Tech news in 3 minutes
Saving Two Old Heathkit Frequency Counters
When I decided to purchase a “for parts” copy of a Heathkit SM-2420 frequency counter from an eBay seller for $60 back in 2023, I had no idea what I was getting into. The unit was listed as not operable. Its line cord had been cut off, leaving a short and very sad looking pigtail. Nevertheless, it was a Heathkit frequency counter. I’d wanted one for decades. When the counter arrived, I unpacked it and was immediately taken aback by how scratched and dirty it was. The screws fastening the counter’s enclosure together were encrusted with rust. Little did I realize what waited for me inside the unit. I unscrewed the five rusty screws holding the top cover and slid it back to reveal the dirtiest circuit board I’ve ever seen. When restoring classic cars, you frequently encounter the term “barn find,” and this Heathkit SM-2420 was a true barn find. The dirt on the circuit board was thick enough to obscure the component designators in places on the circuit board. There was clearly moisture damage and corrosion. Even the long shafts on the rotary switches were deeply pitted. I didn’t want to even power this counter up, fearing the worst. My Heathkit SM-2420 contained the dirtiest, most corroded circuit board I’ve seen. Note the thick corrosion on the two long switch shafts. Image credit: Steve Leibson At this point, I didn’t expect much from this derelict frequency counter and put it aside for several months. More than a year later, I spotted a front panel and circuit board assembly for a Heathkit IM-2420 on eBay. An SM-2420 is a factory assembled unit, and an IM-2420 is supposedly a kit version, but the board listed on eBay looked like a factory assembled unit. It appeared to be in great condition. Perhaps it was a factory repair of a failed kit-building attempt. Although it was listed for $103, more money than I’d paid for my derelict SM-2420 counter some sixteen months earlier, I snapped it up. The likelihood of finding another clean IM-2420 board like this was near zero. When it arrived, the board included an unexpected bonus: three dead brown marmorated stink bugs stuck to the board. Not a great sign, but with the second board in hand, I felt that I was prepared to attempt a restoration. I formulated an action plan for trying to restore this classic Heathkit counter. First, clean as much crud from the barn-find circuit board and its switches as possible. Second, try to force as much Caig Labs’ Deoxit D5 contact cleaner into the various rotary and slide switches as possible. Third, separate the power transformer from the truncated power cord and try to power the unit up using an autotransformer to slowly bring up the line voltage. For step one, I used isopropyl alcohol, more than two dozen cotton swabs, and some gently applied elbow grease. For step two, I sprayed Deoxit into every crevice I could find on the rotary and slide switches, and then I exercised each switch several times. Finally, I dug into the rear of the unit, detached and extracted the remaining short piece of useless line cord from Heathkit’s incredibly convoluted strain relief, and temporarily attached a short line cord to the counter’s power transformer using lever connectors. Then, I slowly applied line voltage. As I raised the line voltage, the SM-2420’s display lit up, to my very great surprise. At least part of it was alive. The next step was to apply a signal to the counter’s 5-60 MHz input to see how much life it had. I applied a signal from my Wavetek signal generator, and the counter counted. What’s more, it provided readings that agreed with my HP 5314A frequency counter. Given that this Heathkit counter already had an oven-controlled crystal oscillator (OCXO), I felt no need to replace the oscillator with a SiTime Super-TCXO. (See my previous article, “Adventures with SiTime’s MEMS-based Super-TCXOs – Super Accurate Clocks for the Future – Part 2.”) To my great surprise, the Heathkit SM-2420 frequency counter powered up after I cleaned the circuit board and connected a working line cord. The initial readings even agreed with those taken by my HP 5314A frequency counter, more or less. Image credit: Steve Leibson Because the SM-2420 was now working, I had no need of the extra mainboard assembly that I’d purchased on eBay, so I did the right thing and bought another SM-2420, listed on eBay “for parts.” From the listing photos, this second unit appeared to be in worse shape than the first. The paint was badly scratched, it had a silver on/off knob that was clearly not a Heathkit knob, and there was an odd rubber foot stuck to the top of the case. However, this counter showed a display in the eBay listing photo, so it was at least half alive. Between this new unit and the previously purchased mainboard, I might be able to cobble together a second working SM-2420. I bought a second Heathkit SM-2420 frequency counter based on this eBay listing photo. Cosmetically, it looked worse than my first one. Note the silver knob on the left. It’s not supposed to look like that. Image credit: Steve Leibson When I received the second SM-2420, it was clearly in rough shape, and someone had replaced the counter’s rotary power switch with a potentiometer/switch combo. That’s likely why the power knob on this second counter looked wrong. I slowly powered up this counter but swiftly found out that its OXCO was dead, so I turned my attention to the extra IM-2420 circuit board assembly. I brushed off the stink bugs and swapped in the board. I plugged the counter into a power outlet, switched it on, and received a frozen display. Applying an input signal had no effect. The digital section of Heathkit’s SM/IM-2420 design is based on SSI/MSI TTL chips. All the TTL chips on my IM-2420 board assembly came from Texas Instruments, except for the DM9368 BCD-to-7-segment decoder drivers, which were made by National Semiconductor. The counter’s 5V power supply is split between two 7805 linear regulators, which supply different sets of TTL chips on the main board. The output of one of my board’s 7805 regulators was low due to a bad 2200 μF axial filter capacitor. I replaced the capacitor, which restored the 5V power supply, but the counter still refused to count, so I plugged the 10MHz output of my GPS Disciplined Oscillator (GPSDO) into the external clock input, and the counter started operating. The internal 10MHz OCXO was dead. Heathkit’s OCXO design in this frequency counter has a home-brewed nature. It consists of a small single-sided, phenolic circuit board containing a crystal oscillator, which is encased in an oven that’s made from a heater coil wrapped around a 2-piece metal shell. The heater coil uses an 8V power rail in the counter to heat the crystal to a high, constant temperature. This oven sits inside a two-piece Styrofoam box. Seven short pieces of solid wire pass through the base of the Styrofoam box and connect the oscillator’s circuit board to the counter’s main circuit board. After cooking inside this oven for several decades, the copper foil traces on the single-sided circuit board no longer adhered to the phenolic substrate, and all of the wires connecting the oscillator board to the main circuit board had pulled free. I disassembled the OCXO, but the damage to the single-sided circuit board was extensive. The story was the same for the SM-2420’s original main board, so I opted to replace the Heathkit OCXO with a 10MHz OCXO module from China. These modules cost $15 or less, depending on the vendor, and are available from the usual sources including eBay, AliExpress, and Amazon. I drilled four mounting holes in the IM-2420 main board and connected the module. After giving the OCXO time to warm up, the counter started working, and I was able to tune the OCXO to within 0.1 Hz referenced to my GPSDO using the board’s 10-turn potentiometer. With its 8-digit display plus an overflow LED, 0.1 Hz is the SM-2420’s resolution limit for a 10MHz signal. This 10MHz OCXO board is based on an OCXO module that’s supposedly salvaged from obsolete cellular telephone equipment. The OCXO may have aged out of spec for cellular telephony, but it works well in my Heathkit SM-2420 frequency counter. The 10-turn potentiometer allowed frequency tuning to within 0.1 Hz, the SM-2420’s maximum resolution at 10 MHz. Image credit: Steve Leibson During burn-in, one segment in my first Heathkit counter’s 7-segment displays started to flicker. I applied the usual engineering fix by slapping the side of the counter, and the flickering stopped, briefly. Then, the segment went permanently dark. A little probing demonstrated that the obsolete DM9368 BCD-to-7-segment decoder driver to that LED display was working properly, so it was the 7-segment display itself that was bad. Heathkit’s IM/SM-2420 frequency counters use 5082-7760 common-cathode, 7-segment, LED displays. These displays were originally made by HP (the part number is a dead giveaway for those who know HP’s parts numbering system), then by Agilent after it split from HP at the end of 1999, and they are now made by Broadcom after that company acquired HP/Agilent’s optoelectronics product line in 2005. These parts are still available from a main line distributor like Digikey, but I opted to order four replacement displays from Electronic Surplus of Ohio via an eBay listing because Electronic Surplus is an old friend. The company operated a surplus electronics storefront called Western Salvage in downtown Cleveland when I was getting my BSEE from Case Western Reserve University in the early 1970s. I misspent many Saturdays exploring the store’s wares. Back then, we affectionately called the store Western Garbage. Surplus electronics stores like Western Garbage are essential to keeping old equipment like my Heathkit frequency counters in service. Sadly, these stores are disappearing as component-level demand by hams and hobbyists dwindles. Although many Chinese vendors offer these obsolete components, my experience with these part