- 2025
- Mar
- 28
A PACO G-30 RF Signal Generator - Part 2: Revisiting
Late last year, this PACO G-30 RF signal generator landed on my bench. You can read that original article here: https://wereboar.com … rf-signal-generator/. Since I’m waiting for parts for the Heathkit AF-1, I decided to take a fresh look at this guy and get it ready since it’s next up on the plate.
This was a purchase from the 2018 Cuyahoga Falls hamfest, and I purchased a number of devices from this vendor. He said he had been using the device up to the time of sale, but needed to clear out some space. Seeing as how it was in really great shape, I bought it and some other devices he had for sale. He has also indicated that it had some work done to keep it running, but not much.
Physically, this unit resembles all other PACO devices, and is in pretty good shape for the age. I’d rate this a solid 8/10.
It goes to a decently high frequency range, although the output at those higher bands is probably not of much use. We’re more interested in the lower stuff, primarily the AM band - even though this says it’s a sweep and marker generator as well. Sure thing.
Inside, it’s pretty clean, and relatively sparse. This uses two tubes, a 6AU6 as the RF oscillator, and a 6C4 for the internal audio modulator. Power is provided by a solid-state rectifier, and everything else is passives.
Looking at this unit, you’d think that some of the components had been replaced - and some have. There are a couple of modern film capacitors in there, and a modern…stack? of electrolytics. Everything else is original, including the ceramic disc capacitors and the suspiciously modern looking resistors. There are carbon comps in there, however, so they’ll probably need to go. I’m not sure what the other style of part is.
So what needs to go here? Well, that electrolytic stack is ugly. While it works, I’d rather replace it with single units of the proper values. That’s easy enough. The selenium has to go, as does the three-leg safety capacitor. Why that thing hasn’t split and burnt is beyond me. That’s literally all we’d need to keep it going, and someone has already determined that the resistor in the power supply after the rectifier probably should be ~3.3kΩ. So, the following parts are needed:
1x 1N4007
1x 3.3kΩ 2W
2x safety capacitors
A note on Selenium rectifiers: I've seen varying things about these little devices. Some will say they're probably never going to go bad on you, and the big ones are your worry. Others state that you should replace them regardless. Some are hysterical in their need to change them out. I'm going to change it as a matter of course, because I can and someone else has already done the legwork. YMMV, of course.
While I’ll probably replace all the resistors eventually, the short parts list is all we need for now.
What’s the output of this device look like? Well…it’s not great.
Yeah. That sucks. And that’s 100% correct for this device. Why? This uses harmonics to generate it’s signals, and that’s what you get. It’s fine for an older device like an AA5 radio, but if you’re going to be doing modern work you’ll want to not use this. This is probably going to be more a labor of love than a useful, everyday device.
I have a second unit of this type, which was purchased at the Columbus Hamfest in 2024 for a buck. This unit is a parts donor (should any be needed) and is in pretty bad shape overall. You can tell it’s been used. A lot. It’s dirty, the front dial is gritty and bent, and it has the former owner’s markings on it. (That’s kind of cool, wish it was in better shape! Depending on how the dirt cleans up, this may go on the working model.)
But, you can also clearly see that most of the parts in the good unit are of the correct vintage.
Those green paper caps are all Dumont branded. That’s also kind of cool.
Once I take inventory of everything I’ll need - and possibly obtain a manual as the one online is of miserable quality and must have been scanned with a 100dpi potato - I’ll get started on this one. I don’t expect it to take long, stay tuned!
Next part of this series: Coming soon.
Previous part of this series: https://wereboar.com … rf-signal-generator/
- 2025
- Mar
- 26
The Heatkit AF-1 Analog Frequency Meter, Part 9 - Testing the rebuild.
It lives! After a quick check of my work, I plugged the device in to the isolation transformer and received this wonderful sight:
For testing, I used the manual’s suggested 3/4 scale for all bands. Here’s what I discovered - everything except for the 300Hz band will dial in as expected, with the parts that are currently installed in the unit. I was able to get close to the specified value on the 300Hz band, but I ran out of potentiometer before getting to the right value.
So - in order to get where I need to be, these are the potentiometers that should be specified:
100Hz - 200Ω
300Hz - 10kΩ
1Khz - 200Ω
3Khz - 5KΩ
10Khz - 200Ω
30Khz - 5KΩ
100Khz - 200Ω
The device probably could have sufficed with just the four 1xxx bands.
I’m currently waiting on a 10kΩ potentiometer of a similar style to what I purchased for the other bands, as soon as that arrives I’ll install it and do a final test and calibration post, as well as a wrapup. I’ll have a final corrected schematic at that point, stay tuned!
Next part of this series: Coming soon.
Previous part of this series: https://wereboar.com … ding-the-rest-of-it/
- 2025
- Mar
- 25
The Heatkit AF-1 Analog Frequency Meter, Part 8 - Rebuilding the rest of it.
Since rebuilding the filter didn’t fix some of the issues, and I was planning on replacing everything else - let’s just go ahead and do it.
The same film capacitors and high-tolerance resistors were used for this section as was used in the filter. I simply started at the top and worked down, replacing each part in turn.
This just leaves the potentiometers, but those may require some special attention. After a work check, I’m planning on testing things out. Stay tuned!
Next part of this series: https://wereboar.com … testing-the-rebuild/
Previous part of this series: https://wereboar.com … building-the-filter/
- 2025
- Mar
- 25
The Heatkit AF-1 Analog Frequency Meter, Part 7 - Rebuilding the filter.
Before doing any testing, I wanted to rebuild the filter section of the unit - if the parts in this area are leaky, then nothing is going to work properly. I’m not as concerned with the rest of the circuit - that’s because if it’s repeatable, then I don’t care what it’s doing as long as it does it every time. This may be a bad assumption, but since it’s going to get rebuilt at some point there’s no need to worry about it now.
I started by identifying what needs to be replaced. That amounts to 4 capacitors, and one resistor (hidden on the switch.) There’s another capacitor underneath the four I need to replace, since it’s in the way it will get replaced at the same time even though it’s not part of this circuit.
They took a little bit of doing to get out because I didn’t want to cut things - I’d like to keep these for leakage and other testing as to verify my equipment is working. “Calibration standards,” so to speak.
I did some basic testing on these with the equipment at hand. They all register fine at low voltages, mostly close to their expected tolerance ranges of +/- 20 or so percent. Even the resistor was good at 221.5Ω. No leakage tests were performed, nor were any frequency tests performed. Vloss was quite good for old parts, however.
The area was cleaned and prepped for new parts:
For this rebuild, I chose my go-to IC film capacitors rated at 630V. The max voltage in this unit is around 235VDC, so 400V units could have been used - but I’ve tried to stay as true to the original as possible. The only similar-for-similar replacement was the mica part on the end.
The new parts are significantly smaller than the old ones, and look nice in their new homes:
With the replacement of a damaged piece of wire and a quick check of my work, it’s time to move on to the potentiometers, which will be used in this round of testing.
Now that the problem preventing testing has been cleaned up and the business end of the unit has been rebuilt, I went ahead and put two new potentiometers in the unit. Both are 10-turn wirewound 2W units. 200Ω for the 100Hz band and 5kΩ for the 300Hz band - as was in the unit when I received it.
The schematic states that there should be 200Ω in all positions, so I’m not sure what’s going on here.
I’ve removed the old parts. These will go into the parts bin.
And new ones are installed:
(No, these aren’t Bourns pots…the manufacturer stole the logo, of course.)
100Hz dialed in as expected as was decently linear across the band. 300Hz, however, ran the pot all the way out and could have probably used another 1K of resistance.
I’m not sure what’s going on with that, but I do know there are several more leaky parts in here. Since I’m planning on replacing all of them anyway, I’m going to go ahead and do that before poking at this with a scope. Stay tuned!
Next part of this series: https://wereboar.com … ding-the-rest-of-it/
Previous part of this series: https://wereboar.com … ng-the-power-supply/
- 2025
- Mar
- 25
The Heatkit AF-1 Analog Frequency Meter, Part 6b - Testing the power supply.
The power supply for the Heathkit AF-1 has been re-done, and all of the relevant components replaced. I’ve checked my work, and don’t see anything that would cause the smoke to be let out of the device. Therefore, it’s time to plug it in. It, of course, goes on the isolation transformer/variac for safety, with the master power switch close at hand.
First thing to do is identify the test points I need to check. Input voltage and Output voltage are going to be important here:
As part of the rebuild, terminal strips were added for various parts. That made it easier to lay the new parts in (since the capacitor’s pins are no longer there,) and it makes it easier to pick voltages off.
Input voltage is nothing unusual.
Ripple is now as expected. 120Hz @ ~1VPP. Certainly much better than the 11VPP we had before.
Voltage after the regulator is as expected as well.
And there’s the purple glow of the VR150 tube.
I’m planning on rebuilding the filter section of the unit before proceeding. Stay tuned!
Next part of this series: https://wereboar.com … building-the-filter/
Previous part of this series: https://wereboar.com … -6-the-power-supply/
- 2025
- Mar
- 25
What’s inside one of those can capacitors?
Pretty much exactly what you’d expect. Nothing really exciting…
I decided to take the top off the capacitor removed from the Heathkit AF-1 that I’m working with, since it’s completely dead and as light as can be. There’s probably no wet electrolyte left in this thing at all.
My pipe cutter made short work of the aluminum can, and it separated into two pieces.
As expected, this thing was dry as a bone. It looks to have used some sort of tar or other dark potting material, which was also dry and hard.
The material itself was just paper and foil, wrapped with old tape. I assume that somewhere in there are dividers for for the different sections.
There’s not much else to be said about this device. I’m going to pull the old material out and keep the base for other things, but the device itself gets replaced on the AF-1’s chassis.
Stay tuned for the power supply rebuild.
- 2025
- Mar
- 13
The Linksys WML11B media player - it’s time to say goodbye.
A few years ago, I wrote a piece detailing some fun I had with a Hamvention find, an ancient Linksys WML11B media player. You can find that article right here.
This is a device from the very beginnings of the Consumer Internet Age, and is evidenced by it’s early Wireless-B connectivity. Fortunately, it also offers an Ethernet connection. But…that’s about it. Modern browsers refuse to communicate properly with this thing’s web interface, seeing as how it was designed to be used with IE6. Modern HTTPS audio streams don’t work here because HTTPS was just starting to show up in the consumer space when this was made.
The weirdest thing is that a gigabit Ethernet port, which is usually more than capable of negotiating downward, doesn’t work well with this device. The device will simply stop playing and claim it has no connection after a short period, usually around an hour. A 100base connection will play for hours until the relatively fragile buffering in this device hits a hiccup and it stops.
That’s a real shame too, because this is actually a pretty good sounding device, probably on par with the Squeezebox Boom that came much later. What’s also a shame is, while this did have a now long-gone backend that would feed it station lists, it also had it’s own internal presets. You could use this thing and not have to worry about someone pulling a Reciva on you.
Seeing as how those squeezebox devices are still viable due to the continuation of the media server by Lyrion, I’ve decided to replace this unit with a used Squeezebox Radio that I purchased online for cheap. I’m not sure what to do with it, but the amplified speakers could be useful since they’re of decent size and good sound quality.
Who knows…you may see it at a hamfest near you!
- 2025
- Mar
- 3
The Heatkit AF-1 Analog Frequency Meter, Part 6 - The power supply.
Since this device came to me with a bad power supply, the first thing I want to do is get that supply back into operation so I can do some further tests on the device. It’s a relatively simple affair, consisting of the transformer, rectifier, filter, and a somewhat unusual voltage regulator tube. There’s nothing going on here you haven’t seen before save maybe the VR150, so let’s jump right in.
Here’s all the parts that were removed:
1x 3-section 10μF capacitor @ 350WVDC
2x 470Ω 1W resistors
1x 470Ω 1/2W resistor
1x 2.7kΩ 1/2W resistor
1x 2.5kΩ 5W resistor
1x 3.3kΩ 2W resistor
The whole reason for this initial exercise, is the original capacitor being bad. The can feels like it has nothing in it at all, and 103Ω certainly isn’t good for an electrolytic capacitor!
The new parts went in to the bottom of the chassis without issues except for the leads being shorter on a few.
Spaghetti comes in handy and is used liberally here.
The top looks a bit different now:
I’m not really concerned with the one capacitor being close to the VR150, it’s a cold cathode tube and doesn’t make a lot of heat. The one close to the 6X5 rectifier may need a heat shield for longer life, however. I think it’s far enough away, but you never know.
A new power cord was added, but I want to check my work before applying power and it wasn’t plugged in yet. That’s coming in the next part, as is some testing with new potentiometers. Stay tuned!
One last thing is the size difference between old and new parts. Here’s some new 3.3k 2W resistors replacing the one next to it. The new ones are almost the same size as the old 1/2W parts, and certainly smaller than the 1W and 2W parts shown. It’s amazing how technology has changed our ability to pack more into a smaller space.
Part 7 on the way, Stay tuned!
Next part of this series: https://wereboar.com … ng-the-power-supply/
Previous part of this series: https://wereboar.com … -different-approach/
- 2025
- Feb
- 27
A new documents library home here at wereboar.
I’ve wanted to create an on-site documents library for some time, but just never dug into a simple web-facing file browser. Well, I just found Directory Lister, and it does exactly what I need. it presents all of the files with no fuss, no uploads, and no login needed - just pick what you want and download.
In order to save a little space, and prevent your browser from trying to open the file instead of downloading it, I’ve zipped up everything. If I can, schematics and documents of what I work on are all posted within this library. Check it out here: https://wereboar.com … r=wereboar-documents.
If you want a copy of everything in the archive, you can get that here: https://app.box.com/ … ide0trkeb0wi4j6nta20. This URL will change as I upload more documents and archive them, but I’ll try to keep at least two older versions available. I’ll post any new updates to the collected archive when they’re available.
More documents will be uploaded as I obtain them.
- 2025
- Feb
- 26
My misadventures with a Home Assistant Green
What happened?
Recently, I decided to give Home Assistant another try. I had have given it a go in the past, but at the time it simply didn’t give me enough functionality to be “worth it.” Now, with the proliferation of cheap zigbee and LoRA sensors, I decided to see what it can offer.
To facilitate this, I purchased an official Home Assistant Green CPU with the official ZigBee dongle. Why? Well, here’s what the website says:
Ready. Set. Go. — The affordable Home Assistant Green is the easiest way you can start using Home Assistant. It’s plug-and-play and comes with Home Assistant already installed.
Here it is on the site itself, as of February 26 2025:
All cool, right? Remember that it’s already installed.
So, I order from a company that I’ve dealt with before, and a week later it shows up on my desk at work.
I read the little booklet inside the box. It apparently has to be connected to the internet in order to work the first time. Why? Well, remember that little bit about the OS and system being already installed?
It’s not.
It has to download an install it from github. So the very first thing I encounter is that it’s not actually like it is on the website. It, in fact, is not already installed on the device. It simply has some sort of bootloader stub that captures the image and installs it. No indication from where, no indication of how long it might take, nothing. The lack of information here is disturbing.
I plug it in, apply power, and walk away to make something for dinner. A while later, I come back and hit the indicated webpage on the device, and all I get is a big screen telling me that the system failed to install. There’s a log, which I grab and copy here: https://pastebin.com/gdYK38d6
I comb forums, make posts, contact the vendor. The vendor is very responsive, but can’t really offer anything except that they’ll take it back and try to flash it themselves. I get no reply from forum posts, but there are plenty of other people who have the same issue. The creator of the Home Assistant system only indicates that I don’t have an internet connection.
I wonder how the software downloaded and tried to install. Must be that wireless wire I hear about?
Anyway, after spending the better part of half my day on this thing, I set it aside.
Something occurs to me. It says look closely at the log. And there it is:
[supervisor.dbus.timedate] No timedate support on the host. Time/Date functions have been disabled.
[supervisor.host.services] Updating service information
[supervisor.host.sound] Updating PulseAudio information
[supervisor.host.sound] Can’t update PulseAudio data: Failed to connect to pulseaudio server
[supervisor.host.network] Updating local network information
[supervisor.host.apparmor] Loading AppArmor Profiles: {’hassio-supervisor’}
[supervisor.utils.whoami] Whoami service failed with SSL verification: Cannot connect to host services.home-assistant.io:443 ssl:True [SSLCertVerificationError: (1, ‘[SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed: certificate is not yet valid (_ssl.c:1000)’)]
[supervisor.core] Whoami service SSL error
[supervisor.core] System time/date shift over more than 3 days found!
[supervisor.host.control] Setting new host datetime: 2025-02-21T17:58:06+00:00
[supervisor.core] Fatal error happening on load Task : Automatic time synchronization is enabled
[supervisor.docker.monitor] Started docker events monitor
[supervisor.jobs] ‘Updater.fetch_data’ blocked from execution, no supervisor internet connection
Pay attention to the things about time, then go down to the part in the middle about certificate verifications. The SSL cert onboard is in the future. That tells me the time onboard is wrong, in the past, and probably whatever the hardware gave the Home Assistant install process! Only then does it check the time, and yeah - it’s way off. After that, it’s all failures because the SSL checks failed. Of course the supervisor doesn’t have an internet connection, it’s certificate is invalid because the device didn’t check the time.
Why didn’t the Home Assistant install process check the time first thing, instead of in the middle of the process? I don’t get that at all.
There’s a reset procedure. It wipes the drive. I try that. No luck, now I don’t even get the failed install web page. I try it a couple of times, no luck. Forum posts suggest seeing what’s on the screen, so I connect a monitor to the device and find that it’s sitting at “Waiting for Home Assistant CLI to be ready.” It never “becomes ready” so it dumps me into an emergency console which doesn’t really give me much power.
I do notice that there’s a message about the wipe process failed in the boot messages. Hmm…
There’s a “You can flash this with an SD card” procedure. I try a couple of different SD cards with several different flashing tools. None of them work. It just tries to boot normally.
I’m at a loss, and running out of patience. The vendor has agreed to take it back. I set it aside and sleep on it.
So why didn’t this thing get the time from the network. Is it trying to get it from my router instead of outside the network? That can be an issue. I run a Mikrotik router with RouterOS6. It doesn’t offer the time by default because Mikrotik’s philosophy was that your router isn’t a clock. You can’t depend on the router being correct. You want a clock, get a clock. It does have an internal SNTP package, but that wasn’t really doing much because I didn’t care. The router routes packets - it doesn’t care if it’s 2011, 2025, or 2100. It just routes packets.
Mikrotik does offer a NTP package that can serve time to the local network. I install that, set it up, and verify with “ntpdate -q 192.168.1.1” that I’m getting a response. Ok.
I try to boot again. Nothing. Reset - still giving me the can’t wipe error. SD card flash, nothing. I notice some jumpers on the board itself, and decide to go for broke and open the case. I’m sure the warranty is now void, but it’s going to be void when I hit it with a crowbar. But, I digress - there is indeed some headers, one has a jumper, and it’s marked “NOR / REC” and is in the “NOR” position.
Normal and Recovery? There’s no information about it. I move to “REC,” apply power, and…
I see the machine downloading the OS and installing. All of the error messages about kernel time synch that were there are gone, indicating that it indeed was trying to get the time from somewhere. It installs, powers down. I put the jumper back into the “NOR” position and boot. And there we are, the Home Assistant setup and login appears.
I put it back together, let the vendor know, and write a WTF Man? letter to the creator of Home Assistant, a company called Nabu Casa. It seems to be working just fine.
So WTF Man?
When I emailed Nabu Casa originally, the only thing they said was “You don’t have an internet connection,” completely ignoring the fact that I indeed had an internet connection because the software downloaded and tried to install. That was worthless. When I was done, I emailed them and told them what fixed it - no response so far, nor do I expect one because, while I wasn’t necessarily impolite, I wasn’t holding back either.
To me, it looks like this device expects your router to provide a correct time. If it does not, or does not provide the time at all, it fails silently and even the manufacturer doesn’t know this. They simply assume that your device is their device and all devices act the same. Forum posts that had this same problem were solved by taking the Green unit somewhere else and trying again - it works, probably because they now had the right time. Others indicate they had to mess with DNS setting in their router - was that the fix or did changing those (and the subsequent restart) simply force the router to properly update the time? Probably yes, a little of that and a lot more of this.
The vendor, while extremely responsive, wasn’t a software troubleshooter and could only offer me help with the hardware. They could attempt flash or give me a new one.
So the TL;DR here is make sure your router can provide the correct time, otherwise - you’re in a world of hurt!
But one last thing here - why didn’t the reset process complete, and why couldn’t it wipe the drive? (This may also be related to the flash failure.) It seems, from the posts I’ve read, that if something fails and the drive doesn’t get into the exact expected state, the wipe process doesn’t know what to do and fails. That’s extremely fragile, and I don’t understand why it does that. I will assume that it’s only wiping the portions of the OS that contain the Home Assistant containers, and if they aren’t there (in some cases I’ve read where the drive name being wrong causes this!) it just fails. It doesn’t stop, it just continues to boot and destroy itself more. Same with the failed install process - it just keeps going and destroys itself.
My advice to Home Assistant?
The time is obviously very important, since you’re using self-served SSL certs. CHECK THE TIME BEFORE STARTING THE INSTALL PROCESS - If it’s before 2025, STOP. The system does not have the correct time. Or, check to see that it’s within the range your certs are good for, you’re providing them so you know the dates!
If it gets into the SSL process and one of the SSL checks fails, STOP. Since this is important to the supervisor that runs Home Assistant, continuing will only cause problems.
You need to explain what’s going on better, how long you expect things to take, and what that jumper is for inside the case. You need to explain certain things like it’s not installed, and that you need the correct time on your router before continuing.
Last, the install process is way too fragile. This is a consumer device, if you’re not able to recover from a simple bad install without using unpublished information you’re doing it wrong.
Final thoughts
If you’re planning on getting one of these for yourself, do yourself a favor and make sure you have a good, wired internet connection and your router can provide the correct time.