Well Happy New Year to me! (and to you reading this, of course...)
CM108 Soldering
I was quite discouraged, if you recall in my adventures with ASL, that I could not solder to save my life a CM108 sound card. Well today was the day I finally did it successfully!
Here's what I did - I first took a piece of painter's tape (masking tape should work, but painter's is a bit heavier duty) and placed it over the IC chip. I left Pin 48 exposed as much as possible, but made sure to leave the other pins below the straight piece of tape. With a screwdriver or knife or any other thin-bladed item, I CAREFULLY drew a scored seal between pin 48 and the nearby pin. I finally took to soldering, except I tack soldered. This is the only way I could do it, and as far as I can tell the VOLDN pin and ground has sufficient contact to activate your node's PTT!
In a nutshell, this should prevent any solder bridging between two SOIC pins, provided that you tack solder, utilize a very fine-tipped solder tip and do a good job of separating the two pins as much as possible. As I'm writing this I'm actually running tests with this soundcard to see how it works, and so far so good! I can key up the test parrot node but I need a mic to make sure the sound input works.
My original ASL node is great and basic, but I want to make it better. So for the new year, I think I'll focus on completing this project first.
A Radio's Repurpose
I was given some heavily parted-out CB radios a while back. Initially, I only needed the basics like the ferrite cores, finals, etc. However, in taking a second look, I believe I can give this radio a second life. Not on the air, but rather the internet.
The radio in question is a Midland 13-882B. When I got it, the S-meter was removed, the 23-channel selector indicator was gone, and it didn't have a meter. The 23MHz crystals were gone too! No biggie, because again this was just a parts radio. But this got the wheels in my head turning...what if I built my base station AllStar node into this chassis?
Immediately I set out researching and drafting designs of what would and wouldn't work, and here's what I settled on:
1. Cutting out the channel selection area - we'll use this area to accommodate an OLED screen that can display many things, such as IP addresses, transmit status, etc. It has an inner diameter of 1.65 inches. I will soon post a custom-made bezel over on Thingiverse for free. Code will come from mattmelling's GitHub page here.
2. KA2284 VU meter - in place of the original S-meter to monitor audio, and for cool lights too :). My plan is to tack this on to the tip connection of the CM108's headphone adapter since the ring will be fed to the audio output amplifier. From what I'm told, ASL is 2-channel output (L+R). Similar to the OLED bezel, I've also designed a custom-made bezel for this on Thingiverse, coming soon.
3. Switches - the 882 had 3 feature switches:
* NB-OFF: I'm planning to use this as a mic gain switch, to add on an additional boost should my mic need it. Optional at this point.
* ANL-OFF: For this one, I'll install a low-cut filter (capacitor to ground) for the receive audio. Again, also optional.
* CB-PA: An obvious one, used to switch between an external or the internal speaker
4. Knobs
* Volume: Will be used for volume adjustment. Power management seems a bit complicated and I can't really find any circuits that would fit my constraints.
* Squelch: ???
* Delta Tune: Not sure, but could be a tone control knob (hi-cut vs low-cut?)
5. Keeping the 3B+ - most have recommended the 4B, but I have what I have already. It does just fine using ASL3 as long as you keep the board itself cool with both heatsinking and active fan cooling. Since it will be in an enclosed case, I'm planning on blowing air out instead of in.
The microphone that I'll use with this will be the stock Midland 4 pin microphone. I can survive without the DTMF feature as I usually use Allmon/Supermon to control connection/disconnection and other commands.
At the rear, we'll only have 3 connections - a 12VDC connection (dropped to 5VDC internally via an LM2596 regulator board, then fed to the appropriate circuits), an ethernet port in place of the SO239 connector (if WiFi isn't an option or hardwired security is preferred), and an external speaker port (ext. spkr). The 12V jack will be located in the PA speaker plug.
To mount everything inside, I'll mount all of the components to a piece of aluminum sheet metal and place electrical tape over sections of the metal that contain items on top, such as areas below the Pi, the power supply board and others. I'll use standoffs to ensure no shorts, but the tape also ensures additional electrical and RF shielding should the supports fail. The motherboard will face upside down during normal operation so I don't think this will happen. Double-sided 3M tape will be used for the mic and audio amplifier as well as the buck converter.
Build Timeline
1-4-2026: I purchased aluminum sheet metal and some mounting hardware from my local tool store place. Total cost here was around $20.
Here are some photos I took of the inside:
2mm bolts and nuts were used to affix the metal mounting plane to the chassis. Note that not all panel-mounted hardware is installed as I have not received them yet. Currently they're in transit.
1-8-26: Ordered heatsinks, PCB standoffs and LM2596 buck converter.
1-13-26: Heatsinks, standoffs, and buck converter came. We can now build the node barebones.
1-20-26: The rest of the products arrived. But guess who forgot to order hookup wire? >:(
1-22-26: 22AWG hookup wire arrived...
Now let's put this all together at last!!!
And the result?
This!
But here's the thing...it doesn't really sound the best. Despite twisting wires in pairs, it sounds terrible. On receive at least.
So what went wrong and where am I improving?
For one, I ditched the MAX9814. In cracking open the Midland mic I found that these mics were speaker mics. Connected directly to the soundcard this produced loud-enough audio, so no further amplification was needed - just some tweaks via the CLI line. There's a slight 300ish Hz hum along the line but with filtering, this could be taken care of. Come to think of it, a potentiometer in place of the squelch could be used for mic gain. Just set the RX audio in the CLI to 999 and adjust on the front accordingly. I also have RX boost on as well.
Now the audio output is a mess.
I ditched the PAM8302 for a basic LM386 module. I have DOZENS of these lying around so why not put one to good use? It works and amplifies but it also amplifies awful hash noise, CPU noise, and the same 300Hz buzzing. I'm currently working on this at the moment but for right now the node has line-level audio. Sounds fine but this STILL has 300Hz riding along the audio line.
I might know what is causing this and how we can fix it.
First, I have lots and lots of audio lines strewn across the circuit. These act as antennas, and if you connect a long piece of wire to any audio amp (yes, even guitar amps), local AM and FM stations come in crisp and clear - softly, but clear. The same goes for this little amp. It's picking up every instruction cycle, every ripple...everything.
We can solve this in many ways. One way is by using shorter wires. Not impossible but that defeats the purpose of this chassis.
Another way is by shielding the wires using a 2-conductor shielded wire setup. This is probably the best and worthwhile solution. To do this, we'll ground the shield at the soundcard side and leave the shield floating on the mic or audio amp side. I did this with the Si5351 when I converted a 23-channel CB over to a digital VFO - grounded the coax shield at the 5351 side and left the crystal input floating. What this does is that it allows ground to find a common grounding place instead of grounding at separate spots. This problem introduces ground loops, which is what I think I'm experiencing.
The next step is filtering. A basic RC high-pass filter will work just fine for this, with a cutoff near where our problem frequency lies. We'll apply filtering to both the mic and the audio amp. In addition, the LM2596 would most certainly benefit from additional filtering too since it is a switching mode power supply. Or I could ditch it entirely and get a 5V linear power supply.
Finally, power conditioning. The LM2596 buck converters are handy but they are noisy. Conditioning voltages with bypass capacitors work amazingly and can give impressive results.
I did decide to ditch the internal speaker and use the speaker grille as a passive vent for cooling. We'll have a fan at the back of the case anyway so this will most definitely help air flow.
Fear not though, I'm putting this AllStar node on the network soon! In fact I tuned in to a Massachusetts repeater network where they were talking about the recent winter storms that passed through. I also think in the latest update they fixed the node uptime stats too.
Would I do this project again? 100%, yes. This was not only fun, but definitely a learning experience that even a Technician could do. AllStar may not 100% be true radio, but it sure is fun if you don't want to freeze putting an antenna up :)
Stay tuned for an update as well as a possible YouTube video on this! Man, my channel is really dusty...
No comments:
Post a Comment