Tuesday, 17 June 2014

New box for GB3LU

Our old Icom repeater unit at GB3LU is starting to show its age - already the internal PSU has died and it was running on an external 12V PSU, and recently it's being turning itself off, for no obvious reason. Time for an upgrade....

I had a Simoco PRF1050 in my "rainy day" cupboard under the stairs that would do the job nicely, as long as it would interface with the existing Logic (a G8CUL designed controller I built some years ago). Checking it out on the bench showed that RX & TX Audio, Squelch (COR) and PTT were easily accessible via the 37-way facilities plug:

1Tx600 Ohm-14dbm nominal for 60% dev at 1Khz (Range -37dbm to 0dbm)
Transformer coupled to pin 20 or AF high if unbalanced
3Rx600 Ohm-14dbm nominal for 60% dev at 1Khz (Range -37dbm to 0dbm) Transformer coupled to pin 22 or AF high if unbalanced
4No connection made
5Common Station AlarmOpen collector low, when active
6Not assigned reserved for future use
7Not assigned reserved for future use
8CTCSSUnprotected input to TX modulator. Level and input impedance to be advised shortly
9+13.6v unswitched1 amp available to power external devices
10TX KeyPull down. Typical current 1.5mA. Open circuit voltage typically 5V
11TX Lock AlarmOpen collector low when activated
13RX Lock AlarmOpen collector low when activated
14CTCSS Decoder DefeatPull down to activate. When activated allows squelch to open with any signal
15RSSIRange as follows but can be negotiated if difficult to implement Graph attached
16Channel Line 6Binary C6, Normally high, pull low
17Channel Line 4Binary C4, Normally high, pull low
18Channel Line 2Binary C2, Normally high, pull low
19Channel Line 0Binary C0, Normally high, pull low
20TX600 Ohm-14dbm nominal for 60% dev at 1Khz (Range -37dbm to 0dbm)
Transformer coupled to pin 1 or AF return/ground if unbalanced
21No connection made
22RX600 Ohm-14dbm nominal for 60% dev at 1Khz (Range -37dbm to 0dbm)
Transformer coupled to pin 3 or AF return/ground if unbalanced
23External LS HighExternal loudspeaker typically 16 Ohms
24Squelch StatusOpen collector, low when squelch open
25Squelch Relay NC1 amp DC maximum at 48V DC. Floating
26Squelch Relay NO1 amp DC maximum at 48V DC. Floating
27Squelch Relay CommonFloating
28DC Input under voltage alarmThe PRF10 can charge external 12V Sealed Lead Acid battery. The battery can be damaged is subject to a deep discharge. The PRF10 automatically disconnects at a predetermined voltage. An alarm is signalled at 100mV prior to disconnect. Open collector low, when activated
29Talk Through EnablePull down to activate
30Squelch DefeatPull down to activate
31No connection made
32No connection made
33RF Output AlarmOpen collector low, when RF fails to meet power
34Unfiltered and unmuted RX AFEffectively discriminator output. Level and impedance to be defined shortly
35Channel Line 50Binary C5, normally high, pull low
36Channel Line 3Binary C3, normally high, pull low
37Channel Line 1Binary C1, normally high, pull low

The plan to interface with the existing GB3LU Logic :

TX Audio : PRF10 pin 1 - Logic pin 12
RX Audio : PRF10 pin 3 - Logic pin 10

Also Earth PRF10 pin 20, 22 (to PRF10 pin 12) for unbalanced audio

SQL - low on Squelch open :
Earth PRF10 pin 27 (SQL relay common)
SQL signal : PRF10 pin 26 - Logic pin 4

PRF10 pin 10 - Logic pin 7 (logic pulls this low to transmit)

Power for the Logic unit from PRF10 pin 9 - Logic pin 1
Ground : PRF10 pin 12 - Logic pins 14-25

This worked fine except for 2 problems....

1) The TX line-input greatly attenuated the CTCSS tones sent from the logic unit and it was not possible to achieve a sensible deviation for the 77Hz CTCSS on transmit.

2) The RX line-output also had a reduced amplitude of the RX CTCSS tone, which is used in the Logic unit to enable the talkthrough (we don't use old fashioned 1750Hz access, but require "continuous" CTCSS)

Problem 1 was solved by finding that there is a direct feed in from pin 8 for injecting external CTCSS tones. Initially this proved not to work as expected until I discovered there's a handbag link on the main PCB that connects the input from pin 8 through to the analogue circuits of the PRF10. Installing the missing link got the external CTCSS input working. I modified the Logic unit to feed a separate TX CTCSS signal from the CTCSS encoder chip into the PRF10 / pin 8. A pot on this feed allowed the deviation to be adjusted to around 500Hz. The driving factor with wanting to use the Logic unit's own CTCSS signal, and not simply configuring the PRF10 to transmit a 77Hz CTCSS tone, is that we carefully adjusted the timing of the Logic's CTCSS to aid the use of personal "cross-band repeaters" - and users who avail themselves of this were keen to maintain the function. This requires that the outgoing CTCSS is muted shortly after the "pip/K" end of over signal, allowing x-band repeaters to drop back to "idle". If the CTCSS was enabled continuously during the time the box is "open" then the only way to operate a x-band repeater is to let the box close down at the end of each over, and then re-access it once your x-bander has dropped out. Keeping the custom-timed CTCSS is therefore an important requirement. Also, as a pleasant side effect, the CTCSS is also muted during the regular repeater idents, so anyone choosing to monitor with "Tone Squelch" will only hear actual traffic.

GB3LU Logic Unit

Re-programming the controller

PRF10 under test

Problem 2 was investigated further....

I found that a normally modulated input signal (with > 300Hz deviation of the CTCSS) would trigger the Logic's CTCSS decoder, even though the 'scope showed the amplitude of the CTCSS tone at the input to the Logic to be only around 14mV rms, and a long way below the amplitude of the main audio signal. A problem still existed though - if an over-deviated signal was present I found that the CTCSS decoder dropped out on speech peaks, and then the talk-gate would close briefly- leading to breaks in through audio on speech peaks - but only with either very noisy or over deviated input signals.

There is a feed of un-processed "discriminator" audio from the PRF10 at pin 34 and this had ample CTCSS level, but also the full audio signal too. I built a 2 pole active LPF with a 300Hz cut-off and put it in series with the "raw" RX audio, and then paralleled this (via a 1k resistor) with the normal RX Line input to the Logic. This improved the detection of the CTCSS tone, it would open with a CTCSS deviation of around 200Hz, but simply tee-ing the two audio feeds (LPF feed of CTCSS and the de-emphasised main RX audio) wasn't the right solution. I would need to modify the Logic unit to provide a separate "CTCSS Input" directly to the CTCSS chip - fed from the LPF'd raw audio.

I decided to revert to the initial condition- one audio feed - and to re-programme the Logic back to "classical 2m repeater behaviour" in that CTCSS is only needed to initially open the repeater, and not to maintain talk-through. This seemed to be a decent compromise, and I also restored the 1750Hz tone-burst access too. The repeater was only configured for continuous CTCSS to overcome a problem with local interference on the repeater's input frequency, generated by a site-sharer at the repeater site, and it appears that the device responsible for this interference is no longer in use.

Bench measurements show that the RX sensitivity is good:

Squelch opens at -113dBm / 20dB SINAD with the PRF10 set to its minimum Squelch threshold (parameter 105.xx = 1). The squelch threshold at this setting  is nominally "9dB" but this isn't actually the case in reality. I experimented with varying PRF10 parameter 921 (factory set squelch calibration) and found:

As found : 077
Threshold : -113dBm / 20dB SINAD

Adjusted to : 020
Threshold : Squelch permanently Open
-117dBm  / 15dB SINAD

Adjusted to: 028
Threshold : -115.5dBm / 18dBm SINAD

I reset the parameter to the default value (077) for the time being. Squelch threshold can be revisited once the box has had a bit of on-air use, and after the antenna/feeder is replaced.

For reference the higher squelch settings:
Parameter 105.xx:
2 (nominally 12dB SINAD) : Threshold = -111dBm / 23dB SINAD
3 (nominally 15dB SINAD) : Threshold = -109dBm / 24dB SINAD

The various timings for the Logic are currently:

"Pip" time (from squelch close to sending the "pip/K" ) = 1 second
CTCSS off time (from squelch close to cutting outgoing CTCSS ) = 2 seconds
Off Time (from squelch closing to closing repeater if no further input) = 9 seconds
Toneburst length (minimum 1750Hz toneburst to be recognized) = 500mS
Latch Time (length of carrier required after toneburst to cause the repeater to open fully) = 1 second
Squelch Delay (minimum length of carrier required to cause an "ack") = 1 second
Timeout Time (overs longer than this will cause the repeater to "timeout" and then to closedown) = 10 minutes
Timeout End Time (period of "pips" marking the timeout before closedown) = 10 seconds

Beacon Time : 15 minutes
Send Locator with beacon : every 4 (i.e. once per hour)

To access : at least 500mS of 1750Hz toneburst or a valid 77Hz CTCSS and one second of carrier/audio
The repeater will key up briefly on input signals for shorter than this duration.

To prevent 10 minute Time-Out - ensure you wait to hear the "pip/K" at the end of the over before you start to transmit. This indicates the 10 minute timer has been reset. If a station transmits for > 10 minutes the repeater will cut the through audio and send a series of 10 one-seconds pips before closing down. Once the input signal has gone the repeater will key up and send "OK" in morse, and will then be available as normal.

A stronger signal can re-set a timeout that's in progress by sending a valid 1750Hz toneburst, or 77Hz CTCSS, as long as they can overcome the signal that's currently transmitting.

There is also a remote DTMF Closedown and Startup command - to be circulated to the Repeater Closedown Operators - which will cause the repeater TX to be inhibited or restored, by sending specific DTMF codes. This assumes you are able to overcome any signal on the input frequency that might be causing problems, but at least it will allow a remote shutdown, even if you have to travel nearer to site. If it means not actually having to get up the hill, and into the hut in a Shetland winter storm, that's a good thing!

GB3LU is back on the air as of June 13th 2014, using the PRF10 - notionally on "soak test".

GB3LU in the hut & on the air

Coverage is reduced at present due to antenna/feeder problems - and these will need to be addressed before the end of the summer.

No comments:

Post a Comment