Monday, 26 September 2016

A 70cm DATV TX filter using a cheap Chinese duplexer

A 70cm DATV DVB-T 7 MHz band-pass filter using a cheap Chinese duplexer


DATV transmitters for DVB-T are notorious for "spread" outside the channel, to the point that keeping it 30 dB or more below the signal becomes a limit for power output, typically 10 W out of a 70 W module amplifier.

Even with -30 dB spread, it is desirable to have a band-pass filter before further amplification or transmission. Usually an interdigital filter is used, but they are either expensive to buy or a bit difficult to build.

VK4JVC suggested using a cavity filter duplexer instead. I tried a four cavity notch duplexer, but the pass-band losses were too high, more than 20 dB. I had bought a cheap, ~A$100, Chinese Jiesai duplexer, but had put it aside as the response looked bad. After try other duplexers (notch and pass-reject types), I tried the Chinese one again, this time successfully.

The result is that the Chinese filter seems to provide a good pass-band for the 7 MHz DVB-T DATV signal with acceptable pass-band losses and steep skirts. The next test is to try it with my 10 W amplifier from Darko in Austria.

The filter

The duplexer is a typical mobile device available on eBay for about A$100 delivered, taking a week or so to Australia. The store insisted I supply some tuning data, even though I was immediately going to change it. I specified 440.5 and 446.5 MHz to keep them happy. It came with notches at those frequencies, but had not been well tuned.

The filter has three square cavities for each of RX and TX. The only adjustment is a screw at the top that capacitively alters the cavity's resonate frequency. There are no other adjustments. Each cavity has an cable in and out, but they seem to be notch filters rather than pass-band.

I am not sure what the power handling capacity is, 25 W, from memory. That would make it an adequate final TX filter for most DATV applications.


The Australian 70 cm DATV band is 7 MHz wide centered on 446.5 MHz, with edges at 443 and 450 MHz. I use the whole 7 MHz as it makes reception on conventional TVs easier and I want high quality 1080p.

Using a HP 8591A spectrum analyser and tracking generator, re-tuning is quite easy. Three cavities at a time first (RX-ANT, TX-ANT), then checking all six with the input and output through the RX and TX connectors, ignoring the antenna connector.

I have tuned the cavities to about -3 dB at the channel edges. The loss through all six cavities is about 2 dB which indicates quite reasonable construction. The side slopes are quite steep.

I am not sure this tune will be adequate to suppress the channel spread, but I will re-tune to find an acceptable compromise. I may need to narrow the width, but without affecting the TX signal.


A cheap Chinese duplexer has bee re-tuned to produce what appears to be an acceptable TX band-pass filter for a 7 MHz wide channel on 70 cm.

Despite their reputation, the Chinese duplexer seems of reasonable quality and has typical characteristics for this type of device. The original tuning was indifferent, so should always be checked.

Friday, 15 April 2016

Airspy, Spectrumspy, noise source and UHF cavity filter characteristics; a low cost spectrum analyzer?

Airspy, SpectrumSpy, noise source and UHF cavity filter characteristics; a low cost spectrum analyzer?


A basic spectrum analyzer/tracking generator for less than $250? Yes. Can it be used to do a demanding task like tuning a UHF cavity filter from a repeater? Seems so.

The "proof of concept", spectrum analyzer software, SpectrumSpy, can be used with the Airspy SDR and a noise source to show the characteristics of a pass-reject UHF cavity filter.

SpectrumSpy and Airspy

SpectrumSpy, "proof of concept", spectrum analyzer software is a new addition to the SDR# download for use with the Airspy SDR. It has the potential for a new direction with low cost SDRs, spectrum analyzers.  Spectrum analyzers are expensive; $1500 then skyward. SpectrumSpy: (separate executable in SDR# folder).

Airspy has a 24 – 1800 MHz native RX range, but down to DC with the SpyVerter option. $199  and US$59

An earlier post is of SpectrumSpy used as a spectrum analyser:

However, a spectrum analyzer needs a tracking generator to be really useful and to test radio filters. A noise source can be used to achieve much the same purpose.

Noise source

A low cost  Zenar diode based noise source is available for about $20. It uses three Mini-Circuits ERA-5+ wideband amplifiers (DC-4 GHz) to get the noise to a usable level.

I did a quick check of the white noise with SpectrumSpy right up to 1 GHz. The level dropped a little with frequency, but that could be either the noise source or Airspy. and many others.

Noise source output

Cavity filter characteristics 

A cavity filter is part of a duplexer that allows a radio repeater to simultaneously transmit and receive with the same antenna, an amazing feat in itself and one of my other interests. Duplexers are fairly complex in design and setup.

Pass/reject is one type of cavity filter that passes the receive signal on one frequency, but rejects the re-transmit of the repeater on another frequency with a notch for the receive side of the duplexer and the opposite for the transmit side.

An expensive spectrum analyzer/tracking generator is needed to adjust the pass and reject frequencies and to minimize losses.

However, SpectrumSpy with a noise source does a pretty good job. The shape, frequencies and depth of notch (>30 dB) are about right in this very preliminary test.

Characteristics of single cavity filter using SpectrumSpy and noise source

Spectrum analyzer/tracking generator plot of similar but different filter.

Typical characteristics from half a duplexer (two cavities) with my old HP spectrum analyzer and tracking generator ($12K in 1990s and about $1000 secondhand now). (I will do it for the same filter another time and edit.) It has different tuning to the one in this post, hence the mirrored shape.


Airspy with SpectrumSpy does a good job in this demanding task. It probably could be used to tune a duplexer; amazing for the very low cost.

Two immediate issues however. First, SpectrumSpy is not calibrated or necessarily linear. Second, it does not have the bells and whistles of a spectrum analyzer such as digital analysis and data, or amplitude/frequency markers.

SpectrumSpy is proof of concept of an SDR used as a spectrum analyzer. This is a very good additional application for low cost SDRs, normally used as receivers. Similar software could be developed for other SDRs, as has been done, but less successfully, for the RTL-SDRs.

Thursday, 14 April 2016

Low cost spectrum analyser/ scanner with AirSpy and RTLSDR

Low cost spectrum analyser/scanner software for the Airspy and RTL-SDR


It is not often I am amazed at new technology, especially for free, but the Spectrum Spy software, a spectrum analyser/scanner for the Airspy SDR, impressed me. It is a poor man's spectrum analyser.

However, it is preceded by at least two spectrum scanners for the RTL-SDR hardware; rtl_power and RTLSDR Scanner.

This post will compare the two devices and three software packages, scanning the entire FM band and the 100 MHz of the local TV band.

The software

The three programs all run under Windows, Windows 10 in my case. All three installed and ran without much difficulty.

Spectrum Spy is part of the SDR# software package. It is a separate program to SDR#, but in the same folder. Spectrum Spy has a spectrum and a waterfall. It updates every few milliseconds, depending on the span. Spectrum Spy only works with the US$199 Airspy. I have a V1 Airspy.

RTLSDR-Scanner is a stand-alone program; Use the "setup" version and it will download all the code it needs to install. No extra files are needed for the RTLSDR, provided its driver is installed. RTLSDR-Scanner is a single pass with spectrum but no waterfall.

Rtl_power is a small program that has both spectrum and waterfall. Some additional RTLSDR files are needed to be copied to the program directory.

The hardware: apples and oranges

The Airspy and RTL-SDR both use the same tuner, but after that the Airspy is much more sophisticated. A significant difference is the frequency span, 10 MHz for Airspay and about 1.2 MHz for the RTL-SDR. The span makes a big difference to the scan rate of the scanner/analyser programs.

The question is whether the $20 RTL-SDR is useful as a scanner, compared to the Airspy.

FM band

A handy source of signals is the local FM band. The screen shots show how the three programs perform.

Spectrum Spy performs very well, showing a 20 MHz span. It shows both spectrum and waterfall. The waterfall is very useful for intermittent signals.

RTLSDR Scanner does a reasonable and useful job. The spectrum is comparable with Spectrum Spy. However, it only does a single scan and takes 20 seconds of so for the scan.

RTL_power had an indifferent result. The spectrum is quite different to the other two programs. While it does a continuous scan, the waterfall does not align with the spectrum.

TV band

The local DVB-T TV is in a 100 MHz band from 600 MHz. Displaying the TV stations is quite a test for a spectrum analyser.

The Spectrum Spy does a very good display in this demanding task.

The RTLSDR Scanner similarly does well in displaying the spectrum, although it takes quite some time to do the scan.

RTL_power does better with the TV than the FM, but is still not great.


The Spectrum Spy program with the Airspy hardware does an awesome job comparable to some spectrum analysers, for low cost,

It would be ideal for doing repeater site surveys, especially with the waterfall as well as the spectrum.

RTLSDR Scanner could be used in a similar manner to Spectrum Spy, just taking longer with a single scan.

Saturday, 5 March 2016

200W DATV power amplifier- 50V 20A power supply

200W DATV power amplifier- 50V 20A power supply


I am builder a DATV power amplifier, about 200W maximum. The power supply is unusually powerful as the amplifiers are only bout 25 per cent efficient. Commissioning such a power supply is not simple.

DATV power amplifier

I discovered through one of the Yahoo groups that it was possible to buy UHF DVB-T pallet amplifiers, broad band from 470 MHz to 900 MHz. I bought one through the Italian eBay for 400 Euro. It has a pair of BLF888A LDMOS transistors. The pallet is a pair of amplifiers with a splitter and combiner to allow single in and out. The BLF888A are each a match pair of transistors in push-pull.

Power supply

The power requirements are about 48V at 20A as the amplifiers are only about 25 percent with the ultra linear DVB-T.

I remembered that a lot of telecommunications equipment use 48V, so had a look on eBay. I managed to buy a new ELTEK Flatpack2 2000W 48V HE Power Supply for $150. These are state of the art telecom power supplies and very small for 2000W output.

The first complication is connecting wires to them. They are designed for an in-chassis slot connection of both mains and DC. I opened it up and soldered wires to the back of the connector. Not a good idea, but will do for the moment. I will make up a slot connector and have no wiring inside the power supply.

The next problem is testing the power supply. I bought four 12V headlight globes and mounted them together with wiring. The globes represent a load of about 220 W. The power supply easily achieved this. The voltage is 53V and is not easy to reduce to 48V, it has a CAN network for control.


For the first part of this project, the power supply, it seems to be working well. Next the heat sinking and testing the amplifier, then some filtering.

Monday, 15 February 2016

Three cavity notch RX filter for 2 m with 1.6 MHz RX/TX spacing

Three cavity notch RX filter for 2 m repeater with 1.6 MHz RX/TX spacing


A three cavity notch filter was constructed and tested. The RX filter has over 80 dB of TX rejection.


The club has a 2 m repeater that is to operate on the new 1.6 MHz RX/TX spacing, as opposed to the usual 600 kHz spacing. The existing high/low pass reject cavities could not accommodate the wider spacing, limited to about a 1.1 MHz spacing.

The club has five 1968 band pass filters, old but well made. A pair of them had been used with a phasing harness to act as notch filters, but the rejection was not high enough for TX reject on RX. Two are probably adequate for the TX filter to reject spurious noise at the RX frequency.

I decided to try and make a three cavity notch RX filter, necessitating a new phasing harness. Each of the cables are a quarter wavelength, with those going to the cavities a little shorter to allow for the probe inside the cavity. I used LMR-400 ultraflex for the double screening, plus I had some at hand. RG-214 is preferred but it is very expensive (~$25 per m).

My first attempt was to use soldered tee joins, but I had trouble with shorts. I think the aluminium Mylar film melts when soldering the braid. I changed to N connectors and tee joiners. That was worked.


The RX filter gives over 80 dB TX reject, which is adequate.

I also measured the SWR to see where the minimum was, expecting it to be at the TX frequency, however it is quite some distance away. Previously I had thought that the RX and TX filters would need separate RX and TX antenna, as I thought the RX notch would be a short-circuit for the TX, but the SWR measurement suggests otherwise. I need to investigate this further.


An effective repeater RX notch filter can be made from three band pass cavities and a phasing harness. Further work is needed to see if a common antenna can be used.

Sunday, 14 February 2016

Duplexer isolation: limits of instrumentation

Duplexer isolation: limits of instrumentation


With high isolation duplexers, about 100 dB, conventional spectrum analysers and tracking generators are beyond their capability of about 80 dB. The rejection of a high isolation duplexer can be checked using a RF signal generator and spectrum analyser.

High isolation duplexers.

I have a couple of sets of four cavity 70 cm pass reject duplexers. The two cavities give about 70 dB rejection, which is ok but more would be better for RX. A third cavity can be added for greater isolation.

Each cavity is individually tuned with a spectrum analyser and tracking generator. The two cavity duplexer can then be assembled and tested. A nice plot with about 70 dB rejection at 438.1 MHz and a pass of 433.1 MHz.

Adding the third cavity results in noise at the rejection notch. The spectrum analyser cannot show the full depth of the notch; it does not have enough dynamic range.

RF generator and spectrum analyser

A RF generator can be used instead of a tracking generator as it has a greater output at the TX or reject frequency.

When connected with two cavities there is quite a spike.

 However, when the third cavity is added the spike almost disappears with the extra 35 db rejection.


A RF generator can supplement a spectrum analyser and tracking generator when aligning high isolation duplexers.

Sunday, 31 January 2016

Tuning the receiver front-end of a Kenwood TKR-750 repeater

Tuning the receiver front-end of a Kenwood TKR-750 repeater


I bought a second-hand Kenwood TKR-750-1 VHF repeater. However it was not clear what band the receiver was tuned to. The actual receiver and transmit frequencies can be programmed with a computer. However the the receiver band width is tuned with a spectrum analyser and tracking generator.

This post describes how to re-tune the receiver front end.

The band pass filters for a 2m repeater RX are quite critical as they are the only selectivity. With a 600 kHz spacing, the cavity filters are usually just isolating the TX and RX when sharing the same antenna, but providing no bandpass to strong nearby signals.

Starting point and instrumentation

I have an old HP8591A spectrum analyser and tracking generator, circa the 1990s, but still a capable instrument. The basic procedure is to connect the tracking generator output to the RX antenna input and the spectrum analyser input to a test point after the RX bandpass filters and amplifiers, just before the first mixer.

The pass-band of the receiver will then be shown. Before adjustment, the centre of the pass-band was about 160 MHz and the bandwidth about 10 MHz at 6 dB. This pass-band would make the receiver useless for the 2m amateur band, in Australia, 144 to 148 MHz. I was apprehensive if I could move it that far.

The receiver front end is relatively simple. From the antenna through two filters L2 and L3, a single RF amplifier bipolar transistor, then three more filters, L5, L6 and L7. The test point is CN1, at the bottom left.

The HP8591A is connected to the antenna and test point. However, before connecting the instrument, the output of the tracking generator must be reduced to about -30 dB. It took some searching through HP manuals to see how to do that, but it is easy when you know how.

Just in case I triggered the repeater, I connected a 50 Ohm dummy load to the TX antenna output.

The test setup shows the devices connected together. The five filters can be seen clearly.

Amateur radios use an odd connector, TMP from Taiko Denki, for interconnecting boards and for test points. Fortunately I had chased some up for when I did the IF tap on my IC-7410 TRX. They are $1-50 from the USA, so I bought ten.

The other must have is a RF coax connector adapter kit. The gold connectors going from BNC to N. and They are a must-have for working with modern radios, from HF to microwave. The two connector ends screw together with a threaded barrel.

The adjustment was suck it and see, not having done it before. Using an insulated tuning stick, I started with the first filter, screwing it in as I wanted to lower the frequency. This adjustment showed up nicely on the spectrum analyser. I then went to the next in the chain and so on. I needed to change the centre frequency on the analyser so I could see where I needed to go. I went though the chain of filters trying to get the right shape, right frequency and minimal losses.

The end result: success

Eventually I had a nice pass-band centred on 147 MHz with the same 10 MHz bandwidth, pretty much the same shape as the original. I was relieved I was able to manage to move it sufficiently, otherwise the repeater was of no use.

RX selectivity and cavity filters

The main purpose of a duplexer is to separate the RX and TX to allow them to share the same antenna. As a reader of my blog will see my interest in building cavity filters. I found it quite fascinating that a repeater could transmit 50 W into the same antenna where it was simultaneously receiving micro Watts of RX signal.

On 2m, the spacing is only 600 kHz (recently increased to 1.6 MHz in Australia, but that is another problematic story). Bandpass cavity filters do not have enough selectivity for this narrow spacing. Special pass reject cavity filters must be used. The TX bank of three cavities pass the TX signal but have a deep notch at the RX frequency. Similarly, the RX bank pass the RX frequency but have a deep notch at the TX frequency.

However, while the duplexer allows the repeater to work, the duplexer is not a bandpass filter for the RX. All the selectivity is left to the little filters in the RX itself, making them critical if there are strong nearby signals.

In Australia, there is a pager band starting at 148 MHz, right next to the 2m repeaters.

On UHF, 70 cm, the spacing is higher and allow the use of simpler bandpass cavity filters. These high Q filters do the two tasks of providing selectivity as well as RX-TX isolation.


It is realtively easy to retune the RX frontend of a repeater using a spectrum analyser and attenuated tracking generator. Tuning the repeater RX on bands with close RX-TX spacing is quite critical as it is usually the only bandpass selectivity.

PS UHF TKR-850-1

Before; about 460 MHz centre, 8 MHz wide

After: OK but not quite as good, centre 434 MHz and 10 MHz wide; a 26 MHz move!