Setup to measure transmitter spectral purityThese measurements were made with Linrad running on a PentiumIV 2.7GHz computer with a modified Delta44 soundcard. The RF signals were converted to audio by use of the WSE converters RX2500, RX10700, RX70 and RX144 or RXHFA. The RX144 and RXHFA units were prototypes, the others were regular units. |
|
|
Me (SM5BSZ, Leif) in front of the test equipment.
The WSE converters are side by side to get adequate cooling
except for the RXHFA which is on top of the RX144 unit.
These two are not powered simultaneously.
The three generators are for IM3 measurements using the
crystal notch filter on which I hold my finger.
The filter is not quite good enough to use for IP3
measurements on the IC7800 as you can see below.
Image by Len, SM4LLP
Spectral purity of a continuous carrier.The unit under test was connected to a high power attenuator followed by an ordinary step attenuator which was connected to the RXHFA or RX144 unit. The step attenuator was adjusted to place the signal approximately 1.5 dB below A/D saturation.A screen dump from the Linrad display in tx test mode was saved for each unit showing the spectrum centered 25kHz from the carrier. To see these spectra, click on the links below: Crystal oscillator 14 MHz (reference) DSW40 7 MHz DX70TH 14 MHz DX77 14 MHz FT1000MPMV 14 MHz IC718 14 MHz IC756PROII 14 MHz IC7800 14 MHz MFJ9020 14 MHz SW-30+ 10 MHz TS50 14 MHz TS711E 144 MHz The spectral purity of a continuous carrier is listed in table 1. In this table spurious signals are not included. For spurs, look at the screen dumps. Note that the mirror image that is produced because RX2500 is a direct conversion receiver is located symmetrically with respect to the spectrum midpoint. This receiver/Linrad spur is suppressed by 70 dB in these measurements except for the crystal oscillator which was measured several weeks earlier, immediately after Linrad was calibrated. |
Model Band Noise floor in -dBc/Hz
(MHz) 5kHz 10kHz 15kHz 20kHz 50kHz
DSW40 7 130.5 130.5 130.5 130.5 130.5
DX70TH 14 105.1 113.8 118.0 120.5 127.7
DX77 14 101.4 112.1 117.4 120.5 128.9
FT1000MPMV,200W 14 114.8 123.7 126.8 128.4 130.0
FT1000MPMV,20W,AB 14 112.3 114.8 115.0 115.0 115.5
FT1000MPMV,20W,A 14 112.1 114.2 114.4 114.2 114.2
IC718 14 111.7 118.6 122.2 124.3 130.6
IC756PROII 14 117.4 125.4 129.8 131.7 136.3
IC7800 14 120.9 131.9 136.1 137.8 142.4
MFJ9020 14 127.3 133.1 135.3 136.6 138.7
SW-30+ 10 134.6 136.1 136.8 137.0 137.0
TS50 14 109.6 114.2 115.2 115.9 116.7
TS711E 144 114.0 121.1 124.3 126.0 130.8
Table 1. Noise floor at different frequency separations
from a carrier.
|
Spectral purity of voice SSB transmissions.All except one of the tested transceivers produce broad SSB signals with a lot of splatter in surrounding channels when operated in "normal" mode with the ALC active. The TS50 is an exception, it does not show signs of an "overactive" ALC.One or several screen dumps from the Linrad display in tx test mode were saved for each unit showing the spectrum centered 25 kHz from the SSB signal. To see these spectra, click on the links below: Unfortunately two spectra were taken with too high signal level which caused the A/D converter to saturate at some point in time. I do not think this has affected the result, but those spectra are not reliable. A/D saturation does cause spurious signals and although the saturated spectra do not look like they have contributions from saturation it can not be excluded. (I will add a warning message in Linrad to make sure saturation is not easily overlooked so I will not repeat this mistake.) The margin to saturation is shown in the lower left corner of each screen dump and it should not be zero. DX70TH 14 MHz DX77 14 MHz FT1000MPMV 14 MHz FT817 14 MHz IC718 14 MHz IC756PROII 14 MHz IC7800 14 MHz TS50 14 MHz TS711E 144 MHz Two transceivers, IC718 and IC7800 allow manual transmit gain reduction. These rigs give clean signals when the ALC is made inactive but they do not give acceptable interference levels when the ALC is acting. Table 2 shows the peak splatter level in 2.4 kHz bandwidth in dB below the peak power in 2.4 kHz bandwidth. Since the mechanism of the interference is too much gain in the ALC loops with an associated oscillatory behaviour of the ALC action, the interference sometimes has equidistant maxima. To give a true representation of the interference, table 2 shows the highest interference level at or above each frequency separation. |
Splatter level below PEP at
Model Band 5kHz 10kHz 15kHz 20kHz 30kHz 40kHz 50kHz
(MHz) (dB) (dB) (dB) (dB) (dB) (dB) (dB)
DX70TH 14 15 15 32 32 51 55 68
DX77 14 31 50 51 51 51 59 68
FT1000MPMV 14 33 35 35 46 46 59 66
FT817 (uncertain) 14 11 11 11 11 11 25 25
IC718 alc,proc.off 14 41 52 56 58 61 62 63
IC718 no alc,proc.on 14 49 58 69 75 84 84 85
IC7800 alc, proc.off 14 38 46 54 61 71 81 88
IC7800 no alc,proc.on 14 43 68 88 88 89 89 89
TS50 14 50 67 77 82 85 85 85
TS711 proc.off 144 16 25 32 42 50 61 65
TS711 proc.on 144 17 27 32 42 53 58 67
Table 2. Peak splatter level in dB below peak power at
or above different frequency separations from a SSB voice signal.
For details, look at the spectra.
|
Spectral purity of keyed CW transmissions.An over-active ALC causes keying clicks with a characteristic spectral pattern. The DX70TH is a typical example. The IC7800 shows the same phenomenon at a lower level, but when the IC7800 is operated without ALC, the clicking sidebands disappear completely as can be seen in the link below. All transceivers were keyed by hand in a normal fashion. For some of them the spectrum of a single key-down and perhaps a single key-up was recorded. Other rigs were tested with a 20 Hz automatic keyer. The length of the space from the end of the previous mark may influence the voltage swing on the ALC line, something that depends on the time constants in the design. This is why it is important to test the units with normal hand keying with variable speed and word spacings. Testing with only a 20 Hz keyer that gives 50 % duty may not tell the whole story.DSW40 7 MHz DX70TH 14 MHz DX77 14 MHz FT817 14 MHz FT1000MPMV (modified?) 14 MHz IC718 14 MHz IC756PROII 14 MHz IC7800 14 MHz MFJ9020 14 MHz SW-30+ 10 MHz TS50 14 MHz TS711E 144 MHz Keying clicks are seen in these spectra as a separation between the peak power (green) and average power (red) above 12 dB. As can be seen in these links, the DSW40 is unacceptable because the key-down is abrupt. Probably the antenna switch is forming the envelope. The key-up is beautiful. The DX70TH and the FT817 are unacceptable because of ALC generated keying clicks. The MFJ9020 produces unacceptable clicks both at key-down and key-up for unknown reasons. Some of the transceivers were also tested in a narrow frequency range. These spectra are linked to below. FT1000MPMV (modified?)14 MHz narrow IC718 14 MHz narrow IC756PROII 14 MHz narrow SW-30+ 10 MHz narrow TS50 14 MHz narrow TS711E 144 MHz narrow The level of the keying clicks is listed in table 3. The data is from the wideband spectra and shows the peak power of the keying click in dB below the carrier. |
Model Band Keying clicks. -dBc in 2.4 kHz.
(MHz) 5kHz 10kHz 15kHz 20kHz 30kHz 40kHz 50kHz
DSW40 7 25 33 37 42 49 54 59
DX70TH 14 28 28 47 47 69 77 80
DX77 14 33 57 62 62 66 76 81
FT817 ser.1K45401 14 14 14 14 14 14 23 23
FT817 ser.4K81.. 14 23 23 23 23 23 33 33
FT1000MPMV (mod?) 14 N67 N78 N83 N86 N88 N89 N89
IC718 14 N63 N75 N78 N81 N84 N86 N86
IC756PROII 14 53 69 78 86 N90 N92 N92
IC7800, ALC on 14 44 58 74 85 N94 N94 N94
IC7800, ALC off 14 N75 N87 N91 N93 N94 N94 N94
MFJ9020 14 23 31 31 34 34 38 42
SW-30+ 10 46 66 71 75 75 75 80
TS50 14 N67 N71 N72 N72 N73 N73 N74
TS711E 144 53 58 61 65 68 81 81
Reasonable >60 >67 >70 >74 >76 >78 >80
Table 3. Peak power in 2.4 kHz bandwidth in dB below the carrier
for keyed transmitters.
The table gives the worst level at or above the listed frequency.
Numbers preceeded by N is peak power values that are consistent
with the average power level which is an indication that keying
clicks are not present.
For details, see the spectrum graphs.
|
|
Table 3 is discouraging.
Only IC718 and TS50 have proper keying but the TS50 suffers from strong
sideband noise as can be seen in table 1.
The keying clicks from the IC756PROII are not very strong and they are
probably generated by the ALC loop.
When the IC7800 is run without ALC, the keying is excellent,
but amateurs probably do not use it like this and with the ALC enabled
the keying is not good.
The line "Reasonable" in table 3 represents what I think are reasonable numbers for modern transceivers. These numbers do not allow for much keying clicks, they represent what modern rigs with a sideband noise level of -120 dBc/Hz at 20 kHz would perform like if they incorporate reasonable keying. Table 4 shows sideband noise levels at 20 kHz from those transceivers that I have measured and that are not modified for lower sideband noise. The data is from this page, from the Scandinavian VHF/UHF meeting in Gavelstad Norway June 2003 and the RS-03 meeting August 21-24, 2003. More data is here: FT1000D and here:IC706MKIIG at 14 MHz |
Model Band Noise
(MHz) (dBc/Hz)
IC7800 14 -137.8
SW-30+ 10 -137.0
MFJ9020 14 -136.6
IC756PROII 14 -131.7
DSW40 7 -130.5
TS850S 28 -129.2 (Incl. conv. 144 -> 28)
FT1000MPMV 14 -128.4
FT1000D 14 -127.9
TS711E 144 -126.0
TS-450S 14 -125.6
IC718 14 -124.3
TS-2000 14 -123.1
TM255E 144 -122.3
IC706MKIIG 14 -122.2
IC970H 144 -121.6
DX70TH 14 -120.5
DX77 14 -120.5
FT100 144 -119.0
FT817 144 -117.2
IC706MKIIg 144 -117.2
TS50 14 -115.9
FT847 144 -115.0
IC821H 144 -113.1
IC706 144 -108.3
Table 4. Sideband noise at 20 kHz frequency separation from
continuous carrier.
|
Receiver dynamic range of the IC7800Receiver measurements take more time than transmitter measurements. At the SSA 2004 meeting I only looked at the receiver of the IC7800.The noise figure of the IC7800 was measured as 18.4 dB with the preamplifier off. The power level of an off channel signal required to reduce the S/N of a weak signal by 3 dB is listed in table 5. I did not see any sign of compression so the receiver is limited by sideband noise in the range of frequency separations tested. The LO sideband noise levels corresponding to the observed power levels are also listed in table 5. The weak signal was placed at 14.1599 MHz, the center of my notch filter. |
F dF P noise (MHz) (kHz) (dBm) (dBc/Hz) 14.1499 10 -20.7 -134.9 14.1449 15 -15.5 -140.1 14.1399 20 -12.9 -142.7 14.1349 25 -10.7 -144.9 14.1099 50 -5.0 -150.6 14.0599 100 +1.2 -156.8 14.0099 150 +3.5 -159.1 13.6599 500 +6.5 -162.1Table 5. Dynamic range of the IC7800 receiver. P is the level of a strong signal that reduces S/N of a weak signal by 3 dB. The corresponding LO sideband noise level is listed in the last column. This table originally contained data points down to 1 kHz frequency separation. The data was inconsistent with the LO sideband noise as observed in the transmitter test at 1 and 2 kHz separation, therefore all the points below 10 kHz have been removed since the error may have been that I have forgotten to take the effect of the notch filter on the strong signal into account. The notch filter is flat with 1 dB attenuation above 10 kHz separation and the values of this table are probably too good by 1 dB. The inaccuracy of the signal generator adds an uncertainty of perhaps 2 dB because it enters both the power measurement and the noise figure measurement. The true value at 10 kHz should be in the range -131.9 to -135.9 dBc/Hz based on these error estimates. All the data points have the same error limits -1 to +3 dB. |
|
The transmit side is (as usual) not quite as good as the receiver.
The wideband amplifier noise of the transmitter sets the limit at
about -142 dBc/Hz.
LO phase noise is the dominating tx noise only at frequency offsets
below 20 kHz or so.
When trying to measure IP3 at 20kHz for the IC7800 I get the result listed in table 6. The measurement is done with a 14 MHz crystal notch filter which is not quite good enough as can be seen from the data. The data in table 6 is consistent with the published value of +40 dBm. |
Level of Level of Level assuming 2 tones IM3 IP3=+39 dBm (dBm) (dBm) (dBm) -1 -81.2 -81 -4 -90.8 -90 -7 -97.4 -99 -10 -103.2 -108 -13 -103.2 -117 -16 -108.2 -126 -19 -111.6 -135 -22 -112.0 -144 -25 -118.6 -153 -28 -118.6 -162Table 6. Attempt to measure the third order intermodulation of the IC7800. It is obvious from this table that the crystal filter used is not good enough. The data is consistent with the published IP3 value of +40 dBm. |
| To SM 5 BSZ Main Page |