Behringer Ultracurve review

[fcserei]

Read: so the lower frequency - below 1kHz - component of the clock jitter is negated

[tiroth]

fcserei,

How feasible would it be to DIY the digital I/O? From reading other dribbles online it seems as if the digital card is

Reciever
6dB digital cut
->to unit

from unit->
Transmitter

Can you provide any information about the interface between the digital I/O card and the motherboard? I am somewhat tempted to work with this unit but I'd prefer not to have to drop $300 just to play with the interface.

It does seem that for the ~US$70 cost of the digital I/O board one could probably gin up a much better replacement; is the digital output transformer coupled?

Thanks!

[fcserei]

As I said elsewhere the internal timing is 64fs i2s in the behringer.
The interface of the receiver card is unknown, so is the test points for the firmware to recognize the card.
You can feed the unit with proper i2s signal at the ADC output and switch manually between analog and i2s input, but then you need to drop the digital signal by 6db ( the firmware is actually 5dB) externally and have to reclock the i2s signals, as without the digit card, the internal timing of the behringer is very close to the allowable limits of i2s.
The digit if card installed timing is much more in spec.

Digit out is also problematic without the card, as the CS8490 is in self clocking mode, means it has no bclk input, rather the BCLK is divided and sync-ed from MCKL. This circuit is available on the EBU card (as the 8402 needs the BCLK) but not on the main board.

The I/O is transformer coupled.
I don't think you can build the same functionality for lot less, not even considering the time.

[ThorstenL]

Hi,

How feasible would it be to DIY the digital I/O?

Rather little I suspect, even with a full service manual.

From reading other dribbles online it seems as if the digital card is
Reciever 6dB digital cut ->to unit from unit-> Transmitter

The 5db Gain cut in the digital domain is part of the DSP, to give some headroom for boosting. In theroy, if you have 0dbfs in and a slider on full, that is 16db, boost you will have to attenuate the input signal overall by 16db as the boost in the digital domain really can only work by attenuating the rest. 0dbfs is a "brickwall". There is NO headroom.

It does seem that for the ~US$70 cost of the digital I/O board
one could probably gin up a much better replacement;

Maybe, if you are really, seriously very experienced, have decent Multilayer SMD PCB fabrication capabilities etc. The Chips on the I/O are standard Crystal, probably exactly the same you would use and the layout and implementation is competent enough. The biggest difference is that the interface is AES/EBU and thus has 110 Ohm balanced impedance, not 75 Ohm single ended.

is the digital output transformer coupled?

As this AES/EBU standard both digital input and Digital Output are transformer coupled. I suspect the biggest improvement would be the use of a set of know "low jitter" transformers and perhaps some tweaking on the PLL Loop filter and PSU (though decoupling is again competent enough to make improvements difficult).

I think in the context the Digital I/O is perfectly fine, if you swap the cheap 'lytics for something decent. One option would be an external clock link for Input and output, where the external low jitter clock would replace the PLL recovered clock and drive both transport and DAC. In fact, I suspect if you drive the Transport and DAC from this clock you may able to leave the Ultracurve unmodified and simply run it on external sync....

Sayonara

[jean-paul]

Decoupling on the digital I/O is poor, only 2 low quality electrolytics. Some smd caps are soldered. There is space enough for some more electrolytics. The small 22 uF NX HiQ's will fit perfectly. There is also enough space for the Monkeysects PLL filter when you use a 3,3 nF SMD cap.

Strange thing is that the board is from 1998 while the 8024 is newer. It also has room for optical TORX/TOTX units but these are not mounted. Optical is less in quality than coaxial but it makes connectivity simpler.I would probably only change the digital output to optical and the input to 75 Ohm SPDIF. I couldn't find any information on the used 110 Ohm transformers in search for a 75 Ohm equivalent. In the 5 Behringers I examined they were not the same in all of them. The Behringers themselves were also different inside. Mostly different electrolytics, but all cheap and unknown brands. Some had other and even more RAM onboard, all reused old SDRAM. They like to use the glue gun at Behringers because the stuff was dripping all over in some of the devices !


But if you want to mount these TORX/TOTX they will direct to the wrong direction. I think this board was used for another product before. You have to extend it with flatcable to the backside of the unit.

IMO the I/O board is not up to par with the 8024 in terms of built quality.
I had 5 of them and some even had the chips soldered beneath the pcb tracks. They all were covered by a abnormal thick layer of flux which I had to remove. They all did work however.

[ThorstenL]

Hi,

Decoupling on the digital I/O is poor,
only 2 low quality electrolytics.

If you say so. Actually, where it matters (close to the IC pin's teh decoupling is quite good with SMD Cap's. If you draw out thw complex networks involved you will find the decoupling "sufficient".

Yes, I suppose one could add more 'lytics, but "more" is not neccesarily better. If you really want to go to twon give the IO Board, Receiver and Transmitter their own seperate supplies (all the way to seperate low leakage mains transformers), there is enough space inside.

And as I said, better interface transformers together with better (correct charateristic impedance) wiring to the XLR Jacks plus a "monkeysecrets PLL Filter) would likely bring a bigger improvement than the whole extra PSU decoupling while leaving the original wiring, transformers et al in place.

Strange thing is that the board is from 1998 while the 8024 is newer.

It is a "standard" item that fits all or the 8000/9000 Digital "mainframes.

I would probably only change the digital output to optical and the input to 75 Ohm SPDIF.

For anyone with a CS8412 & 8402 Datasheet and a pair RCA (better BNC) to XLR Adapter plates such a conversion is a 15 Minute (including disassembly/reassembly) job.

Sayonara

[jean-paul]

Mmm, even if I look very good at the thing and analyse it I can't really say decoupling is overwhelming but I am very neutral to the Behringer unit.
Let's keep the qualification "sufficient" although I am sure that ( done with care ) some more decoupling with better parts will make the unit better. Did it with several CS8412's in other applications and feeding it locally with a regulator does make a difference too.

Even though I like it and still think it's very nice technology it is very cheaply built. Today's price is OK for this unit.

It's good that you mention the wiring, that came up in my mind too. I also suspect a better mains transformer can give good results.

It all depends on how far one wishes to go.

[Nielsio]

Slightly different question:

I'd like to have my equipment on all the time (day and night). Is there any harm in that?

.. considering my Behringer UC (used digitally), Gainclone (left channel 7 mV DC, right 5 mV) and 98 db speakers (Supravox 12" and AER 8").

My computer, which is the main audio source, already is on day and night.

[jean-paul]

It harms your wallet a bit and the eventually the environment. The pc and the Behringer draw most current respectively. The Gainclone the least. My ( small ) concern would be the lifetime of the Behringer's display.

Gear that is always on generally sounds better. If you are sure your amp is built safe you can leave it on for sounds'sake.

BTW why are a lot of people using their pc as their main audio source ? Best results are obtained with a separate cd player.
Normal pc's are sometimes RF radiating and jittery. Most soundcards are not really comparable with a good external source. Please don't come up that you're playing MP3 on your Gainclone....

[fmak]

BTW why are a lot of people using their pc as their main audio source ?
---------------------------------------

Why indeed; but it is boring when computing and one is less critical of deficiencies when doing something elase.

I tried very had to build an audio computer for music and measurement and have concluded that mid fi is possible, given the noise and the cheapo components used. Howeve, it is also a painless way of playing DVDA where all you need to so is to load the disc! Much less hassle than a DVDA player.