Christmas project - Can I injection lock a Mk2 DB6NT transverter?

The excellent G4DDK Anglian transverter range has the facility to "injection lock" it's 116MHz Butler Crystal oscillator to a stable 116MHz source such as a synthesiser. This allows the whole transverter to be locked to a stable reference such as a 10MHz GPSDO. Injection locking does this but still retains the clean, low phase noise of it's Butler Crystal oscillator. Using just the synthesiser as an external LO would have meant that the whole transverter's noise performance would have been governed by the synthesiser's phase noise performance. Using this method, the crystal is just "nudged" on to the exact frequency by the synthesiser.

This got me wondering if I could do the same with the single FET oscillator used in all the previous-generation Kuhne G2 series of microwave transverters. I have tested this out on a 5.7GHz G2 with a 117MHz crystal LO, but the oscillator circuit used in the G2 is common to all the Kuhne transverters below 24GHz so should work in all of them given the correct crystal frequency.

Most microwavers know that the G2 series have the facility to remove the crystal and inject an external LO. It's a standard mod used by Kuhne themselves in the "external LO" version of their oscillator chains. The injection point is via a 100pF to one end of the crystal position, so I tried to inject a signal at that point while the crystal was in position and oscillating and see what happened.
I used  a +5dBm from a G4JNT LMX2541 fractional N synthesiser. This board is simple has good progamming support and has an on chip VCO. It is not though, in the top class for phase noise!
I monitored the transverter LO at the output of the first  tripler at 351MHz with a spectrum analyser and a Rubidium - locked frequency counter. As you would expect, a nice clean signal for the crystal but slightly LF of the required frequency.  On connection of the external 117MHz synthesiser,the oscillator immediately locked up showing 351.0000MHz but sadly the output noise spectrum immediately degraded to match the synthesiser, not the nice clean crystal. I reduced the synthesiser level and watched the spectrum and frequency. As I reduced the synthesiser drive level, the crystal stayed in in lock and the output noise reduced.
"result!"
With about 20dB attenuation - a drive level of -15dBm, the frequency was still 351.0000 and locked but the noise now looked like the unlocked crystal.
So there we have it.
Without removing the crystal, just like the Anglian, you can injection lock your Kuhne G2 transverter to a synthesiser such as the LMX2451 or ADF4350/1 and not spoil the phase noise performance!


Have a Happy and quiet (low phase noise)  Christmas!

73 John

Isn't being an Engineer brilliant?

This afternoon I've been working on the switching for my 5760MHz EME dish feed electronics. Nothing complicated, just a couple of high side MOSFET switches. A 300ms delayed 12 V high current one and a switched, non-delated 28V rail to activate the coax relay. Standard P channel switch circuitry with the 12V  input to the Source of the FET and the output from the Drain. 6.2k resistor from source to gate to keep it off and a NPN transistor from Gate to Ground to turn it on with. CR delay circuit in the base of the driver from the 12V TX out from my DB6NT transverer.  Worked fine, so I used the same circuit for the 28V one but with no delay ahead of it.
Built it, connected it up tried it and BANG! the driver transistor fried. Transistor was taking over an amp and current limiting the 28V supply .FET had also gone phut. Gate Source short circuit dumping the full 28V across the driver transistor. 
Whoah! Why do that with the 28V supply and not the huge 8A 12V supply?
Much head scratching and a couple of changed devices later (good job I have apenty of both)  I thought I'd better check the datasheet of the FET. Yep, maximum drain to source voltage was 60V so that's OK. Ah..... maximum drain to GATE voltage +/-20V Lower than VDSmax.... I didn't know that!
Grounding the Gate in this circuit with a transistor would put 28 - Vcesat = 27.8V   from Source to gate. Yep that would be more than 20V then. Failure would short gate to source connecting the full 28V across the collector to emitter juunction of the driver transistor taking that out as well!

I  do like to understand why things happpen.

Solution? Add an 8.2k between the transistor collector and the FET Gate. Limiting the Vgs of the FET to 28x8.2/14.4 = 15.94 V below 20V now so that's fine.

Reason for the title of the Blog? Every problem is a learning experience, and Engineers are all about solving problems.     Ergo, Engineering is all about learning.... BRILLIANT!! !