The AT-1 is the first amateur transmitter offered by Heathkit, in 1951. It could be operated with accessories like the familiar VF-1 VFO kit. To obtain reasonable output power, it had to be coupled via the AC-1 antenna tuner. The AT-1 link coupled output in stock for is non adjustable. Otherwise, the power output was subject to the vagaries of the reactance of the antenna. In most cases, it was unlikely to exceed 9 watts actual output in stock form. The matching network, an L type with limited matching range, included a rudimentary low pass TVI filter and a bulb which purported to be a coupling indicator (but did not work as well as a meter). The AC-1 was not a balanced coupler either; it could only drive coax fed or single wire antennas. Some hams added a series capacitor to the link output coil, to allow some range of adjustment. The rig shown below seems to have this modification. With an appropriate knob, this could be a good idea, and left in place, as it is a common mod. The rig has a pilot light added on the upper left, which is not stock, but looks OK. It also has a round bat handle toggle switch, which is out of place with the stock slide switches. It is for switching from crystal control to VFO mode, a needed function. I think if I put a rotary switch with the right shape matching knob that will be OK as well.

Inside, the story is not so pretty. This rig has a 6BQ6 black and white TV sweep tube for a final, and a neutralizing capacitor, which includes some clear tape to keep the terminal strip nearby from arcing to ground. There is an additional questionable coax connector on the back panel. This appears to be a sloppy attempt to install the McCoy mods from QST October 1955. This is a well written article. It claims significantly more power output from the sweep tube. It also had an intermediate step using the original 6L6, with performance data, but no schematic documentation on that incarnation. One of the reasons the AT-1 has such low output is that the 6L6 stage is not neutralized, and therefore operates as a frequency double on all bands except 80 meters.

Here's what I started with (oh, boy!):
     

My goal in acquiring this rig was to have a classic 6L6 rig of the period, with a few necessary changes to provide the best performance obtainable by the basic design. Update: I also aquired an unmodified AT-1 (except for a different meter). I plan to use that one in stock condition as a classic, since it looks pretty good and requires little rework. I will move the meter from this basket case one into the stock unit. Then I can get radical about modifications on the butchered AT-1, to see what can be done for improvements. A table with the data from the "shoot out" between the two will be provided.

What is the correct tube to use in the AT-1 modifications? The 6BQ6 has less plate dissipation than the 6L6. A 2E26 would have provided better 10 meter performance and been easy to neutralize. Its output at the lower plate voltage would have been similar on 80 meters. Cost of the 2E26 would have been high at the time. Even an inexpensive 6DQ6 (like the DX-20) would have been a better choice than the 6BQ6. The DX-20 ran a higher plate voltage than the AT-1 for 50 watts input instead of 25 to 30. At the AT-1 plate voltage, the 6DQ6 might have been just right for the existing power supply and transformer. It also would have been easy to neutralize, if needed, and had better output on 40 through 10, since the 6L6 is operated as a less efficient doubler on all bands other than 80. Maybe Heathkit even tried the 6DQ6 mod in an AT-1 and decided to build the DX-20. It is unclear what guided McCoy in his design choice on this modification, possibly cost. For more information on the DX-20, see:
http://www.wirelessgirl.com/Projects/AMTransmitters/HeathkitDX20.html

The VFO switch and "loading control" will stay, either with link or PI network coupling. Neutralization was not unknown technology at the time; a new Novice operator might not have been capable of adjusting it correctly. I plan to use contemporary neutralization circuits to obtain better output suggested by the McCoy data. Neutralization allows operation of the 6L6 as a straight through power amplifier. Ten meters is a bit of a reach for a 6L6, and there is only one oscillator/buffer/quadrupler stage, the 6AG7. With propagation at this time, it is unlikely this rig will find use on 10 meters. So maybe this rig will have 160, 80, 40, 20, and either 15 or 30 meters. Note that on 15 meters, 5 MHz crystals were used, according to this source: http://webpages.charter.net/aj8mh-radio/heathkit.html.

Since this 5 MHz frequency is not output by a stock VF-1, I wonder how they got it to operate in the 15 meter band, which was assigned to Novices at the time. But Novices were not allowed VFOs, only crystal control at that time. True, surplus crystals were cheap back then, but still expensive to a high school kid like me. Also, the 15 meter band was added to the amateur spectrum at the time 11 meters was given to the CB band, in September 1958. In May 1952, CW was allowed on the new 15 meter amateur band, after the introduction of the AT-1. Heathkit possibly did not consider it wise to retrofit the AT-1, since limited functionality could be had by setting the band switch to 10 meters, and tuning up on 15 meters with the appropriate crystal (using the 6AG7 to double a 5 MHz xtal to 10 MHz, and the 6L6 to double it again to 21 MHz). My VF-1 VFO has already been modified to output on 5 MHz, using the defunct 11 meter position. I have another rig which does nicely on 30 meters that way. Simply padding the 11 meter trimmer with more capacitance affords the necessary output frequency. Of course, the VFO dial has no calibration, so a frequency counter is necessary. I add a BNC connector on all my vintage transmitters for this essential monitoring feature. The DX-20 was introduced in 1956 as a replacement for the AT-1. It had a 6DQ6 in the final and 50 watts input; I get around 35 watts out on 80, 40, 20, and 15, much better than the AT-1, even with the McCoy mods.

It may be possible to obtain 30 meter (10.1 - 10.15 MHz) output by using 5.05 - 5.075 MHz crystals or a VFO if you set the AT-1 band switch to 20 meters. I have yet to verify this, but it could be a "bonus band." Here is a summary of the required master oscillator (VFO or crystal) frequencies to get various ouput frequencies on the AT-1. Remember, the 6L6 stage MUST multiply by two as a frequency doubler. It is too unstable to operate straight through on any band but 80 meters, where there is no tuned circuit, only an RF choke in the grid circuit. In fact, the ratings of the 6L6 are specified only to 10 MHz in RF class C service. Class C RF 6L6 tube data

Any attempt to use 3500 KHz crystals on 10 meters will likely not produce sufficient useful output. Any attempt to use 7000 KHz crystals for 15 meter output will fail to conform to the multiplication scheme used in the Heathkit AT-1.

Anyway, the operation of the refurbished AT-1 will become evident as this project unfolds. Step 1 is getting all the burned wiring, cold soldering joints, and sloppy modifications OUT and back to stock condition. Data will be provided on the stock rig, as a bench mark. One of the web authors claims 12 to 14 watts with a stock AT-1 and an AC-1. That might just be good enough to satisfy me. I got 18 watts on 80 meters with my homebrew 6L6 rig, so maybe I can squeeze a little more out than he got, by running the final straight through, not doubling, on all bands. If the modified AT-1 works properly on 160, 80, 40, and 30 or possibly 20, I will call it good enough.

There is a nice restoration article with a lot of good photos at: http://webpages.charter.net/aj8mh-radio/heathkit.html.

Here is another good historical article with a front panel photo: http://p1k.arrl.org/cgi-bin/topdf.cgi?id=99969&pub=qst.

Here are a couple really good articles on the AT-1 transmitter and the AC-1 tuner accessory.

https://www.w6ze.org/Heathkit/Heathkit_025_AT1.pdf
https://www.w6ze.org/Heathkit/Heathkit_013_AC1.pdf

It is important to realize that the Heathkit AC-1 tuner is designed for one specific antenna, an end fed wire. Any other antenna such as a dummy load or a standard dipole will not light the neon bulb. The bulb is not defective; it needs about 70 volts to fire. This implies a higher impedance than 50 - 70 ohms. You will need to devise a field strength meter or other means to establish proper tuning, if indeed the AC-1 tuner works with your particular antenna at all. The link coupling of the AT-1 worked a lot more efficiently with the AC-1 antenna tuner and the antenna it was designed for.

I finally got around to working on this project, just in time for Novice Rig Roundup 2024. In addition to the AT-1 shown above, I acquired another in better shape (unmodified for the most part) and decided to refurbish that one using parts from the earlier one. At least it had the correct meter. Fortunately, the original meter hid the extra holes from the wrong meter. The black "Collins style" knobs were used because that was what the first release of the AT-1 used. These later were gray knobs similar to those found on the "DX" series like the DX-20. This front panel did not have visible extra holes and the paint was in pretty good shape.

Step 1 is to test the power transformer using a variac with a voltage and current meter. Alternatively, the inexpensive "kill a watt" plug style meter available at Home Depot can provide the metering function. Use a series incandescent 60 watt light bulb to limit the current if you do not have a variac. With the 5U4 unplugged, no B+ will be generated, so no load there. With the 6AG7 and 6L6 unplugged, there is no filament drain. The AC current drawn by the transformer alone should be nearly zero. If not, the transformer may be defective, and the project is not viable unless a replacement can be obtained. This saves a lot of wasted time and expense on a project to do this evaluation first.

Step 2 is to just get the chassis wiring back to agreement with the schematic. I use a copy of the schematic and a yellow fine tip marker to trace all wiring. These were kits and subject to mistakes by the original assembler. Also, equipment of this vintage is often modified to the extent that it is impossible to figure out what is going on during troubleshooting. I remove all the mods and start with the orignal setup, while preserving the parts and notes on how to reinstall it if it appears to be a viable mod. Many are not. More on this later.

A terminal strip on the 5U4 socket mounting screw broke off, so I decided to install a new one mounted to the main transformer screw, along with a 2 amp fast blow fuse mounted in a clip. Also there was a hole in the chassis for an unwired rca jack; this was repurposed for an AC on/off switch, using a step drill. This routes the AC wiring away from signal wiring, and the chassis has plenty of space for this improvement.

The AC on/off switch was moved to the rear panel to allow the front panel slide switch to be repurposed for the VFO/XTAL switch. While I do have a fairly good "rock collection" it is still nice to be able to use the VF-1 without rewiring the AT-1. The AT-1 manual says to ground the 6AG7 cathode for VFO operation. This removes the cathode bias resistor, not a good idea if excitation is lost. I prefer to make an AC ground via a 0.01 uF capacitor from the cathode to the VFO/crystal switch, which is grounded for VFO operation. This means you can now move easily from crystal to VFO operation without opening the cabinet and swapping wires. Putting it on the front panel makes it convenient to use, and does not add holes to the front panel.

The VF-1 was released a year later, and designed to be used with the AT-1. The 11 meter position can be repurposed to 15 meters for the AT-1. The AT-1 requires the final to double, so a 7 MHz basic oscillator will not work, since 21 MHz is an odd harmonic. Install a 91 pF dipped mica capacitor across the VF-1 trimmer for 11 meters (approximately 6 MHz.) This will result in a range of approximately 5.25 - 5.35 MHz when set to 11 meters, nearly all of the 15 meter band except the extreme top end.

Of course, the original filter capacitors should be replaced, using the original terminal strips. I upgraded from 8 uF to new 22 uF 450 V 105 degree centigrade capacitors; the higher temperature rating gives added life at minimal cost increase. I added equalizing resistors across the input capacitors, two 180 K 2 watt resistors. This keeps the voltage across the input capacitors equal and does not need to be a lower resistance, since the original 47 K 2 watt resistors are the bleeder resistors to discharge the power supply after turning power off.

When you complete the step of capacitor replacement, test the resistance to ground from 5U4 pin 8 to ground. It should be about 75 K ohms. Don't forget to install a jumper between pins 3 & 4 of the MODULATOR INPUT octal socket on the rear panel. This can either be a permanent wire inside the transmitter, or if you wish, a plug as originally designed. I don't think I will ever use this with a modulator, but it is nice to unplug the jumper when troubleshooting grid drive problems (without stressing the 6L6).

The STANDBY switch has no arc suppression across it. Not only does this wear the switch, a study in Electric Radio magazine found that the practice of switching the center tap connection to ground caused fatigue of the transformer insulation eventually in the DX series Heathkit transmitters. In the DX-20 and subsequent rigs, Heathkit installed a series RC network across the switch. Electric Radio magazine #398 July/August 2023 page 42 had an article by W4AMV suggesting 56 ohm in series with 0.1 uF 600 volt will reduce the spike in the AT-1 dramatically. I had some snubber components of that value (in a single package) which were installed to protect the transformer insulation and switch. Both slide switches were good, and did not even need cleaning. Previous articles in this series appeared in ER #382, #384, and #388. You can get reprints from: https://www.ermag.com/product-category/back-issues/?product-page=2

DESIGN NOTE: One may ask why this transformer failure mode was introduced. Heahkit probably specified a transformer design which stated full wave center tap service. They may not have told the manufacturer that the center tap would be switched, introducing transient spikes of 400 volts or more; the manufacturer, assuming the center tap was always at ground potential, did not provide sufficient insulation for switching the tap. In your home brew projects, it would be preferable to provide a separate filament, low voltage B+, and bias supply, and switch the AC primary voltage to the high voltage B+, on a separate transformer. This should also be taken into account when repurposing old transformers with a center tap for full wave bridge service. This configuration is often found in the popular "economy" designs popular with home brewers, which provides dual B+ voltages, one for low level stages, and one for the finals RF stage.

I relaced the original bare wire to the output SO-239 coax connector with shielded coax wire to reduce coupling to the AC wiring. The original mica bypass capacitors from either side of the line to chassis were replaced with modern type X or Y capacitors for safety. These capacitors are designed for this service, and the original capacitors in vintage equipment can become leaky, causing a shock hazard from a hot chassis.
https://www.mouser.com/ProductDetail/KEMET/C971U472MVWDBA7317?qs=xIpY4gsFVQGg6pR1Tc31Cg%3D%3D

It is possible to polarize a non polarized AC plug by simply using a pair of cutters to open the hole at the plug, and spread it to prevent the plug from being inserted in the AC socket in the incorrect orientation. I do this on all my old projects, or install a modern 3 wire AC cord.

The meter switch was damaged and replaced by someone with a ridiculously huge switch. This was corrected, and also prevents damage when putting the radio back in the case as the switch moves past the front lower lip on the cabinet. Don't forget that there are actually 3 positions of this switch. GRID, OFF, AND PLATE. This allows you to use OFF to eliminate the annoying bouncing of the moving vane style meters of this era. This is a good idea to prevent wear on a better D'Arsonval movement; the Johnson Viking II and the DX-100 had an off position. Now you know why.

The original Heathkit design did not key the 6AG7 cathode. Some people rewired this to key it with the 6L6. While this could provide break in operation with the proper T/R switch (such as the Johnson 250-39), it could lead to chirp. The 6AG7 oscillator and the VF-1 VFO provided a load on the B+ to prevent it from soaring during key up conditions. When the key went down, a dramatic shift in B+ could cause chirp, even when using crystals. I rewired the cathodes to the original Heathkit design so that the 6AG7 and VF-1 ran whenever the STANDBY switch was set to operate or the B+ was on. An external relay style T/R switch will be used with this AT-1 to switch the antenna and reduce gain in the receiver to use it as a keying monitor. For a break in system see:
https://wireless-girl.com/Projects/TRswitches/Johnson/

While wiring the cathodes, insert a 27 ohm 5 watt wirewound resistor in series with the 6L6 cathode connection to the keying jack. I found that if excitation fails, the current drawn by the 6L6 without grid leak bias (which is derived from the RF grid excitation) is excessive. Also on 80 meters in particular, the grid drive is insufficient to produce class C operation and properly bias the 6L6. The added cathode resistor keeps the 6L6 safer without any noticeable reduction in RF power output. The original 0.5 uF paper capacitor in the keying click filter is always leaky, and needs replacement.

Everybody loves to hate the cheap bouncy moving vane meters popular in this era. Replacing them was the most popular mod. I found that the maximum output did not always coincide with the plate meter dip, and use an external power meter to adjust for resonance. But you can do something about the bounce. I installed a 1000 uF electrolytic capacitor (+ end to the B+) across the 51 ohm 1 Watt resistor on the meter switch. A different value is needed for the grid circuit; I used a 100 uF electrolytic capacitor across the 4.7 K 6L6 grid resistor (+ to ground, the grid is negative). This does not eliminate the problem, but it does reduce it a lot.

These moving vane capacitors are not very accurate. I found my meter read 15 mA full scale. There was nothing I could do for the GRID current correction, other than noting that the meter was nearly 50% high. To perform these tests, unplug the AC power and discharge all filter capacitors. Your meter may be different. Use a 12 VDC variable power supply and a modern digital multimeter set to current function to check the meter. To test the basic meter function, use a 1 K series resistor with the power supply; start with the lowest voltage and bring it up slowly. This tells you the error in basic movement and the GRID function. You cannot correct this error with the 4.7 K resistor, because the meter resistance is about 500 ohms.

Use a 100 ohm series resistor to limit current, and start with minimum voltage. Connect it with the + lead to the MODULATOR socket pin 4 and the - lead to 6L6 pin 3. For the PLATE meter function, I changed the original 51 ohm resistor to 68 ohms.

If you are not comfortable doing these meter tests, just leave it alone and use an external power meter to peak OUTPUT tuning for maximum output. Same thing with the grid function, just peak the DRIVER knob for maximum GRID current. You might not get any grid current on 80 meters because the grid circuit is untuned on that band. Set the DRIVER capacitor to fully open (10) for that band only.

Lastly, it is necessary to have realistic expectations for power output for the AT-1. I built a similar 6L6 rig under contract for someone that produced well over 15 watts, with an untuned grid circuit, and a PI network output, for 80 meters only. The 6L6 is not well screened and prone to oscillation. It is not rated for full output above 10 MHz. See 6L6 RF data I researched which you cannot get anywhere else:
https://wireless-girl.com/Projects/6L6classCData.html
Moreover, the 6L6 is operated as a doubler on all bands, to alleviate oscillation. This results in less efficiency, probably around 50%. The normal efficiency for class C is maximum of 70%, under optimum conditions. The AT-1 does not operate with sufficient grid current to achieve class C in some conditions, particularly 80 and 15 meters.

This problem is exacerbated by the use of link coupling rather than PI network coupling in the output. The coupling is not adjustable for antenna impedance, which varies with frequency and SWR. Link coupling is also less efficient. That is why it is largely replaced in transmitter designs from the early fifties onward. I have found Pi networks for transmitter outputs (rather than the traditional link coupling) explained and used in Radio Handbook (West Coast) editions 11 & 12 for 1947 and 1949, though that handbook was addressed to an audience with better than average technical skills. The Pi network has adjustable coupling for more output; the popular link output coils had to be experimented with multiple times to get full output. That is why in more modern designs, link coupling fell out of favor. Bill Orr was always a leader. I experimented with a piece of plastic pipe wound with an antenna link coil to see if I could improve output. I got maybe 11 watts instead of 9 watts on 80 meters and abandoned further modification efforts. The coupling could be varied by inserting the pickup coil into the 80 meter output tuning coil.

The AC-1 antenna coupler, which cost nearly as much as the AT-1 was offered to mitigate this problem. A complete redesign would be required to fix this problem. Due to the use of frequency doubling, a PI network (which is a LOW PASS filter) might have been questionable for spectral purity, though some attempted that in ill advised modifications.

If you are an ARRL member, you can access old QST issues on line at their web site. In QST October 1955 the legendary McCoy published a mod which replaced the 6L6 with a black and white TV sweep tube (the 6BQ6) and applied neutralization. This increased the AT-1 power output from approximately 9 watts to around 24 watts on 80 - 15 meters. That is a gain factor of 2.56 or about 4 dB, one S unit. Really, is all that drilling and blasting on a classic transmitter worth it? Not in my opinion, so I went with the original Heathkit design. The AT-1 is in the QRP class of transmitters, but keys cleanly and without objectionable chirp, even with a VFO, when restored as I recommend here.

A mod which employed a more available 807 might have been better. However. the layout of the AT-1 is terrible. The final grid circuit is an inch from the RF output. The lead from the 6AG7 output is a bare wire that goes all the way across the chassis (8 inches!), invites VHF oscillation, and further increases coupling between input and output of the RF stage. This further complicates any attempts at neutralization. Any reasonable modification would require a new blank chassis and use only the AT-1 cabinet. That is not a modification, its a whole new transmitter design. A comparable 807 rig with proper layout is the Johnson Adventurer. Note in the photos how closely the 6AG7 and 807 sockets are located, and the driver grid tuning is located in a separate shielded compartment. This is why any major modification of the Heathkit AT-1 is futile.
https://wireless-girl.com/Projects/AMTransmitters/JohnsonAdventurer/

Here are photos of my work, with explanations for wiring the modifications I suggest.

NOTE jumper plug required in MODULATOR connector

AC power switch relocated to rear panel

I replaced a lot of original plastic coated wire

AC wiring and new fuse detail

6AG7 VFO switch 0.01 uF
and 1000 uF meter damping

100 uF GRID meter damping
27 ohm 6L6 cathode resistor located near keying jack

STANDBY snubber capacitor, 180K equalizing resistors

Finished Heathkit AT-1 and VF-1 restoration

Removed parts, test output link assembly

Schematic for trace and modification

I am using a new toy, a Siglent SSA 3021X spectrum analyzer to characterize the spurious output of my projects, after I complete them. This data should be relevant to any stock unmodified AT-1 with the matching VF-1 VFO. Keep in mind that the VF-1 has three ranges: 160-80-40 with a fundamental output of 1.750 MHz to 2 MHz, 40-20-15-10 with a fundamental output of 7 to 7.4 MHz, and an "11 Meter: range which I have modified to produce 5.25 to 5.35 MHz output which the AT-1 can multiply be 4 to get 15 meters or 21 MHz. It will soon become obvious that a VFO with an 80 meter or 3.5 MHz fundamental output would produce less spurious output when operating the AT-1 on 80 and 40 meters. The Heathkit HG-10 or Hallicrafters HA-5 would be good choices with that specification.

The current FCC Part 97 rules state:
"§97.307 Emission standards (d) For transmitters installed after January 1, 2003, the mean power of any spurious emission from a station transmitter or external RF power amplifier transmitting on a frequency below 30 MHz must be at least 43 dB below the mean power of the fundamental emission. For transmitters installed on or before January 1, 2003, the mean power of any spurious emission from a station transmitter or external RF power amplifier transmitting on a frequency below 30 MHz must not exceed 50 mW and must be at least 40 dB below the mean power of the fundamental emission. For a transmitter of mean power less than 5 W installed on or before January 1, 2003, the attenuation must be at least 30 dB. A transmitter built before April 15, 1977, or first marketed before January 1, 1978, is exempt from this requirement."

The purpose of this part of my restoration page is to alert users of this vintage equipment to the potential for interference from spurious emissions from this vintage gear. While the FCC has "grandfathered" this old equipment, it may attract unwanted attention. One method to correct the spurious emissions would be to build a fancy band pass filter. A cheaper and simpler solution would be to use a Johnson Matchbox to couple the AT-1 to the antenna. This would add a tuned circuit and link coupling and reduce spurious output another 10 - 15 dB or more. Use of monoband resonant antennas might also help. The Heathkit AC-1 Antenna Coupler is an L network in a low pass configuration that might reduce spurious emissions ABOVE the operating frequency, but will not help below the operating frequency. That is the definition of "low pass" filters.

Some of the photos are a bit fuzzy, so I provide a text analysis of the screen shot shown. I need to learn more about my new toy to see if I can do screen grabs that are sharper. In the meantime, here is the data. All spectum photos are with a sweep from zero to 30 MHz, except for the last one, showing spurs out to 100 MHz.

For comparison to a rig with a PI network output, see my DX-20 page, which has spectrum analyzer data also.
https://wireless-girl.com/Projects/AMTransmitters/HeathkitDX20.html

80 METER SPECTRUM with 1.75 MHz VFO

40 METER SPECTRUM WITH 1.75 MHZ VFO

40 METER SPECTRUM WITH 7 MHZ VFO

20 METER SPECTRUM WITH 7 MHZ VFO

10 METER SPECTRUM WITH 7 MHZ VFO

15 METER SPECTRUM WITH 5.25 MHZ VFO

10 METER SPECTRUM AGAIN, EXPANDED SWEEP

I repeated the measurements with a Hallicrafters HA-5. I did not have a Heathkit HG-10, but it would probably be even cleaner, since it does not have crystal heterodyne frequencies. The HA-5 is a heterodyne VFO with a 5 - 5.5 MHz variable oscillator and a crystal controlled mixer. In most cases, a dramatic improvement is shown. Similar results would have been seen using 80 meter crystals for 80 meters and 40 meter crystals for 40 meters and up (contrary to Heathkit instructions). The grid current will then be low on both 80 and 40 meters, but with the 27 ohm safety cathode bias resistor I added, it seems to work and power output is still around 9 - 11 watts nominal on 80 & 40. NOTE: Items with asterisk * are probably by products of the crystal oscillator or mixing scheme which would not be present with the Heathkit HG-10 or crystals.