Power Op-amps Rock - Driving A YIG Oscillator

High on the list of things I need to write up is my radar-building saga; suffice for now to say that I’m (now) using a tiny yttrium iron garnet in a magnetic field as my viola to generate the signal for my frequency modulated continuous wave radar.

As fancy as it sounds, a Stellex YIG oscillator unit (mounted yttrium garnet, heater, tune and modulation coils) can be found on ebay for around $20. I got a Stellex 6755 job that does 8-10 ghz. In short, all one needs to do is apply heater voltage to the thing to get it generating rf, and adjust the frequency by changing the current in the two tuning coils. It puts out +16 dBm, respectable.

Flash back to radar - the goal is to generate a frequency-sweeping rf signal that moves as linearly as possible. First, we need to generate that sawtooth pattern to modulate our YIG. I played with several analog sawtooth producing circuits but couldn’t find one that generated a perfectly linear ramp, and I really don’t want to have to sample the modulation signal to (try to) correct the nonlinearity in software later. Not feeling like pulling out a microcontroller for the task, I threw together this guy from things I had on hand:



A TTL oscillator is cleaned up and complemented by two schmitt trigger inverters a la 74HC14. On the rising edge, a 74LS191 4-bit counter latches its incremented value, and on the falling edge, the DAC samples the value and produces the specified current. Eventually I'll replace the counter with an FPGA so I can do arbitrary waveforms (hint, a triangle wave lets you deduce both range distance and doppler shift at the same time in an FM CW radar), but for now the modulator is at least aesthetically pleasing.



While the DAC in fact outputs its value in the form of current, not voltage, the current range (max 10mA) is far below what is needed to tune and modulate the YIG (around 250 mA). Alas. Thus, we use the resistor network specified in the DAC’s datasheet to convert our current source (hopefully very linearly) into a voltage signal with which we will… drive a power op-amp setup to convert voltage signals to current.



One of my new favorite parts I happened to find in my ICs bin is the L272, a dual op-amp beast that can source or sink 1A continuously. Rather than using a (regular old) op amp to drive the gate of a mosfet or BJT as is common, this beast of an op amp means we can drive the tuning coil(s) of the YIG directly.

Unfortunatly, the huge current capability comes at a cost: the L272 has a slew rate of 1 V/uS and a unity gain bandwidth of 350 kHz, far below what would be required to follow my above modulator signal with the 5 MHz clock on it. Drat. This limits the sweep rate and thus TX bandwidth of the radar considerably, but at least doesn’t suffer the oscillation problems with transistor-driver circuits that limit max slew rate and bandwidth there, too.
In any event, we use a 10 ohm resistor on the negative side of the tuning coil to generate a voltage signal (hopefully very) linearly related to the current drawn,and use the op amp to compare this voltage with the “requested” voltage, adjusting its voltage output until the current monitor’s voltage matches.



So! It works. It even turns out YIG sources tune down in frequency faster than they tune up, thus helping us match the output to the sawtooth modulation signal exactly. More to come on the radar front...