Settings

File → Settings… holds the things that are set once and left alone. Everything to do with an individual capture — the name, the format, the compression, the duration limit — is in the Capture panel instead, where it is visible while you work.

Its Capture tab is below, and it opens with the two that used to be on that panel: Folder and Preferred device. They moved here because neither changes once it is set — a Duplicator does not move between USB ports and captures do not move between drives — and a control that is set once does not earn a row on the panel you work from.

Its Player tab — which model, which serial port, which speed, the ports never to open, and the one coupling between player and capture — is documented on Player control, beside the rest of what it configures.

When each one takes effect

The buffer and transfer settings resize things a running capture is using, so they apply to the next capture. The front-end gain declaration only changes what a number is labelled, so it applies at once, including to a capture already running.

Folder

Where captures are written. Browse… picks one, or type a path.

It defaults to the platform's Movies or Videos folder — a capture is tens of gigabytes of media, and that is the location a machine's backup rules and disk-space expectations are already built around.

The Free space row on the Capture panel is about this folder, and names it in its tooltip.

Preferred device

Which Duplicator to use, when more than one is attached. Whichever is attached is the default and is what almost everybody leaves it at.

The list names what is wrong with a device rather than hiding it — recovery mode, no firmware installed for one whose USB chip has nothing to run, connected at insufficient speed for one below SuperSpeed. Both are still offered, because both are the same physical port: one is a programming away from working — see If an update fails for which of the two kinds of programming — and the other a cable.

The Capture panel's status line reports the same diagnosis for whichever device is actually selected, so a device that cannot capture says so without your opening this dialog.

Buffer queue

How much sample data this machine will hold while waiting for the disk and the encoder.

Choice	Slack it buys
64 MB	0.8 seconds
128 MB	1.6 seconds
256 MB	3.2 seconds (default)
512 MB	6.5 seconds

The seconds are the figure that matters — the megabytes are just how it is spent. This is how long a write stall can last before samples start being lost, so raising it is one of the three remedies for a machine that cannot keep up (the others being a faster drive and a lower compression level).

On Linux there is a ceiling that is not the application's. The kernel's usbfs memory limit caps what can be queued, and a queue above it fails the capture with a specific message and the command to raise it. Lowering the queue size here is the remedy that needs no administrator rights. See If a capture fails.

USB transfers

Many small transfers (recommended) keeps several reads outstanding at once, so the device always has somewhere to put the next packet.

One transfer per buffer is simpler and slightly cheaper, but leaves a gap between each transfer completing and the next being submitted — and the device has nowhere to put data during that gap.

Change it only in response to a specific failure: this machine did not keep a read request outstanding is the message that points here.

Front-end gain

This is the one setting worth reading about rather than just choosing.

Why you have to declare it

The Duplicator's analogue front end is an amplifier whose gain is set by SW401, a four-way DIP switch on the board. The switch is mechanical and has no electrical path to the FPGA or the FX3 — nothing in the sample stream, the USB descriptors or the vendor requests reveals its position. The application cannot discover it and must be told.

Everything else follows from that one fact:

• There is no default. While the setting is undeclared, no display shows a voltage — levels read in converter codes, 0 to 1023. A plausible default would put authoritative-looking millivolt figures on screen that could be wrong by up to a factor of four, with nothing to reveal it. That is worse than showing nothing.
• Nothing about the capture depends on it. Samples are stored and written as the codes the converter produced. This is a display calibration, applied where the numbers are drawn — so getting it wrong and correcting it later re-scales every figure on screen, with nothing to redo.
• Clipping never depends on it. A clipped sample is one whose code reached 0 or 1023, which is a property of the converter. The clip counts stay correct whether the declaration is absent, right or wrong.

Reading the switches

The switch block is written as it sits on the board, left to right: 1010 is switches 1 and 3 closed, and 2 and 4 open. Each picture below is SW401 as it should look, with the closed sliders pushed towards the ON legend.

SW401	Switches	Gain	Aim for (75 %)	Clips above
	1000	×8.50	176 mV p-p	235 mV p-p
	0100	×6.00	250 mV p-p	333 mV p-p
	0010	×4.40	341 mV p-p	455 mV p-p
	1100	×4.00	375 mV p-p	500 mV p-p
	0001	×3.80	395 mV p-p	526 mV p-p
	1010	×3.34	449 mV p-p	599 mV p-p
	1001	×3.04	494 mV p-p	658 mV p-p
	0110	×3.02	496 mV p-p	661 mV p-p
	0101	×2.79	537 mV p-p	716 mV p-p
	1110	×2.59	578 mV p-p	771 mV p-p
	0011	×2.54	592 mV p-p	789 mV p-p
	1101	×2.45	613 mV p-p	817 mV p-p
	1011	×2.27	659 mV p-p	879 mV p-p
	0111	×2.17	690 mV p-p	920 mV p-p
	1111	×2.02	744 mV p-p	992 mV p-p

All four switches open is not a setting: with no feedback path the amplifier has no defined gain. That pattern is what carries the undeclared state.

Aim for is the level to set a source against — 75 % of the converter's range, which is what the Amplitude History strip marks. Clips above is where the converter runs out outright. The headroom between the two is what absorbs the moments a disc is worse than the moment you set the gain on.

Once it is declared

Every level display gains a millivolt figure at the BNC: the Statistics panel's signal level and extremes, the waveform cursor, the amplitude summary. The declaration is also written into each capture file's tags, so the calibration needed to read those samples as volts travels with them.

It is remembered between sessions. A gain switch stays where it was put, and asking again every session for something that has not changed is how a setting ends up ignored.

What is not here

Test mode is in the Capture panel and is deliberately not remembered between sessions. It is a diagnostic, and an application that silently started in test mode because of something you did last week would produce a capture full of ramps.

Where settings are kept

Platform	Location
Linux	~/.config/Domesday86/ddd-gui.conf
macOS	~/Library/Preferences/com.domesday86.ddd-gui.plist
Windows	HKEY_CURRENT_USER\Software\Domesday86\ddd-gui

These are the application's own. The capture application this one replaced used a different identity, so installing this one never disturbed that one's settings.

A settings file that has been edited by hand, or written by a different version, is clamped to sensible values rather than rejected — an out-of-range value produces a working capture rather than a refusal.