Bias Tuning M easurem ents
4.4 D ynam ic Bias Tuning
Measurements of bias tuning characteristics are traditionally done in a quasi-static fashion whether performed by hand or with an automated system. If an oscillator is being used as a modulator which
is being rapidly voltage-tuned, then one might expect that the tuning properties will be different.
Dynamic bias tuning measurements can be made by linearly ramping the oscillator bias and measuring the power and frequency variations with a calibrated (and preferably linear) discriminator. The operation of such a discriminator is discussed in detail in Chapter 8.
One would start by making quasi-static scans with the Auto_Bias system, gradually decreasing the scan duration and then transferring to the linear ramp and discriminator arrangement to continue measurements up to megahertz modulation rates. The modulation bandwidth is mainly set by the power supply - see Chapter 3.
I began characterising an oscillator in this way using the
automated system but found there was no apparent change in the tuning curves. In a scan of 25 points covering 4.3 to 5.7 volts bias, I reduced the scan time from 480 seconds to 90 seconds by reducing the
time interval between data points. An example trace of five scans covering this range is shown below. The same oscillator block and diode were used as before.
Frequency GHz 87.7 87.6 87.5 87.4 87.3 87.2 87.1 87.0 86.9 86.8 « • • « 0 0 0 «1 0 » *%###* 480s 350s 240s 120s 90s 3.8 4.0 4.2 4.4 4.6 4.8 5.0 5.2 5.4 5.6 Bias /volts
Chapter Pour Bias Tuning Measurements
As can be seen, the points are virtually coincident for all five scans. Therefore, at these timescales there does not appear to be any variation in tuning behaviour with modulation rate. Regrettably, time did not permit any further work in this area and I was unable to continue work on dynamic measurements.
However, this is a topic which deserves attention. On the basis of some extremely brief investigations, it is expected that tuning hysteresis will be evident for an oscillator whose bias is ramped up and down. At slow modulation rates, the hysteresis is large and probably due to thermal effects. As electrical power is cyclically applied, the
device temperature varies. Thermal time constants between the device
and its package and between the package and the oscillator block then control the output response. At high modulation rates thermal effects will be too slow to respond but a small amount of hysteresis is likely, due to much shorter electronic time constants within the device.
A number of studies of the bias tuning mechanisms of low frequency Gunn diodes have been made, mostly around lOGHz. They concur that at low modulation frequencies (up to about IMHz for an X- band device) the bias tuning is thermally controlled whereas at higher frequencies the device capacitance becomes the dominant tuning element^’^'^’^. The frequency dependence of the modulation sensitivity disappears above about IMHz, and above lOMHz a dominant phase modulation is observed which is flat up to the relaxation frequency of the rf energy stored in the cavity. This is typically 0.5-lGHz for a coaxial X-band cavity.
DeSa and Hobson have shown^ from quasi-static measurements
that the incremental frequency:voltage change d f / d V is proportional to the incremental frequency: temperature change df / dT. At higher modulation rates, up to about IMHz, they noted two thermal effects. Firstly, a fast device-dependent effect which is a result of the equilibrium between device dissipation and heat flow out of diode and secondly a slower effect due to heat flow from device to heat sink.
Generally speaking, the ac thermal response is largely dictated by thermal response of diode substrate. The use of an integral heat sink
(IHS) in diode manufacture allows the semiconductor substrate to be
thinned substantially, thereby lowering the thermal impedance and increasing the power handling capability^®.
I would like to investigate this subject in detail in the future with respect to our W-band sources as it will shed light on the operating conditions of modulated oscillators. Undoubtedly, the oscillators in use at St. Andrews differ substantially in operation from the ones described in the references above. X-band diodes are usually transit-time devices mounted in fundamental mode cavities, whereas our oscillators extract the second harmonic frequency of very short devices operating in complicated hybrid modes. It may be possible to model the dynamic bias tuning processes and potentially explain the mechanisms involved. Ultimately, an understanding of this property may allow one to compensate for nonlinearities and produce Gunn oscillator modulators which have more ideal characteristics.
Chapter Four Bias Tuning Measurements
REFERENCES
^ Intelligent Interfaces, P.O. Box 80, Eastleigh, Hants S05 5YX.
2 Wild Vision Ltd., 15 Witney Way, Boldon Business Park, Boldon Colliery, Tyne & Wear NE35 9PE.
3 DAI349-4 Harmonic Mixer, GEC Plessey Semiconductors, Crompton
Road, Groundwell Industrial Estate, Swindon, Wiltshire SN2 5AF.
^ Dr. Graham Smith, St. Andrews University, Private Communication.
^ Faulkner, E.A. & Meade, M.L., “Frequency Modulation Sensitivity of Gunn Oscillators”, Electronics Letters, 5 (10), 15^ May 1969, pp. 217- 218.
^ Martin, B. & Hobson, G.S., “High-Speed Phase and Amplitude
Modulation of Gunn Oscillators”, Electronics Letters, 6 (8), 16^^ April 1970, pp. 244-246.
^ Martin, B. & Hobson, G.S., “Angle Modulation of Frequency-Locked Gunn Oscillators”, Electronics Letters, 7 (14), 15^^ July 1971, pp. 399- 401.
^ Hobson, G.S., The Gunn Effect, Oxford, 1974, pp. 99-104.
^ DeSa, B.A.E. & Hobson, G.S., “Thermal Effects in the Bias Circuit Frequency Modulation of Gunn Oscillators”, IEEE Trans. Electron
Devices, ED-18 (8), Aug. 1971, pp. 557-562.
Eddison, I.G., “Indium Phosphide and Gallium Arsenide Transferred-
Electron Devices”, Infrared & MM Waves, 11 (Ch. 1), Ed. K. Button, Academic Press, 1984, pp. 1-59.