Three desks, three numbers near seventy-two. They count different things.
Every figure below is the manufacturer’s own published specification, read from its product pages on October 23, 2026.
- Yamaha CL5 — 72 mono + 8 stereo
Yamaha prints the CL5 mixing channel count as 72 mono plus 8 stereo, with the CL3 at 64 mono plus 8 stereo and the CL1 at 48 mono plus 8 stereo. The stereo channels are counted separately from the mono ones, so the number a channel list has to reconcile against is eighty paths, not seventy-two.
- DiGiCo Quantum 225 — 72
DiGiCo prints “Input Channels: 72”, with up to 36 aux and group busses carrying full processing in mono, stereo or LCR plus a master, and a matrix of up to 12 inputs by 12 outputs, also with full processing. The bus count is the one to read on a festival: monitor sends, side-fills, broadcast and a delay tower all come out of it.
- Allen & Heath Avantis — 64, or 96
The lowest headline figure of the three, and the only one that moves with a software licence. Allen & Heath publishes “64 Channel / 96 with dPack” and “42 Bus / 56 with dPack”, so the same surface is specified twice and the second specification is bought rather than shipped.
The pipe off the stage is the real ceiling, and on one of these desks the sample rate halves it.
Channel count at the surface costs nothing. Channel count across the network is bought, and it is specified at a stated sample rate.
On the CL5 Dante is built in. Yamaha publishes 64 in and 64 out on the CL5 and CL3, 32 in and 32 out on the CL1, and an internal sampling rate of 44.1 kHz or 48 kHz. There is no 96 kHz row to trade against, so the published Dante figure is the figure.
On the Quantum 225 Dante arrives on the optional DMI-DANTE card. DiGiCo publishes that card at 64 inputs and 64 outputs at 48 kHz, or 32 and 32 at 96 kHz, and the console carries two DMI slots, each adding 64 by 64. Running the desk at 96 kHz halves what each card carries. Its native transports are MADI and Optocore: four MADI interfaces at 48 kHz and two at 96 kHz, and an Optocore loop carrying 504 audio channels and up to 14 racks, or 1,008 channels and 28 racks on dual loops.
The Avantis publishes two I/O ports, each rated 128 by 128 at 96 kHz, with Dante, MADI and gigaACE available as option cards and an SLink port for remote expanders. Of the three it is the one that states its port capacity at 96 kHz rather than at 48, which is the comparison a spec sheet makes easy to get wrong.
All three publish a latency figure. Only one of them includes the converters.
The three numbers look comparable on a page. They are measured across different spans, and the spans are printed beside them.
- Yamaha — “Less than 2.5ms, OMNI IN to OMNI OUT, Fs=48kHz”
That span starts at an analogue input and ends at an analogue output, so it contains the analogue-to-digital and digital-to-analogue conversion as well as the mixing. It describes the whole path a microphone takes through the desk.
- DiGiCo — “Processing Delay: 2ms @ 48kHz, 1ms @ 96kHz”
Labelled processing, which is the mix engine alone. Conversion and transport sit outside the figure, and on a console whose native transport is an optical loop that is a deliberate way to state it.
- Allen & Heath — 0.7 ms
Published for the XCVI 96 kHz FPGA core, and described on the same page as class-leading. Again the core, measured without the converters at either end.
Changeover is a surface problem, and these three surfaces are built on opposite bets.
Physical strips and layers trade against each other. Each manufacturer publishes where it landed.
Yamaha publishes the CL5 fader layout as 16 + 8 + 8 + 2, the largest of the three CL surfaces; the CL3 is 16 + 8 + 2 and the CL1 is 8 + 8 + 2. DiGiCo publishes 25 motorised, touch-sensitive 100 mm faders on the Quantum 225, described as two banks of twelve plus a user-assignable master. Allen & Heath publishes 24 faders over 6 layers on the Avantis, stated as 144 fader strips, with two 15.6-inch Full HD touchscreens, 24 assignable SoftKeys and 16 DCAs; the Avantis Solo halves that to 12 faders, 6 layers and one screen.
The festival reading of those three rows is about how many layer changes a ten-minute changeover contains. A surface with more physical strips puts more of a band on the desk at once. A surface with more layers puts the same band behind a layout the engineer has to have built in advance, which is a rehearsal question rather than a hardware one.
The specifications settle more of this than a demo does.
Four of the figures above decide a festival build before anyone hears the desk.
The ceiling on inputs is the network figure and not the channel count, because the channel count is what the surface will mix and the network figure is what the stage can deliver. The sample rate is a capacity decision on one of these desks and a fixed property on another. The latency figures rank in the opposite order to the spans they describe, so the desk that published the most complete measurement reads worst if the four numbers are sorted as printed. And the surface decides the changeover.
All four are printed on pages the three manufacturers publish and keep current, which is where every figure in this article came from and where the next revision of it should be checked.
A console comparison that ends in a winner has usually compared the wrong things. What the published specifications settle is narrower and more useful: how many paths come off the stage at the sample rate the show will run at, how many busses are available once monitors and broadcast have taken theirs, how many strips the engineer has under both hands during a changeover, and what each latency figure was measured across. Those are readable before a desk is booked, and they are the ones that are expensive to discover on site.
Where the figures come from
- 72 mono + 8 stereo; 64/64 Dante; 44.1 kHz / 48 kHz; 16 + 8 + 8 + 2 faders; “Less than 2.5ms, OMNI IN to OMNI OUT, Fs=48kHz” — Yamaha’s published CL Series specifications page, usa.yamaha.com, October 23, 2026. The CL3 and CL1 figures quoted in §01 and §04 are from the same table.
- 72 input channels; up to 36 aux/group busses; 12 × 12 matrix; 25 × 100 mm faders; 48 kHz / 96 kHz; “Processing Delay: 2ms @ 48kHz, 1ms @ 96kHz”; DMI-DANTE 64×64 @ 48 kHz or 32×32 @ 96 kHz; Optocore 504 channels / 14 racks per loop — DiGiCo’s published Quantum 225 console page, digico.biz, October 23, 2026.
- “64 Channel / 96 with dPack”; “42 Bus / 56 with dPack”; XCVI 96 kHz FPGA core at 0.7 ms; 24 faders, 6 layers, 144 fader strips; 16 DCAs; 2 I/O ports at 128 × 128 at 96 kHz — Allen & Heath’s published Avantis hardware page, allen-heath.com, October 23, 2026. The Avantis Solo figures in §04 are from the same page.