Some thoughts on DSP, and an embarrassing confession

[Sydney]

Re: Some thoughts on DSP, and an embarrassing confession

...So what load does the amp see then? Sensitivity of the individual drivers, right? Because you can amp them separately, and to whatever level you want with no need to pad the tweeter?...

That is one advantage, and the differences between a simple load and a conjugate circuit.

FYI: Filter design can get quite complex:

http://www.swarthmore.edu/NatSci/echeeve1/Ref/DataSheet/IntroToFilters.pdf

[mattsk8]

Re: Some thoughts on DSP, and an embarrassing confession

That is one advantage, and the differences between a simple load and a conjugate circuit.

FYI: Filter design can get quite complex:
http://www.swarthmore.edu/NatSci/ech...oToFilters.pdf

A good one that popped in my head that backs up what your saying would be the difference between a volume knob on a preamp, and trying to make a volume knob work between a 500 watt amplifier and the speaker. The volume circuitry on the pre would be fairly simple because there isn't much load at all, but you'd have a huge circuit board if you had to design a volume control that could handle handle that much power. There's also a lot of other factors, like the way resistors would change in resistance when they get hot, etc, that you'd need to factor in when you did the knob after the amp that don't matter at all when you're just manipulating the signal before the amplifier.

And definitely what Paul said is true, you still need to measure and sum drivers, but you gain an infinite number of crossover points with active so you can play with it until it's perfect.

[Sydney]

Re: Some thoughts on DSP, and an embarrassing confession

A good one that popped in my head that backs up what your saying would be the difference between a volume knob on a preamp, and trying to make a volume knob work between a 500 watt amplifier and the speaker. The volume circuitry on the pre would be fairly simple because there isn't much load at all, but you'd have a huge circuit board if you had to design a volume control that could handle handle that much power. There's also a lot of other factors, like the way resistors would change in resistance when they get hot, etc, that you'd need to factor in when you did the knob after the amp that don't matter at all when you're just manipulating the signal before the amplifier.

And definitely what Paul said is true, you still need to measure and sum drivers, but you gain an infinite number of crossover points with active so you can play with it until it's perfect.

( emphasis mine )
Most definitely.
As often stated the 2 biggest disadvantages given to filtering at signal level are cost and complexity.
To me these are relative terms; With the introduction of op-amps ( and Walt Jung's Cookbook ) it was just a question of when I could go that direction. The Old Colony WMJ board was a first step.

[Deward Hastings]

Re: Some thoughts on DSP, and an embarrassing confession

The downside to active is that power isn't cheap

Not so . . . "power" will generally be the least expensive part of an active design. Both chip amps and the newer Class D boards cost less and perform better than the more powerful "conventional" amps required for passive designs . . . going active can actually lower your amplifier costs and give better performance in the process. The only place where that benefit can't be easily realized is if you are stuck with an "integrated" AV receiver . . . but that's another item the days of which are (or should be) well behind us.

[charlielaub]

Re: Some thoughts on DSP, and an embarrassing confession

Not so . . . "power" will generally be the least expensive part of an active design. Both chip amps and the newer Class D boards cost less and perform better than the more powerful "conventional" amps required for passive designs . . . going active can actually lower your amplifier costs and give better performance in the process. The only place where that benefit can't be easily realized is if you are stuck with an "integrated" AV receiver . . . but that's another item the days of which are (or should be) well behind us.

I recently started doing DSP in software on very, very inexpensive hardware like the Raspberry Pi and I certainly will NEVER go back to hardware DSPs. I can get a very good quality 2-ch USB DAC for about $50 (or as little as $12 for a cheap one), and with the R-Pi being about $40 with SD card you can do the math. For me, amplification and its power supply are now the largest percentage of cost of my active hardware. It really depends on what you want/need from that department. If your SPL and power needs are modest (because room size is not large and/or drivers are efficient) you can do a simple 2x LM3886 + PS per speaker and not need to pay much at all.

DIY speaker building is fun again!

[DE Focht]

Re: Some thoughts on DSP, and an embarrassing confession

I'll bite... lets get this war started... :eek:. Yes there are a few benefits to active. First, you're manipulating the signal before the amp sees it, rather than manipulating the power going to the driver, so the amp gets to basically run free. 2nd, sensitivity is no longer an issue. I have heard the ATC SCM-40 (passive), right next to the exact same speaker, only it's the SCM40A (active). Between those 2, the active was hands down better sounding than the passive version.

The downside to active is that power isn't cheap, and instead of needing one 2 channel amp for a pair of passive crossed 3 ways (for example), you'll either need three 2 channel amps, or one 6 channel, etc. Imagine doing an all active 5.1 channel HT system, where each TM was also active.

Be aware that ATC (Billy Woodman, Bob Polley) do not like DSP crossovers. To them active is still analog all the way through.

I think DSP from Lake, XTA and BSS offer some excellent DSP crossover algorithms. The Digmoda dsp works very well also with 3886 based amplifiers and the B&O amplifiers.

Recently I have been torn with the result of some Op amp based crossovers versus the dsp based crossovers I have available. Mostly I agree with ATC's Woodman that inexpensive dsp gives up resolution to the op amps and can add artifacts.

Having owned many ATC loudspeakers. The active version are superior to the passives.

DSP versus Op amp active.....still not sure.

[kenrhodes]

Re: Some thoughts on DSP, and an embarrassing confession

With the LOW cost of chip amps like the TPA3116 and the TDA7297 why couldn't we just start building passive crossovers on the line side? Like a "F mod" Tiny inductors and capacitors are a lot cheaper then big ones. Any ideas on this "pre-active" crossover?

[mattsk8]

Re: Some thoughts on DSP, and an embarrassing confession

Be aware that ATC (Billy Woodman, Bob Polley) do not like DSP crossovers. To them active is still analog all the way through.

I think DSP from Lake, XTA and BSS offer some excellent DSP crossover algorithms. The Digmoda dsp works very well also with 3886 based amplifiers and the B&O amplifiers.

Recently I have been torn with the result of some Op amp based crossovers versus the dsp based crossovers I have available. Mostly I agree with ATC's Woodman that inexpensive dsp gives up resolution to the op amps and can add artifacts.

Having owned many ATC loudspeakers. The active version are superior to the passives.

DSP versus Op amp active.....still not sure.

I don't understand what you mean when you say that to ATC, active is still analog. According to the ATC rep the SCM40A is 100% active crossover as well as the amps. Not adjustable like some of the newer active Emotiva speakers though. Are you saying they use a passive crossover with the inboard amps?

I agree with the concern with inexpensive DSP units though, from the car stereo days I remember the benefit in sound quality using the crossover in the amplifier, verses the lesser sound quality via the crossovers in the head units. I'm sure the same rule would apply in homes, though I don't have any experience comparing.

[Paul Carmody]

Re: Some thoughts on DSP, and an embarrassing confession

I recently started doing DSP in software on very, very inexpensive hardware like the Raspberry Pi and I certainly will NEVER go back to hardware DSPs. I can get a very good quality 2-ch USB DAC for about $50 (or as little as $12 for a cheap one), and with the R-Pi being about $40 with SD card you can do the math. For me, amplification and its power supply are now the largest percentage of cost of my active hardware. It really depends on what you want/need from that department. If your SPL and power needs are modest (because room size is not large and/or drivers are efficient) you can do a simple 2x LM3886 + PS per speaker and not need to pay much at all.



DIY speaker building is fun again!

You're blazing the trail for the rest of us, Charlie. What you're describing is getting pretty darn close to what I'm envisioning.

[DE Focht]

Re: Some thoughts on DSP, and an embarrassing confession

I don't understand what you mean when you say that to ATC, active is still analog. According to the ATC rep the SCM40A is 100% active crossover as well as the amps. Not adjustable like some of the newer active Emotiva speakers though. Are you saying they use a passive crossover with the inboard amps?

Yes, each driver has an amplifier but at no point in the speaker is the audio signal digital or digitized like it is in a dsp loudspeaker.

The audio signal is split and distributed by op amps, look up the TI active speaker project. This is the type of crossover in an ATC loudspeaker, I have a crossover board from an SCM-16A. When I find it I will take a pic.

The speaker takes an analog input and it remains so to your ears.

[PWR RYD]

Re: Some thoughts on DSP, and an embarrassing confession

You're blazing the trail for the rest of us, Charlie. What you're describing is getting pretty darn close to what I'm envisioning.

Me too!

[mattsk8]

Re: Some thoughts on DSP, and an embarrassing confession

Yes, each driver has an amplifier but at no point in the speaker is the audio signal digital or digitized like it is in a dsp loudspeaker.

The audio signal is split and distributed by op amps, look up the TI active speaker project. This is the type of crossover in an ATC loudspeaker, I have a crossover board from an SCM-16A. When I find it I will take a pic.

The speaker takes an analog input and it remains so to your ears.

You helped me understand something new De Focht . When I said earlier in this thread that in the car stereos I did, using the amplifier's crossover always sounded better than if I used the head unit's crossover, I think (based on what you said and what I've read since then) the reason the amp crossover sounded better is the amplifier had a analog (active) crossover, but the head unit had a digital (active) crossover. Now I get it, and thanks!

Aside from aesthetics, in terms of function what exactly is the difference between the Texas Instruments crossover you brought up, and (for example) this Ashly Analog Crossover? Is the TI xover adjustable?

So there's another debate. ASP vs DSP vs passive. IMO ASP would win in terms of SQ.

[mattsk8]

Re: Some thoughts on DSP, and an embarrassing confession

.....

Since you very obviously understand this, I have a question (not loaded, I'm honestly curious). Which of these 2 active examples do you think would be better in terms of sound quality to use in an active 2 way speaker?... and this is assuming the user in both scenarios knew what he/she was doing when they set it up.

Example one... analog equalizer and crossover...



Example two... MiniDSP...

[Sydney]

Re: Some thoughts on DSP, and an embarrassing confession

The T/I is a circuit board* in which you set the parameters by selecting component values. ( these are fixed )
The Ashly is a complete turnkey solution in which operation can be varied on demand, a bigger feature set.

* http://www.ti.com/lit/ug/tidu035/tidu035.pdf

[Paul Carmody]

Re: Some thoughts on DSP, and an embarrassing confession

So there's another debate. ASP vs DSP vs passive. IMO ASP would win in terms of SQ.

Again, I think we're leaving out the most critical component: the designer. I think the DSP has a far higher chance of sounding better--in the hands of the right designer--because it is almost infinitely adjustable, whereas the op-amp based Signal Processors are more limited.