Nice full duplex build for under $200 (FCC cert components etc)

kk6qms · November 25, 2022, 11:33pm

A really nice build provided by NR9V-

https://forums.qrz.com/index.php?threads/how-to-build-a-professional-grade-full-duplex-allstar-node-for-under-200.839842/

"After a bit of R&D I have managed to build a high-quality Full-Duplex AllStarLink Node using readily available FCC-certified components, no mods/hacks or soldering required, that can be built by anyone with basic computer skills (though if your cousin / neighbor / ham buddy happens to be a Linux wiz that could be good too). AllStar has been around a long time and nodes have been built in some interesting ways but I’ve seen no previous mentions online of anyone having been able to do all the above with professional-grade audio and RF quality, power supply, microPC, audio interface, and node radio(s), all for under $200. Short 1-minute demo video:

For more info or any questions post a comment below or email me on my QRZ address. Also see this thread for notes on an earlier node I built using an RPi4 and a Yaesu FT-530 HT for the node radio. To summarize my findings:

There’s no reason to settle for low-quality half-duplex commercial nodes with noisy SA818 RF modules, cheap CM1xxB audio chips, and closed-source HamVOIP software.
There’s no need to use overpriced/hard-to-find Raspberry Pis for a node computer. Dell 3040 and other Thin Client / Micro PCs are widely available for $50 or less on ebay, and have the same low power consumption as an RPi4 (5V 3A) but with more features and built-in eMMC/SSD memory.
AllStarLink.org has excellent documentation and support making it very easy to build your own node on Intel/AMD or RPi platforms. ASL is 100% open-source with a huge ham community including many experts who have used it for years with complex interlinked repeater systems.
There’s no longer any good reason to use non-FCC-certified Baofeng radios or hack them apart to access a COS line. ASL’s usbradio driver provides reliable COS (RX Audio) detection using the carrierfrom=vox mode. (Note that ASL has a hardcoded 2000mS voxhangtime but I put in a pull request to make that configurable, and I set to 500mS in my usbradio.conf files.)
The Retevis RT85 are great radios available for only ~$50 for a pair on amazon with free shipping. They are FCC-certified - ensuring less chance of causing or receiving RFI/EMI, and are known as “Baofeng-killers” due to their excellent performance and super-low price. They are also sold under other names/variations such as the QRZ-1 and TYT TH-UV88.
Full-duplex nodes enable more interactive and more efficient communications, prevent doubles and echoing, with telco-grade VOIP communication quality built on the Asterisk open-source PBX platform and Debian Linux.
Other radios you use to talk to the node can be half-duplex but it’s better if they support cross-band full-duplex such as a Yaesu FT-470, FT-530, Kenwood TH-72A or a number of other models. Full-duplex radios are also great for satellite communications. An online search for “cross band full duplex HTs” will turn up several good links.
AllStar works really well with Full-Duplex radios because it does not echo your transmit audio back into your receive audio. Thus you only hear what’s on the remote node/repeater you’re connected to (minus your own outgoing audio), so it works just like talking on a speakerphone - true full-duplex comms with no echoes, delays, or feedback.
You can also use full-duplex smartphone VOIP/IAX apps with the node or even a VOIP phone.
AllStar nodes also support EchoLink with a simple cfg setting. You can then access the EchoLink network from your HT or mobile radio rather than having to use the phone app.
The DVSwitch distribution of ASL supports bridging to digital modes such as P25, D-STAR, YSF, DMR, etc.
Using HTs for node radios provides high audio quality while having low power consumption and being simple, compact, inexpensive, and putting out only as much power as you need to cover your QTH and within a couple miles. (Some nodes use mobile radios but those use much more power and usually won’t go below 5 Watts.)
The Masters Communications DRA-30 provides excellent audio quality at a reasonable price and is available in various configurations (kit / assembled, different case options).
MicroPCs such as the Dell 3040 have a power button so you can start or cleanly shut down a node simply by pressing the power button, and they have a Real-Time Clock (RTC) IC which enables a node to power itself on and off on a schedule (configurable in BIOS or in a cron entry), which can further reduce power consumption and eliminate the cost of a separate AC timer. Power to the radios can also be gated by the 5V line on one of the USB Ports.

Installing ASL on a 3040 is not hard if you’re experienced in building/maintaining PCs. The 3040’s have a UEFI-only BIOS that needs the boot file in a specific place. To make a UEFI USB drive just format it as FAT32 and copy the files over. Conveniently the Intel/AMD ASL install image is based on Debian Linux which happens to have the boot file where the 3040 expects it. Once the install finishes to the 3040 internal flash, it won’t boot, but with a UEFI clonezilla USB or other Linux tools you can boot the local OS and then copy the /boot/efi/EFI/debian/shimx64.efi file to /boot/efi/EFI/BOOT/BOOTX64.EFI and it then boots fine. After fixing that the rest of the AllStar setup process is the same as on an RPi.

The Intel/AMD ASL image is different from the RPi image in that the latter just goes on an SD Card, into the RPi, and the install is done. The former however is an installer image, intended to boot on a removable device and install Debian and ASL on an internal device (or possibly on a separate removable USB/SDCard). The Debian installer has about 10 different options it will ask, so if you’re not an IT expert you might want to refer to some online forums to make sure you install to the MicroPC’s internal flash drive, overwriting all existing partitions.

Most HTs need a well-filtered DC power supply, as they’re designed to run primarily on battery power and probably don’t have much if any filtering or regulation. HTs will run much cooler on 7-8V and should not be run any higher than that for high duty cycle node TX. It’s easy to find inexpensive linear DC power supplies or inexpensive (~$5) step-down linear regulator circuits on ebay that have a variable voltage regulator IC, heat sink, filter capacitors, and terminal block connections. A larger filter cap eg. 10,000uF can further reduce any power supply noise if needed. I get power to the RT85’s from the power supply using Retevis/TYT “battery eliminator” adapters, which are available for ~$12-15 ea. online, but to save a few $ you could just solder power wires onto the 2 contacts on the back of the radio. You could also run the HTs on their batteries but for a node that will be transmitting a lot it’s probably better to not constantly charge and discharge them, or have the battery die during a longer QSO or net. If you already have a 12V battery system in your shack, that could provide plenty of clean power, but HTs can overheat if run at high duty cycle on 12+V (even if on the low power setting).

Making a cable to go from a DB9 to 2x 2.5mm plugs and 1x 3.5mm plug can be a bit of a hassle if you want it to look pretty, or, just pick up an old DB9 cable, cut it in half and splice on some aux cables. That’s a lot easier than soldering onto DB9 and mini-phone plugs. With a small terminal block there’s no need to solder anything.

Wiring - DRA-30 to Kenwood, Retevis, TYT, QRZ, etc. HTs:
DB9 Pin 1 — TX HT 3.5mm Ring (Tx Audio)
DB9 Pin 3 — TX HT 3.5mm Sleeve (PTT)
DB9 Pin 6 — TX HT 2.5mm Sleeve (Ground)
DB9 Pin 5 — RX HT 2.5mm Tip (Rx Audio)
DB9 Pin 9 — RX HT 2.5mm Sleeve (Ground)

Lastly you’ll need some ferrite cores on at least some of the cables. Definitely one on the USB cable, and one on the audio cable going to the RX HT. The Laird 28A2029-0A2 (Digi-Key Part Number 240-2077-ND) work well and have enough room to allow cables to be wrapped 2 or 3 times. They are only ~$2 ea. in Qty’s of 10, and these often come in handy for other things.

Once your node is up and running you could mount things in some sort of enclosure, maybe a rack-mount shelf, or a clear plastic carry case. There are many ways that could be done. For a home node you don’t really need a case, just set everything on a shelf in the closet and you’re good to go, and can control the node from any web browser using the included Allmon or Supermon web apps.

Be sure to set up your node RX radio with Tone Squelch enabled, to prevent possible QRM from causing your node to TX. I would not use CTCSS on the TX radio though, so you’ll be able to hear other stations if someone happens to be on the same frequency. Also check your local 2m and 70cm band plans (eg. in Southern California TASMA.org and SCRRBA.org) to find simplex frequencies specifically intended for analog FM hotspots so you won’t possibly interfere with repeaters or calling frequencies.

By setting up the MicroPC BIOS to turn On by default after power is applied, and because the RT-85’s have a regular analog volume knob with on/off switch, everything can be placed inside a carrying case with no access needed to the PC or radios once they have been configured. You can then have a single power switch mounted in the case and either a 12V PowerPole or IEC AC inlet. One switch then easily turns everything on and off. If you opt to run it off 12V you’d need a couple step down regulators to drop the 12V to 5V for the PC and ~7-8V for the radios, or if you run AC into the enclosure you can use the power adapter that came with the PC and a 12V 1-2Amp variable linear power supply for the radios.

Feel free to comment or email me with any questions. I can also build custom nodes for reasonable prices, and will be listing some full-duplex nodes on various sites - look for “full-duplex AllStar node” and be sure to only buy products made and supported by reputable ham(s) using high-quality FCC-certified components. 73, NR9V"

NR9V · November 26, 2022, 8:26am

Thanks for the post Clifford. BTW link to the quick demo video:

I’ve posted this article on a few forums/sites and it’s getting lots of responses, pages of comments and 350+ views on youtube just in the first 2 days. Having worked with PBX systems in the early 90s and in satcomm & telecomm ever since I’m very familiar with what Asterisk can do, and I think a lot of people had been overlooking some of its more subtle features. The usbradio dsp works perfectly for RX CD and for filtering, etc, Asterisk & ASL’s full-duplex implementation is flawless, the Dell 3040s work even better than an RPi and can be had for as little as $30 on ebay, and these $25 RT85s sound just as good as any of my other HTs. I had no idea when building this node that all these things would come together so well but I do believe this is a first in a node that costs very little but works very well. Everyone has different ideas of what a node should look like but I think this one can work well for a lot of use cases and hopefully bring more attention to the benefits of full-duplex comms.

Kc3kpc · November 26, 2022, 11:25am

Is this rig on the air? What is the node number?

NR9V · November 26, 2022, 3:37pm

Kc3kpc, It’s my personal node which is usually on in the evenings and at night. Feel free to message me on my QRZ email if you wanted to connect to it some time and we can schedule a demo. 73, NR9V

Kc3kpc · November 26, 2022, 10:48pm

Will do. Thank you for the video. I just switched to ASL, I was having issues with the other option. My next step is better hardware. Raspberry PI works but think I can find something a little better.

My current setup:

Raspberry PI 4
DRA-40
BCM-440

Everything works but don’t trust the SD card.

NR9V · November 29, 2022, 8:29am

After posting this on several other sites and FB groups there were pages of comments and questions and lots of interest. Having worked with PBX systems in the early 90’s and in the Satcom, Telecom, and Pro-Audio industries for decades I know what Asterisk (the open-source phone switch system that AllStar is built on) can do. I think a lot of people had been overlooking some of its more subtle features. To summarize my key findings:

ASL’s usbradio DSP is excellent and works extremely well for providing reliable Rx Carrier Detect and thereby easily supports the use of HTs for node radios ie. no COS line is needed.
ASL and Asterisk’s full-duplex implementation works flawlessly and is a great feature that’s very easy to enable.
The icing on the cake was these $25 Retevis RT85 HTs that sound just as good as any of my other HTs and have very good RF performance - even in full-duplex scenarios where the node Rx HT can receive a 500mW signal from a mile away very well on 70cm even though the Tx HT is transmitting 1 Watt from less than 18" away.

I plan to put this How To on its own web page and in a pdf with as much detail as possible so wanted to assemble the various comments and feedback received so far with additional notes:

Re. FCC compliance, RF performance, and Chinese radios:

All components have FCC logos indicating that at least some level of certification was done. As an example a 2020 December QST review and bench test of the Radioddity GA-510 (a similar HT) found that it was fully compliant with FCC specs. I have not verified the RF specs on the RT85 with precision test equipment but have had no issues such as intermod or degraded receive sensitivity. With this node one RT85 transmits on ~147 MHz and the other receives on ~431MHz and because it’s Full-Duplex, it is always transmitting whenever it’s receiving, thus if the TX spurious emissions were bad or the receive performance not good the RX HT would likely be overloaded and desensed. The RT85 will receive a 500mW signal very well from a mile away on 70cm when the other RT85 is transmitting 1 Watt on 2m at < 18" away, which says a lot about the RF performance.

This is indeed the most important RF performance criteria of a full-duplex node, and whatever radio(s) you decide to use with a node, the first test you should do after your node has been set up is a sensitivity test where you transmit to it with low power from a mile or so away, and using the Parrot test mode confirm that your audio is clearly received. Also make sure your 70cm frequency is not a multiple of your 2m frequency. ie. if the node transmits on 145.0 MHz and receives on 435.0 MHz, it’s not going to work. Your 70cm frequency should be at least a couple MHz away from the 2rd harmonic of the 2m frequency. If you want the absolute best receive sensitivity on your node, the Tx and Rx antennas should be as far apart as possible, but in my case I don’t plan to use the node from more than about a mile away and it’s working very well for me even with the antennas < 18" apart.

Because no modifications have been done to the radios, PC, or audio interface they are indeed FCC Part 15 compliant. If they were integrated into an enclosure with a different antenna setup then the system might technically not be considered FCC-certified but this is a simple node and there’s no need for changes of that sort. And no one would expect a homebrew node using unmodified Part-15 certified components to go through a whole new certification process. That would make about as much sense as saying that if you add a dipole to an HF rig or plug it into a PC USB port that the rig is no longer FCC-certified.

As for generalizations about Chinese radios, up until a couple weeks ago I would have generally agreed with the idea that “all Chinese radios are junk”, however after seeing reputable companies now adopting them I did some research and found that Chinese radios have come a long way in the recent past. If it’s good enough for Yaesu, Gigaparts and QRZ then it’s worth some investigation, which has paid off because I now have some node radios that are performing very well and that cost next to nothing by the time you factor in recent rates of inflation. I can do full-duplex on an FT-530 or TH-D72 from miles away from my QTH and the audio quality is excellent, indistinguishable from local analog FM repeaters.

An additional note on snap-on toroid filters: One is definitely needed with my RPi node on the USB cable going to the audio interface, so I had put one on the Dell 3040 just assuming it would be needed. But after someone commented on this I did a test and it turns out that no toroid is needed between the Dell 3040 and the audio interface. This is good news as it confirms that the FCC-certified Dell does in fact put out much less EMI/RFI than my RPi4 (which is in a high quality Vilros metal case). I also took off the toroid from the 3040’s network and power cables and that also had no detrimental effect. Rx and Tx Audio are still very clean and clear. Thus it would appear that using a Dell 3040 PC (which has about 15 different certification logos on the bottom) is another benefit of the approach I’m suggesting and this saves an additional $2 off the cost vs. using an RPi. The only toroid that is in fact needed is one on the RX audio cable - I would get some occasional false carrier detects without that.

Everyone’s RFI/EMI environment can be different however. As with with any product or project, you may run into issues and need to do some debugging. You might need to add some extra toroid cores on certain cables, or additional filtering/regulation on your DC power supplies, or move some things around on your desk or in your shack in order to get proper good clean clear audio in and out of your node. Any time you have a receiver that’s picking up and amplifying signals in the < 1 millionth of a Volt range, it’s easy for interference to occur if you have a noisy switch-mode power supply nearby, or you’re too close to a cell tower, power lines, or various other possible EMI sources.

Re. the term “professional-grade”, and audio performance:

The use and meanings of the term professional are definitely context-dependent. If this was an academic paper I’d be more precise, but since we’re “amateurs” ie. we don’t get paid to write articles or forum posts, it just makes for interesting discussion.

In the context of ASL where audio streams are typically filtered to a ~ 300 - 3K Hz bandwidth, audio I/O through speaker/mic connections is not going to be any different generally than it would be on a professional commercial radio after the latter’s signal has been properly filtered and encoded for ASL. The codecs used will generally be the limiting factor. In my experience, audio from a good quality HT will sound very good over an ASL-linked repeater and will generally not be distinguishable from someone going direct into the repeater on Analog RF.

Many if not most Analog FM Repeaters are indeed professional-grade systems, built and maintained by professional EE’s with decades of experience in Audio, RF, etc. The key distinction in this context is does a node uphold that level of quality throughout the entire signal chain? In my experience the small/cheap/SA818-based nodes often do not, whereas nodes using real FCC-certified radios do. Thus if a node does not detract from the audio performance of the overall system relative to any other node, it upholds the performance of that professional-grade system. This node indeed has as good or better audio quality than any other node I’ve heard. Not to say it’s perfect, but I don’t hear any significant hums, buzzes, clunks, hiss, distortion, etc.

I would at some point like to precisely measure the specs of various radio options and fully characterize their frequency response, phase response, SNR, mic input sensitivity and dynamic range, etc., however there’s been no need to do that so far because the audio quality and performance of the node is meeting all my expectations, and ASL and FM repeaters in general only have a bandwidth of ~300Hz to maybe 5KHz at the most thus we’re not aiming for same level of performance as pro audio/recording equipment ie. 24-bit 48KHz+ digital with 120+dB SNR. ASL generally has an 8KHz sample rate and is then significantly compressed, which sounds very good for voice. If a repeater were set up with an unusually wide audio bandwidth and analog FM radios used with the widest possible Tx and receive bandwidths, you could theoretically get HD audio quality through a repeater, and thus better audio quality than most VOIP codecs, but I’ve never seen a radio that has much if any options for this on V/UHF FM or heard of a repeater that tried to do that. Typically FM transceivers have very sharp filters on the Rx and Tx audio and few if any configuration options to adjust those. SSB rigs have much more flexibility in these areas.

ASL does support many codecs and could potentially do HD audio if both nodes enabled the appropriate codecs. It may not be easy for ASL to go higher than 8KHz though, that seems to be hardcoded from what I’ve seen.

BTW although this node is intended more for home than portable use, ideally with wired-internet for maximum performance, USB WiFi adapters are available for as little as $2 and it would be very easy to use wifi, and all components are small enough that the node could easily fit in a lunch box, small briefcase, Go-box or Go-bag.

It’s not an industrial-grade off-the-shelf solution with perfectly calibrated specs, but I’m confident that if we had a QSO on an AllStar-linked repeater you wouldn’t be able to tell much if any audio quality difference between this node vs. if I went in on RF on my IC-9700 or TS-2000. (I’d be happy to do a test like this with anyone in the Coastal SoCal area. My QTH is on top of a mesa with a view of pretty much all of coastal SoCal, and I get into pretty much any repeater from Santa Barbara to Tijuana pretty well on RF.)

In the Pro Audio industry the term “Professional” simply means better than consumer-grade. Pro Audio systems do what they do very well and are intended to operate reliably for many years, unlike consumer devices that constantly need firmware updates or become obsolete after a few years. Whereas the term Amateur in amateur radio simply means that we do it for fun, not for profit. Which in no way implies that we are not capable of fully professional level work, or even profiting from it if someone comes up with a product that fills a need in the market. The term amateur is the butt of many jokes but in reality the people doing the work behind the scenes for ASL, various repeater systems, ‘amateur’ equipment manufacturers, etc., often are true innovators in engineering, science & technology.

Re. the SA818 RF module: It was pointed out that there are some nodes that have properly used this module ie. with proper Rx and Tx filtering, etc., thus providing good audio quality. However, even with that being the case, it does not support Full-Duplex, does not have the flexibility of a real radio, may only be supported under HamVOIP, and is known to have other issues, for example using CTCSS tones below ~127 Hz results in loud harmonics that can intrude into the audio passband.

Re. the use of HT speaker/mic jacks:

Someone thought that maybe the filtering done by HTs for the speaker output might conflict with what AllStar expects. Fortunately, the ASL developers thought of this and the driver configurations cover a wide range of use cases. This uses the ASL usbradio driver, which has a configuration setting to specify if the Rx Audio line is speaker audio vs. raw discriminator audio. The HT does the CTCSS decode and tone squelch, and usbradio does not do any CTCSS encode/decode or DSP squelch detection. Usbradio does bandpass filtering of both the Rx and Tx audio, and does Rx carrier detection.

There are some settings in usbradio.conf and in rpt.conf that are important (see below). Usbradio’s Rx and Tx filters may not be necessary since HTs should already do bandpass filtering on the speaker and mic audio but it shouldn’t have any significant impact on audio quality to do additional filtering in usbradio with the wider frequency range options (250 - 3.5K Hz Rx, 250 - 3.3K Hz Tx). This additional filtering will ensure the bandpass filtering is thorough and no residual PL tones get through and no high frequencies get through that could potentially result in aliasing in the VOIP codec. It does not appear there is a way to disable usbradio’s bandpass filters (thoughts anyone?) but if there was it would be interesting to see if that made any noticeable difference in audio quality. The frequency and phase response might be a little more linear with them off.

Someone else was concerned that audio from a speaker jack could be too “hot” or could get bumped or misadjusted. The audio level from the speaker jack is controlled by the HT’s volume control, which once set at an optimum point remains perfectly calibrated with the audio interface trim and usbradio gain settings. Thus it’s very easy to set the level at an optimum point, and there are many simple ways to ensure the volume knob stays where it should, for example:

Put the node somewhere where people are not going to bump into it and turn knobs, such as on a high shelf or in a closet.
Put a piece of tape on the knob once it has been set. Or even better a dab of epoxy or silicone, or pull off the plastic knob and put tape or a lock-nut on the potentiometer shaft.
Put the node in a case. Would look great in a clear plastic case and would keep the controls out of reach and keep dust off of the components. The whole system uses so little power that heat would likely not be at an issue even with limited ventilation. But a case could require a small fan or some ventilation slots depending on ambient temperatures. The Dell 3040 is fanless and seems to have very good thermal design, the CPU temp always stays in the Green for me, but you’d want to keep an eye on that in Supermon.
Regardless of what type of node you have it’s good to periodically check your Tx signal, which can be easily done with Parrot mode. I have DTMF commands enabled for parrot mode in rpt.conf (*921/922).
Even if you did still bump the volume knob and it became too loud or quiet, someone would probably notice pretty quickly and let you know.

BTW if you ever hear anyone on ASL who is significantly too loud or too quiet – let them know. They will probably be happy to fix it, and the system then works better for everyone. Audio levels are pretty easy to adjust on AllStar (definitely much easier than with the low-bitrate AMBE/IMBE digital radio modes where people are using all sorts of random apps and devices and you get a huge range of volumes, glitches, artifacts, etc.)

Comments re. Frequency Coordination from Jim NO1PC:

"…Hotspots are a ‘new niche’ not fully encompassed amid the more well-known realms of ATV, SSB, EME, repeaters, simplex, etc. Some regions have specifically addressed this, some not yet. “The ARRL Band Plan” is wholly inadequate in addressing this, but then they vacated much of the issue a decade or so ago.
…
There are 53 repeater coordinating/spectrum management organizations in the United States. Links to them are provided by at least the following two web sites:

"

When picking your node frequencies be sure to consult the above sources, and verify the frequencies are not being used by other nearby users or systems.

Re. Full-Duplex:

Why go full duplex (when most nodes are not and repeaters might not support it)? ASL-linked repeaters do support true multi-user Full-Duplex. ASL is built on Asterisk which fully supports this by default. This is a subtle detail that can be hard to understand until you’ve tried it for yourself. I use full-duplex on many different repeaters and it works great. The repeater itself on analog will not support more than one analog RF input at a time if the repeater has only one RF receiver, but generally will support any number of simultaneous ASL full-duplex users and at least one analog RF input.

There is only one repeater I’ve used so far that doesn’t support full-duplex but all others (at least 5) I’ve used so far do. I initially had my doubts about how many would fully support it and was pleasantly surprised. By default when a repeater links to ASL it will be with a node whose audio output gets mixed in with the RF receiver audio input(s). Some repeaters may prioritize one over the other and only allow one audio channel at a time, or ASL may be linked through some intermediary interface, but even in that case multiple connected AllStar users can still be full-duplex and talk and hear each other fine, as that’s a core feature of Asterisk’s architecture. Audio streams from each connected node are always routed to all other connected nodes. No different than if you make a VOIP conference call with multiple people using regular VOIP phones, you’ll all hear each other fine and can talk anytime.

Full-duplex is a subtle but useful feature that can significantly improve the efficiency and interactivity of repeater comms. Doubles and echoing become a thing of the past, because you always hear the remote system and if someone starts to double or you get looping, echoes, timeouts, interference, etc. you can just unkey until the repeater (or node / hub / bridge / etc.) is clear. This is a big step up in efficient, responsive, and interactive repeater use. Go onto any repeater net that has more than a half dozen AllStar/EchoLink users and you’ll frequently hear doubles, echoing, etc. This wastes minutes of everyone’s time for no good reason. But if you use full-duplex it’s not an issue because you have full situational awareness. This kind of goes back to the whole “amateur” vs. “professional” subject. Amateur radio is what it is because we’re not in it for money – not because we don’t communicate well. Or at least that’s what I think the distinction should be. So if you can now very easily monitor your communications and prevent doubling and other issues, why not do it?

Another nice detail about full-duplex is that once you start transmitting from your cross-band full-duplex HT (I use an FT-530 or TH-D72) it’s nice to see your node transmitting back to you right away. I can be over a mile from my house on my HT and start transmitting, its red Tx LED lights up, and then ~25mS later you see the green Rx LED light up which confirms that you’re in range, the node is receiving you and transmitting back to you, and you hear if there is any path noise on the Rx side. This is another small example of how it improves your overall situational awareness.

Recommended usbradio.conf settings:

Your node number (should have been set in asl-menu/initial setup)
rxboost = 0
rxctcssoverride = 1
carrierfrom = vox
voxhangtime = 500
ctcssfrom = no
rxdemod = speaker
txprelim = no
txlimonly = yes
txtoctype = no
txmixa = no
txmixb = voice
rxlpf = 2
rxhpf = 1
txlpf = 1
txhpf = 1
duplex = 1

Recommended rpt.conf settings:

Your node number (should have been set in asl-menu/initial setup)
rxchannel = Radio/usb_[node#] ; Comment out all other rxhannel lines
duplex = 3
hangtime = 100
althangtime = 100
linkunkeyct = none ; prevent extra courtesy tones and hang time
nounkeyct = 1
parrotmode = 1 ; 1 = Parrot On Command
Uncomment following lines:
921 = cop,21  ; Enable Parrot Mode
922 = cop,22  ; Disable Parrot Mode

NR9V · December 4, 2022, 5:40am

Additional notes on some relatively minor details:

• Re. Squelch settings (applies to any nodes and repeater comms in general), Yaesu HTs (FT-530, FT5D) have very good squelch that drops very quickly (maybe ~20mS max) and quietly, and my Icom-9700 is even faster with basically no squelch tail at all, but my Kenwood TH-D72 and D74 are not so quick and have a much more noticeable squelch tail (seemingly more like 50-100mS and fairly loud). So I enabled Tone Encode on the node TX HT and Tone Squelch on my other radios to minimize squelch tails after the node unkeys. Both Kenwoods now drop squelch immediately ie. no squelch tail at all. (The FT-530 doesn’t drop squelch any faster with Tone Squelch on but that’s fine as the tail was short and quiet to begin with.)

• Cross-band full-duplex nodes do work fine with half-duplex radios if they are dual-band with dual-receive. For example my TH-D74 won’t do full-duplex but I have the node’s 70cm Rx frequency on one memory channel and the 2m Tx frequency on the next memory channel, with dual-watch enabled and when I transmit the D74 will not receive at the same time but once I unkey there’s then no difference vs. a half-duplex node. Thus you can use any radios you’d like with a full-duplex node as long as they are true dual-band radios, which pretty much all modern HTs are. (I rarely the use the D74 any more since it’s only half-duplex, really is a great HT otherwise but I’ll probably end up selling it.)

• Some of the Chinese HTs do partially support cross-band full-duplex (FDX), with the limitation that you can only Tx on 70cm and Rx on 2m – so that their 70cm Rx doesn’t get de-sensed/overloaded by the 3rd harmonic of their 2m Tx. I have seen several models such as the TYT 8000 that support this and that could work well for use with this node, and the 8000’s are available new for as little as $65. Haven’t tried one myself as I already have some good Japanese radios that fully support FDX, but it is nice to see the Chinese supporting it in new radios to at least some extent since there are few if any current production Japanese models that do. These would not be ideal for node radios though since the RT85’s are only $25 and have no limitation in transmitting on 2m while another one receives on 70cm.

BTW a size comparison of a TH-D72A and RT85:

• The Repeater-Builder RIM-Lite V2 Radio Interface Module looks like a good alternative to the Masters Communications DRA-30. DRA-30’s fully assembled with no case currently are $65 plus another $10.50 for shipping, whereas RIM-Lite V2’s are currently $60 and $5 for shipping. I just ordered 2 of the RIM-Lite’s, they were in stock and shipped today, and I’ll be testing those to see if there’s any significant difference vs. the DRA-30. Both use the same CM119A chip and have similar circuitry but the RIM-Lite is significantly more compact as it’s a fully surface-mount PCB with Mini-B USB jack. The DRA-30 has trim pots on the audio outs which is definitely helpful, whereas the RIM-Lite has a extra low-pass filter. A ~$2 trim potentiometer can easily be added to the wiring if needed to provide wider output audio level adjustment/dynamic range.

• It appears the ASL USB audio drivers use a fixed 8KHz sample rate (decimating the USB stream 1:6 from 48KHz) at 16-bits ie. a raw bitrate of 128Kbps. Once compressed by the codec it’s probably more like 32 or 64 Kbps sent over IP. (Which is about 10-20x more bandwidth than what’s used by AMBE/IMBE codecs ie. DMR, D-STAR, C4FM, P25, etc. - thus why AllStar sounds far better than all the digital radio modes.) There are a number of codecs that can be enabled in modules.conf, by default ADPCM, A/u-law, G.726, and GSM. It would be interesting to play with those on a test node and confirm if ASL nodes are going with the best-sounding codecs by default, or how much improvement could be made and what the differences would be in internet bandwidth. The 8KHz sample rate is probably the limiting factor though as that limits the frequency response to ~3.5KHz and my guess would be that the ASL-defaults are probably already pretty optimal.

• To have the node be self-contained and look somewhat “professional” all components could be mounted on a piece of ~8"x12" pegboard (eg. “Dura Board”) rather than in an enclosure, so that everything is easily accessible and well ventilated but is also well-secured and cleanly organized. The MicroPC can then be turned on and off with its power switch, and a separate power switch for the HTs can be used, and the node is then easy to move around, set on a shelf or hang on a wall. (Just add a picture frame and you’ll have an instant conversation piece.) Pegboard is very inexpensive and the holes every 1" are perfect for nylon zip ties to keep the components and wiring in place, along with velcro or twist ties for anything you might want to be able to take on and off more easily.

• Power to the radios will probably be best provided by a small switch-mode power supply. Most devices now use these small switching AC wall adapters and they’re very efficient and compact. AC adapters have come a long way in recent decades, they used to have just a transformer, rectifier and small filter cap providing unregulated output with a lot of ripple, which can easily cause 60Hz hum and harmonics if used to directly power an HT. But these supplies now have better regulation and go through many certifications. I had originally tried to avoid switching supplies as they can cause RFI (birdies) particularly in the HF bands on harmonics of the switching frequency (eg. 100KHz typically but as high as 1 MHz), but if it can be confirmed that’s not an issue for a specific adapter that should be the way to go.

My measurements of the RT85’s show that at 7.5V they have the following Current Draw & Power Output:
Off: 3mA
Standby: 100mA
Low Power: 475mA ~1.5W
Mid Power: 906mA ~2.8W
Hi Power: 1020mA ~3.7W

A 7.5VDC 1.5A power supply should thus be fine for 1 HT in receive and 1 transmitting low power. Note that a node Tx radio should only be used on Low transmit power because of the high Tx duty cycle. The RT85’s Low power setting is already significantly higher than most HTs (1500mW vs. more like 500mW on most other HTs or as little as 100mW on HTs with an Extremely Low power setting). Thus 1.5 Amps is over twice the current actually needed, ensuring the power supply should also stay cool. These power supplies are widely available on ebay/amazon for as little as $9 for what appear to be reputable brands that should have good performance. I ordered a couple of these and will be testing them to insure they work as they should without causing RFI to the HTs or on HF. Because switching-supplies switch at a high frequency there should be little or no 60Hz harmonics on the output and large filter capacitors should not be needed. Switching supplies are used with many devices such as internet routers and other electronics and they do not seem to cause any RFI issues at my QTH so I suspect this will work much better than linear supplies which are larger, less efficient and have more ripple.

A case for the audio interface might also be nice, but as these are fairly small simple PCBs I suspect they will work fine just zip tied or screwed onto the node backboard (though ideally not directly on top of the PC or power supplies).

• Re. wiring, Since the RT85s are very cheap and to keep the node as simple and inexpensive as possible the simplest thing to do is just solder 2 (~24 ga.) power wires onto the battery contacts on the back of the radio. These are very easy to solder to. Just pre-tin the wires and the contacts; the wires can then be tacked on in less than a second, and can be just as quickly and easily removed later if need be. These should then go to a small terminal block or a couple small wire nuts. If you use a DB9 with a terminal block (eg. this listing on ebay) all node wiring from the HTs, power supply, and radio interface can be connected with a minimum of wires and no soldering required.

The node ideally should have either one AC power cord (or one 12VDC power cord) that connects to the 2 power supplies for the PC and HTs. This type of full-featured high-quality node will probably make more sense for use on 120V than 12V since it is significantly larger and less portable than the small SHARI-type nodes. It’s pretty portable if properly secured in an enclosure or on a peg board type of mounting surface, but even then the node will take up approximately 10" x 15" x 1.5" of space which is ~5 times the size of the average portable node. If it were desired to run it on 12V the only thing that would change is that DC-DC converters would be used instead of AC-DC switching adapters, which are also widely available for less than $10 ea. online.

The above minor details will result in a nicely-integrated node, with a single power cord, separate power switches for the PC and HT power supply, that is easy to access, power on/off, change frequency settings, etc.

Updated drawing showing all components. Note that the power adapters can be placed behind the MicroPC and audio interface further reducing the overall size of the node. And the RIM-Lite is quite a bit smaller than the DRA-30 thus the overall dimensions can be significantly smaller than shown below:

With the PC’s Real-Time-Clock IC supporting powering on and off on a schedule, it would also be nice to have the PC be able to control power to the HTs. That can be done very easily with a USB cable and a small 5VDC relay. The 5V line on any of the 3040’s USB ports will only be on when the 3040 is on, thus with a $2 USB cable and a $2 relay the power to the HT power supply can be automatically controlled if desired.

Update 12/2/22: Received the RIM-Lite V2 and 7.5VDC 2A switching power supply today. After testing the power supply I found that for the best audio quality ferrite core filters should be placed on both of the switching power supplies, with the cord wrapped around the ferrite a few times. Without that there was significant buzz coming through from the 7.5V switching adapter. This is a known issue with nodes in general, relating to Tx RF being picked up by the power cord and then causing 60Hz harmonics when the diodes in the power supply are near their zero-crossing points - where they can be modulated by induced RF. This results in a broadbanded buzz on 60Hz harmonics that gets louder and quieter as you move antennas or cables around or move around with your HT that you’re talking into the node with. Sometimes it won’t be noticeable at all but then move 1 foot away and you start hearing buzzing. Ferrite cores knock this down by 99+%. With a linear power supply it’s not much of an issue because there’s a lot more capacitance between the diodes and the power supply output, since a linear supply’s filter time constant needs to at least 10x the 120Hz ripple period, whereas switching supplies are doing things up closer to 100KHz. Adding a couple good bypass capacitors to a switching-supply might be just as effective if added in or by the case, I may do some further experimentation there but a ferrite seems like a simple-enough solution. Thus as a general recommendation for any node, a few turns of power cables from switching-supplies should be wrapped around a ferrite core as close to the power supply case as possible. I found that adding a ferrite on the Tx audio cable helps a little also. (I was listening to the node on an SDRplay RSPdx with full audio bandwidth, whereas a normal radio bandpass filters everything to ~300 - 3K Hz. Thus an SDR can help with even further reducing quiet noise that wouldn’t be audible on a normal radio.) In total you’ll want 4 ferrite cores for a node with 2 switching supplies (one for the MicroPC and one for the HTs). The same would probably apply to a node that ran off 12V using switching DC-DC converters.

The RIM-Lite V2 looks nice, they are very compact, sealed in clear heat shrink and they come with a bonus DB9-M plug and shielded DB9 housing which is not mentioned on their site (which saves about another $5) as well as a USB A to MiniB cable. Once I get some more parts in the mail I’ll be building a new node with the idea of it being fully self-contained and “professionally” wired with all components mounted on on a ~8" x 12" piece of “Dura Board” or similar pegboard, plexiglass, etc. This should meet my goal of being the least-expensive possible way to build a self-contained full-duplex node using only high-quality off-the-shelf FCC-certified components. I plan to sell one or maybe a few of these on various sites, not really to make money or with any plan of making a lot of them, just to create and document the whole process and to know that the node is sufficiently well-built to be able to be sold as a turnkey product, for anyone who might prefer to pay a little more but have the node already fully assembled, configured and tested. And, so that anyone who builds one themselves will have all needed info and know that all the important details have been thought through and tested. Making a significant number of these is not something I’m interested in (I was a pro-audio OEM in the 90’s and then decided that wasn’t for me and went into software development instead) but because all the software is open-source and the node uses only commonly available off-the-shelf parts, anyone could make these and sell them on ebay/qrz/eham/etc., while adding their own unique touches, other features, options, or optimizations.

NR9V · December 26, 2022, 9:31am

A few updates:

Just to see how easy it would be I installed a COS wire into an RT85, which is not the ideal approach because the pad to be soldered to is very small and requires great care. (The wire should be routed and secured prior to being soldered, and ideally no larger than 28 ga. wire used.) This approach is not recommended unless you have a microscope, fine tip soldering iron, and experience in working with things on this scale. I used a 24 ga. wire which is pretty hefty compared to a 10 mil wide pad. Fortunately good stereo zoom eg. 7-45x microscopes are available for at little as $150. Definitely an essential item to have around for working on SMD electronics. The RT85 Green RX LED is tied to Gnd and fed 3.3V through a 330 Ohm resistor. Thus to get COS ideally a 2N2222A and 4.7K resistor should be used, or potentially the simpleusb/usbradio usbinvert setting could be used, but the RT85’s green LED has a 2.5V forward voltage drop, thus a pulldown resistor would be needed for going into an input that expects an open collector output, and at that point it’s only 34¢ for a transistor and you then have a proper standard polarity COS input. I’ll probably use this for my personal node but would prefer to avoid modifying more RT85s. If someone else wants to do it though here’s some pics:

Location of Green RX LED:

COS wire attachment location:

Wire routing. I made a small notch in the bottom corner of the case for the wire:

Then added the power wires:

Inverter circuit:

Also I put together a menu settings sheet:

In doing more detailed timing testing I found that the hang time when in carrierfrom=vox mode is about 1200mS from the time you stop transmitting into the node until it registers COS/CD=0, and then about 2.5 seconds until the node TX unkeys. When in full-duplex (duplex=3 in rpt.conf (not 4)) mode the hang time on the local RF TX after COS=0 is longer but the node is not actually transmitting audio to the network during that extra time. It probably stays keyed a bit longer to prevent more squelch tails, since it’s full-duplex and you can key up anytime, thus it seems App-rpt took that into consideration and makes it work a little more like a repeater in this case, giving extra hang time after each TX is complete so that if you or a remote node key up soon there are then fewer carrier drops and squelch tails. This works great for me, the extra 1.2 secs of space is not noticeable or annoying during a normal QSO on a repeater, where there’s already several seconds of hang time and a courtesy tone. When voxhangtime was 2000mS (vs. 500mS now) it was much more noticeable and the time to CD=0 was closer to 3 seconds. I would not expect a major difference in hangtimes in carrierfrom=usb mode and duplex=3 other than the 500mS less time for CD->0.

And an important note for any node, make sure all your port settings match on the ASL site for both the server and the node, and that those match the port settings in iax.conf (bindport), rpt.conf ([node#] = radio@ip:port…), and in your router/cable modem. This is 5 different places the port has to be set. If it’s your first node then you probably have the default of 4569 and didn’t need to change anything, but if it’s not your first node then you have to use a different port number (eg. 4570, 4571, …), and make sure all those places have the same setting and that your router/cable modem has a port forwarding rule to forward traffic on that port# to your node. The node can still look like it’s working if those settings don’t match but the audio will then be more likely to have stuttering/dropouts.

I’m now starting on a new node build that should look pretty nice, basically tying all the above together into a fully integrated turn-key solution.

UPDATE Jan. 1, '23: I have compiled all my previous posts into an article with more detail on my latest node build (see pics below), a complete parts list and pricing, more details on the setup and configuration steps, and more pictures. Link:

How To Build Your Own High-Quality Full-Duplex AllStar Node for Under $200

I will continue to update the above article over time with parts updates, additional info and options, FAQs, etc. I am also able to provide kits with the various smaller components needed and/or can provide fully assembled & tested nodes. Preferably everyone would DiY but I can help with any or all steps of the process if need be.

k4sip · January 11, 2023, 12:52am

Thanks for this post and all the updates, David.

(edit - thanks Clifford for the initial post - whoops)

You got me back into building homebrew nodes and tinkering with ASL again.

Byron