Test of the Large Signal Behaviour of some 144 MHz Radios

DF9IC & DARC OV Durlach A35 - 27. 2. + 19. 3. + 23. 7. 2005 in Pforzheim / Germany

Disclaimer: this web page expresses the personal opinion of the author and is not authorized by any organization. The reader is encouraged to make his own mind based upon the information presented here. All measurement results have been carefully evaluated but stem from a single test session on a single sample of the radio.

The purpose of this test was to gather some information which radios are best suited for 144 MHz operation in large signal environments like VHF contests. Typical signal levels from other (high power large antenna) stations in such a situation are:

-50 dBm (90 dB ref. to noise in SSB BW): a moderately strong station, maybe up to 100 km away near LOS, or up to 30 km away behind a hill
-30 dBm (110 dB ref. to noise in SSB BW): a really strong station, maybe up to 30 km away near LOS
-10 dBm (130 dB ref. to noise in SSB BW): an extremely strong station, e. g. 3 km away LOS

These figures assume that the antennas point to each other which may be true if the interfering stations uses a multi-antenna system and is located in your main direction. Otherwise signals are typically 20 dB weaker when one of the antennas is pointing completely off the other station. These are real world levels - the author measured at his home site in the Nov. 2004 Marconi contest signals from one station with up to -10 dBm and from three stations with up to -30 dBm, using a calibrated HP8558B spectrum analyzer directly connected to the antenna.

If you plan to use a radio for serious VHF contest operation both RX and TX should allow nearly interference-free operation with levels up to 110 dB ref. to noise in SSB BW. The test shows that there are radios on the market which reach this figure in the RX 50 kHz off the carrier, in the the TX 200 kHz off. This may be just acceptable while at least the TX should be improved. Nevertheless it will result in a strong interference if another station is very closely nearby, and your antennas point to each other. For an interference-free operation in all situations at least 130 dB ref. to noise in SSB BW must be handled - but there is not and was never any radio on the market with such a performance.

The IP3 is less important than LO and TX noise performance as there are usually only few strong and very strong signals on 144 MHz so that only few frequencies are corrupted by the resulting intermodulation products. This situatuion is very different from the situation on the lower HF bands where many stations including broadcast transmitters are present. On the other hand the necessary dynamic range (difference between the smallest and the largest signal) is bigger on 144 MHz - therefore the need for the best LO noise performance. Radios that are well suited for 160 m CW DX with good close-in large signal behaviour are not necessarily performing well with a transverter on 144 MHz. This page is on 144 MHz useability only.

In the following tables some measurements are summarized which we did in early 2005. It is not more than a first step, and gives some information which radios must be excluded for serious operation - in fact none are left when you are very serious :-).

There are quite some other critical points left like ALC operation, keyclicks, and discrete spurii which must be considered also but could not be evaluated systematically because of the lack of time. Especially ALC action can cause severe distortion in the transmitter by additional transmitter noise and transient intermodulation at high levels so even radios with good nominal performance in the tables below would need some modification in this point before they reach a similar performance when modulated with voice or CW.

The measurement procedures and the test equipment is decribed at the end of this page.


144 MHz Allmode Radios:

TRX
Owner
NF
IP3

RX Blocking in USB mode
(3 dB S/N reduction in typ. 2,5 kHz BW)
dB

TX sideband noise level in 2,5 kHz BW
(spurii not included)
dBc
dB
dBm
20 kHz offset
50 kHz offset
200 kHz offset
20 kHz offset
50 kHz offset
200 kHz offset
IC275E DF9IC
5.6
-7.5
98
110
117
-97
-104
-109
IC7000 DD9WG
5.6
-7.5
91
94
102
-87
-93
-93
IC706 - measured by DL2KCK DL2KCK - - - - - -91 -95 -103
IC746 DJ0QZ / DK1VC 3.6 -7.5 87 95 109 -82 -91 -105
IC821H DK9VZ
3,4
-9
80
90
101
-77
-88
-97
IC910H DK9IP
3.7
-8.5
81
89
100
-78
-88
-98
IC202 DL3IAS
7.7
-14
100
104
107
-100
-102
-102
Hohentwiel DL3IAS
11.4 (?)
-5.5
96
97
100
-96
-97
-101
FT225RD + MuTeK + mods DK9VZ
6.1
+7
90
90
110
-85
-92
-106
FT817 DK2DB
5.4
-12
87
96
106
-83
-91
-96
FT857D DK9VZ
6.1
-2
86
96
106
-84
-93
-99
TS700G mod. with GaAsFET DK8SG
4.9
-13
100
108
111
-102
-106
-107
TS700S (preamp off) DB6IR
6.6
-7
100
107
111
-96
-102
-104
TS790E DJ5IR
4.5
-14.5
103
104
109
-84
-94
-95
DK2DB homemade 1976 DK2DB
-
-11
109
110
112
-103
-107
-110

Comment:

The table shows that there has been a substantial decrease in TX performance in the past decade(s). The oldest radios (TS700 - mid 70s design, and IC275E - mid 80s design) are 10...20 dB better than newer or currently available transceivers (TS790, IC821H, IC910H).

The IC821H and IC910H have a very poor LO design. They are well suited for FM repeater operation connected to an indoor HB9CV antenna. Other use should be prohibited. FT817, FT857, IC7000 and IC746 are only slightly better. The IC706 has been measured by DL2KCK. It shows an increase of the absolute noise power when the output power is decreased only slightly, so at 20 W output the relative TX sideband noise in 200 kHz offset is a poor -93 dB instead of -103 dB at 40 W.

The TS790E can be used as RX in a large signal environment like a contest but please do not transmit with this radio.

IC202 and Hohentwiel (home construction kit from Germany) have low phase noise close to the carrier but do not improve for larger spacing, resulting in mediocre overall performance. The Hohentwiel is still some dB worse than the IC202.

Other radios currently produced have not yet been tested. Published test results from e.g. ARRL indicate that their performance may be in the same range with the IC910. The RX blocking in 20 kHz offset can be approxinately derived from the ARRL BDR which is defined differently, by subtracting 34 dB (for BW conversion from 1 Hz to 2.5 kHz) and adding 6 dB (for the correction from 1 dB noise increase to 3 dB noise increase) - in total subtracting 28 dB from the ARRL BDR value. This conversion should be correct as long as noise increase is the limiting factor which is supposed to be true.

Data for some of these radios taken from DK9VZ's web page is listed here (compare this with RX blocking in USB mode at 20 kHz offset in above table):

- IC7400: 86 dB
- IC910: 78 dB
(within 3 dB what we measured)
- IC706MKIIG: 83 dB
- TS2000: 87 dB
- FT817: 80 dB
- FT847: 75 dB
- FT857: 74 dB

None of these radios is useable for serious VHF operation. If you own any "modern" 144 MHz radio you are invited to join me for a measurment to gather more data about it (write an e-mail to <call sign>@adacom.org).

I use an IC275E and know why, though being aware of its limitations. It is good enough at least for 23 cm transverter operation :-)

 

HF Allmode Radios with transverter:

TRX
IF
MHz
Owner
NF
IP3

RX Blocking in USB mode
(3 dB S/N reduction in typ. 2,5 kHz BW)
dB

TX sideband noise level in 2,5 kHz BW
(spurii not included)
dBc
dB
dBm
20 kHz offset
50 kHz offset
200 kHz offset
20 kHz offset
50 kHz offset
200 kHz offset
Elecraft K2 + XV144
preamp in the TRX "On"
28 DJ5IR + DJ5IR
6.0
-26.5
95
100
101
-93
-92
-93
Elecraft K2 + Kuhne TR144H+40 14 DJ0QZ + DJ0QZ
1.0
-9
96
103
114
-90
-95
-96
Orion main RX + Javorrnik
Orion sub RX + Javorrnik
14 DK9IP + DK8SG
-

0
-7

-
-
-
-93
-
-88
-
-99
-
TS850 (preamp off) + LT2S
TS850 (preamp on) + LT2S
28 DL6WT + DL6WT
-
3.7
-1,5
-26,5
101
100
104
102
107
104
-93
-100
-103
TS870 (preamp off) + LT2S 28 DK8SG + DK8SG
4.9
-6
98
104
112
-95
-100
-104
TS870 (preamp off) + Javornik 14 DK8SG + DK8SG
1.9
-1.5
95
103
112
-92
-97
-99
IC735 (preamp off) + LT2S 28 DF9IC + DK8SG
-
-
101
106
113
-
-
-
IC735 (preamp off) + Javorrnik 14 DF9IC + DK8SG
-
-3.5
106
115
117
-
-
-
IC746 (preamp off) + Kuhne TR144H+40 14 DJ0QZ + DJ0QZ
1.2
-5.5
99
106
119
-
-
-
FT1000 Mark V main RX (preamp off/on) + Kuhne TR144H 28 DK9VZ + DK9VZ
6.0
1.7
-9.5
-12.5
97
106
118
-91
-99
-101
FT1000 Mark V main RX (preamp offf) + Kuhne TR144H 14 DK9VZ + DK9VZ
1.6
-7.5
103
110
116
-95
-97
-100
FT1000 Mark V main RX (preamp off/on) + Kuhne TR144H+40 14 DK9VZ + DJ0QZ
1.3
2 (?)
-3
-10.9
104
113
120
-
-
-
FT1000MP main RX (preamp off) + LT2S 28 DK9IP + DK8SG
-
-
97
104
113
-
-
-
FT1000MP main RX (preamp off) + Javornik 14 DK9IP + DK8SG
DK8SG + DK8SG
1.4
0.9
+1
+1
100
104
115
113
118
120
-98
-98
-106
-105
-110
-110
FT1000MP main RX (preamp off) + Kuhne TR144H+40 14 DK8SG + DJ0QZ
1.4
-1
-

-

-
-
-
-
FT1000MP sub RX (preamp off) + Javornik 14 DK9IP + DK8SG
DK8SG + DK8SG
2.0
1.2
-4.5
-5
88
88
95
97
109
111
-
-
-
IC7800 + Kuhne TR144H40
measured by DL2KCK
28 DL2KCK
-
-
-
-
-
-98
-102
-108
LT2S has about 17dB gain, 1 dB NF, -6 dBm IIP3 and uses an IF of 28 MHz. Javornik has about 27 dB gain, 1 dB NF, +3 dBm IIP3 and uses an IF of 14 MHz. Kuhne TR144H+40 has about 27 dB gain, <1 dB NF, +9 dBm IP3 with 14 MHz IF.

Comment:

The possible IP performance of a transverter / HF radio system is inferior to that of a 144 MHz transceiver with a crystal filter on the first IF because two frequency conversions are needed until the first narrow filter blocks off-channel signals. When you compare the test results you will find nevertheless that the IP of the transverter / HF radio combos is usually better that that of the 144 MHz radios. This shows the bad design of the VHF radios - using the same quality of the preamp and the mixer as they are used now in mid-class HF radios an IP of +5 dBm could be obtained with a single conversion 144 MHz receiver of 3...5 dB NF.

The LO performance of a HF radio with upconversion to a high IF should be slightly better than that of a 144 MHz LO because its frequency is a bit lower. In practice this difference is quite large which again shows the bad design of most 144 MHz LOs.

The LT2S transverter has an IF of 28 MHz like most 2 m transverters. The Javornik uses a 14 MHz IF because it was optimized for operation with a FT1000, and this radio is substantially better on 20 m than on 10 m. There is no special reason for a such a differenc in performance of the HF radios between 14 and 28 MHz operation as long as the radios use upconversion to a high IF. But in fact some radios perform better on 14 MHz, others on 28 MHz. Thus the transverter should be selected accordingly. Kuhne offers versions for both 14 and 28 MHz.

All transverters have crystal LOs which are so much better than the LOs of the HF radios that they should never contribute substancially to the total noise. Nor should the trasnsverter contribute to the TX noise when driven with the correct level (do not use optional TX IF preamps) but we observed some problems with wideband TX noise on one of the tested Kuhne transverters. This issue has been corrected on samples of the current production. The Javornik transverter is very well matched to the FT1000 in gain and IF band and has a better IP that the LT2S. Nevertheless the performance of any combo is determined mainly by the HF radio.

The TenTec Orion's TX noise was quite bad so that we stopped further tests. Maybe there was a defect in our sample radio or unsufficient internal filtering of the 12 V line power coming in the test from an external switched mode PSU (because the ORION has no internal PSU). But even TenTec's published graph for the LO sideband noise (RX) shows a very moderate performance never reaching more than 107dB @ 2.5 kHz (= 141 dBc/Hz) - this is 10 dB worse than a FT1000MP in 200 kHz offset from the carrier. It stems from the wideband noise floor of the prescaler used as a part of its LO system. Our sample also showed unstable behaviour and had to be rebooted once during the test because the PLL seemed to be unlocked contimously without indication (firmware bug ?).

The Elecraft K2 also has a low IF design using conventional VCOs which should result in a good LO noise supression but does not do so well. We had two radios on the bench, one from 1999 (DJ5IR) and one from 2004 (DJ0QZ). The last one was tested on 14 MHz and showed better results than the older one on 28 MHz. You may compare the ARRL test results of the LO noise that Elecraft publishes on their own website and which is closely within our blocking test result (our measured -95 dB RX blocking in 20 kHz offset is equivalent to -129 dBc/Hz LO noise). The high level of TX noise shows that there seem to be design flaws choosing too low signal levels internally. The AGC threshold is ridiculously high (subjective impression). I also do not understand why it uses low quality ladder crystal filters instead of a filter from monolithic duals like any other radio does. Overall it was the worst HF radio in the test (OK, a 144 MHz IC910H is still worse...).

The IC7800 + TR144H40 has been measured by Christiatn DL2KCK on a different setup with a HP8640 as reference oscillator. It is not completely clear how good the noise performance of the used oscillator is but the real value of the TX noise should be at least not worse than the measurements indicate.

All combinations are worse in the TX noise than in the RX noise which means that the level in some TX stages is too low so that additional wideband TX noise adds to the LO phase noise. This seems to be a general design flaw of nearly all tested radios. Additional tests show that during SSB modulation and CW keying there is often a lot of AGC action which may increase the TX noise level further by more than 6 dB.

The FT1000MP / Javornik is the best available combination. Nevertheless they are still far off what could be realized. We measured two different samples of the FT1000MP which performed very close to each other. The Mark V seems to be also quite close to the MP.


Thanks to Bernhard, DB6IR, Frank, DJ0QZ, Martin, DJ5IR, Horst, DK1VC, Ewald, DK2DB, Helmut, DK8SG, Winfried, DK9IP, Wolfgang, DK9VZ, Nino, DL3IAS and Jürgen, DL6WT, for their kind support.


References:

SM5BSZ has done many similar measurements and reported about them both in magazines (DUBUS, UKW-Berichte) and in the web. Please note that his TX noise and RX blocking values are normalized to 1 Hz and look therefore 34 dB better.

Take a look also into his file list where you can find many interesting topics.

A description of the Javornik transverter is available on S53WWs website.

The author has made a presentation on this topic at the 29th GHz meeting in Dorsten 2006 with additional material. The proceedings are available by mail order.


How do we measure?

We use the following tools:

HP 8642B as low noise low spur signal generator,
HP ESG-D 4432B as universal signal generator,
R&S FSP30 spectrum analyzer for tests,
HP 34401A as AF RMS voltmeter,
W&G SPM-12 level meter 200 Hz - 4.6 MHz,

13.5 dB ENR diode noise source calibrated against HP346A,
low noise overtone crystal oscillator (XO) 145.2 MHz with 20 dBm power amp,
2 isolators for 144 MHz,
resistive combiner with 2 x -40 dB to a common output,
+17dBm mixer with diplexer and AF amplifier with built-in highpass filter (20 kHz -1 dB, 4 kHz -50 dB),
several attenuators etc.

Some of the parts have been specially built for these measurements. The best standard test instruments like spectrum analyzers are too bad for such challenging narrowband tests.

The instrument readings are directly entered into an Excel sheet which calculates the performance figures, to avoid mistakes and miscalculations.

'DUT' = Device under Test.

 

Noise Figure (NF)

The DUTs RX input is connected to the noise source, the DUTs AF output to the HP34401A which is set to 2 s average time. The DUTs AGC is deactivated either by switching it off or by decreasing the RX gain. The noise source is switched on (+28 V) and off, and the AF level ratio ('Y factor') is measured.

The DUTs NF is calculated from the Y factor using the knwon source ENR and the well known formula NF = (ENR-1)/(Y-1).

 

Intercept point of third order intermodulation distortion (IP3)

The IP3 was measured at 50 kHz signal spacing. Usually there are no strong signals very closely to each other, and if, you should avoid them anyway for LO noise. So close-in IP is of little interest.

We use the XO and the HP8642B as generators and combine them through isolators and the resistive combiner. A signal of 2 x -20 dBm with more than 100 dB IM rejection results. We try to measure at 2 x -40 dBm RX input to the DUT but change this level between 2 x -50 dBm and 2 x -37 dBm according to the DUTs IP. The DUTs AGC is disabled again.

The IM signal is measured by replacing it through a signal from a 2nd signal generator inserted through a calibrated directional coupler. The level of this signal is adjusted to give the same voltage at the AF output at the same frequency. The AF output is measured with the analog SPM-12 in wideband mode.

 

RX blocking

The DUTs AF output is connected to the SPM-12 in wideband mode and AGC is disabled. The DUTs input is connected to the HP8642B through a 10 dB attenuator (w/o the 8642B wideband noise is less attenuated - seems the electronic attenuator is a bit noisy). The signal of the 2nd generator is inserted through the same directional coupler as in the IP3 measurement.

The SPM-12 measures S+N. First the signal of the 2nd generator (E4432B) is adjusted such that 10 dB (S+N)/N is measured. The DUTs frequency is adjusted to maximize SNR. Then the HP8642B is switched on and its level is increased until the (S+N)/N level falls to 7.55 dB. Then the S/N power ratio has decreased from 9 to 4.5 (by 3 dB). For this procedure the SNR must be measured multiple times switching the E4432B on and off. The 10 dB SNR level of the E4432B and the blocking level of the 8642B are recorded and used for evaluation.

This measurement is repeated three times with 20, 50 and 200 kHz offset of the HP8642B.

The method ensures that both noise increase and gain compression are recognized. Usually noise increase is dominating. For the measurement of the FT1000MP / Javornik combo the XO with an external step attenuator was used instead of the HP8642B which reaches its own noise limit at 200 kHz offset.

 

TX sideband noise attenuation

The DUT is operated in constant carrier mode (CW with key activated). The TX output is attenuated to about +3 dBm and fed into the mixer whose LO port is connected to the overtone crystal oscillator. The filtered mixer AF output is fed into the SPM-12 used in the low distortion selective mode. First DUT and crystal oscillator are offset by 20 kHz and the resulting beat signal is adjusted in the SPM-12 to give a reading of +10 dB. Then the DUT is tuned to zero-beat the XO and the resulting double sideband AF spectrum is analyzed with the SPM-12.

 

Limits of the test equipment

The precision of the NF measurement is around +-0.5 dB, IP3 around +-2 dB, Blocking and Noise around +-3 dB.

The HP8642B noise has been measured to -114 dBc at 20 kHz, -116 dBc at 50 kHz and -116 dBc at 200 kHz offset with the XO as reference. These values are for 2.5 kHz BW and thus compare directly with the values in the tables above; increase the numbers by 34 dB to get normalized dBc/Hz values.

The reference XO has been evaluated against the HP8642B with an extra 144 MHz full size cavity filter (DK8SG homemade) and is better than -115 dBc at 20 kHz (supposed to be considerably lower), -126 dBc at 50 kHz and -128 dBc at 200 kHz offset. You can expect that the LOs of the transverters are of similar quality.

Further tests

In the most recent test session we also used a downconverter with the crystal LO and mixer with no image rejection down to an IF of 315...515 kHz and a notch filter built from ceramic resonators at 415 kHz. With the modulated carrier centered to the notch the output can be measured by the FSP spectrum analyzer to see ALC and modulation effects in a bandwidth of +-100 kHz. With MAX HOLD transients can be found. Results are not yet published.

 


Bernhard DB6IR (back), Helmut DK8SG (front) and Henning DF9IC (right)

(photo courtesy of DK9IP)

 

 


RX Blocking measurement on the Elecraft K2 with transverter

(photo courtesy of DK9IP)

 

 


Some of the radios waiting to be measured

(photo courtesy of DK9IP)

 

 


Ewald DK2DB with his homemade 144 MHz transceiver dating back to 1976 but still going strong

(photo courtesy of DK9IP)

 

 


How to switch the 2nd RX to the transverter input, and outputting its AF to the left channel of the headphone front jack, while disableing its AGC? - Winfried DK9IP in his endless battle against the internals of TenTec's Orion.

(photo courtesy of DK9IP)