Extending the Behringer Wing by 48×40 in a 4U rack

In this post I describe how we extended our Behringer Wing with an additional 48 inputs and 40 outputs via AES50 in a 4U rack case.

The Summary

The rack case consists of a Behringer SD16 that connects to a Behringer S32 via AES50. In addition, we connect a Midas DN4816-O via its Ultranet input to the Behringer S32. From there all 48 inputs are upstreamed to and all 40 outputs are downstreamed from the Behringer Wing via AES50. All rack parts are powered through the Adam Hall PCL 10 Power Conditioner.

The Configuration

As AES50 supports 48×48 channels we needed to find a combination of stage boxes and modules that would come close to this maximum. We started with a Behringer S32 as the main item of the extension, as this stage box supports 2 AES50 ports and an Ultranet output. We then added a Behringer SD16 (instead of a Behringer S16) for two reasons:

1. We already had a spare SD16 catching dust in a shelf for quite some time.
2. In contrast to the S16, the SD16 supports XLR/TS combo input jacks.

The downside to this is the larger form factor (3U vs 2U with the S16). The 4 Ultranet ports are not needed either. By combing these two stage boxes we reach a 48×24 channel count. That is why we added a Midas DN-4816-O reaching a final channel count of 48×40.

Note: In order to reach the maximum 48×48 channel count one *could* have added a Behringer ADA8200. However, this would have consumed another unit in the rack and added roughly 25 CHF/30USD per output port along with some additional complexity and latency.

To make this setup work we need to configure a few things:

1. On the DN4816-O we need to switch to Ultranet and Individual;
2. The SD16 connects via its AES50-A to the AES50-B port of the S32;
3. On the SD16 we need to switch OUT to 17-24;
4. We connect the S32 via AES50-A to a free AES50 port on the Behringer Wing;
5. The console needs to define a sync source and provide it to the S32; we set our to the DANTE module (or INTERNAL if we had no Dante);
6. The console needs to route (up to) 40 output channels downstream via its AES50 port to the S32;
7. The console needs to route (up to) 48 input channels upstream from its AES50 port from the S32.

As we are using the SD16, which comes with 4 powered Ultranet ports, we could use these to connect an additional DN4816-O to use as a signal splitter for ports AES50-33 to AES50-48. Or we could run a Cat5e cable to a P16-M and connect remote speakers via unbalanced 1/4″ TS line out.

With this setup, the routing and port assignment would look similar to the tables below. The good thing: we do not need a Wing for this. We can use our X32 RACK or the Midas M32C (or any other X32/M32 with AES50 for that matter) or even a Klark Teknik DN9630 as well.

For additional information on how these stage boxes can be connected and route their AES50 channel, have a look at the Behringer documentation. There you find useful diagrams like the one below:

The Test

So, one question remains: what is the performance (as in latency) hit we get by using such a stage box? Can we use it as a digital patch bay? Let’s do some testing …

The Setup

With our test case below we have 3 input signals and generate 18 output signals (1A, 1B, 1C, …, 1E, 2A, …, 3E) all running at 48kHz:

Connect a Roland TR8-S with

• ( 1 ) BD/ASSIGN1 to the Behringer Wing,
• ( 2 ) SD/ASSIGN2 to the input of the S32,
• ( 3 ) LT/ASSIGN3 to the input of the SD16 (all instruments being a 909 Bass2 with the same parameters).

Route the signal

• ( A ) to the output of the Wing,
• ( B ) to the output of the S32,
• ( C ) to the output of the SD16,
• ( D ) to the output of the DN4816-O and
• ( E ) to the DANTE output

Connect all these outputs to the Sound Devices Scorpio as line input and record the sound.

The Result

Below you find a summary with the results. Some observations:

• Not surpisingly the direct connection to and from the Wing is the fastest.
• Connections to and from the SD16 have twice the latency as connections from and to the S32 due to the fact that the signal must change from one AES50 port to the other (adding 22 samples).

I did a couple of recordings and uploaded one of them so you can have a look for yourself.

Relative Latency	Inputs
Output	( 1 ) Wing	( 2 ) S32	( 3 ) SD16
( A ) Wing Analogue (CH01 – CH03)	0smp / 0us	22smp / 458us	44smp / 917us
( B ) S32 (CH04 – CH06)	4smp / 83us	27smp / 562us	50smp / 1042us
( C ) SD16 (CH07 – CH09)	11smp / 229us	34smp / 708us	56smp / 1167us
( D ) DN4816-O (CH10 – CH12)	8smp / 167us	30smp / 625us	52smp / 1083us
( E ) Wing Dante (CH13 – CH15)	33smp / 687us	55smp / 1146us	77smp / 1604us

The Parts

1. Behringer S32
   3U, 32 mic pre-amps XLR inputs, 16 balanced XLR outputs, 2 AES50, 1 Ultranet output
2. Behringer SD16
   3U, 16 mic pre-amps XLR/TRS combo inputs, 8 balanced XLR outputs, 2 AES50, 4 powered Ultranet outputs
3. Midas DN4816-O
   1U, 16 balanced XLR outputs, 1 Ultranet input, 1 StageConnect Master, 1 StageConnect Slave
4. Adam Hall PCL 10 Power Conditioner
   1U, 10A, 1 AC IEC input, 8 fused and switched AC IEC outputs, 1 unfused and unswitched AC IEC output
5. Thomann Rack Case 4U
   4U, 400mm
6. Cat5e Cable
   10m shielded cable with Neutrik etherCON connectors to connect the extension with an AES50 clock master
   

The Price

Below you find a price indication of the components used in the build (with exchange rates at the time of writing).

	1	1.04136	0.89012	1.13897
	CHF	EUR	GBP	USD
S32	880	916.3968	783.3056	1002.2936
SD16	580	603.9888	516.2696	660.6026
DN4816-O	300	312.408	267.036	341.691
Power Conditioner	135	140.5836	120.1662	153.76095
4U Rack Case	85	88.5156	75.6602	96.81245
Cat5e Cable	25	26.034	22.253	28.47425
Sum	2005	2087.9268	1784.6906	2283.63485

The Conclusion

For roughly 2000 CHF (2100 EUR, 1785 GBP, 2280 USD) we get a total of 88 ports with 48 (!) microphone pre-amps (and yes, their quality is not outstanding, but defninitely usable) and a whole bunch of XLR/TRS combo jacks.

The overall latency is still incredibly and unnoticable low with the inputs and outputs on the SD16 being the slowest. It is interesting to see how fast Dante is in comparison to AES50.