Repeater Uplink Failure and a Parity-Mismatched Teleprinter on the Same Evening
At 19:40 on a July evening the control room of a central Delhi district loses reliable voice contact with PCR vans in three sub-divisions. The repeater is a 25 W VHF set at a rooftop site: duplexer loss 2 dB, feeder loss 3 dB, and a 6 dBd antenna on a 30 m mast. Operators in the control room are heard clearly by every van, but the vans' replies break up or vanish, and the complaint persists after the repeater transmitter is turned up. At 19:52 the leased-line teleprinter delivers the flood-alert message with a strange corruption: characters such as C, E, F, 1 and 7 print as accented or wrong glyphs, while A, B, D, 3, 5 and 9 print correctly. The local teleprinter is strapped 8N1 at 50 baud; the sending end was changed that afternoon to 7E1, also at 50 baud. Loop current measured at the local end is 18 mA against a nominal 20 mA. The Executive Engineer's flood warning must reach all three sub-divisions before 20:30.
1. From the figures given, calculate the repeater's EIRP in dBm and in watts, and explain why raising the transmitter power did not cure the complaint. What does the pattern 'they hear us, we do not hear them' tell you about which side of the link is weak?
Read the worked answer
Calculation
- 25 W = 10 log10(25000 mW / 1 mW) = 43.98 ≈ 44 dBm
- Duplexer 2 dB + feeder 3 dB = 5 dB loss → 44 − 5 = 39 dBm at the antenna port
- 6 dBd = 6 + 2.15 = 8.15 dBi → EIRP = 39 + 8.15 = 47.15 dBm ≈ 52 W
- In dipole terms, ERP = 39 + 6 = 45 dBm ≈ 31.6 W. Quoting one figure as the other is a 2.15 dB error, enough to mislead a planning note.
Why more power did not help Every doubling of transmitter power buys only 3 dB, roughly half an S-unit on a narrowband channel. A fault that takes a signal from readable to absent is far larger than 3 dB, so turning 25 W into 50 W was never going to fix it. Worse, it attacks the wrong side of the link.
What the asymmetry proves Audibility in one direction only means the downlink (repeater transmitter, 44 dBm, mast-top antenna) is healthy. The failing path is the uplink: a 5 W handheld with a short whip, body loss and building clutter arriving at the repeater receiver. The limiting element is the receive side - receive feeder and connectors, duplexer, receiver sensitivity (measure 12 dB SINAD), desensitisation from a nearby transmitter, or water ingress in a connector after monsoon rain.
Ranked remedies
- Inspect and remake receive-path connectors, measure VSWR and RX sensitivity (recovers several dB cheaply).
- Mast-head receive preamplifier (10–15 dB, but check intermodulation risk at a crowded site).
- Raise antenna or add a receive-only fill-in site nearer the three sub-divisions.
- Increase handheld power only as a last resort; it drains batteries and does not address receiver desense.
2. Explain exactly why only some characters are corrupted on the teleprinter, and decide whether the 18 mA loop current is the cause. What is the correct fix and how would you prove it?
Read the worked answer
Rule out the red herring first Both ends run 50 baud, so framing is intact - a baud mismatch would garble every character on the page. The loop current at 18 mA is a marginal 10% deviation: it can raise the error rate on a long line but it cannot produce an error that depends on which character was sent. So the loop is not the cause and should not be touched until the format is aligned.
The signature The corrupted characters are exactly those whose 7-bit code contains an odd number of 1s:
- C = 1000011 (three 1s), E = 1000101 (three), F = 1000110 (three), 1 = 0110001 (three), 7 = 0110111 (five). The clean ones contain an even number: A = 1000001 (two), B = 1000010 (two), D = 1000100 (two), 3 = 0110011 (four), 5 = 0110101 (four), 9 = 0111001 (four). That is a character-class pattern, and only one fault produces it.
Mechanism The sending end transmits 7E1: the even-parity bit is 1 whenever the data bits contain an odd number of 1s. The local set is strapped 8N1, so it reads eight bits as data. The parity 1 lands in bit 8 (0x80), pushing the character into the extended range and printing a wrong glyph. Characters whose data had an even 1-count received parity bit 0, so they came through unchanged. The afternoon maintenance change is therefore the cause.
Fix and proof
- Restrap the local teleprinter to 7E1 (or restore both ends to the agreed standard, recorded in the log).
- Re-run the message and print a test string covering the whole character set - a fox message (THE QUICK BROWN FOX JUMPS OVER THE LAZY DOG 0123456789) - and check it letter by letter.
- Require a clean run of several consecutive messages, then check loop current and adjust if still below tolerance.
- Inform the far end in writing so the strap is not flipped back at the next shift change.
3. You have roughly 30 minutes before the flood warning is due. Sequence the three problems - flood warning, teleprinter format, radio uplink - and justify the order.
Read the worked answer
Order: disseminate → repair the 5-minute fault → investigate the long-lead fault.
1. Push the flood warning out now, by voice, on the downlink. The requirement is that three sub-divisions are warned by 20:30. The downlink is the one path proven to work - every van hears the control room - so a broadcast alert achieves the operational aim immediately. Do not hold the warning hostage to the uplink fault. Collect acknowledgements through the nearest police station landline or mobile numbers held in the control room, and log who acknowledged and when.
2. Fix the teleprinter strap (2–5 minutes, no site visit). This is a configuration change with immediate full restoration and it produces the time-stamped hard copy that relief teams and any later inquiry will rely on. Re-transmit the flood message as a printed message once the format is corrected.
3. Open a fault for the uplink, but do not expect a fix tonight. Mast work, connector inspection and sensitivity measurement need daylight, rigging and the technical wing. Log the fault with time, symptoms and the tests already done, and record the interim workaround (voice broadcast plus telephone acknowledgement) so that nobody assumes the link is healthy.
Principle applied: sequence by consequence multiplied by lead time. Warning the public cannot wait at all; the strap change is minutes; the uplink is a hardware investigation with a workable workaround. Also inform the senior officer on duty of the interim arrangement in writing - an undocumented workaround is what turns a technical fault into an administrative one.