Tuesday, May 16, 2023

Notes On The XHData D-219 Analog DSP Radio

My D-219 came a couple of weeks ago and I've been checking it out. A fair amount of discussion has ensued about this radio, and I think feelings about it are generally positive. It is an analog DSP radio, using the Silicon labs Si4825 DSP chip. That to say, the tuning "mimics" the analog tuning of old school radios, but uses digital technology to do it. Amazon had it on sale for about $12, so I jumped at the chance to try it out.

The XHData D-219

The fit and finish on this radio is very nice for a $12 radio. The tuning wheel tunes butter-smooth with no noticeable backlash, better by far than many. The radio measures 3 inches tall, 1-1/8 thick, and 5-3/16 inch wide, just about 1/2 inch wider than the RadiWow R-108. The whip antenna is 21 inches long as opposed to the 19 inch whip on the RadiWow R-108. The radio takes two AA cell batteries and they last admirably, even with the speaker engaged. It presumably will use rechargeables as well. A 5 volt adapter plug is at the top of the radio next to the headphone jack. A stereo headphone jack is also at the top (only mono delivered), right next to the radio's sliding ON-OFF switch. On the right side are the tuning wheel and volume control. A crisp, well marked dial scale faces frontward and beneath it is the sliding band switch, similar to those found on other inexpensive analog radios. The band switch looks and feels more robust than others I've owned. A nice hand strap also come with the unit. An English manual is provided, describing minimally how to use the radio.

The radio tunes mediumwave, shortwave, and the FM band. Note that this radio is made predominintly for the Asian markets. It only tunes mediumwave from 522 to 1620 kHz, and that at only 9 kHz channel steps, used outside of the western hemisphere. Two FM ranges are present - 64 to 87 mHz, and 87 to 108 mHz. There are nine shortwave bands, covering broadcast segments from 4.75 mHz to 22.0 mHz.

The following comments refer to the mediumwave band. In particular, I'll discuss the tuning quirks of this radio. I have other analog DSP portables which tune similarly.

I'm impressed by the mediumwave sensitivity, it's very good. Examples: On weak daytime signals, it was better than my other DSP analogs, the Sangean SR-35 and the inexpensive ($13) Dreamsky pocket portable. It was better than old school analog Sony ICF-S10MKII. Against the old school Tecsun R-9700DX, a much larger radio with longer loopstick, it was no match of course. It wasn't quite up to par in sensitivity with the DSP all digital RadiWow R-108, but close.

Tuning is interesting to say the least, pitting a 9 kHz channel step against a 10 kHz band, and the already existing tuning wonkiness of these analog DSP chips. Daytime tuning may seem pretty normal. Nighttime is a different story. On moderate to strong stations, there may be three, five and sometimes even seven tuning peaks as you tune through a signal. This to me suggests a wide AFC (automatic frequency control) bandwidth in signal selection, possibly up to 27 kHz (3 channel steps) either side of the actual tuned center.

Note that these analog DSP radios don't use a traditional stepped tuning encoder. They are tuned with a simple 100K ohm (the SiLabs versions) or 10K ohm (most Chinese chip versions) potentiometer. The chip reads a voltage across the potentiometer and determines the tuned frequency from that. Software then decides what to do thereafter.

Checking the software manual for the Si4825 chip, one encounters the UNI-AM software switch on the DSP chip, named for "Universal AM". This is likely what is used to defeat the normal default AFC range of 1.1 kHz. I don't see the extreme 3/5/7 peak characteristic in the Sangean SR-35 or the Dreamsky. Each peak as you tune off signal center is reduced in audio by a couple of dB, mimicking the analog tuning of old. It is apparent that a signal's strength needs to be above a certain threshold to engage this multi-step tuning curve. Weak signals won't engage it. To-wit, a signal above this threshold is essentially "captured", forcing the tuning to its channel. This presents a problem in trying for weak stations between stronger channels.

See the example chart just below. I'll describe a typical experience in tuning this radio at night where many signals at varied strengths are present.

9kHz Offset Station we are trying for
---- ------ -----------------------------------------
612  2 kHz  610 WTEL Philadelphia, PA (medium strong)
621  1 kHz  620 WSYR Syracuse, NY (very weak)
630  0 kHz  630 CFCO Chatham, Ontario (medium strong)
639  1 kHz  640 WNNZ Westfield, MA (weak)
648  2 kHz  650 WSM Nashville, TN (very strong)

My listening post is in western NY near Rochester, about 75 airline miles west of Syracuse. With the D-219, catching WSYR-620 or WNNZ-640 in Massachusetts is hit and miss. This is because they are sandwiched on either side and between two much stronger signals. The stronger signals engage the 3/5/7 step tuning algorithm.

Example: In this 610-650 kHz range, WSM-650 was the strongest of all during this test. I centered WSM-650 for strongest audio, approaching from a higher frequency. This will occur at 648 kHz on this radio. Tune left one peak, and WSM's audio reduces by a couple of dB. Tune left again and WSM's audio reduces again a couple of dB. After two tuning steps, we should be tuned to 630 kHz, but the radio is still tuned to 650 kHz due to the AFC capture effect. Tune one more step to the left (621 kHz actual), and suddenly the radio tunes to CFCO-630. Why? AFC capture effect again. We have skipped over WNNZ-640 because it is extremely weak and below the capture threshold. The same for 620 WSYR, the station we should be tuned to - too weak. And since WSM at 650 kHz was the strongest signal, stronger than CFCO, when we were actually tuned to 630 kHz, the radio remained on WSM (648 kHz actual, with audio reduced). Important!!! - the tuning direction is important here - we are tuning downward and off the strongest signal, we are not approaching the strongest signal. More about this phenomonym explained below.

Turning the radio off, then on, sometimes has an interesting effect. In the above situation where we landed on CFCO-630 coming from 650 WSM, turning the radio off then back on may indeed land you on 620 kHz!!

In another case of downward tuning, where CFCO at 630 kHz was in a deep fade and virtually non-existent signals at 620 and 640 kHz, tuning progressed from WSM-650 directly to WTEL-610, bypassing 630 CFCO. Nothing was heard in between. It was the capture effect at work again.

Notice for the above scenarios we are tuning downward in frequency. Tuning upwards from 612 kHz will have a different result because we are approaching adjacent channels from a different direction and adjacent channel signal pair strengths will differ, with the strong signals capturing first if above the software-set threshold.

On old school, traditional analog radios it was fairly easy to figure out what frequency you are tuned to. We'd go to a known channel and count the channel "bumps" as you tuned up or down. It's not totally possible with the DSP analogs, at least the D-219. AFC capture may force-tune to a stronger channel one, two, or three steps away.

This is where a passive loop or helper antenna might benefit this radio, by increasing signal strengths of weaker channels so they trigger the AFC to capture. Care should be taken not to overload the radio with too much signal. My 18 inch passive, tuned loop is too much for this radio. Possibly a 12 inch passive loop might not overload. My testing a few years ago showed that a 12 inch loop was roughly equivalent to an 8 inch ferrite coil. It will provide plenty of signal.

Shortwave is a problem with this radio if you are in a high signal area near one or more mediumwave transmitters. Intermod and overload bleed through is severe throughout the shortwave spectrum. I am unable to test this radio on shortwave anywhere near the city of Rochester because of it. My next trip to the country I'll do that. I'm not an FM fan, so no FM testing was done.

All-in-all, I like this radio quite a bit. It has a great amount of what I like to call "fun factor". Mediumwave sensitivity is good and better than most. If XHData decides to make a 10 kHz step version for the North American market, I'd go for it. It's perhaps the best but least expensive radio I've bought.


I've studied the software manuals in pretty good detail for these analog chips and also the digital ones. Here's what I have gleaned from them.

Document AN610.pdf covers the American SiLabs 48xx analog chips. Asian clones of the DSP chip are slightly different in their operation. I hesitate to use the word "clones", as the Asian chips have a few more bells and whistles than the SiLabs ones. To our advantage, I might add.


Channel step size (9 or 10 kHz) is set by software at power on time.


"3. The channel spacing is configurable for the AM band mode only. System controller can select between 9 (9 kHz) and 10 (10 kHz) channel space. Note: SW is set to 5 (5 kHz) by default, FM to 100 kHz."

The document does not indicate that any other value can be set, other than 9 or 10 if in AM band mode.

The analog tuned DSP chips don't have a traditional stepped tuning encoder. They tune using a 100 K ohm linear potentiometer. A voltage is impressed across the pot and when you rotate the pot knob the DSP chip reads the voltage, and from the voltage, the controller software calculates a frequency to tune to.

If we are set to a 9 kHz step, tuning proceeds to the closest 9 kHz boundary. If we are set to a 10 kHz step, tuning proceeds to the closest 10 kHz boundary.

Note: The analog CCrane Radio EP Pro has a switch on the back to select 9 or 10 kHz step. It must force the radio to go through a power up sequence to accomplish this.

Filter bandwidth for AM looks to be fixed, i.e. not able to be set at power on, probably at about 6 kHz. Remember, we are talking about the analog DSP chip here, not the 473x digital version.

The digital series chips (the 473x models like the PL-380, etc.) are tuned totally differently. Tuning (internally at the software level) is in finer graduations and by default is to 1 kHz. The user usually has control of the step size in the AM band - 1, 9, or 10 kHz. They use a regular mechanical encoder, not a potentiometer.

ASIAN ANALOG CHIPS, commonly the KT0932m, KT0936m, KT0913:

Again, step size is set by software at power on time.

The document I have on these indicates the step size can be set at not only 9 kHz or 10 kHz, but also the 1 kHz step size. The tuning mechanism is a 10 K ohm potentiometer here. The KT0913 supports up and down channel buttons, the other two do not.

Notable in the Asian documentation is the raw sensitivity of these chips is claimed to be a bit better than the SiLabs chips, by about 6 dB. 16 µV @ 26 dB signal-to-noise ratio.

Filter bandwidth for AM has some flexibility. It can be set between 1 and 5 kHz at power up. I don't see a 6 kHz filter anywhere.

The Asian equivalent to the SiLabs Si473x series is the KT0935r.

Bottom line-

If a radio uses the Asian analog DSP chip, 1 kHz step size on AM is possible if the designer sets it at power up time. American SiLabs chips can only step at 9 or 10 kHz.

An interesting YouTube video on the XHData D-219 by Todderbert can be seen here:

Todderbert D-219 Review

There is a comment by a Anna P in this video link that explains that these chips have a wide AFC and can retune the frequency when the signal was not at the center by itself. I'll paste her full explanation here, as it's an interesting read. I'm not sure I'm in total 100% agreement, but there is merit to much of what is stated. Thanks to Jay Allen for the tip on this comment.

"Anna Plojharová - I don't think you need a dedicated "10kHz"  version. These analog tuned DSPs use rather wide AFC function which retunes the radio to the exact signal it sees. It first tunes to the exact frequency corresponding to the knob position and looks for carrier there. If not found, it gradually widens the frequency range till it finds one and then retunes properly on it (tunes the antenna). If this carrier is more than half of the tunning control step away (so there is other position of the dial), it artificially reduces the volume in order to get the "correct tuning" feeling, but that is 100% artificial behavior, the real reception is tuned exactly onto that carrier found. Usually the total search range for the given dial position is way wider than channel spacing, often 30..50kHz (so +/-3..5 channels). And the artificial "detune volume reduction" is often temporary - if there is really just that single carrier, after few seconds many chips just take that as "the desired station" and bring the volume back up."

The only issue could be, if two stations are the same frequency difference from the dial position, then it becomes tricky to convince the radio to select the weaker one, there the correct tuning step does help.

Todderbert's AM bandscan shows that the radio's tuning does not automatically retune to center peak of 10 kHz station boundaries. That implies to me that a 10 kHz spacing model would be preferable and work better here in the western hemisphere.

AFC width is selectable in software (the so called "UNI-AM" parameter), but only two choices - a default value of 1.1 kHz and what they call "Universal AM Band", which is an unspecified but wider AFC width. My suspicion is if "UNI_AM" is chosen, the radio may respond well to either 9 or 10 kHz spacing. HOWEVER, important to note here, the AN-610 software document says this is only available in the Si4827 chip models. The D-219 has the 4825A, per Kelly (thanks for checking Kelly).

On my CCrane Radio EP Pro, the one with the 9-10 kHz spacing switch, if I set to 10 kHz and tune to a station, then switch the slider over to 9 kHz, the tuning is off and does not recenter. This is dramatically shown on WYSL-1040 here, a very strong station. Tuned perfectly to 1040 kHz in 10 kHz spacing mode, then switching to 9 kHz where the new closest channel would be 1044 kHz, WYSL virtually disappears as the radio is now tuned 4 kHz away from 1040 kHz. Clearly, 9 kHz spacing does not recenter to a 10 kHz boundary. The wide AFC statement clearly does not apply to all analog DSP radios.

No comments: