Behringer DEQ2496 Ultracurve Pro DSP Review

[Robbo99999]

Simple rule to remember is: if your upstream device has 10 dB better performance, the effect on lowering the downstream device is 0.5 dB or negligible.

A rule like that would be good to remember, but I'm not sure I understand that, as there are probably a few ways to interpret what you said there. Are you saying re SINAD that say 114dB DAC combined with 60dB JBL 308 will result in 113.5dB SINAD? Or thinking about it more deeply I think you're saying that as long as the DAC is 70dB or above when combined with a 60dB JBL 308 that overall SINAD can only range from 59.5dB to 60dB? Thereby diminishing returns setting in if your upstream device is any greater than having 10dB better SINAD than your downstream device. EDIT: Yep, I got it after thinking about it, you're saying in practical terms that you certainly don't need better than 10dB better SINAD in your upstream device, as overall SINAD is limited by the worst performing link in the chain and as long as the upstream device has 10dB or better SINAD than the downstream device then it won't do anything to make SINAD worse (worst case 0.5dB worse).

So to clarify for others reading:

114dB DAC with 60dB downstream device = 60dB overall SINAD

70dB DAC with 60dB downstream device = 59.5dB overall SINAD

(can't get better than the weakest link in the chain)

A minidsp 2x4 HD has a SINAD in the neighborhood of 86 to 90db depending on who is measuring it. It hasn't been measured here with the latest instruments and methods.

Ah, I'm talking about the non-HD version so it's this one:
https://www.minidsp.com/products/minidsp-in-a-box/minidsp-2x4

 

[pkane]

1. Current production 2496 series units have switched over to CS5381 as an ADC as AK5393 has been discontinued.
2. The main culprits in terms of distortion are a number of tiny ceramic caps in the signal path - no NP0 in sight even down in the hundreds of pF (a puzzling design decision), though the ones in the nFs are the most critical ones. Some further improvement can be achieved thereafter by upgrading the NJM2068s. Also, the output stage is a single-ended impedance-balanced job stock, so one opamp is seeing very high levels. Some have entirely replaced the analog stages, which is one way of dealing with the issue I guess.

That's interesting. I wonder if there were some other circuit changes, as well. I measured the ca 2008 DEQ2496, and the overall analog (and especially noise) performance was very much worse than what Amir measured with the newer unit.

 

[amirm]

A rule like that would be good to remember, but I'm not sure I understand that, as there are probably a few ways to interpret what you said there. Are you saying re SINAD that say 114dB DAC combined with 60dB JBL 308 will result in 113.5dB SINAD?

No, the other way around. A SINAD of 70 dB will make a 60 dB JBL 308 into a 59.5. If the source is any better than 10 dB like in your example, the fractional loss of the downstream device gets smaller and smaller until it becomes just 60 dB.

In no case you can improve the downstream device beyond the best it can already do by itself (i.e. 60 dB).

 

[Robbo99999]

No, the other way around. A SINAD of 70 dB will make a 60 dB JBL 308 into a 59.5. If the source is any better than 10 dB like in your example, the fractional loss of the downstream device gets smaller and smaller until it becomes just 60 dB.

In no case you can improve the downstream device beyond the best it can already do by itself (i.e. 60 dB).

Thanks for clarifying, yep, like my EDIT explained, I understood what you meant once I had thought about it.

 

[sam_adams]

Below are some output stages of various Behringer devices:

CX3400:

DCX2496 rev. B:

SRC 2496:

All pretty much the same circuit that they call servo-balanced. It's all based on the NJM4580 dual op-amp. Since the output stages are all just copy-and-paste from one design to the next, performance really won't be that much different from device to device.

 

[andreasmaaan]

Ah, I'm talking about the non-HD version so it's this one:
https://www.minidsp.com/products/minidsp-in-a-box/minidsp-2x4

Yep, SINAD seems to be 80+dB:

Source: https://www.diyaudio.com/forums/minidsp/175810-measurements-minidsp-3.html

 

[LearningToSmile]

JBL M552 maybe? Unless you're looking for new.
There is also the BSS FDS360 that works with pre-set frequencies and slopes by replacing resistors and caps on the add-in cards. The op-amps are socketed so it might be worth a shot.
The absolute best active xover I've owned is Bryston 10B.

I'll keep an eye out, they show up on the used market here sometimes but not very often from what I can tell.

 

[Robbo99999]

Yep, SINAD seems to be 80+dB:

View attachment 89399

Source: https://www.diyaudio.com/forums/minidsp/175810-measurements-minidsp-3.html

Well that ain't too bad! Given what I've learnt just now about combining SINADs then there is theoretically no detriment in combining this with JBL 308 speakers.

 

[andreasmaaan]

Well that ain't too bad! Given what I've learnt just now about combining SINADs then there is theoretically no detriment in combining this with JBL 308 speakers.

Yeh, the only time I'd be more cautious about combining SINADs as a simple number would be when the relative harmonics or IMD are significantly different between components, for example, where an upstream device has a better SINAD figure than a downstream component, but it consists of significantly higher-order HD (which is unlikely in reality).

In this case, all the distortion is H2 and H3, which suggests it will be relatively benign. I have used this MiniDSP and the more-or-less identical (same DSP and ADC/DAC chip) Wondom ADAU1701, and subjectively don't find much to fault with it. Possibly not 100% transparent in all cases (more likely due to noise than distortion), but close enough for most purposes.

 

[infinitesymphony]

I bought mine in May 2003 for €390. It's quite a bit cheaper now. I think it's one of the longest living Behringer products, in production for 17 years now and no end to see. Still its list of very useful features is broad and quite unique: 10 x parametric EQ, 31 band graphic EQ, feedback destroyer, stereo width, compressor/expander, 61 band analyzer with microphone input, auto EQ, ... whatever. It's like a Swiss army knive for fixing audio problems.

Exactly, the #1 target audience for the DEQ2496 is portable broadcast / live sound racks. It really does have quite a few features even if you don't use all of them (or more than one of them). I used to work on a crew that used one of these just for audio delay before hitting the video switcher.

 

[bobbooo]

Yes, I previously remember a discussion related to this where there was an online calculator where you could add SINAD's together as if they were in an audio chain, and thereby you could work out what the combined effect would be. I tried googling it just now but didn't really come up with a definitive source. Well, let's say miniDSP 2x4 is 70dB SINAD, and what do you reckon SINAD would be of the ADC/DAC within the JBL 308, say 60 dB SINAD? What would happen if we combined a state of the art DSP solution of 114dB with said 60dB JBL 308. If we compared the result of the two scenarios how would the overall SINADs of the audio chain compare?

Here's that online calculator (bottom of page):
http://www.sengpielaudio.com/calculator-thd.htm

Incredibly useful site, with loads of calculators and converters. So with two components with distortion at -70dB and -60dB it gives a combined distortion level of -59.59dB. Replacing the -70dB with -114dB gives a combined distortion of -60dB (to three decimal places). What's often neglected here however is the distortion of the speaker's transducer. Here's @RayDunzl 's measurement of the JBL 308's total distortion (ADC/DAC + transducer):

Being generous and taking a best-case of -50dB, and combining with a -70dB distortion level for the miniDSP 2x4 this gives -49.96dB. This 0.04dB difference is imperceptible, so the miniDSP 2x4 would be effectively a transparent addition to the playback chain here. Most speakers aren't much better, so this all means we can get away with using equipment with relatively high distortion upstream of the speakers without any audible penalty, and why most competently designed smartphone/laptop/AVR as part of the playback chain will not introduce any audible distortion in practice. Chasing audio electronics SINAD numbers way above transducer SINAD is a purely academic exercise, and has nothing to do with audible sound quality, so should not affect judgements of these products for consumer use.

 

[digicidal]

SHD Studio unfortunately had no A to D converter. Strictly digital I/O. Even the 2 X 4 HD does A to D.

As I've written before, it seem miniDSP always either gives you something you don't need that you have to pay through the nose for, or leave something out that you really want. They never just give you the full suite of I/O ports. The 2 x 4 HD with spidf outs and Dirac Live would be perfect, but they won't release one that way without charging $1K.

Yeah, that was what I was referring to when I said "less flexible" - although the step up (SHD) for that, if necessary, is there... as you noted it's over $1K. I'm one that is perfectly fine with everything happening in the digital domain... I don't do analog sources anyway. However, my problem is that I want HDMI switching and passthrough... so I'm in a different (but similarly poorly handled) case as far as minidsp offerings go.

I'm still looking for that holy grail... but there's always something missing... even with a budget of 10X what the DEQ2496 costs.

The upcoming Bryston BR-20 preamp is getting dangerously close... but still missing REQ (not to mention it's likely to be ~$5K). As someone that's up to 30-40% movies/TV... but couldn't care less about 5 channels and surround decoding - and with limited vertical rack space - I understand I'm a unicorn apparently. Just give me 2.2/3.2 channels with REQ (with good bass management) in hardware, in a 2U appliance, with at least 3 HDMI inputs and take my money now!

 

[JeffS7444]

View attachment 89396

Thanks, I was being lazy and hoping for a DEQ2496 schematic so I'd have the exact part numbering, but if I must trace the circuit, so be it. In the case of the SRC2496 above, I'm thinking that C86 (100p) and C43 (20p) might be all I need to upgrade to C0G types. And although much larger than the stock part, I could probably squeeze a film capacitor in there, replacing the 47 uf coupling capacitor C94 (but is it worth the bother @ $7.33/ea?) Have I missed anything else which might give me lower THD?

 

[AnalogSteph]

Below are some output stages of various Behringer devices:

CX3400:

View attachment 89394


DCX2496 rev. B:

View attachment 89395

SRC 2496:

View attachment 89396

All pretty much the same circuit that they call servo-balanced. It's all based on the NJM4580 dual op-amp. Since the output stages are all just copy-and-paste from one design to the next, performance really won't be that much different from device to device.

Well, I'll be... Floating balanced output, huh? Must have misremembered that.

That said, they're not entirely copypasta. In the SRC, IC23B has a fairly easy life but capacitor distortion caused by C55 and C87 will be quite high. The rev. B DCX is much harder on IC6A's output stage, but given the substantially lower R56/R57 values distortion may still be a fair bit better.

Oh, and built-in Pin 1 Problem, as par for the course on Behringer products back in the day.

There has got to be a better circuit. They're basically converting differential to SE and back, mostly to save one opamp and misc. filter passives (which could admittedly pose some matching issues).

In the case of the SRC2496 above, I'm thinking that C86 (100p) and C43 (20p) might be all I need to upgrade to C0G types.

You've got that the wrong way round. The ones that should make the biggest difference are the highest values, i.e. the 2n2s, followed in quite a distance by the 100ps, and the 20p is so small that it may very well be NP0/C0G already. So if you want to be lazy, C55/C87 it is.

And although much larger than the stock part, I could probably squeeze a film capacitor in there, replacing the 47 uf coupling capacitor C94 (but is it worth the bother @ $7.33/ea?)

Not worth the mess. A nice polar Panasonic FC or bipolar Nichicon UES perhaps.

Do verify which version of the circuit you have. 390p would be higher on my list than 100p.

 

[Robbo99999]

Here's that online calculator (bottom of page):
http://www.sengpielaudio.com/calculator-thd.htm

Incredibly useful site, with loads of calculators and converters. So with two components with distortion at -70dB and -60dB it gives a combined distortion level of -59.59dB. Replacing the -70dB with -114dB gives a combined distortion of -70dB (to three decimal places). What's often neglected here however is the distortion of the speaker's transducer. Here's @RayDunzl 's measurement of the JBL 308's total distortion (ADC/DAC + transducer):

Being generous and taking a best-case of -50dB, and combining with a -70dB distortion level for the miniDSP 2x4 this gives -49.96dB. This 0.04dB difference is imperceptible, so the miniDSP 2x4 would be effectively a transparent addition to the playback chain here. Most speakers aren't much better, so this all means we can get away with using equipment with relatively high distortion upstream of the speakers without any audible penalty, and why most competently designed smartphone/laptop/AVR as part of the playback chain will not introduce any audible distortion in practice. Chasing audio electronics SINAD numbers way above transducer SINAD is a purely academic exercise, and has nothing to do with audible sound quality, so should not affect judgements of these products for consumer use.

Ah, thanks for that link re the online calculator (just have to remember to use negative numbers) and that tallies with the 10dB rule that Amir told us about. You did a typo in your calculation though, the number I bolded & underlined in your post should say -60dB.

Yes, it does seem that we're a bit overharsh with our SINAD numbers for our electronic devices on this forum, but it does help to have standards and if it doesn't mean increased cost then no harm in holding manufacturers to account for not producing excellent performance numbers.....but it is important to keep SINAD of electronics in context & proportion like you've mentioned & showed with your speaker distortion graph.

In terms of thinking about the relevance of high SINAD electronics, headphones have lower measured distortion than speakers (I've only just kinda realised this contrast!), with headphones often having less than 1% distortion which I believe is unheard of in speakers. Could it be that the low distortion of some headphones could mean that higher SINAD numbers are more relevant when it comes to headphone listening? Is there a way of calculating a relevant SINAD target in conjunction with headphone listening given headphones low distortion.....it would be quite useful to have a ballpark practical target for SINAD for electronics when used in combination with headphones? On the same token headphones don't suffer from room modes (which cloud the sound) and headphones also don't have the long decay times that speakers have in rooms (which are both particularly applicable in the bass area where distortion in speakers is often highest), so I'm thinking this all adds up to headphone listening being a higher resolution experience in general that might be able to define greater nuances (SINAD) in electrical gear? (Of course there's the HRTF/HpTF problem with headphones that would decrease resolution (which speakers don't suffer from), but some of that can be EQ'd out).

EDIT: an additional thought re the application of high SINAD electronics, when you EQ then you need to apply a negative preamp, which effectively means you're reducing your SINAD, so you need to start off with a higher SINAD product if you're taking EQ into account. I guess you're typically talking about a -5 to -10dB preamp when EQ'ing, so I guess this translates to removing 5 to 10dB from your SINAD. So taking this into account your want your DAC to have about 15-20dB better SINAD than your downstream devices.

 

[solderdude]

Is there a way of calculating a relevant SINAD target in conjunction with headphone listening

(passive) speakers and headphones usually have rather decent distortion at around 1kHz and chances are, when measured, you will be running into noise floors of the measuring equipment.
Speakers/headphones can have nice SINAD numbers (actually the only noise produced is air being pushed through ports/vents) when playing at say 80dB SPL but it may only appear to be bad when measured as the noise floor of the measuring equipment and above all surrounding noises in the entire audible band will skew that number.
That produced noise is amplitude dependent and won't be there when nothing is playing where noise will be there in electronics.
SD would be more appropriate and even that only says something about the behavior at 1kHz which will also differ per amplitude.
You can measure THD at 1kHz at say 100dB but wil probably never reach that in music as bass could be 10-15dB louder and most headphones won't like 115dB swings in low frequencies which due to compression will also modulate the 1kHz.

For upstream I also think the SIN part is very important when using digital volume control.
If one wants to reach 115dB peaks, so you can listen impressively loud for a few seconds, but don't want to hear audible hiss later in the evening your SIN should be low.
Not much an issue when you control the volume with an attenuator after the DAC.
The AD part is much less of an issue with music as our 'dynamic' range is much smaller than our absolute dynamic range and is also freq. dependent as well.
This means that when the importance of the SIN part is also application and SPL dependent.

It is easy and cheap to buy a DAC with a good S/N ratio and good multitone as well as frequency response and this is what is measured and reported as well. Look at those numbers rather than the SINAD ranking.
Of course, chances are that when the SINAD is high the other aspects also may not be poor.

 

[Robbo99999]

(passive) speakers and headphones usually have rather decent distortion at around 1kHz and chances are, when measured, you will be running into noise floors of the measuring equipment.
Speakers/headphones can have nice SINAD numbers (actually the only noise produced is air being pushed through ports/vents) when playing at say 80dB SPL but it may only appear to be bad when measured as the noise floor of the measuring equipment and above all surrounding noises in the entire audible band will skew that number.
That produced noise is amplitude dependent and won't be there when nothing is playing where noise will be there in electronics.
SD would be more appropriate and even that only says something about the behavior at 1kHz which will also differ per amplitude.
You can measure THD at 1kHz at say 100dB but wil probably never reach that in music as bass could be 10-15dB louder and most headphones won't like 115dB swings in low frequencies which due to compression will also modulate the 1kHz.

For upstream I also think the SIN part is very important when using digital volume control.
If one wants to reach 115dB peaks, so you can listen impressively loud for a few seconds, but don't want to hear audible hiss later in the evening your SIN should be low.
Not much an issue when you control the volume with an attenuator after the DAC.
The AD part is much less of an issue with music as our 'dynamic' range is much smaller than our absolute dynamic range and is also freq. dependent as well.
This means that when the importance of the SIN part is also application and SPL dependent.

It is easy and cheap to buy a DAC with a good S/N ratio and good multitone as well as frequency response and this is what is measured and reported as well. Look at those numbers rather than the SINAD ranking.
Of course, chances are that when the SINAD is high the other aspects also may not be poor.

Well thanks for that, I understand some of it. What do you think the S/N Ratio target should be that we look for so it doesn't cause limitations? And what characteristic or value (if value is applicable) do we look for in multitone for that not to be a limiting factor either? And how do you "calculate" / "arrive at those figures" that when looking at an audio chain? What I like about the "SINAD Rule" of making sure your devices upstream are 10dB better than the devices downstream is that it's an easy & practical way of visualising/choosing an audio chain. If there is indeed a link between SINAD and the variables you mention then might be just simpler to remember the "SINAD Rule".

 

[solderdude]

What do you think the S/N Ratio target should be that we look for so it doesn't cause limitations?

That depends on what the application is, what the spectrum of the noise is, efficiency of transducers and desired dynamic range.
Safe is as low as possible which may not be what is needed for all circumstances.

What I like about the "SINAD Rule" of making sure your devices upstream are 10dB better than the devices downstream is that it's an easy & practical way of visualising/choosing an audio chain. If there is indeed a link between SINAD and the variables you mention then might be just simpler to remember the "SINAD Rule".

You can have 2 exact same SINAD numbers where one is distortion dominated and the other is noise dominated.
The number doesn't say anything about the harmonic spread nor about the noise spectrum nor about IM.

 

[Robbo99999]

That depends on what the application is, what the spectrum of the noise is, efficiency of transducers and desired dynamic range.
Safe is as low as possible which may not be what is needed for all circumstances.



You can have 2 exact same SINAD numbers where one is distortion dominated and the other is noise dominated.
The number doesn't say anything about the harmonic spread nor about the noise spectrum nor about IM.

Hmm, ok, I was after some more easily implemented practical guidelines re audio chain setup, but your posts throw up more questions than answers as per usual. Thanks for your replies but I won't ask you more on it, as I don't want to contaminate this thread with a tangent. EDIT: I'm sorry for that comment, but it's always the same whenever I enter into conversation with you, either we communicate differently or have different motives, probably better if I leave your posts alone (most of the time!).

 

[Xyrium]

These are really neat units. When did you test the Minidsp DDRC-88? I can't find it in a search of the site.