De-Feedback Setup and Hardware Configuration

Step 1

What you need and how it connects

SNUC option 2 mini PC, the configuration Alpha Labs specify, preconfigured with the De-Feedback VST host
Focusrite Scarlett interface, 2 or 4 channel, fourth generation
USB cable from the interface to the SNUC
XLR cables for your microphone sources and your outputs
Monitor, keyboard and mouse for setup
Power, on the same protected circuit as the rest of the rack

Microphone→Focusrite input→USB→SNUC and VST host→Focusrite output→Console or amp

Connect the Focusrite to the SNUC before you power the SNUC on. Windows will enumerate the interface at boot and the driver will be there when the host launches, which avoids a common cause of a missing audio device on the day.

Step 2

First boot and driver installation on the mini PC

Press the power button on the SNUC mini PC. Windows boots on its own and no login is required on a preconfigured unit. If this is a fresh mini PC, complete the Windows setup and connect it to a network long enough for driver updates to come down.

Install Focusrite Control from the Focusrite site rather than relying on whatever Windows finds on its own. Focusrite Control is what gives you per-channel gain, phantom power and the routing view, and the Focusrite ASIO driver installs alongside it.

Once the audio interface appears in Focusrite Control with its inputs metering, the mini PC is ready to configure and you can launch the De-Feedback VST host.

Step 3

Focusrite audio interface configuration

Set the audio interface before you touch the VST host, because every level inside the host depends on what the Focusrite is handing it.

Connect the Focusrite to the mini PC over USB and confirm it is powered
Open Focusrite Control and confirm the interface is detected
Enable phantom power only on the channels that need it
Set input gain so speech peaks between roughly minus 18 and minus 12 dBFS
Turn direct monitoring off, so nothing bypasses the De-Feedback processing
Confirm the Focusrite outputs are routed to the console or amplifier inputs you patched

Direct monitoring is the one that catches people out. If the Focusrite is passing the input straight to its own outputs, part of your signal is reaching the speakers without going through the VST host at all, and the processing will look like it is doing nothing.

On the four channel audio interface, set gain per channel independently. Inputs 1 and 2 usually carry the microphones being processed, and inputs 3 and 4 carry line-level sources you would rather leave alone.

Step 4

De-Feedback license activation and deactivation

Open the De-Feedback VST host
Go to Settings, then License
Enter the license key supplied with the system
Click Activate License, then restart the VST host
Confirm the license status reads as active before you go any further

License activation needs an internet connection once. After that the mini PC runs offline, which is how it should be run at an event. The De-Feedback license is tied to that specific hardware, so keep the license key somewhere separate from the machine it is installed on.

To move the license to another mini PC, deactivate it first: Settings, then License, then Deactivate License, and confirm. Once the license shows as deactivated it can be activated on the replacement unit. Deactivating before a hardware swap saves a support call later.

Step 5

Audio device settings in the De-Feedback VST host

This is the part that decides whether the system is usable, so set it deliberately rather than accepting the defaults.

Audio driver: Focusrite USB ASIO
Sample rate: 48 kHz
Buffer size: 16 samples
Input device: Focusrite Scarlett, the channels you patched
Output device: Focusrite Scarlett, the outputs feeding your console or amp

16 samples at 48 kHz is about a third of a millisecond per buffer, and it is the setting this system is specified at. On most rigs a buffer that small is asking for trouble, because the machine underneath it is an unknown. Here it is not: the SNUC option 2 mini PC is the configuration Alpha Labs specify for the De-Feedback VST host, so the buffer size was chosen against hardware that was tested at it. That fixed pairing is what keeps the whole processing path under 2 milliseconds.

Raise the buffer only if you have put other software on the machine and are hearing crackling, and treat that as a sign something else is competing for the CPU rather than as the normal setting.

Leave the Windows sound settings out of it. Set the Focusrite as the Windows default so nothing else grabs the interface, turn off audio enhancements, and then do all your real work through ASIO.

Step 6

Patching and gain structure

Where you insert the processing decides what it can do. These are the three configurations we use.

Insert across the console output

Everything at once

The simplest patch and the one used for most demos. The whole mix passes through, so every open microphone benefits. Playback and program material go through it too, which is fine because the processing only acts when a frequency is building.

Insert across a subgroup

Microphones only

Bus your live microphones to a subgroup and insert there. Music playback and video audio bypass the processing entirely. This is the configuration to use in a room where the loop is only ever caused by the microphones.

Insert per channel, 4 channel unit

Individual sources

With the four channel interface you can process the two or three problem sources independently, for example a pastor's lapel and a roaming handheld, and leave the rest of the desk untouched. Keep a spare output as an unprocessed monitor feed for comparison.

In every configuration, set gain at the interface first and levels at the console second. Aim for speech peaking around minus 18 to minus 12 dBFS at the interface. Then verify the unprocessed path and the processed path sit at the same level, so bypassing is a comparison rather than a volume change.

Step 7

VST host controls and what each one does

Input gain minus 12 to plus 12 dB

Trims the level arriving at the processing. Use it to land in the right range if you cannot get there at the interface, and leave it at zero if you can.

Sensitivity 0 to 100 percent

How readily the detector decides a frequency is building. Between 40 and 70 percent covers most rooms. Higher settings act sooner and are more likely to catch a sustained note that was meant to be there.

Response time 1 to 100 ms

How quickly suppression is applied once a frequency is flagged. Faster catches a ring earlier. Slower rides over short transients and sounds more natural on music.

Bypass on or off

The comparison control. Level-match the two paths before you use it in front of anyone, because an unmatched bypass will make the processing sound better than it is.

Step 8

Startup routine and troubleshooting

Startup routine

Connect the Focusrite audio interface, then power the mini PC
Launch the De-Feedback VST host and confirm the Focusrite ASIO driver is selected
Check meters on every input you are using
Send a test signal through the full path to the speakers
Flip bypass once to confirm both paths are level-matched

No audio through the system

Confirm the USB connection and that the Focusrite is powered
Confirm the VST host is set to Focusrite USB ASIO rather than Windows audio
Confirm the VST host input and output devices match the patched Focusrite channels
Confirm Focusrite input gain is up and the channel is not muted
Restart the VST host before restarting Windows

Crackling, dropouts or high CPU

Confirm the buffer size is still at 16 samples and nothing has reset it
Close anything else running on the machine
Confirm Windows is not installing updates during the event
Check the CPU meter in the VST host while a real signal is passing
Make sure the unit has airflow and is not sitting on a warm amp

Between events

Apply Windows updates on a weekday, never on a show day
Keep the Focusrite audio interface driver current
Write down the settings that worked in each room you return to
Keep the De-Feedback license key stored away from the machine
Run a test with the microphones you will use on the day

Questions

Setup and troubleshooting questions

What buffer size should I use?

16 samples at 48 kHz. That is the setting the system is specified at, and it is roughly a third of a millisecond per buffer. It runs there because the hardware is fixed: the SNUC option 2 mini PC is the configuration Alpha Labs specify for the De-Feedback VST host, so the buffer size was chosen against known hardware rather than against whatever machine happened to be available. Raise it only if you have added other software to the machine and are hearing crackling.

Why ASIO and not Windows audio?

The Windows shared audio path adds tens of milliseconds and gives you no control over buffer size. Select the Focusrite USB ASIO driver in the De-Feedback VST host and leave Windows audio out of the signal path entirely.

The plugin is not passing audio. Where do I start?

Check the four things that cause it most often: the Focusrite is connected and powered, the VST host is set to the Focusrite ASIO driver rather than Windows audio, the input and output devices inside the host are pointed at the Focusrite channels you patched, and the input gain on the interface is up. If all four are right and there is still nothing, restart the host before you restart Windows.

What input level should I run?

Set interface gain so normal speech peaks between roughly minus 18 and minus 12 dBFS. That leaves headroom for someone who gets excited. Watch the output meters for clipping separately, because the processing can raise level in passages where it removes a suppressed frequency.

How do I move the license to another machine?

Deactivate it on the current machine first, in the De-Feedback VST host under Settings and then License. Once that license shows as deactivated it can be activated on the replacement mini PC. Keep the key somewhere other than on the machine it is installed on.

Still stuck? Email info@dallaseventaudio.com and you will get a person who has run this rig in a room.

De-Feedback setup and hardware configuration