Capture control¶
The Capture panel is the one that starts things: what the capture is called, what it is written as, and the buttons. Which device and which folder are set once in Settings — see below.
Monitoring and capturing¶
Two buttons, and the difference between them is the whole model.
Start monitoring opens the device and runs the stream with no file on the end of it. The samples are validated, measured and drawn, and nothing is written anywhere. Use it to set a player's RF output, to check a cable, or to decide whether a disc is worth capturing. It costs nothing to leave running.
Start capture attaches a file to that stream. From idle it starts the stream as well, so the common case is one press rather than two. From an existing monitoring session the file is attached at the next buffer boundary and the stream is not interrupted — the device never knows a capture began.
Stop capture detaches the file and finalises it, and leaves the stream running. That is what makes both sides of a disc possible without reopening the device: stop, turn the disc over, start again.
Stop monitoring ends the stream. It is disabled while a capture is running, because a capture is this stream with a file on the end of it — stopping the stream would end the recording too, which made this a second, unlabelled stop button sitting directly above the real one. Stop the capture first; that leaves the stream running, so nothing is lost by taking the shortcut away.
Both buttons take on a colour while they are doing something, so the state reads from across a bench rather than only by reading the label: green while monitoring, red while capturing. The red says recording — it is a normal state, not a fault.
Automatic capture… is the third button, and it is greyed out until player control is on and a player is connected — present rather than absent, so that a panel with no player attached still says the path exists. It opens the four-page workflow that examines the disc, names the capture from what it found, takes the side, and reports what was written. It is a way into that window rather than a mode the panel enters: the format and sample rate below are the same settings the workflow's own second page shows, and the two cannot disagree.
Both buttons have a command-line equivalent, for when something other than a person has to press them — an audio recording that must start at the same moment, or a capture nobody is going to be in the room for. A capture started from a script can be stopped from this panel and one started here can be stopped from a script; they are the same capture either way. See Scripting captures.
Where the device and the folder went¶
Both are on the Capture tab of File ▸ Settings… rather than on this panel. 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.
Which device. Preferred device in Settings, which offers Whichever is attached by default and is what almost everybody leaves it at. The list names what is wrong with a device rather than hiding it:
| Entry | What it means |
|---|---|
| A bare path | A capture device, ready |
| — recovery mode, no firmware installed | The device's USB chip has no firmware it will run. It cannot capture. Tools ▸ Firmware ▸ Bring up a new or legacy board… programs it whatever state it is in; Update firmware… repairs it if it was working before |
| — original firmware, too old for this application | The device is running the firmware it had before this application existed. It works, but nothing here can talk to it — see Bringing up a new or legacy board |
| — connected at insufficient speed | It enumerated below SuperSpeed. It is on a USB 2 port or through a hub that is, and cannot carry 80 MB/s |
A device with no firmware is listed rather than hidden. Reporting "no device attached" to somebody looking straight at one is exactly the moment they most need to be told what to do next — which is why the Status line at the bottom of this panel says the same thing without your having to open Settings at all.
Where the file goes. Folder in Settings, with a Browse… beside it. 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 below stays on this panel, because it is live and it is about the capture you are about to take. Hover it and it names the volume it is talking about.
Test mode lives on the Tools ▸ Test data submenu, for the same kind of reason: it is a diagnostic rather than part of setting up a capture. The full procedure, and how to check a test capture afterwards, is on Test mode and integrity checking.
What gets written¶
Name¶
Leave it empty and each capture is named after the time it was taken —
RF-Sample_2026-08-16_14-30-00 — in local time, which is what keeps a folder of captures in
order and matches what the person who took them remembers.
Type a name and it is used instead. Characters that would not survive a copy to another
machine are removed: Windows rejects <>:"/\|?*, treats a trailing dot or space as
invisible, and reserves a list of device names. A name that works where it was typed and
fails on a colleague's machine is prevented here rather than discovered later.
In test mode the field is disabled, not merely ignored. A field that accepted text and then did not use it would be a lie about what the application was going to do.
Naming… beside the field opens the other way of naming a capture: say what the disc is — title, type, standard, side, notes, mint marks — and let a name be built from it. All of it is recorded in the metadata file written beside every capture, whether or not any of it reaches the file name. See Naming and metadata.
A name typed here wins outright over anything set there, and the dialog says so rather than leaving it to be discovered.
The Naming… button colours itself when nothing at all has been said about the disc — no
name typed, no field ticked — so that a capture is not filed under a timestamp by accident.
It is a nudge and not a warning: an unnamed capture is a legitimate way to work, the colour
is the theme's accent rather than an error red, and Start capture is never held up by it.
The colour clears on the first keystroke in the Name field rather than when the field loses
focus, so it is not arguing with you while you type. Test mode suppresses it, because there
the name is forced to TestData_ and there is nothing to offer.
Format¶
| Choice | What you get |
|---|---|
FLAC — .ddd.flac |
Mono 16-bit native FLAC, roughly half the size, carrying the capture's provenance in its tags. The default |
Uncompressed — .ddd.s16 |
The same samples with nothing wrapped round them: signed 16-bit little-endian, no header |
Uncompressed is twice the disk for none of the encoder, which is the trade worth having on a machine that cannot sustain the encode or when the output is going straight into another tool. Nothing in the file says what it is, what rate it was written at or which build produced it — that is the format's nature, and the reason FLAC stays the default.
An uncompressed capture can be encoded to FLAC afterwards, which is the usual thing to do when the format was chosen because the machine could not sustain the encoder live:
flac -8 --force-raw-format --endian=little --sign=signed \
--channels=1 --bps=16 --sample-rate=40000 \
capture.ddd.s16 -o capture.ddd.flac
The raw file says nothing about itself, so every one of those has to be supplied — including
--sample-rate=20000 in place of 40000 for a capture taken at 20 Msps. The full
walkthrough, and what to do about the tags the conversion cannot recover, is on
Capture files.
Both are read back by Tools → Test data → Analyse test data… and by --analyse-test-data.
Sample rate¶
| Choice | What you get |
|---|---|
| 40 MSPS for LaserDisc | Every sample, the converter's own rate. The default |
| 20 MSPS for VHS | Half the rate, half the file |
The choices are named by what they are for rather than by what they do to the samples: "2:1 decimated" is a fact about the implementation, and the decision being made here is which format is going through the machine. VHS names the common case rather than the only one — Betamax, Video8 and any other tape format are the same choice, because what they share is a bandwidth that is a fraction of a LaserDisc's.
The converter always runs at 40 Msps. Decimation happens in the FPGA, not on this machine, and that is what makes it worth having: halving the rate correctly means low-passing the signal at 10 MHz first, and the gateware does that with a 63-tap half-band filter costing 13% of the FPGA's logic and no CPU at all. The application asks for the rate over the register link and receives a stream that is already half rate — so the USB link carries half the data too.
Without the filter, decimation would fold everything above 10 MHz down on top of the signal: a 15 MHz component would reappear at 5 MHz, directly on top of a tape's luma FM carrier, and nothing downstream could tell the alias from the signal. The filter is flat to within 0.0015 dB up to 8 MHz and 85 dB down at 15 MHz, so that fold lands well below the converter's own noise floor. It also delays every frequency by the same amount, so an FM carrier and its sidebands arrive together rather than being smeared apart.
What no filter can do is protect the band edge itself. The response passes −6 dB at exactly 10 MHz and is symmetric about it, so energy just above 10 MHz still folds down to just below it. A signal with real content near 10 MHz should be captured at the full rate — that is a property of halving a sampling rate, not a shortcoming of this implementation.
Tape RF has a fraction of a LaserDisc's bandwidth, which is what makes 20 Msps enough for it.
The design, the coefficients and the measured response and phase are on The decimation filter.
The setting is written to the device before the stream is opened, so it is fixed from the moment monitoring starts rather than only for the duration of a capture — there is no way to change it under a running stream, and a control that stayed live would appear to work and do nothing. Stop monitoring to change it.
The signal analysis panels follow it: at 20 Msps the spectrum's axis tops out at 10 MHz, the scope's spans cover twice the time, and the bins are half as wide. A display that kept the converter's rate would put a tape's 5 MHz carrier at 10 MHz.
A decimated FLAC capture carries the rate label for 20 Msps and a DDD_DECIMATION tag. A
decimated .ddd.s16 carries nothing at all, because there is nowhere to put it: write the
rate down.
It works in test mode too. The gateware generates its test pattern downstream of the decimator, so a decimated test capture is an unbroken ramp at 20 Msps and Tools → Test data → Analyse test data… checks the decimated path exactly as it checks the full-rate one.
Compression¶
FLAC compression, 0 to 8. The default of 8 gives the smallest file and is what a multithreaded libFLAC sustains at the device's full rate. It is disabled for the uncompressed format, which has no encoder to ask.
Lowering it is the first remedy for a machine that cannot keep up. The Encoder backlog and Buffer queue figures in Statistics are what say whether that is the problem: a backlog that climbs is the encoder, a queue that climbs with no backlog is the disk.
Duration limit¶
Stop automatically after this many minutes, or No limit, which is the default. The stop lands on a buffer boundary, so nothing is half-written. Reset puts it straight back to No limit — a limit tends to be set for one capture, and holding the down arrow back from forty minutes is forty presses.
The limit is read on every statistics tick rather than latched when the capture starts, so both it and Reset stay live during a capture: deciding halfway through a side that the limit should go is a reasonable thing to want. It is a length of time, so a decimated capture still runs for as long as it says.
The default is no limit deliberately: a limit that fired in the middle of a side would be worse than no limit at all.
Warn below¶
Warn when the destination volume has less than this much capture time left. Ten minutes by default; Never turns it off.
It is a warning and not a refusal, and it is raised once per capture rather than repeatedly. The estimate is an estimate, and an application that declined to start because it predicted a shortfall would sometimes be wrong in the direction that costs a session.
Free space¶
How much longer this volume will hold a capture, with the byte figure after it. Hover it and it names the folder it is talking about — that folder is in Settings now, so the row says which volume it means rather than leaving you to remember.
The time comes first because it is the question people actually have. "412 GB free" does not tell you whether this will last the side you are about to play; "2:51:40 of capture" does.
The estimate follows the Format and Sample rate you have chosen: 40 MB/s for FLAC, which is what one writes on average, 80 MB/s uncompressed, and half of either when decimating. The same volume therefore holds four times as much 20 Msps FLAC as 40 Msps uncompressed, and the readout says so.
An unknown figure means the folder does not exist yet, which is an ordinary thing to have on the way to creating it — it is not reported as a full disk. It names the folder when it says so, because you cannot read it off this panel.
Status¶
The line at the bottom of the panel: No capture device attached, Ready, Monitoring, or the full path of the file being written. The whole path, not just the name — somebody who has just started a forty-minute capture should be able to see that it is going to the drive they meant without leaving the application.