Wednesday, 21 March 2018

2m Duplexer unusual design repair tune pager-reject

2m Duplexer design repair tune pager-reject


My local radio club's main 2m duplexer had an intermittent fault. The duplexer was fixed and retuned to a new channel, 1600, not 600 kHz split, partly to move the RX further from a Pager TX.

The Telewave TPRD-1556 Pass-Reject duplexer has an unusual coupler that I had not seen before that works very well, 45 db reject each. They should be possible to DIY. The fault was a defective piston capacitor used to tune a cavity.

An unexpected benefit of the pass-reject RX cavities is a significant 72 db reject of the 148 MHz pager TX. More is possible, if needed from notch cavities at 25 db reject each.

Tuning high performance cavities, 45 db each, highlights the limitations in instrumentation, particularly the dynamic range, 85 db, with the reject response lost in the noise floor.

Design of the Telewave TPRD-1556 Pass-Reject duplexer 

The Telewave TPRD-1556 Pass-Reject duplexer looks like any other six cavity pass-reject duplexer, except that the cavities only have one coax connector not two.

The mystery is in the design of the coupler. The coupler use semi-rigid coax as capacitors as well as the main part of the coupling. There is a piston capacitor to adjust the spread. The design allows a spread from 400 kHz to over 1600kHz, whereas another conventional pass-reject cavity could only go out to 1100 kHz spread.

The couplings are different lengths in order to mirror the RX-TX responses, something that can be difficult to do. I don't know how the do it, presumably with phasing.

Telewave TPRD-1556 Pass-Reject duplexer

RX coupler and detail; ~150 mm long

RX coupler detail showing one end of the centre conductor connected to the coax connector pin but the other earthed. One end of the shield is connected to the capacitor and the other end is connected to the centre pin

TX coupler; ~ 130 mm long

Repair- intermittent piston capacitor

The piston capacitor, while quite reliable, had failed as the ceramic was loose in one end.

Fixing it was simple. It was not possible to just unsolder the capacitor as that would disturb the coupling. I removed the inner piece, then using bolt cutters, crushed the ceramic. Then the bottom portion could be unsoldered and the top part removed. I only had a shorter, smaller value capacitor, but as the screw of the original was most of the way out, I thought it would work, as it did.

New capacitor and original one, showing construction with concentric cylinders.

Tuning the duplexer

Tuning is straight forward with a spectrum analyser and tracking generator. Repeater RX 145.6 MHz, TX 147, 1600 kHz split. The cavities are done one at a time. The length of the resonator is set first and shouldn't need to be readjusted. The spread is then adjusted. The cavities work very well giving 45 db reject and less than 0.5 db loss; within specification per data sheet.

Having done all of one side, I then join two, check, then join the third. The photo shows two cavities, with the reject a bit lost in the noise. The reject should be 90 db, but the analysers dynamic range is only about 85 db. The second photo in each pair is the SWR for three cavities; not sure were the second dip comes in the RX chain, but of no consequence. Between the pairs, the mirroring can be seen.

Two RX cavities showing TX reject, better than 85 db.

SWR of three RX cavities, odd second dip.

Two TX cavities showing RX reject, better than 85 db.

SWR of three TX cavities

Pager TX rejection

The repeater site is near two 2 kW pagers at about 148 MHz. There was a major concern for desensing the repeater's RX. Fortuitously, the club had this faulty pass-reject duplexer that could be fixed and retuned for the site.

An unexpected outcome is that the pass-reject duplexer has about 72 db reject on RX and similar for TX. This can be supplemented by notch filters on the pager frequency to give a further 25 db reject for each cavity, if necessary.

Perhaps obvious, but I was not aware of the potential for pass-reject duplexers to attenuate unwanted signals other than the desired RX and TX for the repeater. This is a significant advantage over other designs, such as notch cavities that reject either RX or TX, but let everything else through.

The repeater's licensed split was changed from the standard 600 kHz to 1600 kHz to further separate the the RX and pager. Ironically, the pager rejection for be 600 kHz split would be 20 db or more, the reject being in the noise and outside the spectrum analyser's dynamic range of about 85 db.

I will report more on this when the repeater is installed. I will probably do another post on this subject as it is an important topic, but incidental to the main purpose of this post.

148 MHz pager rejection for RX 145.6 MHz, 72 db with pass-reject duplexer

Home-made pass-reject cavity filters

I had been experimenting with notch filters for home-made duplexers, very easy to make, whereas I will now try to make pass-reject couplings based on those used in this duplexer. They are particularly good in that they use a single connector , simplifying construction. They seem to be out of patent and should not be a problem copying them, not that I want to sell any. More in another post.

Limitations of instrumentation tuning pass-reject duplexers

Tuning duplexers highlights the limitations in instrumentation, particularly dynamic range. Each cavity has about 45 db reject. Two give about 90 db reject, but the instrument shows noise at the reject frequency, indicating it is outside the instruments dynamic range. All three cavities give about 135 db reject, way down in the noise.

The response can be improved a little using a low resolution band width (RBW), 1 kHz in the last photo. But it takes some time to plot a 10 MHz span at 1 kHz and is only useful for static devices.

I use a Siglent SSA3021X spectrum analyser, a "cheap" Chinese instrument and it has about 85 db of dynamic range. Name-brand ones do maybe 10 db better; not a lot. The specification of the instruments give the noise floor, about 150 db, but don't give the dynamic range. The dynamic range is largely determined by the analogue to digital converter (ADC).

One way of checking the TX reject at the RX is to use the repeater's TX, with the antenna port connected to a dummy load, and measure TX reject signal at the RX port. It is prudent to use a cheap power meter and a variable attenuator to first check that the signal is low enough. The spectrum analyser can then be connected to the RX port and the TX leakage measured; notionally 145 db down on TX output. To a point, the check needs to be done as the repeater's RX is to be connected to the port. The TX reject signal is more indicative than accurate as there can be stray RF from leaky cables and enclosures. Test leads and the repeater cables should be double shielded coax to minimise this.

The other check is the TX output of the repeater and through the duplexer to make sure losses are within expected losses.

I will do another post on this subject as it is important and include some work with wide-band software-defined radios (SDRs) as real-time improvised spectrum analysers.


No comments:

Post a Comment