DSP and not DSP

[brig001]

Given the success of Colin’s Swansong speakers, I got thinking - dangerous I know
Colin uses DSP for the bass and analogue for the rest, so what if I did the same with a single driver?

I could split the signal into two, use mini DSP for the bass, a second order high pass for the rest and sum them to feed to the power amp. I would set the high pass similar to Colin’s crossover (300Hz) and the DSP would have a matching low pass

The only issue I can see is the latency of the DSP, but as long as the “crossover” is low enough, it shouldn’t matter

What does the team think?

Brian

Diet, makes more sense now

[rowuk]

I think that the first step is always to identify the problem and then find a solution instead of having a solution for a problem that one does not have.

The ability to power a woofer with a different amplifier than the rest has some very compelling arguments but also some caveats. If we are talking about a subwoofer, the efficiency issues alone dictate the need for separate amplification. When we talk about the mid- or upper-bass, the discussion becomes far different.

My experience is that there are people very sensitive to phase - where DSP latency could be a big issue - but there are others that are not seriously affected. There are also speaker designs that mask phase issues - generally by always being annoying to the sensitive listeners.

In my case, I do not want my analog playback to be digitized at all. This made me choose inherently very well behaved drivers mounted in cabinets that are complementary to those drivers strengths and minimising the usual issues of diffraction, phase shift or baffle step. This keeps the amount of work for the analog crossover "minimal" in the pass band where music is important. Adding a digitally controlled subwoofer below 40Hz would not be a problem, but also not a necessity as the subwoofer would only need to produce 1 octave of "pressure" - something easily achieved in analog.

That all being said, I use the MiniDSP HD for all of my initial testing and listening. It simply offers almost unlimited possibilities during development. At the end of the process, the optimised slopes are created in a passive analog crossover between the pre and power amps. The differences between final development stage with the MiniDSP and the analog crossover are subtle, not night and day.

If I only had digital playback, I would leave everything up to the power amplifiers in the digital domain. There are very high end PC solutions like Acourate that have impressed me.

The MiniDSP is not a major investment and I certainly recommend having one - if only "just for fun".

[brig001]

Thanks for your reply
The problem I have is from about 200Hz down, and room related. I am using full range drivers down to 50Hz, and the last two octaves are tricky. One because of the small driver, and two because of the room

For example, there is a notch in the left speaker caused by a bay window, so I fill it in with the right speaker. I use DEQ for that so I don’t overdrive it

Mini DSP doesn’t have DEQ, hence the need to split the signal and use the mini DSP compressor in place just for the bass

Above 200Hz would be purely analogue

Brian

[simon]

At the end of the process, the optimised slopes are created in a passive analog crossover between the pre and power amps.

I'm interested how you do that? A separate box of caps and resistors between the pre and power amps? Incorporated into the end of the pre? Something else?

I'm currently using LR4 active crossovers which are simple enough to implement but I'd like to try passive to compare. But using a power amp input impedance of say 100k in on line calculators gives huge impractical chokes.

Sorry for the thread jack Brian.

[brig001]

Sorry for the thread jack Brian.

No problem Simon

How are you doing your LR4 crossover?
I could use that for the mid - high section

Brian

[simon]

I got a couple of cheap pcbs from KMTech on ebay. He has all sorts, it's worth looking through what he has.

This looks to be what I have, a lot easier than drafting my own. Then I used one of Andrew's general 3 leg reg boards I had in stock for a bipolar PS. A lot simpler than passive I think.

https://www.ebay.co.uk/itm/124982772474 ... media=COPY

[rowuk]

Thanks for your reply
The problem I have is from about 200Hz down, and room related. I am using full range drivers down to 50Hz, and the last two octaves are tricky. One because of the small driver, and two because of the room

For example, there is a notch in the left speaker caused by a bay window, so I fill it in with the right speaker. I use DEQ for that so I don’t overdrive it

Mini DSP doesn’t have DEQ, hence the need to split the signal and use the mini DSP compressor in place just for the bass

Above 200Hz would be purely analogue

Brian

The MiniDSP has digital EQ - several parametric channels.

200 HZ is right in the middle of deeper male voices, cellos, trombones, an important part of the melody range of a piano. Integration could be a real challenge.

[rowuk]

At the end of the process, the optimised slopes are created in a passive analog crossover between the pre and power amps.

I'm interested how you do that? A separate box of caps and resistors between the pre and power amps? Incorporated into the end of the pre? Something else?

I'm currently using LR4 active crossovers which are simple enough to implement but I'd like to try passive to compare. But using a power amp input impedance of say 100k in on line calculators gives huge impractical chokes.

Sorry for the thread jack Brian.

Actually, chokes for small signals are very small. That is why LCR phono amplifiers work. Alternately, RCRC or CRCR can give you higher order so yes, a separate box of caps, resistors and chokes.
As we have a fixed resistance at the amp input, many times the passive modelling is VERY straightforward. I use CR, RC, CRCR or RCRC.

[brig001]

The MiniDSP has digital EQ - several parametric channels.

200 HZ is right in the middle of deeper male voices, cellos, trombones, an important part of the melody range of a piano. Integration could be a real challenge.

DEQ = dynamic EQ. Think PEQ with a compressor built in. This means that I can’t overdrive the speakers, but also that it’s ineffective at high volumes - fine by me as that’s only when we have a party

Good point about 200Hz, so I might have to go lower. Thankfully it only gets really exciting below 80Hz, above that I only need a gentle slope. Just checked and Colin et al settled on 100Hz, so it’s definitely doable there

Brian

[Nick]

200 HZ is right in the middle of deeper male voices, cellos, trombones, an important part of the melody range of a piano. Integration could be a real challenge.

I know that you were not directly talking about what Colin has built, but I was very very impressed on the Saturday evening listening to Marks wonderful cello recording just how good the integration was and how natural it all sounded.

[IslandPink]

I think we should first credit the cello recording to Janos Starker and Robert Eberenz
Secondly, although I would say it's one of the best DSP bass integrations I have heard, I could hear the effect of the crossover when the cello went down to the lower registers and it wasn't entirely natural.
I can't claim that the OB's I use with the (series) crossover in the 180Hz region are blameless either.
There is definitely a case for having that area free from crossovers and getting the bass, at least down to 60 or 70Hz to integrate fully with the midrange, and in that respect i'd like to have heard the Starker Bach recordings with the Purifi monitor I have worked on..... which I have just given over to my client ... hoping I can get the finished pair back for a quick review, if I ask nicely ...

[simon]

Thanks rowuk, food for thought. Do you have any suggestions for the choke suppliers?

[Nick]

Don't think chokes are simple and easy. I tried several chokes in the LCR and ended up with S&B because the others had problems. The hardest was constant L with signal level. I could simply adjust the signal level on test and watch the eq change. The solution was a very oversized core. Using a 15k riaa made it worst as it required higher L's

[simon]

Fair point, it would be interesting to hear rowuk's experience on this.

Perhaps easier to stick with resistors.

There's a lot to be said for active crossovers...

[rowuk]

Thanks rowuk, food for thought. Do you have any suggestions for the choke suppliers?

Simon,

Nick is certainly a much better source for advice than I am!

As I personally have only used caps and resistors, chokes are uncharted territory for me in regards to passive crossovers (I have only used them in phono preamps and FM tuners). I generally order parts from DigiKey and Farnell but have wound my own power supply and speaker level chokes/coils and probably would do the same for the crossovers if I heard something that would steer me away from what I already have. I am not an electrical engineer, I just have built and listen to a lot of stuff. I am not a cap hater

As far as the passive crossovers go, my current system is first order at around 450 Hz and second order Bessel at 8K. The power amps have an input resistance of 20K. I am using a passively crossed over subwoofer up to 40 Hz. I have not yet high passed the stereo pair yet (and maybe will not as the nodes are well filled in with the sub and stereo pair). Mundorf caps.

Maybe as a further comment: for a passive crossover in my view to work, we need a preamp with very low output impedance to "drive" the crossover. I ended up with a Pass B2 as all of my valve attempts did not do a good job of driving the crossover/fairly long interconnects.

In my view, one of the hardest things to do in audio is to create a perfectly neutral sounding buffer. Active crossovers however put MANY buffers in the signal path - second order 2amp stages, 4th order 4. That is why selection of the best behaved drivers is critical. We simply then have less crossover work.