Friday, November 15, 2019

DSP Radio Chip Notes

A recent re-inspection of the Silicon Labs DSP radio chip datasheets left me with some new impressions of their quirks and capabilities. Additionally, a question was recently asked about observed differences in sensitivity between ultralight DSP portables. I thought I'd talk a bit about this and also some other general DSP chip notes related to chip radios.


Consumer chip-based DSP radios are really quite simple in their design. Many if not nearly all of the portable radios we DXers use are employing the Silicon Labs Si4734/35 or Si4831/35 chips. The Si47xx is first generation and the Si48xx is the second generation or newer chip. Their basic functionality is the same.

Circuit-wise, in most consumer-grade radios the chip is generally wired right to the input coil, that is, it is connected directly to the two wires of the ferrite loop. Simplicity! Largely gone are the days of superheterodyne local oscillators and ferrite loops with touchy and complicated multiple windings which are difficult to align. The entire radio is right there on the chip.

The core-sensitivity of every "radio on a chip" is identical. Manufacturing defects or minor chip-to-chip manufacturing differences should be near nil. Connected directly to the ferrite loop, the sensitivity differences in radios therefore are defined by the length and efficiency of the ferrite loopsticks and also their shielding and positioning away from noisy circuitry.


The dBµ signal strength display "19" on the front of this radio (RSSI reading, actually in dBµ/V) is reading the voltage off the ferrite loop as input to the chip. 19 dBµ represents 19 dB above one microvolt. Converted, it is 8.91 microvolts. So the dBµ signal strengths you are seeing are really a measure of antenna voltage and not differences in sensitivity of the chips, all other things being the same. Essentially you are seeing the sensitivity and efficiency of your ferrite loop, measured by the radio. How cool is that!

Another interesting fact to consider is the AM sensitivity of the 1st generation Si4734/35 chip and the 2nd Si4831/35 chip differ by 5 microvolts. Surprisingly the older Si4734/35 is the more sensitive - 25 µV vs. 30 µV (lower is better). But this difference (about 1.5 dB) will be hardly perceptible in reception. Still, a little improvement often makes the difference when signals are at the noise level.

The renowned CCRadio EP Pro uses the same 2nd generation Silicon Labs Si4831/35 chip we are discussing here. It achieves its nearly on-par sensitivity with the famed Panasonic RF-2200 because of its 200mm twin-coil, tuned ferrite loopstick. You can make a Tecsun PL-380 as sensitive as a Panasonic RF-2200 simply by removing the ferrite loopstick and substituting a 24 inch air core loop. A great trick except for the signal overload which cannot be handled adequately by the 380's DSP chip.

The second generation improvements in the Si4831/35 chips seem to be mostly added bells and whistles in control circuitry. However the sensitivity of the FM section of the newer Si48xx has been improved considerably, from 4 µV down to 2.2 µV (again, lower is better), amounting to nearly a 6 dB improvement.

It should be noticeable as it's almost a doubling of the signal sensitivity. Consequently if you are an FM Dxer, be on the lookout for newer DSP radios that use the second generation Si48xx chips.


Does your radio's sensitivity seem off? My PL-880 seemed so right from the start. An additional item to consider is channel centering. In other words, is the channel you are tuned to centered in the filter passband? One would assume that in a DSP-based radio the channel would always be centered. Not necessarily true. I bought a Tecsun PL-880 about a year ago. I was disappointed with it's sensitivity on mediumwave and found it was no better than my PL-380 or Sony SRF-59. I did the SSB centering tweak, noting that this was fairly close right out of the box. Still no improvement, but that effects SSB reception only. What to think? Did I get a bum unit?

Tecsun PL-880

The PL-880 has a wonderful fine tuning control available all the time - right under the main, or coarse, tuning control. It's a separate dial which allows tuning in finer 1 KHz steps. Now a year later, recently after tuning to a fairly weak mediumwave station, I happened to bump the fine tuning up 2 KHz and noticed something very unusual. The signal's audio got stronger and more bassy and the RSSI strength on the meter (dBµ) increased by +5 dB. The next day at mid-day I performed the same test on distant groundwave station KFI-640, Los Angeles, about 240 miles distant. KFI is one of my test stations when I test and compare radio sensitivities. I had only been able to get about 16 dbµ out of KFI at mid-day, with difficult reception just barely above noise level. Tuning up 2 KHz to 642, KFI now shows about 21-22 dBµ with clear audio and acceptable reception, well above the other two radios mentioned.

So check your channel centering in AM mode on these radios. Note that with some DSP radios this test might not be possible. Unfortunately what happens on some models is the RSSI reading will drop to zero since you are off-channel, making checking signal strength impossible. Check your radio to see.


I hear talk all the time about connecting up modern portables to longwire antennas. Some portables even have a separate antenna jack for shortwave and even mediumwave. My advice: be careful. And be even more careful if you live in a dry climate like the desert where humidities are very low.

Many years back, I blew out two Sony ICF-2010s with attached longwire antennas. Understand, the '2010 was notoriously sensitive to external antennas even though it was set up for it. The overload, whether it be signal or static pulse, would zap the input FET transistor. It was replaceable and you could even get one at Radio Shack.

My Kaito KA-1103 was very sensitive to static discharge, though technically the original is not a DSP chip radio. The PL-880 is turning out to just as sensitive to static. It's common here in the desert at certain times of the year when it's extremely dry to get a static discharge off a door knob or other metal object, or radio, grounded or not. Recent static discharges to the PL-880 are resetting it just by picking it up while I'm slightly charged. No harm so far. The '1103 got to the point where it corrupted the ROM and wouldn't work anymore. And guess what you have when you connect your little radio up to a longwire? A giant atmospheric static discharge receptor fed right to the radio's input circuitry. You don't have to be in a thunderstorm with lightning to have a static pulse.

Tecsun PL-380

So on to the Sangean ATS-909X. About a month into owning the '909X the AGC started acting up. Sensitivity was way down as observed on the field strength meter. Reception then became intermittent and within minutes would eventually  fail altogether. At first, resetting the radio - but only by removing the batteries - corrected the problem. The problem would reoccur and I went through the same routine each time. In the end it became permanent and resetting the radio had no effect. The radio was dead. What had I been doing the week before all this started? I had been messing with a longwire antenna connected to the antenna port. Coincidence? Maybe, maybe not. As I have witnessed in the previous paragraph, static pulse can not only zap the front end of the radio, it can corrupt the radio's ROM operating system. Resetting the radio does not always fix this. I sent the '909X in to Sangean for repair and they reprogrammed the ROM. The radio has been fine ever since.

I've done a lot of experimenting with directly-connected air core loops, particularly with the PL-380 and Eton Traveler III. It is common to overload or even static-zap these radios in this way because of your direct connection to the input pin on the SiLabs DSP chip. In these cases the radio has so far always recovered. The symptoms are de-sensing of the radio to zero signal for perhaps ten minutes or so. Shut the radio off and let it recover if it will.

Again, be careful. Modern transistorized or chipped portables will not stand up to longwire antennas and static pulse like the old tube radios of yesteryear did. Ground the longwire temporarily at first  before connecting to the radio.


Sangean ATS-909X
On rare occasion, a design engineer will use one of these chips in a different fashion and to greater advantage. Sangean was one of them. In what I consider one of the most highly underrated radios - the Sangean ATS-909X - engineers used the SiLabs DSP chip at the back end of the radio, in the I.F. section. This, to take advantage of the chip's superior DSP filtering in the I.F. stage, running the I.F. signal through the chip.

So, incorporating this into a traditional PLL-designed radio is a stroke of genius, combining the best of old front-end design with new technological filtering abilities at the I.F. stage. The result is a radio with premium selectivity.


I am guilty as most, I continue to buy this DSP stuff hoping for a more sensitive radio. But the truth of the matter is that they are all basically the same in their electronics when they use the same chip, unless of course they use some kind of additional input or bandpass filtering ahead of the DSP chip or pre-amplification of the signal off the ferrite loop or are engineered into a different design like the Sangean ATS-909X. Few do. Otherwise the difference lies only in the base output off the ferrite loopstick, its length, antenna factor (efficiency), or positioning in relation to surrounding circuitry (a source of birdies and display noise).

To wrap up, go for the radios which have the longest ferrite loopstick. Until newer 3rd generation DSP chips come out, hopefully with greater sensitivity and lower noise floors, the only difference is in your ferrite or external antenna.


Silicon Labs Products page:

From the Silicon Labs site. Modern DSP chip usage in commercial and consumer radios:

Si4730    FM/AM Receiver
Si4731    FM/AM Receiver with RDS
Si4734    FM/AM/SW/LW Receiver
Si4735    FM/AM/SW/LW Receiver
Si4736    FM/AM/Weather Band Receiver
Si4737    FM/AM/Weather Band Receiver with RDS
Si4738    FM/Weather Band Receiver
Si4739    FM/Weather Band Receiver with RDS
Si4820    Mono FM/AM Receiver
Si4824    Mono FM/AM/SW Receiver
Si4825    Advanced Mono FM/AM/SW Receiver
Si4831    Stereo FM/AM Receiver
Si4835    Stereo FM/AM/SW Receiver
Si4836    Advanced Stereo FM/AM/SW Receiver
Si4822    Mono FM/AM Receiver
Si4826    Mono FM/AM/SW Receiver
Si4827    Advanced Mono FM/AM/SW Receiver
Si4840    Stereo FM/AM Receiver
Si4844    Stereo FM/AM/SW Receiver


Troy R. said...

Excellent & insightful, thank you.

hamrad said...

OMG I can't believe I did not know about your website. It is awesome and I found the link this morning via a post on the UltraLightDX group post. I have a YouTube (hamrad88) and have about 11,000 subscribers of which I will definitely let them know about your website.


Tom Stilse (KE4GSK).

hamrad said...

I was so excited about finding this website, I spelled my name wrong. It is Tom Stiles and I just posted a video on YouTube about this great website.

Tom Stiles (kE4GSK)

Unknown said...

Hello.thank you for some great information. However i take exception with your comments on dsp chips acting alike perhaps but not exactlyon 31 radios.example xhdata d 808. Vsccrane ssb the 808 is far better then the ccrane etc.


Hi Unknown,

You're right, the sensitivity of the XHData D-808 is better than the CCrane SSB. This is due to the longer ferrite loopstick length in the 808. The 808's loopstick is 3-7/8 inches and the SSB's loopstick is 2-3/4 inches generating greater signal voltage. The radio's chips are the same 4735 chip (though I have seen some reports of the 808 having a Chinese knockoff of the SiLabs DSP chip).




Thanks for your enthusiasm! Glad you like the site. Hopefully I'll have energy and ideas to keep it going for a few more years!


Unknown said...

Hello, this is Thomas! I saw this website mentioned on a YouTube video by Tom Stiles. I love AM radio so this info is very interesting to me.

My newest DSP radios are a Tecsun PL 310et and the Eton Elite Traveler...the latest version of the Traveler lll mentioned on this website. These radios are only slightly better on AM and shortwave (reception wise) than my old non DSP (as far as I know?) Grundig G6 Aviator Buzz Aldrin Edition radio. For such a tiny receiver it`s amazing. It`s much smaller than the other two and has no muting while tuning (or soft mute) making scanning the entire shortwave band for active signals a breeze. I was amazed by its AM and shortwave sensitivity after I bought it for only $50.00 new at Radio Shack

It has a tuning wheel that lays flat against the front of the radio that you can quickly rotate with a finger. The Tecsun has both muting on tuning and horrible soft mute while listening to weak AM signals and irritating pumping of audio on strong stations. This can be defeated by tuning 1kz off frequency. It`s best to use headphones to do this because the volume drops considerably. The Eton has no soft mute while listening to weak signals which is a blessing. I use these radios as a team due to their various functions and different audio.

The muting while tuning using these DSP radios is a horrible curse. Why would they build a shortwave radio with such a terrible "feature"? Using the tuning knob to find active shortwave frequencies is a nightmare but they don`t even allow you to use the search features outside the broadcast bands! Why??? So I use the Grundig often to quickly find active frequencies outside the broadcast bands and then listen using the Tecsun since the bandwidth can be adjusted for a better sound. If the station is very clear I`ll use the Eton since it has a nice sound on strong signals. The Eton has the best sound on AM though with much less noise than the DSP radios. I linked to this website on my new radio blog. Feel free to join and post anything you like:

Unknown said...

This is Thomas....I meant to say the GRUNDIG has the best sound on AM...not the Eton.

Unknown said...

This is Thomas. You can easily eliminate DSP birdies on AM by using a passive loop like the Tecsun AN-200. It even completely clears them up on my digital but non DSP Grundig G6. The G6 has a terrible distorted audio hum on frequency 750 and on 850 and the loop clears it up perfectly. The Eton Elite Traveler also has several birdies on the AM band that are cured using the loop. Anybody who is serious about AM listening should have one. Before I eliminated the sources of static interference in my home I also used the loop to null it out on my favorite stations. I`m so glad I have one!