SMALL LOSSES. DEMONSTRATION OF PRINCIPLES ON 144 MHZ.
(Jan 29 2013)

The DUTs for 144 MHz.

With 0.25 and 0.125 sections of 50 and 75 ohm cable connected between a source and a load one can present five different impedances to a load. while having low and similar dissipative losses.

DUT1 = 0.125 wl (126mm) 75 ohm
DUT2 = 0.25 wl (328mm) 50 ohm
DUT3 = 0.125 wl (152mm) 50 ohm
DUT4 = 0.25 wl (318mm) 50 ohm
DUT12 = DUT1 and DUT2 in series with DUT 1 connected to load.
DUT123 = DUT1, DUT2 and DUT3 in series with DUT 1 connected to load and DUT2 in the middle.
ETC ETC.

To allow the measurement of insertion loss in a 50 ohm system a network analyzer was used to establish a source impedance near 50 ohms, the reference impedance for this series of measurements. That was done by tuning the reference until the impedance was (nearly) unchanged when DUT3 was connected. All the DUTs were inserted and the impedances measured to allow a computation of mismatch losses when a DUT is inserted between a source and a load, both having the same impedance as the reference. For details look here: Impedances of source, load and DUTs. The results are used in table 1 below.

If the load is matched to 50 ohm, the insertion loss will be the sum of the mismatch loss and the dissipative loss but if the load is a low noise amplifier its gain as well as its noise floor will become different for different source impedances and as it turns out, mis-match losses do not affect S/N much. For non-matched amplifiers the gain and noise floor will change much more than for matched amplifiers, but the NF will not.

Measurement of insertion loss is performed by inserting different DUTs between the Tx and the Rx port. for details look here: measurement of insertion loss

Another way to measure small dissipative losses is to measure the return loss when the DUT is inserted between the short and open of the calibration kit. The instrument will show zero when either one of the cal kit parts is inserted directly on the network analyzer. If a DUT is inserted, the signal has to pass through it twice and for that reason one should expect a somewhat smaller signal. In real life the signal may be very different from that expectation due to instrument errors, but if one takes tha average of both open and short one can get an accurate result because the errors cancel. For details look here measurement of reflected power

Yet another way to measure dissipative losses is to measure the influence on the S/N when a DUT is inserted in front of a low noise amplifier (LNA). For this to work the LNA has to be tuned for optimum NF at 50 ohms source impedance. By use of the different DUTs, placing DUT1 at different distances from the LNA one can measure the S/N at a SWR of 1.53 with four equispaced phases. Together with the direct connection one can get 5 points by which one can fit a paraboloid to the NF that can be deduced from S/N and known losses. Details here: LNA for measuring small losses.

Some of the DUTs that give a small impedance transformattion were measured by the influence on the NF they give. For details look here: Dissipative losses measured by NF degradation


device         Zre  Zim     Mismatch   Insertion  Dissipat.   Dissipat.
                             loss       loss     loss/S12    loss/s11    loss/NF
                (ohms)       (dB)       (dB)      (dB)         (dB)       (dB)
Rx port      49.57  -0.85      -          -          -           -         -
Tx port      49.41  -0.05   0.0004        -          -           -         -
DUT1         56.11  19.97   0.1565     0.1753     0.0196      -0.008       -
DUT2         49.47  -0.06   0.0004     0.0597     0.0600       0.080     0.0591 
DUT3         50.59   0.90   0.0005     0.0290     0.0288       0.031     0.0279
DUT4         50.13  -0.02   0.0005     0.0590     0.0581       0.078     0.0588
DUT13        57.82  19.71   0.1576     0.2041     0.0484       0.022
DUT31        70.06 -12.06   0.1796     0.2294     0.0484       0.109
DUT123       54.44  20.10   0.1558     0.2696     0.1084       0.055
DUT132       57.64  19.68   0.1566     0.2631     0.1084       0.063
DUT312       70.02 -12.29   0.1810     0.2930     0.1084       0.158
DUT213       38.09 -12.63   0.1766     0.2815     0.1084       0.148
DUT231       34.03   5.81   0.1680     0.2802     0.1084       0.051
DUT42        50.15   0.01   0.0005     0.1188     0.1181       0.119
DUT24        48.70  -0.11   0.0008     0.1195     0.1181       0.119
DUT34        50.65   0.18   0.0007     0.0874     0.0869       0.089
DUT43        49.05   1.22   0.0002     0.0897     0.0869       0.089
DUT32        50.55   0.84   0.0004     0.0897     0.0888       0.090
DUT23        48.29  -0.92   0.0022     0.0903     0.0888       0.089  

Table 1 Impedances and losses for the different DUTs.

Losses of interesting things.

The above investigations show the feasability of using NF degradation to measure the dissipative losses of adapters, relays and other low loss things we might want to use in front of our low noise amplifiers.

Figure 1 shows a set of DUTs for which the losses would be interesting to know. Not really at 144 MHz where this study has been done, but at higher bands where losses in these things would be greater and more harmful. Remember, this page and is about methods...


Figure 1. Interesting things to measure the losses of. From above:
DUT5: BNC to N and back plus two N to N adapter pairs.
DUT6: BNC to N and back plus a N to N adapter and a CX-520-D relay.
DUT7: BNC to N and back plus a N to N adapter and a 8761B relay.
DUT8: BNC to N and back.
DUT9: Five BNC to BNC adapter pairs.




Device       Loss
             (dB)
DUT5        0.0233
DUT6        0.0372
DUT7        0.0243
DUT8        0.0057
DUT9 new    0.0612
DUT9 bent   0.0593

Table 2 Losses as determined from NF degradation.



DUT9 gives the losses of a BNC female to female adapter plus a BNC male to male adapter as 0.0120 dB. It seems reasonable to assume that the loss is similar for both adapter types which means that a single BNC to BNC adapter should be counted as a 0.006 dB loss on 144 MHz. In this case bending the ground fingers for higher contact pressure does not give a significantly lower loss.

DUT8 gives the loss of a N to BNC adapter pair as 0.0057 dB. It is reasonable to assume they have similar loss so a single N to BNC adapter should be counted as a 0.0029 dB loss on 144 MHz.

The difference between DUT5 and DUT8 is 0.0176 dB. That is the summed loss of four N to N adapters which means that each N to N adapter should be counted as 0.0044 dB on 144 MHz.

DUT6 is the CX-520-D relay plus a N to BNC adapter pair and a N to N adapter. The adapters sum up to 0.0101 dB so the dissipative loss of the relay should be 0.0271 dB on 144 MHz.

DUT7 gives the loss of the 8761B relay as 0.0142 dB when the adapter losses of 0.0101 dB are subtracted.

Conclusions.

This study points to methods that should be useful on higher frequencies where traditional measurements are more difficult and where really low losses and noise figures are essential when antennas point into the sky.

Small mismatch losses are totally harmless in narrowband systems since they generate no noise and do not change the noise figure of low noise amplifiers.

The losses in adapters are very small on 144 MHz provided, of course, that they are of good quality.

These are typical losses to count for some typical components on 144 MHz:

Adapters:
BNC to BNC 0.006 dB
N to BNC 0.003 dB
N to N 0.004 dB

Relays:
CX-520D 0.027 dB
8761B 0.014 dB

Remarks.

Cheap connectors and adapters may use inappropriate dielectric materials and perhaps also inappropriate metals. They may also have poor tolerances and sometimes a cheap, too thick, male connector pin can destroy an expensive quality connector.

The measurements linked to on this page all used the signal source described here: Impedances of source, load and DUTs. which has attenuators immediately on the adapter that is the male connector of the signal source. This was a mistake, there is no need for such a well defined impedance, some decimeters of a good quality cable would have eliminated the problem demonstrated in figure 2.


Figure 2. Holding my hand around the 3 dB attenuator nearest the LNA for 10 seconds gives a S/N loss of about 0.012 dB.



It takes something like an hour for the temperature to stabilize after the attenuator was warmed by 0.012 dB (corresponding to about 1 degree temperature increase.) Most of the measurements on this page were performed with a fan blowing on the attenuators. The effect of holding the attenuator nearest the LNA for 10 seconds is far more short-lived. See figure 3.


Figure 3. Holding my hand around the 3 dB attenuator nearest the LNA for 10 seconds gives a S/N loss of about 0.012 dB. With a fan blowing on the attenuator the excess temperature has disappeared in about 15 minutes.



It is obvious that several of the measurements on this page have lost accuracy because of attenuator warming. All the raw data is available so the reader can verify that errors of this type do not affect the general conclusions.

The attenuation results for 144 MHz are not of much interest so there is no reason to repeat this study with better accuracy. Knowing the problems will however be helpful for studies on higher frequencies where very small losses could affect S/N significantly.

Figure 4 shows the typical stability of the S/N measurement.

For further studies on higher bands I will place the attenuator nearest the LNA in an ice/water slurry to keep the temperature stable.


Figure 4. Stability. S/N values are accurate with 3 decimals on a dB scale with the parameters used for the measurements on this page.