In this thread, we hack Heathkit's -104 lineup

[N8YX]

The "stock" -104A mentioned above is currently tuned to 3843 awaiting the Possums. That the receiver and control circuitry could use some updating is fairly obvious, though it isn't a terrible performer as-is.

Next month I'll get the regulators and update both the +5 and +11v sections. Then on to one of the junker units as a board-test mule.

[KC4TIR]

You might want to try AF4K.com for your crystals.

[N8YX]

You might want to try AF4K.com for your crystals.

If I wanted to spend twice what ICM or the others are charging per rock, yes.

[KC4TIR]

Do you need a downeast 144/28

[N8YX]

Do you need a downeast 144/28

And a 50/28, 222/28 and a 432/28...

[KC4TIR]

Never had a 432/28, sold the 50/28. So where do I send the 144/28. I got the deal of the century on two ssb electronic transverters. 144/28 was last used with a Ren Tec Omni six plus and worked fine. I have some extra mirage amps, however some need repair.

[N8YX]

The prototype SB-644 synthesized VFO is now fixed.

When I bought the thing and first hooked it up, the PLL looked to have some dead-banding issues. That is, even though the synthesizer would lock the output warbled by ~1KHz.

First order of business was to replace all power supply capacitors and all on-board tantalum caps. Luckily, I had a stash of the latter but waited for a week on the former. Once finished, this didn't help stability much. Next, a cleaning of all IC sockets with DeOxIt followed by a re-seating of the devices. No dice. However, I did catch a few chips with marginal pins and replaced them as a preventative measure.

Next was into the loop lifter/error line driver section. This design is a textbook example of a late 70s layout. That is, a timebase serves as a reference input to Pin R of an MC4044 Phase Detector IC. A presettable Divide-by-N counter (whose preload values are defined by the VFO's counter modulus and display circuitry) is used for the V input. Outputs U1 and D1 (run in non-charge-pump mode) feed an LM301A set up as an integrator. Two more '301As serve as an error line amplifier/driver pair. They supply a control voltage to the varactor diode which forms part of an LC tank required by the VCO chip, an MC1648. The output of the VCO feeds a pair of MC12012s: One drives the Divide-by-N counter; the other is used to derive the actual VFO signal.

The loop filter employed a bunch of carbon-comp resistors. All of these were replaced as they drift in value, and if timing is critical then they may not allow for proper loop lock owing to out-of-resonance conditions. All of the polystyrene capacitors in the error amplifier/driver section were also replaced.

What happened next was that - upon powerup - the VFO would slowly drift from 4.0ish MHz up through its capture point of ~5.245MHz...stabilize momentarily...then rapidly proceed to drift to its upper limit. This implied inability of the loop filter to produce a proper output waveform. Inspection of the area revealed...two more overlooked tantalum caps in the +/- input circuits. Order some replacements, wait a week, solder them in...no joy, but upwards drift seemed to occur a bit more rapidly now. So the circuit was integrating the U1/D1 pulses correctly. Hmmm...

A few more back-and-forth component swaps were tried, and ultimately the thing would sit on 4.00xMHz on powerup but refuse to lock. It was as if the Divide-by-N input was absent, but when scoped it was there...albeit with a lot of noise on the line. Puzzling, to say the least - and since I have no schematics or "known good" examples of the device I was unsure of what the waveforms on the pins should look like.

Desparation set in...I began fingering around the board with the VFO powered up. During one of these "touch and feel" sessions the device went into lock perfectly and remained stable. Very interesting. A bit more experimentation showed that if Pin 1 of the PD chip (MC4044) were touched directly the setup would behave.

Capacitance to ground? Resistance to ground? A bit of experimentation led to the tacking of a 22pF ceramic disk capacitor from that pin to a nearby ground connection and the device stabilized nicely. I reset the timebase reference oscillator, threw the covers back on, powered it up and have been running it for 14h straight on the bench. Once thermal stabilization has occurred (likely due to timebase warmup) there is MAYBE a 5Hz random variation noticed on the counter - and I attribute that to circuit noise.

It looks as if Heath's engineers may have missed that one crucial bit about noise on the reference line: If said noise component makes up a significant percentage of the applied signal voltage there's no apparent way to make the loop lock cleanly.

My design will use a DDS setup and the counter/readout board from a junker SB104 but it'll incorporate the same look and feel as the (never produced for sale) Heathkit offering. Meanwhile, I've got to modify the SB-104A so it can output T/R information to the VFO for RIT and split-frequency tracking purposes. Then the prototype goes in-line.

Not a bad design, and there's at least one other version of the setup floating around. Mine has a big Molex connector mounted to the rear bulkhead and this looks to be used for interfacing a logic analyzer to the tuning modulus section.

[N8YX]

It's not every day that you run across the "spark plug" which ignited your desire to become a ham. While searching an archive for a radio-related construction article I happened to remember something I once read...punched in a few keywords...and voila:

http://www.americanradiohistory.com/...-Page-0047.pdf

http://www.americanradiohistory.com/...-Page-0048.pdf

Those of you who have been following this thread will likely pick up on the fact that the one thing Heathkit didn't offer in the way of a 104-line accessory is an antenna tuner, though at least one prototype "SB-670" is known to exist. When I saw that Heathkit lineup many years ago I knew I wanted to build their stuff...and to get my ticket.

And there you have it...why I love the '104. And why I ended up here, of all places.

[N8YX]

Dancing back around to the transceiver, I thought I had an "oops!" moment when I was only seeing 25-50w out of the thing on 40 and 20M, a bit less on 75 and 15...and 10M appears to be plagued with a spurious PA problem. Running the rig on its HP-1144A supply instead of via the shack's RigRunner/RS-50M seemed to verify good PA health on 40 and 20M with ~100w out - but the other bands need some help. Methinks I've got some cooked toroids in the low-pass filter assembly, which leads to absorption and loss of filter efficiency. Fortunately I've got several more of the assemblies which can be swapped into the unit.

I lost the receiver section earlier in the week. In the process of fixing things I made the S-meter and AGC mods as detailed here:

http://www.nostalgickitscentral.com/...B-104-8-98.pdf

The "roofing filter" will be added to each of these rigs as I acquire parts. The converted receivers will use a dual roofing filter scheme - ~4KHz B/W for AM mode and the other a regular Heathkit SSB filter. A couple weeks ago I managed to grab 6 Trio YG-3395S SSB filters (~2.4KHz B/W) and these will be used in the transmitter signal chain. Not quite ESSB but a bit richer sounding than the stock Heathkit parts allow.

[K7SGJ]

<snip>
I lost the receiver section earlier in the week.

Did you look under the bed?