TEF6687 vs TEF6686 vs Others

[WHJ]

[WHJ]

I recently completed my tests of the various FM tuners that I have available. My final tests were measuring how strong an adjacent frequency needed to be before reception was blocked. I measured how much stronger a 100% modulated adjacent frequency signal needed to be than before audio reception of a weak (0 dBuV) desired signal was prevented. That point was obvious. The SNR would decrease at first and then suddenly the received audio would be badly distorted to completely blocked. I added a new similar test of measuring the adjacent frequency signal level required to block RDS decoding.

In making the adjacent frequency tests you need to be sure that your test equipment does not limit your measurements. If I simply had a tee connecting my two RF generators together, the stronger one would create intermodulation distortion products in the other. Even the 20 – 30 dB isolation of the average combiner is inadequate for combining two signals differing in level by over 60 dB. So, special techniques were used to prevent this problem. The outputs from my HP 8643C generator and Skyworks Si4713B test transmitter were combined. The 91 dBuV output from my Si4713B was severely attenuated to provide the weak 0 dBuV desired signal. That meant that little of the other 60+ dBuV signal would be getting into it.



The TEF6686/7 specification sheet showed adjacent channel selectivity figures reference that adjacent station having 30% modulation. My tests were with adjacent stations with 100% modulation. Consider what that looks like on a spectrum analyzer:


Modulated signals are obviously much wider than an unmodulated carrier. I show both 30 and 100% modulated signals for comparison. That is why my selectivity figures with 100% modulation are lower than the TEF6686/7 spec sheet figure of 70 dB since they used 30% modulation. An important point is that both signals on 99.3 MHz are 60 dBuV! That is what the figure on the right would show if I had my spectrum analyzer set with a resolution bandwidth of 200 kHz as well as on a broadband RF wattmeter (such as my R&S NRP-Z11). Regardless of that bandwidth setting, my analyzer still sees that modulated signal as being 60 dBuV. If I set that for a range of -50 to +50 dBuV with the modulated signal it would still show that as being around 40 dBuV but it would be overloaded and create IMD with a 60 dBuV input signal. Add a band full of other 60 dBuV signals and the RSS sum of signals would be in the 65 – 70 dBuV range. Remember that when you are setting up your home receiver’s spectrum display. Probably for similar reasons, the RSPduo still overloads where the Overload light first goes out.

dBm has frequently been the spectrum analyzer measurement unit. Those figures for a received level of signals of -100 dBm seem counter-intuitive and confusing. The FCC uses dBuV/m (dBu) for field strength measurement. Because of that I prefer to use dBuV for receive voltage measurement in my receiver. Most of those figures are positive. 0 dBuV = -107 dBm.

The Sensitivity row shows that a typical TEF6686 will produce a 26 dB SNR audio output when its RF input level is -5 dBuV. 26 dB SNR was used since that is the standard used for most specification sheets. Due to the way that FM and its "capture effect" works, the SNR initially goes up more quickly than the radio's input signal level. So, a lower figure (more negative number) is better. Good RF sensitivity is not that important these days if the selectivity and dynamic range is inadequate. You won't receive a weak signal if your receiver overloads and causes that strong signal to splatter all over your dial. Most regular FM radios have detectors which initially increase the SNR by a significantly greater amount than the input signal level increases. I found this to be true on the TEF668X and the Skyworks. Both SDR radios had more of a linear relationship, with those two figures increasing at the same rate. The SDR radio manufacturers do not appear to want you to compare the specifications of their radios with regular radios. Basic spec sheet parameters, such as FM sensitivity and selectivity are missing.

As the data on the chart shows, the sensitivity of all tuners is similar. I am not sure that the sensitivity difference between the extremes in that will influence a reception. If it does, adding a good preamplifier should solve the problem. Eliminating noise and/or interference is generally not as easy. The TEF6686/7 definitely had better audio and RDS selectivity. The Skyworks tuners were particularly weak on that.

When RDS was first used its injection level was frequently 3 - 5%. Since that amount increases the total modulation by the same amount, the audio modulation level must be decreased. Stations trying to sound louder than the other guy don't like that. The FCC compromised. They said if you have a 4% RDS level you can increase your total to 102%. The same is true for SCA injection. More recently stations have found that 2% RDS injection is fine. Some stations effectively say, "I don't care about DX listeners," most in my prime coverage area decode a 1% RDS injection level fine. So, I documented the minimum dBuV signal required for multiple levels of RDS decoding. With weak signal reception, the SDR radios took a bit longer to decode RDS data while the Skyworks radios took still longer to decode them compared to the TEF668x radios.

The simple part of "Display with x dBuV In" shows how accurately the radio reports the receive level on its display. The displayed value should go up 10 dB for 10 dB input signal increase. Note how on the AirSpy HF+ with 10 dB increases past a 50 dBuV input, the readings are of 50.4, 54.4, 54.6 dBuV. Due to the obvious overload I stopped my measurements. Yes, with those high signal levels you can turn the gain down or activate the AGC, but then you won't be able to receive that -7 dBuV signal reported for its sensitivity on another frequency.

My new radio containing a TEF6687 tuner IC recently arrived. It looks like a typical TEF6686 radio.



It is shown just after it received CKSB-FM-2 Saskatoon via MS. I tore off the TEF6686 label on it and added my own TEF6687 label as a reminder.

[WHJ]

Continued from above...

I confirmed that it had a 6687 inside.

The TEF6687’s are labeled F8705 while the 6686’s are labeled F8602 and the TEF6686A as F8605.

I like having the keypad on the DP-666.

With the keypad, I don’t have much use for its touch-screen capabilities. Its performance appeared to match that of the other TEF6686 models. I’m glad that I had many of the silver radios with small buttons. If I only had one of them, my rating of it might depend on which I had. That might be the limiting factor in the measurements on all the other models tested. Hopefully I don’t have an extreme example of one of those. While not retesting it, I added old Sony XDR-F1HD data to the chart for reference.

I’d rate the TEF6686/7 radios as the best of those for FM DXing. My one TEF6687 did not have any advantage over the TEF6686 for DXing. Its spec sheet only mentions increased audio fidelity. The Skyworks radios did not meet my hopes and do not seem worthy of consideration. The SDR’s tested were inferior for weak signal reception on a typical crowded band. And, for all radios, if you compare the audio sensitivity dBuV figures with those for 3% RDS injection decoding, you will see that you can receive over 10 dB weaker signals based on an audio ID compared to relying on RDS decoding for the ID. We just need to have the future TEF6686’s recognize audio ID’s and slogans instead of relying on RDS data.