Passive Balanced Mixer - problems.

Hello again all,
I have been tearing mu hair out for these mixer problems and I have got the answers. Now everything is on/track with the project. As usual, family life (shopping, Easter food, washing terrace, oiling terrace, digging etc) took priority.

1 - The signal generator I have is an older generator telecom that can run AM, FM, PM etc. and it can even synthesize a GSM radio base (HD not fitted) BCCH. Modulation tone covers 0.01Hz to 499.999KHz and FM deviation from 1Hz to +/-350kHz. Loads menus in it, and needs a mouse to navigate all the different menus. It has a 5V / 12 V DC bias function at the output. I have been using the generator for about 12 years and never knew the DC shift function existed. The DC superimposed on the RF out cobbled the diodes operating point. As soon as I disabled the DC, the mixer started behaving as expected, but with a low output (13.5dB conversion loss).

2 - Now the mixer is behaving as a text-book mixer, so I changed the mixer diodes for Schottky diodes. I have a small stock of Hewlett Packard spare parts - HSCH-3486 Zero-Bias Schottky Diodes. It made a world of difference. Conversion loss reduced to about 9dB.

3 - Experimented with ports. Swapped ports in the previous diagram. Gained another dB: RF is now 130MHz. IF is now 14MHz and LO is now 144MHz.

4 - Experimented with ferrites. I used a pair of 2-hole "binocular" (pigs-nose) ferrites. Worked a treat. Gained another dB and now identical loss in both directions- 7dB. With +18dB drive the conversion (up and down) is linear (dB for dB) up to about +5dB input. I did reduce the 130MHz drive to +18dBm from the -23.5dBm I was using earlier. Didn't want to burn the Schottky diodes.

These ferrites are 7mm x 7mm x 4mm, with 1.7mm Dia. holes. They are used in many 2-port RF power dividers for up to 2.7GHz.

So if one is playing about with passive mixers, there is a big difference between HF receivers using old radios as IF, or classroom demos, but as soon as you start using HF and VHF then the components all become critical. Real Schottky, matched, VHF ferrites, and of course, NO DC from the signal generator!

I have been working in telecoms since about 1970 and I have never seen a signal generator that is intended to deliver DC. But I suppose one should "always expect the unexpected".

Very best regards from Harry - SM0VPO

Hello gentlepeople,

I wonder if anyone else can give me some input ideas? I have been building passive mixers for decades and this is the "standard" building block I nearly always use:

I have been working on a 144MHz to 14MHz transverter and as usual I used "old faithful" mixer as shown above. I am using some lovely ferrite rings that are capable of handling VHF. I have the same ferrite rings for transformer-coupling in the output of my 200mW FM transmitter. The losses are very low. I chose ferrite rings because I want the FM transmitter and the transverter to be wide-band - ie. no tuning required within the band.

https://www.elfa.se/Web/Downloads/_t/ds/RND%20165-00182_eng_tds.pdf

The transmitter (high-power local oscillator) circuit works perfectly from 85MHz to 145MHz with these ferrites.

The problem I have with the transverter it is that it is not bi-directional. When I inject a 144MHz signal into the IF port at +4dBm and I get a beautiful 14MHz at 0dBm (1mW) at the RF port. But if I stuff 14MHz at +4dBm into the RF port then the output from the IF port is -15dBm; almost 20dB (99%) loss.

Has anyone else any experience of this?

In older projects where it worked, the only difference is can see is the frequency. Perhaps I am "pushing mu luck" to use the same circuit to work at HF and VHF simultaneously? Also, I am using standard 1N4148 switching diodes, but they have always worked for me. I never bothered with schottky diodes before.

I have tried ferrite ring transformers with different turns, but the losses are exactly the same. I also increased the original local oscillator circuit from 10mW to 200mW (which is how the 200mW FM transmitter was born to make sure the drive level was sufficient.

I know that I could use an active mixer, the same as other published designs, and I could eliminate the transformer solution. But an active mixer is mono-directional and it would need some form of TX/RX switching, which I was really trying to avoid.

Finally, a comment to Ivan. Your comment on the FM transmitter regarding filtering was really good. I am finalizing a 7-pole elliptical filter that now seems to be working fine. The filter "knee" losses start at about 2dB around 107.5MHz, rising to 30dB at 118MHz. The Cauer-Chebyshev filter is quite critical for coil winding and it needs to be well documented to make it repeatable and get the three notches at exactly the right places.

Very best regards from Harry - SM0VPO