Taking a Break!

I have been consumed by Project X and I need to take a break. The stark realization is that this rig even when 100% was likely chewing gum and bailing wire. 

Yesterday I struggled trying to get the transmitter working with just the lower-level stages. The 20 and 40M bands are much improved but 80M still has a problem with a large carrier signal riding along. 

The SSB signal going into the transmit mixer is the same circuitry for all bands. Yet in the mixing process on 80M the Ghost in the Machine wants to add a lot of carrier. 

I need about a week thinking of skimpy bikinis on a beach somewhere, so the background brain power can focus on the problem with Project X. So, blog readers you need a break too. 

A Pig in a Bikini

See you in about a week. The above photo really gets you focused on thinking about rig problems.

73's

Pete N6QW

Yet Another Mystery on Project X.

 Before investing a huge time sink in building / reworking the FP-200 power supply for Project X, I wanted to assure myself that everything short of the Final Amp is working!

To date most of my energy has been spent on cleanup, fixing some of the mechanical issues and in general that the receiver is more or less working. To that end the Receiver works pretty well on 40 and 20 Meters. The 80M band is a work in progress. I then looked at could Project X generate an SSB signal. That was my trip down the rabbit hole yesterday.

CIO, Bal Mod and Prod Detector

First question: Does it generate SSB signals? That is a hard question to answer as it depends.

The CIO (Carrier Insertion Oscillator) and Balanced Modulator are textbook and no rocket science here. The Bal Mod is a Diode Ring and includes both phase and amplitude balance. The CIO at 5.173500 MHz regrettably includes no "netting" the frequency so that with crystal aging it is a crapshoot to place the carrier at just the right place on the crystal filter passband. An issue but not the root cause of what I am about to describe.

The SSB generation is "variable" to best describe the results. On 20 and 40M it does generate SSB but on 20M you see more of the carrier present in the SSB envelope. 

On 40M you get the closest of a typical carrier null (adjusting both for phase and amplitude) almost a flat line and nice sharp Christmas tree pattern on the DSO. It sounds good on an outboard external receiver. Incidentally, 40M is the best on Receive.

On 80M you see a RF Signal Filling the Screen and very little SSB envelope visually or audibly.

So what is causing that condition? My first foray into a solution was to use 40M to adjust the Bal Mod both for phase and amplitude (the 500 Ohm pot and 5-18pF Trimmer). Note the only input signals for any band are the audio input from the mic amp and the carrier input at 5.173500 Megahertz. There is no band switching of anything. The resultant signal going into L1 is just that and not band dependent.

But the place where the mystery may have a test point is in the transmit mixer stage. Today I would put an ADE-1 at this spot but that was not an option 60 years ago.

Transmit Mixer Stage

Trust me there is no manual explaining the theory of operation for Project X so much of what I share may be from the seat of the pants.

The two transistors Q4 and Q14 are the Local Oscillator where on 20 and 80M the LO is in the 9MHz range, but in 40M Q14 is a tripler circuit for a Q4 generated LO at 4 MHz which becomes 12MHz at the output of Q14. 

On Receive, the output following the L10 trap feeds the Receiver Mixer stages through the RLY-1 relay contact. Band Switch contact S1-D controls whether Q14 is a tripler or operates straight through.

On transmit, The LO output from the L10 trap is now injected into the Emitter of Q15 and way off in the distance the SSB signal having passed through the 4 Pole Crystal Filter and the 2nd and 3rd IF stages is routed to Base of Q15 to initiate the mixing process. The switched output off of the Collector selects the proper mixer output. If we have the 12MHz LO and the 5MHz SSB (having passed through the filter which and now USB) we get two products from Q15, with signals at 7MHz (LSB from sideband inversion) and 17MHz (USB). The network consisting of L12 and the 82pF cap pick off the 40M signal and pass that result to the Driver stage.

The 40M results where we are able to see and hear a somewhat normal SSB signal is telling as all of the stages up to Base of Q15 are at 5.173500 MHz and not subject to any band switching.

The Q14 stage when operated as a tripler does what it is supposed to do. The problem shows signs of something is not right on 20M and in the extreme on 80M. This leads you to switch S1-D where on 20 and 80M a parallel combination of a 33pF Cap and 3,3K Resistor connected to the Collector of Q14 are Grounded. 

A further note if there was an issue with Q14 with the Grounding on 20 and 80M and especially 80M the 5.173500MHz SSB is not that far removed for 4 MHz with the tank circuit of Q15. I also note that 80M Receive sucks which get back to the Switch S1-D which converts Q14 to a tripler. 

So we need to look at Switch S1-D and the 33pF and 3.3K resistor on Q14 and the signal output from the Collector of Q15 on the three bands. This may not be the final journey, but certainly "The" starting place.

Them that know can make things go.

I am getting more convinced each day that the Project X radio died as a product line as its design was too dependent on all of the stars aligning in a straight line. Some great ideas here but very complex and prone to failure.

73's

Pete N6QW

Done in by Technology!

Most but not all SSB Transceivers have RF Amplifier Stages. This stage can be troublesome as it is a high wire balancing act to provide a sufficient amount of gain but not so much that the atmospheric nose amplification is so great as to negate the gain stage.

The RF Amp stage can be untuned so that it is a wideband amp and a follow-on Band-Pass Filter limits the beefed-up signal just to the band you are operating within. But most of the designs some 60 years ago used a continuously tuned RF Amp Stage. Frequently a multi-section variable cap would have one section to the VFO while another section tuned the RF Amp stage. Those with Ten Tec gear like the Triton series manually tuned the RF Amp stage with a slick named "Resonate" control.

The Project X has a separate control, but it is not a slug tuned coil (Ten Tec) nor even involves a mechanical variable cap. It was a slick design back in 1965 but not so today. I have been wrestling with this circuit. 

Project X RF Amp Stage

Designs of this era (Swan, SBE and others) used the transmitter Pi Network as part of the Receiver RF Amp circuitry as it acts like a band pass filter on Receive. The relay in the lower left-hand corner shows a connection to an 8-pF capacitor that is connected on the other end to the transmitter Pi Network. On Transmit it is grounded to prevent RF from smoking the front end. 

Now to the RF amp stage itself and the tuning is done by biasing a pair of Varactor diodes, which of course changes their junction capacitance depending on the voltage applied.

Let's examine what happens. Slug tuned coils L5 and L6 are in the circuit for all three bands. In the 80M position the Varactor diodes are in parallel with the 390pF caps. On 40M the Varactors are in parallel with the 82 PF. Note the ends of these caps not connected to the band switch are grounded, which provides a tank circuit that is changed with frequency.

 The Varactors are biased by the 5K pot and for these two bands essentially the same value of swing capacitance is added to either the 390pF or 82pF. Peaking of the circuit for each band is not at the same value of bias but within the range (in Theory). 

The first problem is that these older capacitors have drifted and going through the pot range with a fixed value for L5 or L6 one band will peak while the other does not. I have tuned L5/L6 so that on 40M the signals pin the S meter (not really) but are very loud. For that same coil setting 80M is deaf. The reverse is true. So, I can conclude that on 80/40M the fixed caps have drifted out of range and the Varactors can't compensate for that change. 

Now to 20M where L3 and L4 form a whole new tank circuit as L3 is in Parallel with L5 (less inductance) and the combo is resonated with an 82pF cap and the Varactor. The same for L4, L6.

Note also that on 20M the tune pot (5K) is lifted from ground and a 2.2K resistor is placed in series with the 5K. This has the effect of narrowing the range of voltage applied to the varactors by a significant amount resulting in less capacitance change.

Now for the real problem as the operating instructions for the Project X rig simply say don't touch any of coils. Well, someone touched the coils before I did. 

Several issues are in the mix to resolve the problem as the defined end state should be you move the band switch, peak the pot and somewhere in its range you get a strong received signal for each band. Typically, you will see instructions that say do the initial tune up on this band as all subsequent settings are made from this baseline.

A big clue is that only L5 and L6 are in line at all times so setting L5 and L6 on 80M is likely the linchpin. A first task is to pull the 390 pF caps and see what they read. Another task is to see the range of voltage supplied by the 5K pot.

If I were to employ a circuit like this today, I would use a lesser value than 390 and 82 and in parallel is a trimmer cap whose value at midpoint would equal 390 or 82PF. This way if the fixed value changed up or down a trimmer cap could be adjusted smaller or larger.

Now I did spot that one of the slugged tuned coils whether on 40 or 80 peaked at about the same point. The other coil needed a significant change in position for a peak. Perhaps only one cap is bad. On this coil the slug has to move out of the core for a peak on 80M which suggests the 390pF has increased in value. 

Coils L3 and L4 are really cheapo. These consist of an outer threaded ferrite core. Around the threads are wound about 6 turns of #16 enamel wire (tightly wound) and no paper form. They are hanging in mid-air and soldered directly to the band switch. There is a relatively sharp peak. Judging by the number of turns, L3 and L4 dominate. The old 100 Ohm resistor in parallel with a 1 Ohm -- you essentially have a 1 Ohm resistor. 

So now you have a view of how a tunable front end was achieved in 1965, but you also see how 60 years later with component drift this design falls apart. Done in by Technology (old technology).

New Technology Mugs

A Father's Day gift form one of the kids, a Villeroy Bosch Hot/Cold Mug. A twizzle on the old Thermos Bottle trick. There is an air gap between the inner mug vessel and the outer shell, The mug does keep your coffee hot longer! A designer at the firm must have remembered Thermodynamics 101.

Them that know can make it go.

73's

Pete N6QW

Well, the VFO rebuild did not work!

After putting all back together, the VFO was inoperative. But now it is working, and the fix will astound you or should I say the implementation will give you cause to say," I'll be damned!"

First, we must acknowledge something about Germanium transistors, like the 2N2092, which is that it is 65 years old. These class of transistors are prone to failure especially one which has a 4th wire ground lead that is internally connected to the transistor substrate.

2N2092 and 2N2093 are Equivalent PNP Devices

 L to R: Collector, Shield, Base and Emitter

I asked Copilot AI "What's up with Germanium Transistors" that fail?

 Common Failure Mechanisms

• Tin Whisker Growth Tiny conductive filaments called tin whiskers can grow inside the transistor casing, especially if the interior is tin-coated. These whiskers can cause internal shorts between the transistor terminals, leading to erratic behavior or complete failure.

• Internal Shorting Some vintage germanium transistors—especially those with built-in substrate grounds—are known to suffer from internal shorting over time. This is often due to poor sealing or contamination during manufacturing.

• Thermal Sensitivity Germanium is more sensitive to temperature fluctuations than silicon. Heat stress or rapid thermal cycling can degrade the junctions, especially if the substrate ground doesn’t provide adequate thermal dissipation.

• Voltage Spikes & Breakdown Without proper input/output protection, transient voltage spikes can damage the base-emitter junction. Substrate-grounded designs may lack sufficient isolation, making them more vulnerable to such stress.

• Aging & Shelf Life Many germanium transistors are decades old. Even unused ones can degrade due to oxidation, moisture ingress, or breakdown of internal seals. Substrate-grounded types are particularly susceptible if the grounding path allows leakage currents or corrosion.

The Internal Shorting and Aging/Shelf Life are the main two culprits. Essentially, this time I saw no voltage on the collector because internally the innards of the 2N2092 were shorted to ground.

Others who are working with the same Project X rig told me about this solution and based on prior work, I felt it would work.

Recall that I tested the pulled 2N2092 in a transistor tester and then plugged it into a working radio. It passed both tests. But in those two cases I only used the Emitter, Base and Collector leads and ignored the substrate Ground Shield. There it was right before me... the answer.

 

I was able to get my nippers down into the VFO chassis and simply cut the ground lead on the 2N2092. Boom we have oscillation.

Them that know can make things go.

So now you know several pieces of critical information. Avoid vintage Germanium transistors especially those with 4 leads. When servicing an old boat anchor with 4 lead Germanium transistors try cutting the shield lead and see if that fixes the issue. Finally, a significant warning to me personally... Stop buying boat anchors!

73's

Pete N6QW

I would have never guessed.

 

Finally Access to the VFO Board!

In looking at the schematic and measuring voltage at the emitter we had power to the circuit but no oscillation. The 1st thought was that the Germanium transistor was bad... I pulled it only to find it tested good. Plugging it into a socket on a working homebrew rig (Plessey Circuit) it was good!

Now that I had access to the components I could make other voltage measurements. I noted there was voltage on the collector BUT none at the junction of RFC and 5.6K and the .01 Cap. An ohmmeter measurement across the RFC (shocking) it was open. Ah hah, found it! 

Not so fast McGee! True it was open, further checks once removed from the circuit affirmed that. But there was no visible sign of an open, it was not crisped nor any leads broken off. It was just plain open. 

Seeing the choke was inaccessible, there was no probing of or any accidental damage to the choke before it died. It just stopped working. The transistor is good so no huge current surge that would have smoked the choke. I must not rule out the Malocchi!

The replacement choke is a 1000 microhenry (AKA 1 Millihenry) and unfortunately, you have to reassemble everything to test if the repair worked. 

The rig designer never thought DFMA or that any rig maintenance would ever be required. I did get a chance to use my Krud Kleaner on the chassis and the Project X front panel. It worked well.

Hopefully I will get it tested today and see if I really did find the problem. A real help to me was I started off by writing down what had to be done and the sequence of steps. I also followed the script as I did the task! RTFM!

Tasks and Sequence!

I stopped last night just short of powering on the rig as I was tired and was concerned about getting this far and making a boo, boo. That also gave me time to rethink anything that I had not covered in the work tasks and the fix. I will start by retracing what I did and verify, no shorts or solder bridges. 

At this stage, I am glad I decided to fix the VFO as I learned a few things along the way. If the choke replacement is not the cure, then that still is good info. We know the device is good so that reduces the set of what is the problem.

Someone was diddling with the dial drive mechanism which I fixed. The main tuning knob consists of a National Velvet Vernier reduction drive that is panel mounted. At the end of the vernier shaft are two discs one is fixed and the other spring loaded like a pressure plate. A large diameter disc is fitted to the shaft of the VFO tuning capacitor and this larger disc rides between the two discs that are spring loaded. In effect the ratio of the two smaller disc radius and the radius of the disc on the VFO capacitor is s further reduction. 

My seat of the pants calculation suggests with the two drives it is like a 50:1 reduction. Translated 50 turns of the tuning knob to complete tuning across the band. The Swan Monobanders and the SBE-33 used this small/large disc to tune the across the band. But these rigs did not have the added vernier drive.

When disassembling the VFO cap I found that the large disc was not riding between the two smaller discs but on the outside of the first disc. Much smoother tuning now! I had to use a long screwdriver to pry open the two discs as I slid the VFO cap disc into position. I spotted this even before I did any disassembly as I noted at times some backlash in tuning.

Them that know can make things go. 

It's been 6 months now so how would you answer the question are you better off today? The super rich are counting their crypto so we have their answer. Are your grocery prices better and is your bank account growing?

73's

Pete N6QW

Reversing course on Project X.

First a caution and a funny story. I hate the type of Manhattan construction where you use Super glue to glue the cut off copper squares to a PC Board. Frankly Super glue and Pete are a volatile combination. Somehow every time I use the product, I get my fingers stuck together. 

Yesterday I had the need to use Super glue to repair something and managed to get Super glue on my index finger. The good news is that I only got it on my finger but avoided the gluing the fingers together. 

When I got the T Mobile Home Internet, they required you use the Biometric function on your Smart phone. That was pretty slick as now you can open your phone with a fingerprint. Don't you just love technology.

After the Super glue event, I cannot use Biometrics to open the phone and had to enter the code. Then it hit me... the Super glue prevents the phone from reading the index finger. So, I will have to wait until the glue wears off. I was laughing that I was done in by Technology! Stupid phone.

The reverse course is that I will repair the VFO on Project X and not substitute the Digital VFO. There were only about 500 Project X radios made and 200 of those went to a foreign government. So, keeping it as close to original actually adds to its value. Mine is S/N 44

The hurdle is that the VFO assembly was installed 1st and everything else was installed around it. So, there is a bit of mechanical deinstallation needed before you can even get to a point of removing the VFO assembly.

The 1st step is to document the removal process and to evaluate the sequencing so that a minimal amount of deinstallation is needed. 

The bonus is a chance to clean up this area so not all bad.

Them that know can make things go.

73's

Pete N6QW

Design Decisions ~ The process.

 Raspberry Sundae

Our ice cream sundae is a perfect metaphor for today's blog. If you love to homebrew, then you likely have a penchant for ice cream. Afterall, who in their right mind would turn down a taste of a Raspberry Sundae.

But to take a bite, does involve a calculated decision. Afterall with the average age of our ham being way above 30 then it is not a simple decision as concerns for health such as lactose intolerance, adding to an already bulging waistline, cholesterol impacts and then there is that leaking filing that rapidly responds to sugar shock on the nerve endings. When you are 16, none of these issues matter for at that age it is YL's and not ice cream that are in the PFC (Pre-Frontal Cortex) and causes sensations in other parts of the anatomy.

But those factors are in the process when it comes to a radical decision such as completely removing the analog VFO from the Project X radio and replacing it with some modern digital electronics.

Let us ponder some of the considerations that would enter our decision process.

The very 1st question is the technology and that is not a science project as the Arduino / Si5351 are mature and easily implemented. We must ask can the fix be done and can it result in a seamless operation that not only replaces the original circuit but adds a better readout and a more stable frequency generation.

Another consideration is how to switch bands without adding more controls on the front panel. Fortunately, there are three switch positions on the original band switch that are sequentially grounded. These can be used to selectively ground three inputs on the Arduino to change the LO frequency. Project X used the same carrier oscillator frequency for the three bands and by shifting the LO (and sideband inversion) results in LSB on 80 and 40M and USB on 20M. The Si5351 would only supply the LO for the bands as the original Carrier Oscillator in the rig would not be disturbed.

In one similar project where I took the Hallicrafters FPM 300 and added a Digital LO the code for that project is the sourdough starter and could be lifted and easily modified for Project X.

Another consideration is space in the radio for the digital electronics. Removing the original VFO would provide an adequate area and the Frequency window on the front panel would easily accommodate a ST7735 Color TFT. 

The decision should also provide a capability to return Project X VFO to stock and the capability to fully test the digital solution before the J and P (Jerk and Pull) of the existing VFO. 

This last consideration gives rise to a two digital VFO solution and one that enables a stepped approach. About two years ago I built a remote VFO for a Ten Tec Triton II as the dial string slide rule dial was real crap. 

That remote VFO had a 12 Key keypad where band selection was made and some of the keys enabled selection of step tuning rates. In addition to the rotary encoder two keys provided up/down tuning of the frequency and two keys could provide a +10 or -10 kHz split for the DX pileups. 

There were but two connections to the Triton II, one being power and the other the LO output. There I used the internal carrier oscillator of the Triton II. 

This remote VFO could be easily adapted to the Project X radio by simply disconnecting the internal VFO at one location in the radio and using an existing hole in the rear panel apron install an RCA jack for the remote LO connection. Rewriting the code is some simple number changes to accommodate the change in IF frequency.

This would enable a full evaluation of the digital LO and provide a modicum of confidence if my solution would work. Then and only then add the internal Digital LO similar to that use in the FPM-300 reincarnation.

Them that know can make things go.

73's

Pete N6QW