1 The architecture
Red-outlined segments are what you're building. The DB9 stays where it is and gets one wire added.
Two problems, one cable. The front-panel 4P4C tap went silent — its level lives behind a software attenuator we can't reach without the Tuner app. And carrier detect has been running on VOX, which only trips for split-seconds and never holds.
J7 fixes both. Pin 7 is a fixed 330 mVrms output with no attenuator in the path. A GPIO set to CSQ Detect gives the node a real hardware carrier signal, so VOX is retired entirely.
2 T568B — both RJ45 ends
Looking into the plug with the gold contacts toward you and the latch down, pin 1 is on the left.
| RJ45 pin | T568B conductor | Our function |
|---|---|---|
| 1 | White/Orange | PTT ground (future) |
| 2 | Orange | PTT (future) |
| 3 | White/Green | RX audio ground |
| 4 | Blue | TX audio (future) |
| 5 | White/Blue | TX ground (future) |
| 6 | Green | RX audio |
| 7 | White/Brown | COS ground |
| 8 | Brown | COS (CSQ Detect) |
Green looks wrong but isn't. T568B splits the green pair across pins 3 and 6, which aren't adjacent. That's correct — 3 and 6 are electrically the same twisted pair. Don't "fix" it to 3 and 4.
3 The whip — RJ45 → DB25
Build from an offcut of the same solid Cat6a. About 2 ft, or whatever reaches comfortably from the coupler to J7.
| RJ45 pin | Conductor | Function | DB25 terminal |
|---|---|---|---|
| 6 | Green | RX audio · 330 mVrms | 7 |
| 3 | White/Green | ground | 9 |
| 8 | Brown | CSQ Detect → COS | 4 |
| 7 | White/Brown | ground | 9 |
| 4 | Blue | TX audio (future) | 1 |
| 5 | White/Blue | ground | 9 |
| 2 | Orange | PTT (future) | 2 |
| 1 | White/Orange | ground | 9 |
J7 — DB25 male, mating face (pins toward you)
You don't need to decode DB25 orientation. The breakout board's screw terminals are printed with pin numbers — just match the number. The diagram is only for sanity-checking your work.
4 The 28 ft run — attic end
5 The DB9 end — one wire + one jumper
The only soldering in this build, and it's at the rack.
| Conductor | Function | DB9 pin | Status |
|---|---|---|---|
| Green | RX audio in | 5 | already wired |
| White/Green | ground | 9 | already wired |
| Brown | COS in | 2 | SOLDER THIS |
| White/Brown | ground | 6 | SOLDER THIS |
| Blue | TX audio out | 1 | already wired |
| Orange | PTT | 3 | already wired |
| whites | ground | 6 | already wired |
DRA-30 — DB9 female, mating face
Install jumper JU1 on the DRA-30 board. Without it, DB9 pin 2 never reaches the CM119A's COS input and the carrier line does nothing. Leave JU2, JU3, JU4 off. This is the easiest step to forget and it fails silently.
Why pin 6, and where "ground" actually is
Do not rely on the metal hood. The DRA-30 schematic (Rev 1.0, Masters Communications) shows DB9 pins 6 and 9 tied to board ground and nothing else. There is no shell, frame, or chassis connection drawn on that connector, and the DRA-30 lives in a plastic case, so even a bonded hood would only reach board ground. Landing the drain on pin 6 gets you the same electrical node, documented, with nothing left to guess.
Pin 6, not pin 9. Pin 9 is the RX audio return, and RX is the path this whole build exists to fix. Keep the shield off it. Pin 6 carries the COS, TX and PTT returns, where a few microamps of shield noise costs nothing.
There is no earth anywhere on the node side. Board ground goes to USB ground goes to the Pi goes to a 2-prong supply. The entire node floats. Its reference arrives from the repeater, up the cable, on DB25 pin 9. That is the ground, and it comes in through the cable rather than out of the wall.
Which is why the network rack downstairs stays out of it. Bonding the shield to the rack would tie it to building earth on a different floor while the conductors inside it sit at repeater ground. A shield referenced to something other than the wires it surrounds is the one version of this that actively hurts. Bond at the DB9, float in the attic, touch nothing else.
6 Verify before it leaves the bench
Plug the whip into the coupler and meter the whole chain, DB25 to DB9.
7 CPS (at the laptop)
The COS wire is dead copper until the repeater is told to drive it. J7's GPIO pins do nothing out of the box — pin 4 has to be assigned a function in CPS. This is the step that makes the whole build work.
Before anything else: read and save the existing codeplug.
The MTR3000 went out of production in 2015 and Motorola no longer supports it. There is no factory codeplug to fall back on. Read the repeater, Save As a dated file, and do not overwrite it. Every later change gets saved under a new name.
Pin #2 must stay unprogrammed. This is a transmit-safety item, not housekeeping.
The whip lands Orange on J7 pin 2 even though PTT is marked "future" everywhere in this document. That means a physical conductor will be sitting on the External PTT pin from the day the cable goes in. If pin 2 is ever assigned Ext PTT, that wire becomes able to key a licensed Part 90 transmitter — including from a floating input or a solder whisker.
This installation is receive-only until transmit is separately authorised. Verify pin 2 reads unassigned, and leave it that way.
Active Level Low and the node config are one decision, not two.
CPS Active Level: Low pairs with carrierfrom = usbinvert in section 9. They are two halves of the same polarity choice. If you flip one you must flip the other, or COS reads permanently keyed or permanently idle. Set CPS first, then confirm polarity at the node against real traffic.
Two things to confirm in CPS while you are in there — neither is verified yet, so check rather than assume:
• Is J7 pin 7 Rx audio squelch-gated or open? The COS wire is what gates the node, so
open (ungated) audio on pin 7 is the intended arrangement. If CPS offers a gated/ungated choice
for aux Rx audio, note which way it is set — it changes what "silence between calls" should
sound like.
• Is the pin 7 output level adjustable? The manual specifies 330 mVrms at 60% deviation.
If CPS can scale it, record the current value before changing anything, because R12 on the
DRA-30 is the other half of that gain budget.
Menu paths above are from the working session. Exact wording lives in mtr3000/mtr3000-installation-and-user-manual-68007024098-h.pdf, Table 2-5.
Writing an output pin is an additive change. It does not touch the repeat path, and the repeater keeps repeating throughout apart from the brief write itself.
8 At the repeater
9 Node config (remote)
In /etc/asterisk/usbradio.conf:
carrierfrom = usbinvert ; was vox — hardware COS now
Polarity check: with nobody transmitting, watch Signal on input. NO at idle and YES on key = correct. YES at idle = flip to usb.
10 What "working" looks like
• rpt xnode 1000 → RPT_RXKEYED=1 sustained through an entire
transmission, not flickering
• Repeater audio audible in the app, clean and continuous
• Level near −10 dBFS on open-squelch noise
Once it's in, tell me — I'll measure and set R12 properly rather than guess.