miniDSP users read this.

[Pallas]

Re: miniDSP users read this.

Pallas, are you out there? ;)

Yes. Turns out I was entirely correct.

But the bottom line is either that
(a) we all discover a potential build defect to look out for in our current and future miniDSP's;
(b) Bob's process of troubleshooting teaches many of us something about how to optimally use the miniDSP, and some of the audible pratfalls of suboptimal setup.

Or do you think that Bob's process of troubleshooting his miniDSP configuration didn't teach many of us ("us" being people reading threads about the miniDSP) something about how to optimally use the miniDSP, and some of the audible pratfalls of suboptimal setup? (In this case, gain structure issues leading to high distortion.)

[Bob Brines]

Re: miniDSP users read this.

Final comments based on my experience with the miniDSP at the Lone Star Audio Fest:

It was my ignorance to purchase the RevA. I should have purchased the RevB. Took 15 seconds to switch the two jumpers, after many minutes of sweating the miniature Phillips head screws. They were extremely hard to back out. I stripped several on one side of the case and was just about to take a saw to the case when I successfully got enough screws out of the other.

The ADC is absolutely intolerant to being overdriven. A bit of care with the input and this shouldn't be a problem. It is also rather easy to clip the DAC with overzealous use of EQ. Since I was using the miniDSP to drive an H-baffle bass, I needed a 6dB/octave boost to counter the dipole rolloff. This was plenty to clip the DAC at full power input from my system DAC. The solution I used (not optimal, just what worked at the moment) was to run the top end amp at full volume ( not for power, but for gain) and control the volume from the music player. The bass amp did not need to be full up due to the efficiency of the driver.

Perhaps there is a better choice of software for the miniDSP in my application. I used the 2-way with EQ. The system hipass was determined by the natural rolloff of the small OB I was using -- 300Hz. I set the highpass to 150Hz LR to remove bass from the treble driver. I set the lowpass to 250Hz Butterworth. Those with trained ears detected a dip at the crossover. With this software I was not able to get a flat XO. Perhaps more time and effort will fix it. I would have liked to be able to choose the XO order, and I think that having Chebechev would have been nice. I will be discussing these issues on the miniDSP forum.

I don't do electronic kits, so I bought the miniDSP-in-a-Box. That means I am forced to use an analog input. Perhaps a pre-assembled UBS or SP/DIF will be offered. Perhaps variable XO order is available with one of the other software packages. If not that would be a welcome feature. I would also be nice if the miniDSP would work at 24/96 as I have a lot of source material at this rate, but I suppose that is a lot to ask at this price point. But then again, mose of the PA processors are also limited to 48kHz.

Actually, the miniDSP performed pretty well once I learned how to use it. I would recommend that if you are willing to DIY it, go with the SP/DIF input module.

Bob

[Deward Hastings]

Re: miniDSP users read this.

The ADC is absolutely intolerant to being overdriven.

ALL ADCs are "absolutely intolerant" to being overdriven . . . this is not unique to the miniDSP, or a "failing", even, it just is what is (in the digital world). It is also why 24 bit is necessary for digital capture (and more for processing), even though 16 bit is sufficient for delivery to almost any listening environment. And if you are going to do any response "boosting" in the digital realm then you have to provide for the available "bits" from the beginning . . . if your input signal goes to full scale then you have to provide a cut equal to or greater than the proposed equalization adjustment before any equalization is applied.

It's not as if any of the above should come as a surprise . . . it applies in the analog world too. But soft clipping and inherent noise in analog circuits hides the mistakes, and lets bad design pass for acceptable. Digital is not so forgiving . . . if you don't get it right it is obviously wrong.

Regarding amplifier gain and speaker "sensitivity" . . . set the former correctly and the latter doesn't matter at all. Simply choose your maximum (peak) output (say 105dB at 1 Meter) and set the amplifier gain(s) so .9 V produces that output from the attached driver(s). Done.

It is also (almost) always a mistake to use driver response irregularities as part of the crossover. Think instead of the driver/amplifier combination as having (or being correctable to) flat response, and the crossover as being, well, a crossover. That conceptualization, or approaching the design from that perspective, leads to a better understanding of where and how to provide necessary equalization, and what the limits on system topology are going to be. Dividing the functions and treating them independently almost always produces a better speaker in the end.

[AJ]

Re: miniDSP users read this.

Dumb question maybe, but are sure the DSP isn't amplifying the signal? As in, are you sure the signal amplitude feeding the final amplifier was the same for all tests? Just curious, as I have no experience with the miniDSP. It would be very odd if you were the first one to notice this.

[dlneubec]

Re: miniDSP users read this.

I don't do electronic kits, so I bought the miniDSP-in-a-Box. That means I am forced to use an analog input. Perhaps a pre-assembled UBS or SP/DIF will be offered. Perhaps variable XO order is available with one of the other software packages. If not that would be a welcome feature. I would also be nice if the miniDSP would work at 24/96 as I have a lot of source material at this rate, but I suppose that is a lot to ask at this price point. But then again, mose of the PA processors are also limited to 48kHz.

I believe miniDSP has a version that does 24/96 or better, though it may not be in the form of "miniDSP in a box". Also, I believe you can program just about any crossover variables you want using advanced biquads, though I don't know which software add on software you use for that.

Charlie Laub is developing a free, open source, spreadsheet solution, called Active Crossover Designer, that I believe will allow you to program the biquads on the miniDSP. See this thread: http://techtalk.parts-express.com/sh...d.php?t=231475
He has the initial release available for download.

[Bob Brines]

Re: miniDSP users read this.

Deward,

Right to everything you say. I was not intending to point out failings in the equipment, just describing my implementation. And, BTW, this is my first attempt at a dipole.

In this particular case the high pass roll-off is determined by the size of the baffle. I wanted the smallest possible baffle for the top end. Aesthetics. Every choice has consequences and the consequence here is dipole roll-off at 300Hz. I could cross higher and use the XO to contour the roll-off, but I wanted to stay out of the telephone band. Choice - consequence.

The bass driver is capable of at least 500Hz, so I am free to cross it anywhere I want. It would seems that a 1st order on the bass would make a perfect XO, but 1st order isn't available in the software version I am using. I will go back to miniDSP and ask the question.

6dB/octave bass boost is a requirement with a dipole speaker. The normal approach to bass boost is to use a plate amp with boost build in. But plate amps that will cross at 300Hz are hard to find. I am using a DTA100 for the bass driver and that's plenty of power for driver in hand. But the boost still needs to be applied. I could apply the boost in the music player, but I chose to do it in the miniDSP. It is now possible to clip the DAC without clipping the ADC. So, I am throttling signal in the music player.

Anyway, I am learning about dipoles.

Bob

[Bob Brines]

Re: miniDSP users read this.

Dan,

Thanks.

Bob

[Deward Hastings]

Re: miniDSP users read this.

The bass driver is capable of at least 500Hz, so I am free to cross it anywhere I want. It would seems that a 1st order on the bass would make a perfect XO

It's a bad idea for so many other reasons that its issues in a dipole configuration almost don't matter . . . but it's a bad idea there too.

The normal approach to bass boost is to use a plate amp with boost build in.

I don't know where that's "normal", but of course it will work, if all the other necessary precautions are taken. But . . .

Getting more than three octaves out of any driver while maintaining reasonable pattern control is difficult without using either bass boost or driver beaming or both, and getting four really stretches the limits and requires steep crossovers on both ends. And then there's overexcursion and cone breakup to contend with as well. There are few really successful designs (and many failures) for reasons . . . reasons mostly to do with poorly thought out design goals. Beginning with "make it good, then make it inexpensive" at least has a chance . . . I've never seen it work the other way around. Every choice does indeed have consequences, and dipoles involve lots of choices . . . and lots of consequences that too commonly get ignored. Almost all the dipole "failures" that I have seen are a result of choices that made no sense in the first place, but the designer "wanted to do it that way" . . .

[Bob Brines]

Re: miniDSP users read this.

Gee, Deward, I had no idea what a failure I had spawned, nor did a several dozen visitors to my room at LSAF, some of which are quite expert at this business.

Bob

[GranteedEV]

Re: miniDSP users read this.

But the boost still needs to be applied. I could apply the boost in the music player, but I chose to do it in the miniDSP. It is now possible to clip the DAC without clipping the ADC. So, I am throttling signal in the music player.

Bob, I`m not quite clear on how you did it, but I would have begun by reducing the output gain a bit (last page of the miniDSP software, or at least the advanced ones I've used if you're going to implement such boost.

Of course, perhaps such extreme amounts of gain as with a wide bandwidth dipole, are best suited to more powerful circuitry than a little pocket sized miniDSP

[Bob Brines]

Re: miniDSP users read this.

Bob, I`m not quite clear on how you did it, but I would have begun by reducing the output gain a bit (last page of the miniDSP software, or at least the advanced ones I've used if you're going to implement such boost.

Of course, perhaps such extreme amounts of gain as with a wide bandwidth dipole, are best suited to more powerful circuitry than a little pocket sized miniDSP

Perhaps. I haven't had the play time to check out options such as this. I used the input gain to balance the speakers -- they seemed to be leaning to the left a bit. I think that the channel output gain is still on this side of the DAC and would not help the problem.

Bob

[charlielaub]

Re: miniDSP users read this.

Bob, I`m not quite clear on how you did it, but I would have begun by reducing the output gain a bit (last page of the miniDSP software, or at least the advanced ones I've used if you're going to implement such boost.

Of course, perhaps such extreme amounts of gain as with a wide bandwidth dipole, are best suited to more powerful circuitry than a little pocket sized miniDSP

I don't think that the MiniDSP is at fault here, be it pocket sized or whatever.

I see this as a gain issue (lack of available gain) or lack of headroom, as others have mentioned. It's a similar issue to when people who cannot drive their pro power amp ( wants +4dB levels) to full power using their home audio preamp ( standard is -10dB levels). You could get more gain by inserting a gain stage after the MiniDSP, but Bob mentioned that he didn't want to add any circuitry or have to build anything. The gain could also be built in to the power amp. Once the gain is raised, the 6 dB/oct shelving boost (increasing at lower frequencies) can instead be a 6 dB/oct shelving cut, or a mix of some gain and cut.

These kind of tradeoffs can be done in my ACD tools and implemented in the MiniDSP via the advanced biquad programming plug in. You can mix and match a filtering function with gain changes and implement it in one stage (biquad). For instance, if you were using a Linkwitz Transform to add a lot of bass boost to a closed box sub you would probably experience a similar issue, but you can solve the problem by reducing the gain applied in the LT stage. For example, if you want 24 dB of boost at low frequency, you could split this between 12 dB of boost at low frequencies and 12 dB of cut at high frequencies, or whatever split you would like. This not the same as adding in a gain stage, but rather it is changing the gain profile of the filtering function itself. It all happens in one step, and this provides additional flexibility to deal with headroom and different amp gains, etc.

If the amp gain could be increased, the component producing the necessary gain is then shifted away from the MiniDSP, and you gain back headroom there. Switching from Rev A to RevB is not getting you all that much more headroom/gain, only 6dB I believe, so while that helpful it is not a great fix in general. I think this is why Sigfried Linkwitz does not see the MiniDSP as a great match to open baffles - by itself it's not, but by increasing gain after the MiniDSP and choosing other gains wisely I think it would work great.

-Charlie

[charlielaub]

Re: miniDSP users read this.

FYI - I just happened upon this related thread over at DIY Audio:

http://www.diyaudio.com/forums/minidsp/175810-any-measurements-minidsp.html

It talks about the noise level of the MiniDSP 2x4 (with measurements) and then goes on to talk about gain structure when using these units.

-Charlie