Clifton Laboratories 7236 Clifton Road  Clifton VA 20124 tel: (703) 830 0368 fax: (703) 830 0711

E-mail: Jack.Smith@cliftonlaboratories.com
 

In addition to viewing signals from a receiver, it's possible to use the Z90/91 to view transmitted signals. With a two-tone audio generator, it's simple to measure a single sideband transmitter's intermodulation distortion (IMD) performance. 

The block diagram shows my test arrangement. I made measurements with two different two-tone generators:

Kenwood SM-220: 1000 Hz & 1575 Hz
Telulex SG-100 DDS function generator: 600 Hz & 2100 Hz

In this mode, the SpectraScan is used as a spectrum analyzer, programmed for the test transmit frequency. It's critical, of course, that the signal input to the Z91 be limited to avoid damage; hence the termination with coupler described at the Loads page.

I've also added a new page showing how you might use the coupler from a Telepostinc LP100 wattmeter, along with a Z90/Z91 to measure the carrier suppression and IMD.

I've also added K2 measured performance data.

I won't go into the details of what IMD is, as VK5BR has a good overview at his web site. In one sentence, IMD is a non-linearity in a single sideband transmitter than can cause the transmitted signal to contain unwanted signals outside the normal speech frequencies that are seen as interference to nearby frequencies.

The traditional method of measuring IMD is to feed the transmitter's audio input with two clean audio frequencies, not harmonically related, and observe the transmitter's output signal with a spectrum analyzer. If the transmitter is perfect, only two RF signals are seen, corresponding to the two input tones, shifted to RF by the SSB modulation process. 

If the transmitter (and any linear amplifiers as well) is not perfect, new frequency components will be generated and will appear in the output signal, as illustrated at the right. The new distortion signals are mathematically  related to the input signal. The degree to which these unwanted signals are suppressed is a measure of the transmitter's IMD performance. 

Before looking at IMD in SSB mode, I first ran a test with a 1000 Hz modulating tone with the transmitter in AM mode. 

As the figure shows, the modulating waveform has noticeable 2nd and 3rd  harmonic content. The 2nd harmonic is down about 30 dB and the 3rd harmonic is down about 40 dB. This corresponds to about 3% harmonic distortion, a rather poor value for a high fidelity audio system, but reasonable for a communications transmitter, where fidelity is not a primary concern.

Here's the TS-940 operating at 100 watts PEP output, with two-tone input. The two tone frequencies are 600 Hz and 2100 Hz. 

Theory says that we should see an "in-band" intermodulation product at 1500 Hz, and multiple out-of-band products. The in-band product can just be seen at about -30 dB with respect to either single  tone. The strongest out-of-band IMD product is about -38 dB with respect to a single tone. (This is probably the 2 * 2100 Hz + 600 Hz product at 4900 Hz above the theoretical carrier frequency)

The ARRL compares the IMD to the transmitter's PEP, which is 6 dB above the single tone level. Hence, the 4900 Hz IMD product is -44 dB with respect to PEP. This is more than adequate performance.

I again started by testing the TS-870 with a 1 KHz tone in AM mode. The output is exceptionally clean, with the 2nd harmonic of the 1 KHz modulating signal about -35 dB with respect to the modulating signal, corresponding to about 1.7% THD.

Here's the SSB transmit test. with the same conditions as the TS-940, except the span is increased to 10 KHz and the signal level is increased to provide for better amplitude resolution.

Out-of-band IMD is excellent, 35 dB below one tone and it falls to the noise level much more quickly than seen for the TS-940. Note the type of out-of-band signal looks more like noise, instead of discrete tones as seen with the TS-940.

But, the in-band IMD is terrible--the 1500 Hz product (2100 Hz - 600 Hz) is only about -3 dB with respect to the input tones. 

Since the output stages are broadband, this in-band IMD must originate in a lower level stage, as if the problem were in the final or driver stages, the out-of-band IMD would be correspondingly excessive. 

I tried reducing audio levels, experimenting with the speech processing option, but I could not find any operating condition that made a significant difference.

I thought part of the trouble might be related to noise output from my SG-100 DDS function generator, so I tried the 1000 Hz / 1575 Hz two-tone generator output from my Kenwood SM-220 station monitor.

The out-of-band noise looks much better, and that confirms my suspicion that the SG-100 is  too noisy for low level audio generation.

However, the same in-band IMD problem is present, perhaps not quite as bad as the earlier test, but it's still there. In this case, the 575 Hz product (1575 Hz - 1000 Hz) is about 8 dB below the input tones.

Note that the strongest out-of-band IMD is 30 dB below one tone, and the 5th order IMD product is about -40 dB from one tone.

These figures are similar to those the ARRL measured in its February 1996 Technical Review of the TS-870, as it found 3rd order IMD at -32 dB with respect to PEP, or -26 dB with respect to a single tone. It found the 5th order product to be -47 dB with respect to PEP, or -41 dB with respect to one tone, which agrees well with my measurements.

There's something wrong here with the in-band IMD, and at the moment, it's not clear to me what it is. The ARRL measurement showed the  in-band IMD product about -35 dB with respect to either tone. It's possible the problem is related to overload in the TS-870's microphone preamplifier, although the TS-940 operated well with the same test signal input level.

I received my K2 today (29 August 2006) and measured its transmitted intermodulation distortion, using a Z91 prototype. The test configuration is similar to that shown at the top of the page, except I used a Bird 8323 100 watt, 30 dB coaxial attenuator instead of the dummy load and coupler. In order to bring the power attenuator's output into a range that would not overload the Z91, I added a 10 and 20 dB Minicircuits BNC attenuators between the 8323 and the Z91's input. I also set the Z91's internal attenuators for 30 dB, thus producing a total of 90 dB attenuation between the K2's output and the Z91's input amplifier. 100 watts output power thus corresponds to -40 dBm at the Z91's input amplifier, which is the optimum level for maximum dynamic range.

I used a Telulex SG-100 digital function generator in two-tone mode to create 700 Hz and 1900 Hz simultaneous test tones.