A project I've wanted to do since I found out this radio isn't transmitting anymore. Can I bring this back into service as a useful APRS and/or 2 meter radio?
Well we'll see, because here are the fruits of over 2 years of brainstorming.
Background
AliExpress is known for selling a multitude of RF amplifier modules. Most of the time, these amps are preamps meant for receivers. However, if you dig deep enough, you'll find amplifiers that will push 3 watts with a milliwatt input signal all the way up to 70 watts!
What went wrong
So I thought that one of these modules would be the answer. But here's why it's not a good idea...
Those modules, at most, push out 2 watts at VHF. Not bad, but this was a 25ish watt radio in its prime. I'd also be stuck with a QRP FM rig, which would only be good for about 3 or so miles.
So no go, back to the drawing board...I knew there were much better ideas out there...
Originally, I thought that the original RF PA module, a Mitsubishi M67741H, went bad. However I found that this module was fine, but I had cut into the front case already so it was trashed, right?
Absolutely not!
It was discovered that the culprit was most likely the APC module, which is used to adjust power based upon the reflected voltage at the antenna output. Sadly, the APC module got destroyed...don't ask why...
So I threw it all in a box and planned to sell it on eBay. Until today.
The Plan Now...
Take a look at this schematic snippet:
And now from the service manual...
Notice Q16. By my assumptions and research, this transistor turns on when the APC's FB pin (forward bias) goes high at 9.85V. The entire module is controlled using the 8 volt signal on transmit at the 8T pin. The resulting voltage drop across the base and emitter of Q16 would go back and deliver 9.15V to the DB pin of the APC module. When the driver module draws power and the voltage travels along the trace and through the 10 ohm R90 resistor, we get around 6.8V to the driver module's DB (drive bias) pin. In addition, this would send 9.15V to Pin 2 of the RF power module and turn on the module's preamp stage.
How does it work? It's simply Ohm's law and the voltage drop of any silicon device at work.
First, let's see where we get 9.15V from:
9.85V - 9.15V = 0.7V
0.7V is the exact voltage drop of any silicon device. So this checks out.
Easiest fix? Inject 9.85V at the APC's FB pin on transmit and hope that fixes things. Like this:
And since we know what pins do what, the final APC will look like this in the circuitry:
You can access the base of Q16 directly from the APC's Pin 7. The voltage can be turned on and off via an optocoupler such as a PC817. Use the 8T pin to turn the opto on and off and feed Q16's base voltage on the output side.
That's only one part of the problem solved, though.
The "Gimmick"
The first circuit I've developed relies on the temperature of the module, not a sampled voltage at the module's output (VSWR).
The M67741H can operate between -30C and +110C, which means that a standard, run-of-the-mill 10k NTC thermistor will work great for our sensor. The thermistor is 10k because this is the resistance value it presents at an ambient temperature of 25C, which is a bit over room temperature at 77F.
Below I have attached the circuit which will sit in the APC's original socket:
Currently the circuit is in the "OFF" status. The logic input represents the 8T line, and right now we are in receive mode.
And here it is again, this time in the "ON" status. The 8T line is high which represents the radio in transmit mode:
Perfect - we have an output voltage of 9.841V, which is very close to our optimal 9.85V we need to drive the base of Q16!
Now, here's the safety feature - right now, the module is at room temperature, but what happens if we hit the maximum 110C that the M67741H can operate at? The answer, below:
WOAH 3.2V??? Yes, which means that Q16 will conduct, however this will shut the module's preamplifier stage off - thereby cutting out transmission and saving you from having to buy another obsolete, hard-to-source module! In fact, in looking at the M67741H's datasheet, I can find no measured values of output power when the Vcc1 pin is set at or around 3V. 4V is the minimum tested and at 150MHz it produced 0 watts!
As mentioned above, this circuit works the same as the factory APC, but instead of measuring the VSWR, we're measuring heat. As VSWR increases into an infinite load, the module will generate heat. Instead of blowing up like what you'd expect from a CB radio from the 70s, the NTC will sense this heat and provide feedback which will lower the voltage that is provided to Q16's base, thereby shutting down the module until it cools back down.
As an added protection, I'm going to add a 12V cooling fan to the chassis cooling fins. This may be useless and overkill, but I want to save this module if I can. I already cut into it, and if it blew up on me, it could be a catastrophic failure.
Now, how will we mount the NTC to the module you may ask? It's quite simple actually: I'll make use of a crimp-on terminal lug and bolt it to the module's fin tab. That's it, quite simple. Make sure though that if you do this, you do NOT mount the thermistor behind the metal tab. This could create an air gap between the cooling fin, thermal paste and the aluminum housing which will guarantee burnout.
As overvoltage or component failure protection, it's been recommended that I place a 10V Zener diode in parallel with the load. A 1N4740 works perfectly. Similar to this:
The Zener is rated at 10V as this will be the maximum voltage allowed to pass to Q16's base. 10V is only a 0.15V difference from the specified rating from the service manual. If the voltage jumps above or close to 10V (seems to open at 9.95V), the Zener shunts open via reverse voltage and dumps that excess voltage to ground. At most, we're looking at about 9.9V being let through. This does not change the thermistor's job - it still senses heat, and drops the voltage down to around 3.5V at 110C!
Keep in mind that the potentiometer/trimmer setting will change as the radio's main power supply changes. Kenwood recommends that this radio be run at 13.6V, and the trimmer setting will change if you choose to run it off of 13.8V or even 12V. It may not seem like much, however this circuit will see it as a drastic change. It's best, if you are installing this circuit, to run the radio on a regulated power supply at 13.6V. When you make adjustments, also be sure to run the radio at 13.6V.
BUT...The only issue with this circuit is that as the module heats up to normal operating temperature (40 to 50C), the potentiometer setting will change as well. A good place to start nonetheless, but I don't think this will work in the long run. In addition, an SWR fault takes at most a second to blow up a final. This circuit needs probably 5 to 10 seconds to even sense that the module is getting hot. And sometimes, SWR faults don't even cause excess heat. Are you seeing it now? SWR is a silent killer.
Back to the drawing board...
The "Gimmick" 2.0
As you saw previously, we don't really want the preamplifier to bias based on temperature (you probably shouldn't...). That's okay, because take a REALLY good look at this section of the schematic:
This is a control voltage that is sampled from the radio's final RF power, usually going no further above 700mV in normal operating conditions. In the original module, as this voltage went up (higher SWR from an increased sampled output from the final fed through D19), the power to the PCV pin was raised, lowering the drive voltage, which in turn lowered the RF power module's output power!
Since we know the PCV pin controls the drive amount, we can use a simple NPN transistor instead of the thermistor circuit as a variable resistor. Fun fact: The word "transistor" is actually two words - transfer resistor, because we are transferring current to the base of the BJT to vary how much current flows from collector to emitter. Basically, as we increase current to the base, the transistor opens up more and pulls the divider closer to ground.
And to show it works, here's a snippet of the simulation during a high SWR event:
At 2.83 volts to the FB pin (about 8 volts VSWR on the PCV pin), the RF module will be completely off. According to the datasheet, and as I previously mentioned, about 4V is ground zero for the module. Anything below 4V will shut the module completely off!
Final Circuit
As a result, here is the final schematic that I've deemed stable and robust to work in this radio. Complete with an LM358 op amp, which compares the voltage coming from the PCV pin with the voltage on the output pin (which is the voltage on the output of the op amp, which will be around whatever voltage is on the PCV pin). This ensures that one can use VR5 to adjust output power instead of relying on another potentiometer, just as the original circuit had intended. Additionally, instead of relying on current, the op amp relies on voltage difference and pulls nano amounts of current on the input pins. An op amp will do whatever it takes to make sure that the inverting and non-inverting inputs are equal.
And the result???
IT WORKS!!!
~9.85 volts applied to Pin 7 of the APC footprint provides a nice, clean 15 to 20 watts output, with a solid frequency of 144.390MHz. I did still see 7.4V on Pin 2 of the mic amp circuit, but for now this shouldn't be an issue. It's better than over 8 volts! I believe this may have to be due to being programmed to TX and RX out of band. When I programmed in 150MHz, the power shot up to 22 to 25 watts, so I imagine the voltage on that pin may have been closer to 6.8V.
I did opt to keep the M67741H module but this time, I added on Corsair TM30 thermal paste to the back. Just a quick swathe on the back of the module and squish it up snug against the heatsink. I did decide to add a fan on the external heatsink fins. The module un-heatsinked got noticeably hot after a 10 second keydown, and to better increase my chances of 25 watts, active cooling is a must.
Now...in my initial tests it did blow up on me in the 8T line. I breadboarded the circuit and clearly that was a mistake, as I most likely shorted the 8T line directly to ground from accidentally bending the power rails together. That or I had pulled current that I wasn't supposed to. I was transmitting and saw the RF power immediately cut off. The culprit in the end was Q6 (mainly) and Q8 (just to be safe), and both were relatively affordable from Mouser or DigiKey. I do use relatively very lightly - electronic parts are way too expensive nowadays, especially for these two basic parts: a PNP and an "digital" NPN - an NPN with a base and base-emitter resistor built in. Regardless, I added these replacements back in the circuit and the 8T line was alive once again.
To make matters worse, I also blew up the main power switch PNP, Q15. God, the luck I have with this radio...
So with this in mind, word to the wise...only use breadboards for initial tests. Once you've confirmed things work, PCBs are a MUST!!!
Breadboards don't suck, but they do for RF-based or related projects and could end up costing you the entire project.
Final Words
In looking at other 90s Kenwood rigs, I don't see why this wouldn't work in them. Especially in the popular TM-241, where the layout is the EXACT SAME as the TK-705.
Kenwood's APC module was quite ahead of its time, and many other land mobile and amateur radio names began using their own implementations of Automatic Power Control around the same decade. In fact my Yaesu FT-2200 has an APC module in it, and I about guarantee that my Yaesu FT-891 has some form of power control and would shut down due to high SWR.
And to think that I was going to sell this perfectly-fine radio on eBay...
No comments:
Post a Comment