# State of the art in solid-state noise sources enhances measurement accuracy.

State of the Art in Solid-State Noise Sources Enhances Measurement
Accuracy

Improvements in noise diode efficiency, on-to-off impedance ratio, supply voltage sensitivity and temperature stability significantly have increased the accuracy of noise figure measurements over the past two years. Modern noise figure measurements are made using microprocessor controlled instruments, which perform most of the calculations necessary to derive the estimated noise figure. This paper shows how the noise figure measurement accuracy depends on the properties of the noise source and how new types of noise sources improve the measurement accuracy.

Generally, noise figure is determined by measuring the ratio of noise power available at the output of the device under test, (DUT) as the available input noise power is switched between two known levels. This ratio is called the Y-factor.

Y = [T(effective) + T(hot)] / [T(effective) + T(cold)]

where

T(effective) = the DUT's effective noise temperature T(hot) = the noise temperature of the noise source in on condition T(cold) = the noise temperature of the noise source in off condition

The available gain of the DUT is assumed to be equal for the two input noise conditions, as shown in Figure 1.

Noise figure can be described as the ratio of the actual noise power available at the output of the DUT to that available from an ideal noiseless DUT of otherwise identical characteristics when the source termination is at the reference temperature of 290 [degrees] K.

NF = [T(effective) + 290]/290 NF(dB) = 10 log {[T(hot)/290 - 1] - Y [T(cold)/290 - 1]} -10 log {Y-1}

The uncertainty of the measurement resulting from errors in T(hot), T(cold) and Y can be evaluated as, [Mathematical Expression Omitted]

The noise figure measurement accuracy may be improved by taking the indicated derivatives, [Mathematical Expression Omitted] where the excess noise ratio (ENR) is defined by,

ENR = T(hot)/290 - 1

From Equation 5, the overall accuracy can be improved when,

Y >> 1

and when T(hot) and T(cold) are held constant.

The last parameter that is important in order to obtain accurate noise figure measurements is the noise source impedance. In Equation 1, the available gain terms were assumed to be equal for the noise source on and off conditions. This is not true. The available gain depends on the noise source impedance. The available gain, as well as the noise figure of the DUT, changes as the source impedance is changed. Expressions for the change in available gain can be used to correct for or estimate the error resulting from the impedance mismatches. The correction for the noise figure dependence upon the source impedance involves measurements using several known source impedances and calculation of the DUT's minimum noise figure, the corresponding optimum impedance and the DUT's noise resistance. Therefore, the minimum change in source impedance is one of the most important improvements in new solid-state noise sources.

To correct for the change in available gain, the Ys in Equations 1 and 3 have to be multiplied by,

Ga(cold)/Ga(hot) =

[Mathematical Expression Omitted]

where

[Mathematical Expression Omitted] Assuming that [[Gamma].sub.2](hot) [is nearly equal] [[Gamma].sub.2](cold) then [Mathematical Expression Omitted]

The summarized results of these calculations set goals for the solid-state noise source improvements in order to enhance the noise figure measurement accuracy, including high Y-factor, hence high ratio of ENR to NF; constant ENR, thereby constant T(hot); constant T(cold); and constant and well matched source impedance.

Generally, noise diodes are designed for operation at certain supply voltages and up to a maximum frequency determined by the diode capacitance and diode package. The high efficiency of the noise diode assures low rise in operating junction temperature and little self heating due to power dissipation. T(cold) is thereby kept constant and close to the ambient temperature. New noise sources use typically 10 mA of bias current for a 10 MHz to 18 GHz unit.

The amount of attenuation from the noise diode to the output of the noise source helps isolating the noise diode junction temperature generated power from the output. The power will be attenuated and T(cold) is,

T(cold) = [eta] Tdiode(cold) + (1 - [eta]) T(ambient)

where

1 / [eta] = the attenuation

The new 15 dB ENR model noise source has a minimum attenuation of 18 dB and the 6 dB ENR model has an attenuation of 27 dB; [eta] = 1.6 percent and 0.2 percent, respectively. Therefore, T(cold) becomes very close to T(ambient), which can be measured and entered in most noise figure meters that then automatically calculate the noise figure.

The large amount of attenuation also results in a constant output impedance. This is the key to controlling the impedance mismatch uncertainties. The higher the ENR of the noise diode, the more attenuation can be applied and the more constant the output impedance will be. The typical ENR of noise diodes is 35 dB. A noise source with 18 dB of internal attenuation has an isolation of 36 dB between the output and the reflection from the noise diode. Therefore, the change of the noise diode impedance becomes less significant.

Temperature Stability and Voltage Sensitivity

The prime function of a noise source is to establish a reference noise power level for noise figure measurements. The reference noise power level when the noise source is on is ideally equivalent to the noise power generated by a resistor with the same impedance and a physical temperature of T(hot) [degrees] K.

k X B X T(hot)

where

k = 1.38 X [10.sup.-23] J/[degrees] K B = the measurement resolution bandwidth (Hz) T(hot) = the physical temperature ([degrees] K), at which a matched resistor would create the same available power.

As the excess noise ratio (ENR) is defined by,

ENR = T(hot) / 290 -1

T(hot) is constant when the ENR is constant. It is therefore important to keep the ENR constant during changing environments.

The ENR of a solid-state noise source varies with temperature and supply voltage. Temperature stability of the ENR of better than 0.002 dB/[degrees] C has been obtained over temperatures ranging from -54 to +100 [degrees] C. The DC current through the noise diode is varied internally to obtain temperature compensation of the RF noise power generated by the diode. Experiments with noise diodes each supplied with different current and tested at -54, +25 and +100 [degrees] show that a variation of the current can compensate for the noise power generated by the noise diode. A common temperature coefficient of the optimum supply current has been obtained from the experimental data in Figure 2. The obtained results from Si noise diodes show that the current through the noise diode can be varied with a linear positive temperature coefficient of 23 [mu]A/[degrees] C to main an almost constant noise output power. The temperature compensating current regulator, shown in Figure 3, has a temperature coefficient of 23 [mu]A/[degrees] C for noise source temperature compensation applications.

The Zener diode [D.sub.1], and the base-emitter diode of the transistor [Q.sub.1], create an almost constant voltage across resistor [R.sub.1]. According to Ohms law, the current [I.sub.1] through [R.sub.1] will be constant if [R.sub.1] is constant. The [Q.sub.1] transistor's temperature coefficient, base-emitter-voltage is -2 mV/[degrees] C.

If the Zener diode voltage and the transistor base-emitter voltage are not constant with temperature but [R.sub.1] is constant over temperature, then the ideal resistor is,

[Mathematical Expression Omitted]

[I.sub.2] through the noise diode is the sum of [I.sub.1] and the basis-collector current [I.sub.b].

[I.sub.2] = [I.sub.1] - [I.sub.b]

In reality, [R.sub.1] changes with temperature nd the basis-collector current [I.sub.b] changes from 6 percent of [I.sub.2] at -55 [degrees] C to 2 percent of [I.sub.2] at 175 [degrees] C (junction temperature) for the transistor.

As first order approximation,

[Mathematical Expression Omitted]

from Equation 16 [delta] Vz/[delta] [degrees] C can be determined and thereby Vz for a wanted [Delta] [I.sub.2]/[Delta] [degrees] C at a given current [I.sub.2]. [Delta] Vz/[Delta] [degrees] C for some Zener diodes can be varied by varying the current through the Zener diode. This is true for Zener diodes with breakdown voltages between 3 and 8 V. [R.sub.2] sets the current through the Zener diode, as shown in Figure 4. Reverse voltage protection up to 70 V is accomplished by diode [D.sub.2]. The regulation of the current also improves the noise source's sensitivity to supply voltage variations significantly. Improvements have been made so that variation of a +28 V DC supply as much as -3 v and +10 V affects the noise output less than 0.05 dB in a test experiment carried out at 18 GHz.

Bent Hessen-Schmidt received his MSEE from the Polytechnical University of Denmark in 1985. Currently, he is director of engineering at Noise Com Inc. Hessen-Schmidt has specialized in design and development of microwave electronics. Present work includes research, design and development of new noise, synthesizer and frequency modulated products. Previously, he worked at Crimp A/S, Denmark, where he started its microwave, fiber optics and radar systems groups.

Improvements in noise diode efficiency, on-to-off impedance ratio, supply voltage sensitivity and temperature stability significantly have increased the accuracy of noise figure measurements over the past two years. Modern noise figure measurements are made using microprocessor controlled instruments, which perform most of the calculations necessary to derive the estimated noise figure. This paper shows how the noise figure measurement accuracy depends on the properties of the noise source and how new types of noise sources improve the measurement accuracy.

Generally, noise figure is determined by measuring the ratio of noise power available at the output of the device under test, (DUT) as the available input noise power is switched between two known levels. This ratio is called the Y-factor.

Y = [T(effective) + T(hot)] / [T(effective) + T(cold)]

where

T(effective) = the DUT's effective noise temperature T(hot) = the noise temperature of the noise source in on condition T(cold) = the noise temperature of the noise source in off condition

The available gain of the DUT is assumed to be equal for the two input noise conditions, as shown in Figure 1.

Noise figure can be described as the ratio of the actual noise power available at the output of the DUT to that available from an ideal noiseless DUT of otherwise identical characteristics when the source termination is at the reference temperature of 290 [degrees] K.

NF = [T(effective) + 290]/290 NF(dB) = 10 log {[T(hot)/290 - 1] - Y [T(cold)/290 - 1]} -10 log {Y-1}

The uncertainty of the measurement resulting from errors in T(hot), T(cold) and Y can be evaluated as, [Mathematical Expression Omitted]

The noise figure measurement accuracy may be improved by taking the indicated derivatives, [Mathematical Expression Omitted] where the excess noise ratio (ENR) is defined by,

ENR = T(hot)/290 - 1

From Equation 5, the overall accuracy can be improved when,

Y >> 1

and when T(hot) and T(cold) are held constant.

The last parameter that is important in order to obtain accurate noise figure measurements is the noise source impedance. In Equation 1, the available gain terms were assumed to be equal for the noise source on and off conditions. This is not true. The available gain depends on the noise source impedance. The available gain, as well as the noise figure of the DUT, changes as the source impedance is changed. Expressions for the change in available gain can be used to correct for or estimate the error resulting from the impedance mismatches. The correction for the noise figure dependence upon the source impedance involves measurements using several known source impedances and calculation of the DUT's minimum noise figure, the corresponding optimum impedance and the DUT's noise resistance. Therefore, the minimum change in source impedance is one of the most important improvements in new solid-state noise sources.

To correct for the change in available gain, the Ys in Equations 1 and 3 have to be multiplied by,

Ga(cold)/Ga(hot) =

[Mathematical Expression Omitted]

where

[Mathematical Expression Omitted] Assuming that [[Gamma].sub.2](hot) [is nearly equal] [[Gamma].sub.2](cold) then [Mathematical Expression Omitted]

The summarized results of these calculations set goals for the solid-state noise source improvements in order to enhance the noise figure measurement accuracy, including high Y-factor, hence high ratio of ENR to NF; constant ENR, thereby constant T(hot); constant T(cold); and constant and well matched source impedance.

Generally, noise diodes are designed for operation at certain supply voltages and up to a maximum frequency determined by the diode capacitance and diode package. The high efficiency of the noise diode assures low rise in operating junction temperature and little self heating due to power dissipation. T(cold) is thereby kept constant and close to the ambient temperature. New noise sources use typically 10 mA of bias current for a 10 MHz to 18 GHz unit.

The amount of attenuation from the noise diode to the output of the noise source helps isolating the noise diode junction temperature generated power from the output. The power will be attenuated and T(cold) is,

T(cold) = [eta] Tdiode(cold) + (1 - [eta]) T(ambient)

where

1 / [eta] = the attenuation

The new 15 dB ENR model noise source has a minimum attenuation of 18 dB and the 6 dB ENR model has an attenuation of 27 dB; [eta] = 1.6 percent and 0.2 percent, respectively. Therefore, T(cold) becomes very close to T(ambient), which can be measured and entered in most noise figure meters that then automatically calculate the noise figure.

The large amount of attenuation also results in a constant output impedance. This is the key to controlling the impedance mismatch uncertainties. The higher the ENR of the noise diode, the more attenuation can be applied and the more constant the output impedance will be. The typical ENR of noise diodes is 35 dB. A noise source with 18 dB of internal attenuation has an isolation of 36 dB between the output and the reflection from the noise diode. Therefore, the change of the noise diode impedance becomes less significant.

Temperature Stability and Voltage Sensitivity

The prime function of a noise source is to establish a reference noise power level for noise figure measurements. The reference noise power level when the noise source is on is ideally equivalent to the noise power generated by a resistor with the same impedance and a physical temperature of T(hot) [degrees] K.

k X B X T(hot)

where

k = 1.38 X [10.sup.-23] J/[degrees] K B = the measurement resolution bandwidth (Hz) T(hot) = the physical temperature ([degrees] K), at which a matched resistor would create the same available power.

As the excess noise ratio (ENR) is defined by,

ENR = T(hot) / 290 -1

T(hot) is constant when the ENR is constant. It is therefore important to keep the ENR constant during changing environments.

The ENR of a solid-state noise source varies with temperature and supply voltage. Temperature stability of the ENR of better than 0.002 dB/[degrees] C has been obtained over temperatures ranging from -54 to +100 [degrees] C. The DC current through the noise diode is varied internally to obtain temperature compensation of the RF noise power generated by the diode. Experiments with noise diodes each supplied with different current and tested at -54, +25 and +100 [degrees] show that a variation of the current can compensate for the noise power generated by the noise diode. A common temperature coefficient of the optimum supply current has been obtained from the experimental data in Figure 2. The obtained results from Si noise diodes show that the current through the noise diode can be varied with a linear positive temperature coefficient of 23 [mu]A/[degrees] C to main an almost constant noise output power. The temperature compensating current regulator, shown in Figure 3, has a temperature coefficient of 23 [mu]A/[degrees] C for noise source temperature compensation applications.

The Zener diode [D.sub.1], and the base-emitter diode of the transistor [Q.sub.1], create an almost constant voltage across resistor [R.sub.1]. According to Ohms law, the current [I.sub.1] through [R.sub.1] will be constant if [R.sub.1] is constant. The [Q.sub.1] transistor's temperature coefficient, base-emitter-voltage is -2 mV/[degrees] C.

If the Zener diode voltage and the transistor base-emitter voltage are not constant with temperature but [R.sub.1] is constant over temperature, then the ideal resistor is,

[Mathematical Expression Omitted]

[I.sub.2] through the noise diode is the sum of [I.sub.1] and the basis-collector current [I.sub.b].

[I.sub.2] = [I.sub.1] - [I.sub.b]

In reality, [R.sub.1] changes with temperature nd the basis-collector current [I.sub.b] changes from 6 percent of [I.sub.2] at -55 [degrees] C to 2 percent of [I.sub.2] at 175 [degrees] C (junction temperature) for the transistor.

As first order approximation,

[Mathematical Expression Omitted]

from Equation 16 [delta] Vz/[delta] [degrees] C can be determined and thereby Vz for a wanted [Delta] [I.sub.2]/[Delta] [degrees] C at a given current [I.sub.2]. [Delta] Vz/[Delta] [degrees] C for some Zener diodes can be varied by varying the current through the Zener diode. This is true for Zener diodes with breakdown voltages between 3 and 8 V. [R.sub.2] sets the current through the Zener diode, as shown in Figure 4. Reverse voltage protection up to 70 V is accomplished by diode [D.sub.2]. The regulation of the current also improves the noise source's sensitivity to supply voltage variations significantly. Improvements have been made so that variation of a +28 V DC supply as much as -3 v and +10 V affects the noise output less than 0.05 dB in a test experiment carried out at 18 GHz.

Bent Hessen-Schmidt received his MSEE from the Polytechnical University of Denmark in 1985. Currently, he is director of engineering at Noise Com Inc. Hessen-Schmidt has specialized in design and development of microwave electronics. Present work includes research, design and development of new noise, synthesizer and frequency modulated products. Previously, he worked at Crimp A/S, Denmark, where he started its microwave, fiber optics and radar systems groups.

Printer friendly Cite/link Email Feedback | |

Author: | Hessen-Schmidt, Bent |
---|---|

Publication: | Microwave Journal |

Date: | Apr 1, 1991 |

Words: | 1558 |

Previous Article: | On-wafer scaling load-pull. |

Next Article: | Using an external mixer with a spectrum analyzer. |

Topics: |