I have had a stereo based on Linn Tukan and AV5105 since 1999, and recently purchased a pair of used Linn Keilidh speakers, manufactured in April 1999. I have been fascinated with Linn's AKTIV setup, and ran a AV5120 Center channel in AKTIV for years, so I really wanted to convert the Keilidh to active crossovers. As folks here may be familiar, the Keilidh have a two-way crossover, but there are two bass drivers, so the ultimate setup has separate amplification for each driver (two bass and one tweeter). Unfortunately, Linn no longer sells the AKTIV cards for their AV5105 amplifiers, and I'm not in the market for an entirely new Linn system, so I had to find an equivalent solution.

Since I use the Metric Halo Labs ULN-8 as my DAC, I realized that I could use their DSP to split the signal with a digital Linkwitz-Riley crossover, feed two pairs of outputs and run those to three amplifiers. I have a Pass Labs XA30.8 for the tweeters, and two vintage Linn AV5105 amplifiers for the bass channels. The AV5105 have RCA inputs with RCA pass-through jacks, making it easy to have the same signal on parallel amp channels.

Some photos have been attached to show the surgery on the passive crossover to convert them to active; the K400 + K200 Linn speaker wire for tri-amp connections to the cabinet; and a partially wired stack of two AV5105 showing pass-through of the stereo bass signals and K400 output.

I started by converting the right speaker, and then using measurement microphones to match the tweeter and bass levels on the active right with the still-passive left speaker. From these measurements, I learned that the original passive crossover burns about 2.5 dB in inefficiency, and that my dissimilar amps (Pass Labs versus Linn) have as much as 8 dB difference in gain. All of this was compensated in the MIOConsole settings, where individual channels can be delayed and adjusted in gain.

I started with Linn's original crossover frequency of 2.8 kHz, which is sometimes documented as 2.7 kHz, but I seem to be getting better sound with the crossover set to 2.5 kHz or even 2.2 kHz. I will have to experiment more on that, perhaps researching the drivers to find out whether they have optimal crossover points.

If anyone has comments or suggestions, I'd appreciate the feedback.

Brian in Los Angeles

Edited January 2Jan 2 by Zebulon
correction

For reference, here is the DSP graph that implements the digital active crossover. The MIO is configured for "5.1" so that I can output multiple channels, and I decided to treat the tweeters as a "front" pair, with the bass drivers as a "rear" pair.

In the graph Left and Right are patched to a stereo Linkwitz-Riley crossover. From there, the Low and High outputs all go through 24-bit dither with high-pass, triangular probability distribution function noise, which are each patched to the correct outputs.

You'll note that the graph also mixes Left + Right, runs that through another crossover at 45 Hz with a limiter and dither to drive my subwoofer (LFE) output. The Keilidh normally reach down to 65 Hz ±3 dB with the passive crossovers, but that spec is extended down to 45 Hz with the active tri-amplification. So, my subwoofer is only filling in the audio below 45 Hz.

Edited January 2Jan 2 by Zebulon
grammar

For those who are curious, my subwoofer is a passive NHT, powered by a QSC SPA2-200 half-rack 200W amplifier.

Interesting.

I have a passive pair of Keilidhs that I bring in from time to time. I think they're great.

Nice write up 👍

Did you do anything to manage time alignment between the drivers?

I'm not an expert but I think the Linkwitz-Riley filter is not linear, meaning that phase shift will vary with frequency.

I've used a DeqX device to do something similar, but it provides a method to time align the drivers as well as implement a digital crossover between them.

3 minutes ago, davm said:

Did you do anything to manage time alignment between the drivers?

I'm not an expert but I think the Linkwitz-Riley filter is not linear, meaning that phase shift will vary with frequency.

I've used a DeqX device to do something similar, but it provides a method to time align the drivers as well as implement a digital crossover between them.

This is a good question, but the passive crossover is also a Linkwitz-Riley design. I assume that the digital version has the exact same phase shift - albeit without the 20% tolerance slop that the analog capacitors and inductors have.

That said, if time alignment is needed, then the MIO DSP has delay building blocks that could be used.

I did some reading about Linkwitz-Riley crossovers, and learned that the tweeter is usually wired with opposite polarity to compensate for the phase shift of the second-order design. For three-way Linkwitz-Riley crossovers, the midrange output is wired with reverse polarity.

I finally decided to stack the Linn AV5105 in my rack, since nothing more needs to change with the wiring. The other components are DAT, MD, QSC 200W (sub) and 60W (center) amps, and a Parasound Halo C 2 for surround decoding.