I have added the Turn Island Systems “Product Satisfaction Policy” to our “About” page. This has always been our intent, so here it is in writing:
If you purchase a Turn Island Systems product and you are not happy with it for any reason (excluding obvious damage or abuse), return it to us and we will refund the purchase price less our shipping and sales tax costs. Even if the product works as specified, if it does not meet your needs you are welcome to return it for the refund.
“Any reason” means just that. For example, if it turns out that a preamplifier or filter doesn’t actually help your system performance (which can sometimes be tricky to determine), or you for any reason decide you don’t want it, feel free to return it.
I’ve been trying to avoid this one because winding and adjusting all those toroids is tedious and time-consuming, but this thing just makes so much sense that I had to do it. The Nine-Band Filter-Combiner (9BFC) is yet another filter-combiner for use with the WSPRSONDE and other QRP transmitters, and it allows a single multi-band antenna to be driven by multiple transmitters. It combines inputs on 80, 40, 30, 20, 17, 15, 12, 10, and 6 meter ham bands. You don’t have to use all the inputs; leaving any unused ones unconnected is just fine.
The 9BFC design is quite a bit more critical than the previous 6-band combiner, as the added 17 and 12 meter bands are extremely close to the 20, 15, and 10 meter bands. This tight spacing requires rather narrow filter passbands, and there is always a tradeoff between filter bandwidth, inductor “Q”, and filter loss. In order to keep the 9BFC loss to a reasonable level, commercially-available surface-mount inductors as used in the 6BFC are not adequate for these five closely-spaced bands. Instead, I am using iron-powder toroids (T50-6 variety) for those frequencies. The 80, 40, 30, and 6 meter filters do use surface-mount inductors.
Not only are these toroids tedious to wind, but the inductance of each of the ten toroids must be carefully adjusted to provide the proper filter shape and impedance match. Adjustment is done by spreading or compressing the windings while observing the filter response with a Vector Network Analyzer — this process requires some iteration since each channel has two toroids and they do interact. Once adjustments are complete the windings are secured with fingernail polish.
Of course, your antenna has to support the bands in use. When using the eight-output WSPRSONDE and the EFHW-8010 (end-fed half wave, 80-10 meters, from myantennas.com) I am using all the 9BFC inputs but the 6 meter port. It works quite well (my two WSPRSONDE locations are shown here, Friday Harbor WA, and Occidental CA). I have a separate 6-meter antenna for the Occidental site.
It’s alive!!! The new WSPRSONDE v2 WSPR/FST4W multichannel transmitters are rolling off the Turn Island Systems assembly line:
There are a few changes from the previous batch of WSPRSONDEs:
Smaller chassis, customized for Turn Island Systems.
Software control of output power: 1W or 250mW
There are a few other minor changes on the circuit board — things like Board Revision ID, allowing for future forward and backward compatibility.
But the WS-v2 is otherwise identical to the units that have been deployed on at least 2-1/2 continents. It continues to generate the clean, precise, and stable signals so necessary for accurate measurement of ionospheric propagation effects.
Along with this new hardware release comes updated WS firmware, with new features and options.
This has been in the works for a while now, and I am very pleased with the finished product!
If you are interested in some details of the WS-v2 assembly process, here is a short video of the Turn Island Systems CNC mill preparing a blank aluminum faceplate for the WS.
The forum has been quiet for some time, but it’s a good place to make comments, suggestions, critiques, and anything else even slightly related to Turn Island Systems.
And if you have one of our RX-888 Clock Adaptors (now available from TAPR), please share your experience with it — good or bad.
Update: Same board, brand new enclosure! Now available in limited quantities.
The SDR front-end filter-preamp is now available. This design takes the previous passive Shelf+Low-Pass filter design and adds a preamp at the output. This will improve the performance of most wideband SDR receivers.
I was recently asked about the filter-preamp, and wrote this in response:
The filter is useful when using a wide-band SDR such as the RX-888, since there is always the challenge of managing dynamic range in an SDR. The RX-888 has a 16-bit Analog to Digital Converter at the input (and some other SDRs use 12 or even 8 bits). A 16-bit ADC provides a dynamic range of about 96dB, and the signal-strength levels at the antenna can often exceed this range. A little bit of overload isn’t fatal, but too much will result is significant distortion and reduce the receiver performance. If you reduce the input gain to minimize the overloading signals then you will not be able to receive the weaker signals.
This dynamic range problem tends to have a frequency dimension as well. In the USA, we have the AM broadcast band (540 – 1700 KHz), and these are often the strongest signals seen at the receiver input. In addition, the regular atmospheric noise is stronger at lower frequencies, so receiver gain is usually not helpful at the lower end of the HF spectrum.
In Europe, the AM broadcast stations are spread across the entire HF spectrum, and mitigating the overload problems caused by these is a much tougher problem!
Note that traditional receivers, having band-filters and preselectors at the input are less prone to this overload, as they only have to work with a relatively narrow slice of the spectrum at any given time. The RX-888 is continuously receiving the full spectrum, from the KHz region up to 30 MHz (or 60 MHz). Also, with the wideband SDR we usually use a wideband (or multi-band) antenna, which doesn’t help the overload situation.
Also, the SDR, being a sampling receiver has an issue with “aliasing”, where signals higher in frequency than 1/2 the sample-clock frequency will be aliased down in frequency and appear as an interfering signal in the receiver range of interest (see “Nyquist Frequency” for details.) Most SDRs have an input filter that attenuates these frequencies above the Nyquist rate, but with the RX-888 we usually run the sample clock at about 66 MHz (half the maximum) The RS-888 internal filter is designed for the faster sample-clock, and so provides no attenuation for signals in the 30-60 MHz range.
So, what to do? The filter I provide has two sections: The low-frequency “shelf” filter, and a high-frequency anit-alias filter. The shelf filter provides a gradual and limited attenuation increase through the HF band (more attenuation at low frequencies), which compensates for atmospheric noise levels and provides significant attenuation of the AM broadcast band. The shelf filter has two identical sections — usually we want both of them enabled, but in a quiet-RF location we can disable one of them for less low-frequency attenuation.
The anti-alias filter provides a sharp cutoff above 30 MHz, with over 50 dB ultimate attenuation. This greatly reduces the aliasing problem.
These filters do add loss, and the RX-888 isn’t particularly sensitive as it has a high input noise figure. The low-noise / high dynamic-range preamp section of the filter-preamp compensates for the filter-loss, and adds about 9dB of additional gain which improves the RX-888 small-signal sensitivity. A few more dB gain would be nice, but that would be a different design. The filter-preamp I have now does provide a noticeable performance improvement in most cases.
Near Friday Harbor (home of Turn Island Systems) we have a RX-888 receiver in a fairly quiet location. Before the filter-preamp was added the AM broadcast stations were the dominant signals and we had to reduce receiver gain to avoid overload. The filter did a good job of equalizing (average) signal strengths across the full HF range.
If you are not familiar with TAPR, this brief introduction comes from their website:
TAPR is a non-profit 501(c)(3) organization of amateur radio (“ham”) operators who are interested in advancing the state of the radio art. The initials stand for “Tucson Amateur Packet Radio” but today the organization is much broader than that: we long ago became an international organization, and while we still support packet radio our areas of interest have expanded to include software defined radio, advanced digital modulation methods, and precise time and frequency measurement.
Turn Island Systems is proud to be associated with the good people of TAPR.
While there are different ways to distribute high-quality reference-clocks to multiple receivers and transmitters, or to general lab equipment, perhaps the best and easiest is with a Clock Distribution Buffer. The TIS-126 has been designed for this job:
This unit can send a square-wave clock to six output ports. Input and output are 50 Ohm impedance, and the frequency range is from 100 KHz to 100 MHz (and down to under 1 Hz in many cases). The input level can range from -20 dBm to +20 dBm.
This has been in use for a while, combining the multiple outputs of the Beacon Blaster and WSPRSONDE onto one SMA jack, allowing the use of a single antenna. The results have been excellent, so this box is now on the Turn Island Systems website, available for purchase:
Applications for the Filter-Combiner are not limited to the WSPRSONDE. Some have used this to combine the outputs of other QRP (1W or less) transmitters. One person is using the Filter-Combiner in reverse as a general preselector, sending the output of a single multiband antenna to several narrow-band SDR receivers (reducing overload from out-of-band signals).
What about eight bands? I glad you asked that question! Fitting additional bands into the combiner has proven to be a major challenge. A 160 meter port can’t be implemented using available (inexpensive) components — too much inductor loss. Squeezing the 17 and 12 meter bands into the gaps between 20, 15, and 10 meters also requires inductor tolerance and loss well beyond that of readily-available parts. However, if there is demand for it a future combiner could include a 6-meter port.
But single-channel filters can be provided for those “oddball” frequencies.
As you may know, the powerful RX888 SDR receiver is truly a game-changer — especially when used in conjunction with ka9q-radio and wsprdaemonsoftware. This combination provides full HF-band coverage and analysis — much more than I can go into here.
But to obtain the desired frequency accuracy and stability necessary for serious propagation analysis the RX888 does need an external 27 MHz reference clock. Fortunately, the receiver provides a small “U.FL” connector on the board, allowing easy interface to the clock circuitry. However, it’s not as simple as connecting the output of a GPSDO (typically a Bodnar, or the upcoming TAPR reference). The RX888 clock input does not provide a 50 Ohm termination, and requires an AC-coupled reference of around 1V P-P. What is needed at this interface is a DC-block, an attenuator, and a termination.
Rather than string together a bunch of little adaptors, we (Turn Island Systems and TAPR) have come up a single-board solution, as well as a replacement back-panel (so you don’t have to drill a hole in the existing panel for an SMA jack), and a short U.FL jumper cable.
That SMA plug on the right-hand side of the adaptor is not installed for the “Inside the RX888” configuration, but the footprint is there so this board can be used as an external adaptor.
As you can see, the circuit is quite simple. Here’s the schematic:
And it’s pretty easy to cobble together an interface board yourself. Here’s my first attempt, which has been in operation for about a month now. This one isn’t exactly the same (it’s missing one capacitor and the attenuator values are slightly different):
I usually do my prototyping with surface-mount components, but for old-times sake I decided to use my old stash of through-hole parts on this. I’m waiting for delivery of the assembled boards, but if all goes as planned these kits should be available by the end of April.
If you want to use off-the-shelf components to interface between a Bodnar GPSDO and the RX888, I recommend inserting a DC-block at the RX888 external clock jack, followed by a 10 dB attenuator (the links are to probably-OK parts, I haven’t tried these particular ones myself). The attenuator reduces the output of the Bodnar to a more appropriate level, and provides a “good enough” termination at the RX888. Since the board-end of the internal jumper cable is unterminated the 10dB attenuator only provides about 8 dB of attenuation, and this is fine. The attenuator also provides a compromise termination to both the cable from the GPSDO and to the internal jumper.
Note that since Glenn wrote this we have discovered that the coupling / DC-block capacitor is desirable. Also, Glenn has included a switch that allows you to enable/disable the RX888 internal clock. Many of us are not bothering with the switch, but just pulling the internal “enable” jumper clip.
