bluefin2 card

Essential Buyers Guides: Don’t get short-changed on DSP

By Mike Reddick, European Sales Manager, Calrec

Most guides to buying audio consoles recommend looking at channel facilities; how many you need and what your future requirements might be. This is of course very sensible, but Calrec's European Sales Manager Mike Reddick tells us that not all channel counts are created equal. If you are asking questions about channels, you need to read this first.

When comparing digital consoles’ DSP capabilities, care should be taken comparing “like for like” specifications. The difference between what you think you have and what you really have may be significant depending on whether the console uses shared DSP resources or not.

Most console manufacturers use off the shelf DSP devices, such as Sharcs, to provide DSP power. These devices have a fixed architecture which lends itself very easily to the concept of DSP sharing whereby a pool of DSP resource is shared to provide input channels and output busses.

This seems like an attractive solution – the flexibility to move processing around between inputs and outputs is a good thing, right?

But this “flexibility” means that working out a console’s capabilities is not always straightforward. If a console is described as having 240 DSP channels, what does that mean?

Well, firstly it probably doesn’t mean channels in the conventional sense. In analog terms, a 240 channel console would have just that: 240 input channels, and a separately specified number of main outputs, groups, auxes etc. With shared DSP, a console defined as having “240 channels” will have 240 DSP channels which can be used as input channels, main / group / aux busses, tracks etc.

Additionally when specifying DSP capacity, manufacturers usually base it on mono channels, so stereo channels will take 2 x DSP channels, and surround channels will take 6 x DSP channels. So if you build a configuration with 8 stereo channels, this will reduce the remaining available channels by 16. Adding 4 x 5.1 channels will reduce the remaining available channels by 24.

The great advantage of Calrec’s Bluefin2 FPGA based DSP is that there is no sharing of DSP resources between input channels and output busses. Therefore a “240 channel” console really does have 240 input channels, and this resource will never be depleted by adding more output busses.

As an example, consider the specification of Calrec’s Artemis Light console:-

240 channels
Up to16 Main Busses
Up to 48 Group Busses
Up to 48 Track busses
Up to 24 Aux busses

At first glance the Artemis Light appears to be a 240 channel console, comparable with other manufacturers’ 240 channel consoles, or even inferior to other consoles which may claim to have more channels. However, the crucial difference between shared and non-shared DSP does not become apparent until you start to build real desk configurations.

Let’s compare an Artemis Light with an alternate console which uses shared DSP. Any configuration you build must have output busses as well as input channels. This is where the difference between shared and non-shared DSP really takes effect.

In a shared DSP system, if you start off with 240 DSP channels and then add (for example) 2 x stereo main outputs, 4 x stereo groups, 8 x mono Aux outputs, the available DSP channels drops by 20. The problem becomes even more acute when you are working in surround. For example, allocating 2 x surround Main outputs, 4 x surround Groups, and 8 x mono auxes will suck 44 x DSP channels from your capacity. Suddenly that 240 channels capacity doesn’t seem so powerful.

The following series of charts show what happens with shared and non-shared DSP when you start to build simple, real-world configurations.

Our starting point is both consoles with a “240 channel” specification:

DSP fig 1

It looks equal. You could even start off with a shared DSP console with a higher number of channels.

Now, let’s add some mains, groups and auxes:

DSP fig 2

Notice that adding 4 x Stereo mains, 8 x Stereo groups and 8 x mono Aux busses, has used 32 x DSP channels and these have been removed from the available input channels of the shared DSP. Non-shared DSP, like Calrec’s Bluefin2, still has the full 240 channels available.

Let’s continue this exercise:

DSP fig 3

The addition of 24 x track outputs has reduced the input channels remaining on the shared DSP console to 184, whereas the Calrec DSP is unchanged. This effect becomes even more pronounced when configuring a console for working in surround sound. Let’s add in 4 x 5.1 surround mains and 6 x 5.1 surround groups to our example:

DSP fig 4

The effect of adding surround busses is quite dramatic. From an original capacity of 240 channels, now there are only 116 shared DSP channels remaining, whereas the Calrec DSP has still has 240 channels available.

And don’t forget, these are mono input channels. If we now change some of these to stereo or surround, this will reduce the remaining channels for both the shared and non-shared DSP consoles. Let’s do that now by adding in 4 x 5.1 surround input channels and 8 x stereo input channels:

DSP fig 5

Both DSP systems have used input resources, but as the Calrec console still had the full 240 input channels available there remains a healthy 200 DSP channels, whereas the configuration of output busses on the shared DSP system has significantly reduced the number of channels remaining – in this example only 76 shared DSP channels remain, less than a third of the original specification.

The difference between shared and non-shared DSP makes a huge impact on desk resources. It is always better to have resources you don’t need when you are building a show, rather than having to count your inputs and outputs and worry about how much DSP you have left. That’s why Calrec desks have all DSP resources on all channels at all times.

These are worthwhile questions to ask when specifying a new console – is the DSP shared or non-shared? Check. Don't get short changed.


Leave a Reply