Restoring my Atwater Kent Model 20 Compact
(And putting it back into working order)

Local links:
Thomas the Cat listening to
                The Coil's Voice
Thomas the Cat listening to "The Coil's Voice"
Shown is an Atwater Kent Model 20 Compact (Type No. 7570) and matching Atwater Kent model H horn-type speaker.
(Since it is impossible to make a cat pose for a picture, I'll try again to get a better one... some day...)

Several years ago, I was in a local antique store.

I'll admit it:  I was looking for radios.  That is, in fact, about the only reason that I ever wander around an antique store.  Generally, I'm turned off by $20 soft drink bottles, empty soda cracker tins, and very ragged teddy bears sporting price tags bearing numbers more resembling lottery winnings.

I was rather surprised to see, on this particular day, an Atwater Kent Model 20C receiver (along with a Model H matching horn speaker) sitting there.  The chassis was quite dirty and rusty and the finish on the cabinet was in pretty rough shape - but the speaker looked reasonably good:  Nothing obviously broken or missing.  I saw the price, gulped a bit, and managed to jaw down the proprietor by a significant amount  - too easily...  I probably paid a bit too much for both (the speaker may be worth more than the radio) but it wasn't unreasonable.  I wasn't even fazed by the description tag that said "Old Radio - Does Not Play."

I don't have a very large collection of old radios but I like to have them around and in working order.  What about replacement tubes?  As it turns out, many hundreds of millions of vacuum tubes were manufactured in the 20th century and there are still a lot of them around.  Sure, the prices are creeping up year after year, but rare is the tube that costs more than if its original purchase price were simply put in a bank account when the tube was made and allowed to gather interest.

After getting the radio and speaker home, it was clear that the radio needed quite a bit of work:  Much of the cotton insulation on the internal wiring was missing having been nibbled away by mice and/or insects, there were a lot of spots on the chassis that showed rust instead of paint, and of the four type 01A tubes that came with the radio (it needs five) three of them were good.

What happened to Atwater Kent anyway?

In the 20's and 30's the Atwater Kent name was virtually synonymous with radios.  Millions of radios were produced during this period bearing the Atwater Kent name. 

The depression put a heavy damper on radio sales:  Even an inexpensive radio represented a very significant investment to the working (or non-working) man and the higher-end Atwater Kent radios seemed out of reach for many. 

After several cost-cutting measures, Mr. Kent decided to simply close the business in 1936 and retire rather than produce (what he considered to be) inferior products.

Fixing the speaker:

I cleaned up the radio a bit (removing the dead bugs and most of the mud) and decided to concentrate on the speaker.  The horn speaker (a Model H) operates in a way very similar to high-Z dynamic headphones in that a pair of high-impedance coils vibrate a steel diaphragm at the base of a horn to produce the sound.  A quick check with an ohmmeter indicated that both of the coils were open.


One of the adjustments for the speaker is to vary the spacing between the diaphragm and the pole pieces of the electromagnets by adjusting the threaded coil assembly in the base of the speaker.  Unfortunately, this adjustment was "frozen" - not much of a surprise considering that this assembly was constructed of cast aluminum - a material prone to oxidation and galling.  I soaked the threads for several days with penetrating oil and tried again.  Still, it wouldn't budge.  It was now time for the proverbial "bigger hammer:"  Carefully clamping the speaker in place and using a large set of adjustable pliers (with some rags in the jaws to prevent gouging of the metal) I tried again - and this time it came loose...  along with a piece the casting.

"Drat," I thought.  Actually, it wasn't as bad as it originally looked:  Only about 60 degrees or so of the side of the threaded part broke loose - and it broke cleanly.  I cleaned out the threads, degreased the pieces, and then used some gray, metal-filled epoxy ("JB Weld") and carefully put it back into place.  The repair was successful and unless you know what to look for, you really can't see it at all.


Atwater Kent model H horn
The Atwater Kent Model H speaker.
This is a high-impedance horn-type speaker intended for "plate output" radios such as the Model 20.

The Voice Coils:

Now, to the coils:  Again, there were two of them wired in series, and both of them were open and a bit of correspondence with another old radio collector indicated that this is a common problem.  The coils themselves were wound with extremely small-gauge wire (#40 A.W.G. perhaps) and thicker wire (#26) was used for the "end" connections.  I carefully unwound the insulation to where the end connecting wires attached to the coil wires.  The solder connection was badly rotted - an apparent galvanic reaction between the flux, moisture, the copper wire, tin, and lead of the solder.  A check with the ohmmeter indicated that the "inner" connection (the inaccessible one near the middle of the coil) was also bad.  There was little hope of salvaging these coils so I set them aside - and decided to look for a workable replacement.

Pouring through the Antique Electronic Supply catalog I spotted some "High Impedance Headphones" in the catalog.  Having seen these things before, I knew that they would have coils similar to these in the horn speaker - so I ordered a pair.  One of the earpieces (each earpiece contained two coils) was dissected immediately upon arrival and only a very slight modification was required to the iron pole pieces (they had to be made slightly narrower) to accommodate the new coils.

As it happens, Antique Electronic Supply no longer sells these high-impedance headphones, but similar headphones are readily available on Ebay - although it does seem to be a bit of a shame to sacrifice a perfectly good pair of vintage headphones!  When it comes right down to it, however, working headphones are far more commonplace than working Atwater Kent horn speakers!  If you don't want to do this, please note again the link above to the company that may be willing to re-do the coils for you!

Recharging the magnet:

Now, I needed to "recharge" the magnet.  The aluminum base assembly contains a large soft iron pole piece - apparently used to provide a magnetic "bias" to the electromagnet and the steel diaphragm. Over the 75 years or so since it was made, it had lost nearly all magnetism - barely showing any evidence when a compass was placed nearby.  This was "recharged" with the aid of a bank of 12 volt deep-cycle batteries that I keep float-charged for emergency power:  About 30 feet of #10 copper wire was wound into a 1-foot diameter coil, the iron pole pieces were placed inside, and the free end of the wire was connected to the battery bank momentarily.  (Do not try this at home - and if you insist on it, make sure you place your wallet and credit cards at some distance from where you are working.)  This caused several hundred amps to flow, creating an extremely strong magnetic field in the coil in which the pole piece was placed on and duct-taped in the proper orientation to a piece of wood - to keep it from flying away when the field was excited.  Several 2-3 second jolts were given (allowing the wire to cool between jolts) resulting in a highly magnetized piece of iron.

In "normal" operation, the plate current from the audio output stage of the receiver flows through the magnet, magnetically "biasing" the diaphragm very slightly.  For this reason, it is important that attention be paid to direction of current flow in the coils and the polarity of the pole pieces. 

It has been suggested that "recharging" the magnet in this way wasn't really necessary as the plate current through the coil would have been sufficient to magnetize the pole pieces:  They may be true, but making loud noises and big sparks is fun!

Comment about using permanent magnets to "re-polarize" the speaker's magnet:

The guts of an Atwater Kent model H horn
                      speaker - showing diaphragm, coils, magnets, and
The "guts" of an Atwater Kent Model "H" horn-type speaker arrayed in the bottom of the speaker.
In the center is the "bottom" of the horn with the "O" ring.  Starting in the 3 o'clock position and moving clockwise is the voice coil assembly, the Speaker Diaphragm, the spacer/washer, and the spring.   In the middle one can see the O-ring used, but not shown is the ring of foam rubber also used.

The Speaker Diaphragm:

Finally, there was the metal diaphragm itself:  It had rusted nearly through.  Looking in my junk box, I saw a piece of tin-plated steel shielding on a VCR circuit board and after verifying that its thickness was nearly identical to the original piece and spending a few minutes with a pair of scissors and using the old diaphragm as a pattern, I had a new speaker diaphragm!

Final Reassembly and testing:

A few "dial-versus-frequency" settings from my AK 20C

While every AK 20 is different, the chart below will give you a general idea as to the tuning position for various frequencies.  The three number correspond to the three dials on the front of the radio as determined off-air from local stations and using test gear.
Frequency (kHz)
Dial positions (left-to-right)             
518 (approx.)
100 (fully-meshed)
As you can see, the tuning is very non-linear and the extreme top end tuning range of the radio is a bit below 1500 kHz.  As you get higher in frequency, stations become increasingly difficult to separate from one-another - a trait of "TRF" receivers such as this!

In general, the 3-position antenna coupling switch is CCW (counter-clockwise) for larger antennas and better channel separation while CW for smaller antennas and weaker signals.  Note that the setting of the first dial can vary slightly with a change of the antenna coupling switch, particularly near the top frequency end.

The order of assembly, starting at the horn, is:  Rubber gasket (I used an O-ring,) Diaphragm, another rubber gasket, washer, spring, and (finally) the coil assembly.

Using a compass, I made certain that both electromagnets were wired to exert the same polarity of magnetic field when current was passed through them.  I then installed the soft iron pole piece so that it exerted the same polarity of field as the electromagnets when current (in the polarity marked on the speaker's terminals) was passed through them.  I then examined the long-since-deteriorated pieces of rubber that had been used in the speaker for spacing the diaphragm and determined that one piece could be replaced with a rubber "O" ring, and the other could be replaced a piece of closed-cell foam rubber.  If I had been more diligent, I could have found two "O" rings that were of the appropriate sizes and avoided using the piece of foam rubber.

Audio was applied to the speaker - using a step-up transformer for better impedance matching - and the coil was adjusted for best sound.. (Note:  For this test, no DC bias current was applied, as it would be if it were connected to the radio.)  If you adjust the coil assembly in too far, it touches the diaphragm (and sticks with a "click") and you get almost no audio.  If it is too far out, one also gets very little audio as the feeble magnetic fields can't work their influence on the diaphragm.  There is a very narrow range where the coil is just close enough to the diaphragm that it works - but far enough away that magnet doesn't cause the core of the coil and the diaphragm to stick to each other.

As you might expect, a horn-type speaker does not produce hi-fi audio, but rather audio that is just "fi."  It is reasonably loud and very listenable even in a good-sized room, but its frequency response is not unlike what you get from a telephone.  With the directly-heated filaments of the type 01A tubes, the radio only takes 1-2 seconds to warm up - fast enough to make you think that some solid-state sleight-of-hand was involved!
Important notes on connecting these types of speakers to these antique radios:

It should be noted that some speakers simply won't work right at all if connected "backwards" while others hardly mind at all.  If you have a speaker with a soft-iron magnet (e.g. an Atwater Kent type H or M speaker) and not only does the polarity of the connection not seem to make a difference - but also that the speaker seems "quiet" to you, you may need to remagnetize the speaker's main magnet!  You can use the technique described above to recharge the magnet, or you can adapt the technique used in the YouTube video, above using small, permanent magnets!  It is also possible that after a few hours of playing, the magnet will slowly magnetize itself owing to the DC current in the coil and the speaker will start working better.

Fixing the receiver:
"You idiot!  You got rid of the patina!!!"

I'm sure that some of the purists cringed when they read about "polishing the brass" and "stripping the chassis down to the bare metal."  Indeed, the "aged" look is one that is prized and well sought-after.  (And no, I don't consider it to look "aged" if it just looks like it was left out in the rain...)

Now, if an antique piece is in reasonable condition, I am disinclined to do much to it along the lines of refinishing.  However, if the condition of the piece is poor because of neglect and its condition can interfere with its operation, then I will go through some trouble to rebuild it - even if that means (dare I say it) refinishing

In this case there was quite a bit of surface rust on the steel chassis and damage to the wiring - so I stripped it down to the bare metal and repainted it. 

The mahogany cabinet fared somewhat better:  The finish on the top cover was in somewhat poor shape - but some work with some finish restorer and steel wool  removed the worst of it.  A few coats of old-fashioned paste wax helped blend in the "rough" spots and make the cabinet presentable. 

(Anyway - it's my radio!)

Click here for the schematic of this radio.

I now had a working speaker but I did not yet have a working receiver.  To remedy this, needed some replacement tubes (at least two of them) - of type 01A.  This is a very early triode - a type that was essentially obsolete by the end of the 1920's.  Fortunately, huge numbers of them were manufactured so they are still readily available, with good, used tubes being in the $8-$16 range. 

At this point, the receiver (and speaker) sat around for about 2 years before I did anything with them...  And suddenly, in February of 2001, I decided to resume work on the receiver.  Upon examining the receiver closely I was reminded why I had initially put off the restoration project:  There was an awful lot of surface corrosion on the steel chassis, the brass rings on the tube socket  were badly corroded, there was some visible corrosion on some of the capacitor plates, and much of the original cotton-covered wiring was missing its cotton covering.

First -  disassembly:

Hmmm...  This was going to require a far more complete "restoration" that I had anticipated.  Fortunately, I'd had the foresight to order some wax-impregnated cotton-covered wiring from Antique Electronic Supply.  This wire was "New-Old Stock" - it had (supposedly) been manufactured in the 1930's and been sitting around for many decades in a warehouse.  They also had some replica "speaker wire" (extremely flexible cotton-insulated wire with fine wire strands woven into it) and some replica "Atwater Kent Battery Wire" (cotton covered cable with multiple conductors) inside - and I got enough of this to provide power and speaker connections and use pieces of it for interconnections.  I also obtained some paint, some very fine-grit sandpaper, and fine steel wool.

Some careful drawings were made showing component placement (there aren't very many components in these radios, fortunately) as well as hardware orientation, wire dressing, and a few other relevant details.  I also took some photographs in case the details of the drawings proved to be inadequate.

Then came the "disassembly."  When I was done, I was left with what was, for all practical purposes, "Model 20C kit."  I stripped the chassis down to bare metal, cleaned contacts, and polished the socket rings.  I even went to the trouble of completely disassembling the tuning capacitors (one unit at-a-time, so I'd have at least one properly assembled unit to look at for comparison) and at one point, I had piles of plates and washers on my workbench.  I then went about cleaning corrosion off the plates, putting the worst plates "inside" the stack where no-one can see the pitting, and carefully reassembling the capacitors.  A dab of silver-conductive grease was put on the rotor's electrical contact point to assure the best, noise-free contact possible.

About conductive epoxy and grease

A few readers have emailed asking about the conductive grease and epoxy that I used.  Well, here is the info: 

Both products are made by Chemtronics - a company easily found via  a web search.  A page with information on these two products may be found here

These products are occasionally available locally at a well-stocked electronics distributor - as well as by mail order.  Mouser Electronics (1-800-346-6873) also carries these parts as follows: 

The conductive grease is Mouser P/N:  5168-7100  and it is $47.20
The conductive epoxy is Mouser P/N:  5168-2400 and it is $111.66
(2012 prices - which have WAY more than doubled since 2006!)

A much less-expensive alternative to the conductive epoxy (which has a relatively short shelf life) may be found in a "Rear Window Defroster Repair Kit" that is probably available at your local auto parts store.  This is a fairly rugged conductive paint that can be used to make permanent twisted wires or mend intermittent connections between dissimilar metals. 

Note:  I have no connection to Mouser - It's just a place to get things...


The chassis was wiped down, degreased, and painted with a glossy black lacquer.  This first coat - both interior and exterior to the chassis - was for complete coverage and resulted in a very shiny, glossy finish - most unlike the original finish.  The interior chassis was "dusted" with a coat of lacquer to break up the gloss.  The front panel was then similarly "dusted" with five coats of paint resulting in a matte-like finish not unlike the original.  The "final" coat was a "dusting" of brown lacquer to "lighten" the finish to make it more closely resemble its original hue.

Using the drawings, a schematic, and the pictures, I carefully reassembled the receiver - checking each individual component to make sure its values were reasonably close to those stated in the schematic. Fortunately, none of the components were bad - even the grid leak resistor (a common component to fail) was within its 2-4 megohm range.  The only component to require any sort of repair was a 600 ohm wire wound "grid resistor" that was in series with the grid of the second RF stage:  The fine-gauge nichrome wire was broken mid-resistor.  Unwinding a turn or two, twisting the ends, and a small dab of silver-conductive epoxy restored this resistor's function.

During restoration I noticed that the 1st AF transformer wasn't the original.  It had clearly been replaced a very long time ago as its surface corrosion was similar to that of the chassis - and it was a "DeForest" brand.  It would be interesting to know the story behind that!


Before I could test the receiver I needed a source of power.  This receiver was originally designed to be powered by a bank of batteries.  Since that battery bank had a barrage of cells, multiple tap points were available for different voltages.  Thus, I was faced with buying and connecting a pile of batteries, or building a supply that provided 90, 67.5, 22.5, and -4.5 volts at low current, and 5 volts at at least 1.25 amps.  I chose the latter.

A page describing the power supply may be found here.

Once the supply was tested, it was time for the radio's smoke test.  I was more surprised than anyone when the radio worked the first time.

After some minor adjustment to the radio's tuning capacitors plates (so they tracked each other properly) I was able to tune the radio as outlined in the manual, being able to hear radio stations every 10 KHz on the dial (it was night-time...) once it had been connected to a rather modest random-wire antenna and a ground.
Two versions of the Model 20 Compact

There were at least two versions of the Model 20 Compact produced.  The early version was the No. 7570 and the later version was the No. 7960. 

These two units were pretty much identical except for these points: 

- In both radios, the left-hand rheostat adjusted the filament voltage of the two RF amplifier tubes. 

- In the 7570, the right-hand rheostat (10 ohms) adjusted the filament voltage of detector tube and the two AF stage tubes. 

- In the 7960, the right-hand rheostat (20 ohms) affected only the filament voltage of the detector tube and its bias. 

- A fixed resistance (1 ohm) was added on the 7960 to prevent full "A" voltage from being applied to the filaments of the two AF stage tubes (such as would occur with a new A battery) since the filament voltage of these tubes was not adjustable. 

- The 1st audio stage on the 7960 operated at zero bias and from the 67.5 volt supply, unlike the 7570, where both audio stages used the negative bias and operate from 90 volts.  The operating impedances are also different on the 1st audio stage of the 7960 as compared to the 7570. 

- The detector tube had a slight amount (2 volts or so) of positive bias on the 7960, from a "centertapped" 450 ohm resistor across the filament voltage. 

- The tubes sockets in the 7570 hale from the "breadboard" radio:  Flat contacts make contact with the bottom of the tube pins. A wide brass "ring" with a notch held the base of the tube in the socket using the bayonet-type pin on the tube's base. 

- In the 7960, more "conventional" tube sockets were used:  The tubes actually plugged into a socket and were held in place with the friction of the socket's contacts on the tube pins. 

The changes to the filament supply were probably made to improve performance of the AF stages:  Having "full" filament voltage probably allowed somewhat better audio quality - especially in the presence of strong signals - allowing the AF stages to work at "full power" while allowing adjustment of the detector's operating gain. 

The changes in the sockets were likely done as a cost-cutting measure:  The "showy" brass tube sockets used in the 7570 were hidden within the cabinet.  While the more conventional sockets don't "hold" the tubes in place as well (if the receiver were transported, for example) they probably provided more reliable electrical contact with the tube's pins.


Operating this receiver is decidedly more complicated than using a "modern" receiver (i.e. one made since the mid-30's) The most obvious difference is the presence of three tuning knobs.  This receiver is a TRF (Tuned Radio Frequency) type.  In this type of receiver, several selective stages (filters, actually) are used to single-out the desired signal.  This is different from the more (now) conventional Superheterodyne in which all signals are converted to a single IF (Intermediate Frequency) where simple fixed filter is used.

What this means is that all three TRF stages must be tuned to the same frequency - manually.  This also means that the receiver's selectivity will be worse on the higher frequencies:  This is because the selectivity of each stage may be thought of as a percentage of the operating frequency:  The higher the frequency, the more "width" that selectivity implies and therefore the "less able" the receiver is to differentiate closely-packed stations.

By comparison, in the case of an superheterodyne, the IF frequency is typically quite low and never changes, thereby avoiding the problem of multiple stages having to be precisely tuned and the selectivity worsening as you go higher in frequency.

The TRF receiver has another interesting property:  At the low-frequency end of the dial, the audio frequency response may be changed by the tuning of the dial - but at the top end, it can be difficult to separate closely-spaced stations.

There are also other operational differences:  Receivers such as these aren't nearly as sensitive as modern ones and fairly large wire antennas are required for good results.  Because of the nature of wire antennas, there is actually an "antenna coil tap" on the receiver to adjust how the antenna's energy is coupled into the receiver's first stage.

Being a simple TRF, there is also no AGC (Automatic Gain Control.)  In a "conventional" receiver the AGC is a circuit that detects the strength of the received signal and adjusts the sensitivity of the receiver automatically:  Sensitivity is increased for weak signals and decreased for strong ones, with the result being that signals of differing strengths have equal "loudness."  Because this receiver lacks this circuit, strong signals are loud and weak signals are... well, weak...  As distant signals fade in and out, they get correspondingly louder and softer.

Finally, this radio lacks much power in the audio output stage.  At the most, only a few hundred milliwatts or so of audio power can be applied to the speaker by the radio.  While the fidelity of the horn speaker isn't particularly good, it is actually quite efficient and directive, making pretty good use of the available audio power.

Despite all of these "deficiencies" (Remember:  The aforementioned circuits had barely been invented when this radio was made!)  this radio actually is very useable - even by today standards.  The tuning is somewhat awkward - but it takes only a little bit of experimentation to get used to it.  Its sensitivity is really perfectly satisfactory for any local radio station.  At night, even with a modest wire antenna, stations may be heard on every "spot" of the dial.

NDB DXing with the Model 20

It is fair to say that the Model 20 is no "DX Machine."  The measured sensitivity was found to be about 200 microvolts at 200 ohms (probably a mismatch - and it does have a high-impedance antenna input, remember!) but, at night, one can easily hear something every 10 KHz on the dial. 

One of the things to remember with the model 20 is that the dials are calibrated from 0-100, with 0 being the highest frequency (or, as the book states, the shortest wavelength.)  Because the dial is not calibrated in terms of frequency, the manual strongly recommends that the listener keep a log of dial settings for each station. 

Ideally, 0 correlates with approximately 1500 KHz (in the mid-to-late 1920's, the AM broadcast band extended only as high as 1500 KHz) and 100 correlates with something below 540 KHz - more-or-less. 

After tuning around for a while (with the radio connected to a 60 foot random wire and an earth ground) I noticed that I could even hear atmospheric noise between stations at the low end of the band. 


On a whim, I carefully tuned all of the dials to zero - then re-peaked them for maximum noise since the dials don't track each other absolutely perfectly.  Listening carefully, I could clearly hear the CW identification of the INE beacon from Missoula, Montana on 521 KHz at a distance of several hundred miles.  Not bad, considering that this beacon is reported to run 400 watts - a power level not unlike that of the "powerhouse" stations of the era in which this radio was manufactured.

Coming up in the future... eventually:  "Solid State" tube replacements: Sure, you can still get 01A's, but if you wanted to listen to your 20C every day, would you still use them?  (Now, I just need to figure out the solid-state equivalent of a filament and it's effect on tube gain...)

For more pictures, go to my Atwater Kent Model 20C gallery.

A few other Atwater-Kent and "vintage radio" related links:

Jim's Antique Radio Page has information on several radios, including technical information and owner's manuals covering the Model 20 here.

Note:  The links on Jim's page have been unreliable as of late - here are versions of those documents:

Please note that these have been converted to .PDF and some clean-up of formatting, OCR correction and layout has been done.  Note that in the above, links to the "original" versions are still included.  Thanks to Jim's Antique Radio page for originally making them available!


Go back to the main KA7OEI page

Any comments or questions?  Send an email!

This page copyright 2001-2014 and is maintained by Clint Turner, KA7OEI.  It was last updated on 20140217

Since 12/2010: