RVFD · Radio Bridge · Field Guide

Fixing the RoIP receive audio

Everything works except the RX level. Audio reaches the DRA-30 but lands about 37 dB too low. This walks you to the cause at the rack.

Do not disturb

Safety, before anything

4P4C pin 2 carries +14.2 V @ 1 A and will destroy the DRA-30. You already metered it and confirmed the splice is clean — so do not rework that connector.

Mic jack (RJ45) pin 2 is a live 0–5 V keypad line sitting directly beside pin 3, your PTT conductor. A short there keys a licensed Part 90 transmitter. Leave that splice alone too.

Never connect: 4P4C pin 1 or pin 2 · DB9 pins 2, 4, 7, 8.

Already fixed — do not redo

Signaling (media_address)FIXED & VERIFIED

rxdemod flat → speakerAPPLIED

rxboost → no (clip guard)APPLIED

Rx Level 100 → 500+9.5 dB RECOVERED

Cable, splice, pinsVERIFIED GOOD

Carrier detect = voxCORRECT AS BUILT

What you're bringing

Small flat screwdriver for the trimmer · your phone (this page) · a handheld to key · laptop with CPS only if you land on Path B. The Pi and DRA-30 are already in the rack. No new cable, no DB25, no attic.

Step 1 — Look at R12 before you turn anything

DRA-30 circuit board showing trimmer locations — **R12** is the large blue trimmer at lower-center with the white slotted screw — that's the RX level. The two smaller blue trimmers upper-right are **R14 / R16**, which are TX levels. **Do not touch those.**

Open the DRA-30 enclosure.
Find R12, lower-center, large blue trimmer.
Note where the slot is pointing before you move it. Photograph it. That's your undo.

Decision point

Where is R12 sitting?

Near minimum or mid-travel →Path A. That's very likely your missing 37 dB. Already near maximum →Path B. R12 is cleared; the repeater isn't delivering.

Path A — Sweep R12 with live measurement

This is the likely path. We tune it against real numbers rather than by ear, and land it on the manufacturer's target in one pass.

Tell me you're ready. I'll stop Asterisk to free the sound card and open a recording window.
Key a steady tone and hold it. A constant carrier reads far more stably than talking.
Turn R12 slowly clockwise while it records.
I read the level back in dB. We stop when it hits target.
I restart Asterisk and confirm the setting holds.
Reconnect the app, key the handheld, listen.

Target

What "correct" looks like

Masters' published calibration: with squelch open and nobody transmitting, the channel noise should read

−10 to −11 dBFS

Calibrating on noise looks strange but it's deliberate — open-squelch noise is about the same amplitude as full-deviation voice, and it's a stable reference you can generate on demand. Set noise there and voice lands correctly without clipping.

Path B — R12 is already maxed

Then R12 is exonerated and the MTR3000 isn't putting out its nominal 500 mV. Prime suspect: a station alignment setting on a codeplug last written in 2015.

Open MOTOTRBO CPS and read the codeplug from the repeater.
Find the station alignment section and locate Speaker Level.
Write down the current value before changing it.
If it's low, raise it, write the codeplug, and re-run the Path A measurement.
Do not touch "Disable Repeat Path." The repeater must keep repeating for the field independently of the node.

If both are maxed

Then that jack genuinely can't deliver, and the answer is the rear J7 accessory connector — pin 7 = Rx audio, 330 mV, 1000 Ω, ground on pin 9. You already have DB25 shells from the studio.

Before assuming that means new cable: J7 is the same chassis in the same attic. Measure whether the existing run has slack to reach the rear. If you do move, move everything to J7 — audio, PTT and a GPIO-programmed COS on one connector — rather than splicing RX alone.

Reference

Getting into the node

ssh asl@100.98.1.10        # Tailscale, works from anywhere
# hall LAN 192.168.0.37 — changes with location, don't rely on it

Useful commands

sudo asterisk -rx "radio show settings"    # live RX/TX levels
sudo asterisk -rx "rpt stats 1000"        # keyups, COS, TX time
sudo systemctl restart asterisk
sudo systemctl stop asterisk              # frees sound card to measure

Active config is /etc/asterisk/usbradio.conf. simpleusb.conf is inert on this node — editing it does nothing.

Wiring as built

Pair	Function	DB9	Repeater
Blue	TX audio	1 / gnd 6	mic RJ45 pin 4 / 5
Orange	PTT	3 / gnd 6	mic RJ45 pin 3 / 5
Green	RX audio	5 / gnd 9	4P4C pin 4 / 3
Brown	spare	—	unterminated

Shielded Cat6a, 28 ft, rack → attic. Shield bonded at the rack end only, floating at the repeater. DRA-30's DB9 is female; cable is male.

Don't be fooled by these

RTP packet counts — a ulaw stream sends ~50 packets/sec of silence just as happily as voice. Flowing packets prove transport, not audio.

The DRA-30's red LED is PTT, not COS. It means the adapter is keying the transmitter — it says nothing about receive.

rtp show settings never prints a TURN line on this Asterisk build. Its absence proves nothing.

Config changes don't persist through a restart if made with amixer. The driver rewrites the mixer on load — edit the .conf.

Why 29 dB SNR is fine here

In pro audio that number would be alarming. Analog FM land mobile is a different medium: 12 dB SINAD is the industry usable threshold and a full-quieting signal tops out around 35–45 dB. Our problem was never the ratio — it's that we're sitting 26 dB too low in absolute level, wasting converter headroom. Fix the level and the FM link becomes the limit, which is where it belongs.