Using HP8970B with circulators.
(Dec 5 2012)
This page is a sub-page to PRECISION MEASUREMENTS OF NOISE FIGURES. Below is details and raw data from measurements on a set of 8 low noise amplifiers for 144 MHz. The amplifiers are presented here.

Figure 1 gives the setup for measurements 1,2 and 4 while figure 2 shows measurement 3 which differs in that the second circulator is omitted.

Figure 1. Arrangement for measurements 1,2 and 4. Blue is for measurement, red is for calibration.

Figure 2. Arrangement for measurement 3. Blue is for measurement, red is for calibration.

The purpose of the circulator on the noise head output is to make sure that the impedance at the LNA input is exactly the same regardless of whether the the noise head is in the on or in the off state. The dual circulator used has an isolation of about 70 dB, it is fine tuned by use of small magnets. The NF errors caused by variations in the noise head can be as large as 0.5 dB. Much less than 70 dB isolation is required to bring this error to below 0.001 dB.

The fixed attenuator at the DUT input serves the purpose of presenting a reasonable source impedance at all frequencies. Some amplifiers might oscillate without this attenuator. It can be omitted if such amplifiers will not be measured.

A temperature sensor was permanently mounted on the noise head with a small amount of silicon grease to ensure good thermal contact. A time interval of about 10 minutes was needed for the temperature to become stable enough to perform calibration or measurements with a temperature difference from start to end of 0.1 degree or less. Care was taken to always use the same cables and adaptors.

An advantage with this setup is that once the temperature has stabilized, the change of DUT does not affect the noise head temperature at all. This helps accuracy.

Figure 3 is a photo showing the measurement of the NE334-S01 amplifier using the setup in figure 1 (the second measurement, figure 5.)

Figure 3. Measurement with two circulators. The temperature probe was permanently fixed on the noise head at the point where the N connector joins the flat surface at the end of the noise head. The Tx module above and slightly to the right of the DUT (NE334-S01) was the source of thermally conducting silicone grease used to give goot thermal contact between the temperature sensor and the noise head. The noise head and the dual circulator were not moved or touched during measurement. The temperature was stable about 10 to 15 minutes after calibration. The circulator on the input of the HP8970B is a similar type as the one on the noise head, but a single one. It is well screened, a requirement in the more noisy environment in my own (SM5BSZ) lab. All amplifiers were powered by the nominal 15 V DC powers the whole day and had a warm up of several hours before the first measurement. The room temperature was 298.5 degrees at the moment of this photo and did not vary much. The noise head was only 3 degrees warmer, the good thermal contact with the circulator keeps it cool.

The amplifiers were measured with the hot/cold method using ice and steam and also by comparing S/N on a highly stable signal: PRECISION MEASUREMENTS OF NOISE FIGURES. The data presented below has the result of those measurements in columns 2 and 3. GAIN/BSZNov2012 and NF/BSZNov2012.

Column 1 is the name of the DUT.

Columns 4 and 5 give the readings of the HP8970B. They are labeled GAIN/DRICirc and NF/DRI/Circ.

Columns 6 and 7 GAIN/diff and NF/diff give the differences between the observed values in columns 4 and 5 and the previously measured ones in columns 3 and 4. Note that the attenuation through the circulator and 3 dB attenuator from the measurement of NOTHING is added in column 6. Avrg below column 7 is the average of the differences in NF for the 8 amplifiers. The measurement on NOTHING, the BNC female to female connector is excluded from the average. It suffers from poor accuracy since the gain is zero and a small error in the NF of the measurement system itself that is produced by the calibration process will have a large impact.

Column 8, Dev-, is the data in column 7 with Avrg subtracted. Ideally this column should be zero and indeed the numbers are small.

Column 9 Dev Sqr. is the sum of squares of column 8. It is used to compute the RMS deviation (from zero) in column 8

Figure 4. Arrangement as in figure 1 with the HP8970B owned by HB9DRI.

Figure 5. Arrangement as in figure 1 with the HP8970B owned by HB9DRI. This is a repetition of the first series of measurements which is listed in figure 3

Figure 6. Arrangement as in figure 2 with the HP8970B owned by HB9DRI. Note that the RMS deviation is significantly larger when the second circulator is omitted.

Figure 7. Arrangement as in figure 1 with the HP8970B owned by HB9Q. The same noise head was used. As expected, the different HP8970B units show the same results.