achievable with a reasonable number of components to also possibly try an analogue crossover in an external enclosure.
The total package to get them built and wired for active use (four terminals - straight to the driver connectors) - was about £100 for the drive units in each speaker, plus wadding, connectors, and some flared ports that fit so exactly that they must surely have been in the original design too. All in, its approaching half the cost of the originals. Saving money wasn't really the intention here though - it was a way to dip a toe into building loudspeakers with reduced risk.
There are plenty of web sites dedicated to DIY loudspeakers. Most of them do focus on analogue crossover systems - after all it will fit better with most people's stereo equipment. Its not really surrising - but a critical part of loudspeaker design is 'voicing' and much of that is achieved in the crossover design, which provides shaping to the response for each driver. Part of the shaping is low and high pass filters to split the frequencies between drivers - but also we need to handle 'baffle step' and suppress any nasty resonances, and ideally we would like to compensate for the different acoustic centres of the drivers.
Often, there will be a choice of electrical networks that create a theoretically reasonable crossover design. Some will be discounted as too complex to implement economically - and some will be discounted because they use components that are not available at a reasonable price.
Choosing the final crossover needs taste and care and iterations - and this is with components that may be toleranced to 5%, and possibly 2%. Either way, you may need a big component bin to match pairs and to be able to get what you actually need to high accuracy. It's time consuming, potentially expensive, and benefits from experience and taste.
Now - I'm not experienced, and I'm short on time and have no desire to invest in a generous parts bin tha will allow fine tuning. This is where active speakers with DSP crossovers come in. The linearity and predictability of half-way decent amplifiers and DACs is astounding - they are much, much more true than analogue components that are typically found in a crossover. And for most reasonably priced components, the actual loudspeaker drivers will vary more too, even within a single production run.
Tuning a loudspeaker with a passive crossover will require some soldering and careful component selection. It may need a complete toplogy change. What a drag! I don't find it an attractive proposition.
But a DSP system - just change some sliders and you're done, ready to listen again. The rate at which one can iterate and experiment is completely different to one for a passive loudspeaker system - or indeed one with an electronic active crossover.
I can't do anything about my inexperience - except practice. And practice is dramtically easier with a DSP crossover. Whether I have good taste and can voice a loudspeaker - time will tell. There is plenty of anecdotal evidence that 'measures flat' is not ideal. I can at least find out relatively easily.
The first thing to do with the 2x4HD is to download the software. As an early customer I had a code from Purite which saved $10.
- After download I:
- installed the USB driver
- installed the provided formware upgrade
- installed the configuration tool (which runs inside the Adobe Air environment)
I mensioned that I had a crossover running in 10 minutes after finishing the software: it is pretty straightforward, and I cheated by not bothering to measure the drivers in the cabinets. Rather I made an assumption that they would have infinite baffle responses similar to those on the reference specification sheets.
The PL18 is remarkably benign with an assumed cone breakup between 3kHz and 5kHz, and about 87db efficiency. The off-axis response falls away inline with drivers of this size and we can see the driver starts to beam above 2kHz.
The 25TFFC is also pretty benign. The spec sheet suggest 90db efficiency but it looks a bit above that to me given the slight rise. More concerningly the resonance is over 1kHz and the natural rolloff is down 6db a shade under 1.5kHz.
The cabinet is 22cm wide, which suggests that the mid-point of the difraction adjustment should be about 520Hz.
This gave a dilemma - if I cross over low using the tweeter's roll-off then it will likely be strained and impacted by the resonance, and if I cross over high to avoid major interference from the rolloff then I will have some power responce issues owing to the beaming from the midwoofer.
In the event, I decided to cheat - its DSP after all. So my first cut 'slap it in the box' crossover strategy was: - add a low shelf to the tweeters to counter the natural roll-off, with an 8db rise at 1kHz - so the tweeter is effectively flat from 1kHz up - low shelf for the midwoofers with 4db 'baffle step' centred at 520Hz - I have the speakers relatively near the walls in my study - a negative-gain peak filter at about 3.8kHz to reduce the rise from cone breakup, so the woofer is effectively flat to 4kHz (at least on axis) - 8th order (48db/octave) LR crossover at 2kHz, before serious beaming starts
In effect I just try to flatten the natural responses of the drivers beyond the crossover point and then use the default crossover for the 2x4HD just before the beaming starts.
My very first attempt padded down the tweeter levels for the estimated difference from the efficiency difference and the impact of the baffle step. In fact this might appear unecessary (possibly as a result of the shelf filter I added to extend the tweeter down to 1kHz is having the same effect.
A quick check with REW shows in-room response like this:
This actually sounds somewhat aggressive to me - while it was a bit dull on first listen with the treble pulled down. Time to play wth voicing! And that is, after all, what the DSP makes so easy.
Here they are, with the measurement microphone in place with a USB ADC attached and a Maplin microphone stand. I've had the measurement microphone (a Monacor/StageLine eclectret) for as long as I've consider building loudspeakers. I'd recommend the direct USB products from MiniDSP (or possibly Dayton) these days since they supposedly come with a frequency response file - I had to transcribe a printed frequency response to FRD format and don't have phase information.
So - what about the 2x4HD?
It does what it says on the tin. Its easy to use and if it has sonic artifacts they are immaterial compared to my room, partially run-in drivers, and inexperience. Currently I have an Audiolab Q-DAC hooked up to my workstation and the output runs to the 2x4HD, which feeds the two stereo Cyrus power amps through a Audiolab 8200AP. Potentially it could displace both Audiolab products and I could use the USB input and its own volume control - something to try in the future. Unfortunately it's not clear how much of that would work from a Linux host, but I can try that too given time.
The software is easy to use but I'd prefer some changes. The crossover, PEQ, FIR filter (that I'm not using), and channel volume are all managed seperately, and it would be nice to se a graphic of the final transfer function per channel, but I don't think that's possible.
It would also be nice if the design tools used to create the crossover and PEQ banks could create a FIR filter with similar amplitude response but flat phase, and perhaps also allow some phase manipulation. The FIR design mechanism is rudimentary. That seems a shame if the CPU is powerful enough to perform reasonable FIR processing.
It would be nice if there were some documentation of the control protocol so that a Linux system (and I'm thinking Raspberry Pi) could send audio data at full volume through the analoge inputs or USB and still control which of the 4 filter presets to use, and what master volume to use. Ideally the design/processing system would run on an RPi or other small Linux or BSD system also.
It would be nice if the inputs were on the back, like the outputs, so that it had an uncluttered 'front' with the IR receiver on it. Or possibly a mechanical control for volume and preset. In fairness these features are available on other models in MiniDSP's range.
This seems like carping though - the 2x4HD is simple to use and a practical and unpretentious tool that will enable anyone with a supported PC to have some fun in ways that would have been daunting previously. It can be used, and reused, in many ways - the flexible routing of source channels to outputs means that a unit can run a 2-wa crossover, or one side or a 3- or 4-way system (albeit not with the USB input).
The experience has been delightful - I enjoyed dealing with Wilmslow, Falcon and Purite and have no complaints at all. And I feel I'm at the start of a journey, fiddling with a project like this is a journey of investigation, experimentation, and learning.