gray green ivory Regency white red black

Repairing a Regency TR-1 — my experiences

Most surviving TR-1s no longer work. Many collectors (such as myself) prefer to keep them in "as-manufactured" condition. That way the authenticity is preserved. But at some point, you may be anxious to hear one play. Before you start soldering, ask one question -- DO I REALLY WANT TO DO THIS? Once it's changed, it's no longer original, and you may end up with a "modified" radio, and still have one that doesn't work!

NOTE -- I'm presenting the information below to document what I did. Unfortunately I'm really not in a position to fix radios belonging to others. There are people who do specialize in fixing "antique" and collector radios and you should be able to track some down on the web or in newsgroups. But unless you have significant experience with electronics repair, I would think carefully about doing it yourself.

Getting Started

When I bought my ivory TR-1, I discovered that it had been modified. So I figured, "well, it's not authentic anymore anyway, so now I've got my 'experimenter' radio". The first clue was when I spotted a replacement oscillator/converter transistor, X1. The "new" one was a GE 2N168A, which dates back pretty far. It "wiggled" when I touched it -- a bad sign, indicating the foil PC board trace into which it was soldered was probably pulled away from the board. This happens when too much heat is used in soldering.

Collecting Documentation

I was going to need good documentation of how the radio should appear, where the parts go, and suggested schemes for doing the repair. Here's what I used:

Back To Top

Taking the Radio Apart

    1.   Turn the tuning dial fully counter-clockwise and then unscrew the knurled nut in the middle of the dial. Use a pliers and a tissue (to prevent scratching the dial) if you must, but try using just fingers first.

    2.   Take off the back.

    3.   Unscrew only the one flat-head screw that was under the tuning dial. Leave the other two screws attached.

    4.   Remove the chassis / PC board structure from the plastic case. Note the hole punched in the chassis (red arrow points toward it). Later you may have to use this access point to unsolder the board from the chassis.

Now comes the next major decision. Should the PC board be unsoldered from the chassis? Many parts, including the commonly-defective electrolytic capacitors, can be replaced without unsoldering the board, and for that matter, without unsoldering the part, if you're careful. To replace an electrolytic cap, you could just clip the leads of the old one and solder the new one to those leads. The result is not as "pretty" as actually replacing the cap on the board, but it may be less risky. But, if you need to remove the board ...

    5.   Straighten tab 1, and tab 2.

    6.   Unsolder (go easy with this) tabs 3 and 4. Pull the board away from the chassis as you do this, but be careful with the fine antenna wire that should remain attached.

    7.   Pull the chassis and PC board apart. The antenna wire and the volume control wires stay attached. Because you desoldered the two tabs, the equivalent electrical connection will have to be restored with clip leads between the PC board solder points and the chassis during repair and testing.

Making the Repairs

You can see the damage on the foil side of my PC board. I cleaned that up and replaced some traces with 30 gauge wire-wrap wire. Note that someone had substituted a green wire for a damaged trace.

The repairs to be made to the circuit depend on what's wrong, and following the PF Reporter article should track things down. My own feeling is that the likely culprits are probably the electrolytic capacitors. I found two of my four to be shorted, one open circuited, and the other one working properly. I replaced all four. The exact values are not too critical. For example, I replaced them as follows: 5uF/25V and 2uF/3V each with new 4.7uF/50V. And the two 40uF/3V with new 47uF/25V. Pay close attention to the polarities before removing the old ones.

I also discovered the on-off switch to be bad. This is just two pieces of metal that touch, so I took a screwdriver and used the edge of it to scrape away any dirt from the contacting parts of the switch. That fixed it. I also bent the battery clip a little tighter, and cleaned the "switch" contacts in the earphone jack.

Back To Top

There are various important test points in the radio. The photo shows connections to the AVC (Automatic Volume Control) line, IF (262 kHz Intermediate Frequency) signal, and audio before and after the AF (Audio Frequency amplifier) stage (at the top of the volume control, and at the speaker, respectively). The ANT and OSC trimmers are shown as I believe they are configured. The various coils are not shown, but the one without the metal can is the oscillator coil, and the others are the IF transformers.

Now, when it's fixed, put it back together. Be careful soldering the tabs between the chassis and the board -- it's easy to create solder bridges.

While it's possible for other parts to be bad, it's probably less likely. However, if "re-capping" the radio doesn't fix it, then the troubleshooting article should help. Here's what the radio looked like after I was finished. The yellow lines point to the new capacitors, and to somebody's replacement transistor (some time later I replaced that transistor with an original TI223).

The current draw is roughly an average of 4 mA at 22.5 volts. A test with a dc voltage source shows it operates well even at voltages as low as 15 volts. That's good news as far as battery life is concerned.


I experimented by making slight adjustments to the cores in the IF transformers and oscillator coil, and adjusting the trimmer screws on the tuning capacitor. It was aligned fairly well to begin with, but the dial calibration was off by about 30 kHz. There is some risk in turning the cores in the coils -- you could break the ferrite material. Another problem I found in tuning the oscillator coil was that the COIL turned instead of the core. If I had turned it any farther I may have broken the delicate leads connecting the coil to the base. A little dab of instant glue solved that problem and allowed the alignment to continue (I followed the procedure in the PF Reporter article). Using instrumentation I found the center of the IF amplifier to be around 273 kHz rather than 262. This is not a big problem (it's a somewhat arbitrary choice) and I could have left it, but I decided to adjust it to exactly 262 kHz for the sake of authenticity. Did the alignment help the sensitivity? Not sure, but I like to think so.

How Well Does It Work?

After reading the Consumer Reports article, I was fearful of bad performance. I was pleasantly surprised at what I discovered. The radio has a "full" sound; not tinny like some transistor radios, and has plenty of volume. I recorded some off-air signals for you to listen to. I put a (cheap) microphone in front of the speaker and captured the audio with a laptop computer as a WAV file, and then compressed it to MP3. This grand experiment was performed in my family room and so the radio got some interference from the laptop and from the DSS satellite receiver about 15 feet away.

I live in suburban Milwaukee, Wisconsin, about 15 miles north of the middle of the city. The radio was oriented so the maximum sensitivity was to the south, toward Milwaukee and Chicago (approx. 100 miles away). I easily heard stations from both cities. Tests 1 and 2 were performed at about 8:30 PM, on Dec. 22, 2000, prior to aligining the radio.

TEST 1 -- WOKY - 920 kHz, Milwaukee (70kB, 17 seconds, MP3)

TEST 2 -- For this, I tuned the radio gradually through the whole AM broadcast band, from 540 kHz to 1610 kHz. It picked up various stations from Milwaukee and Chicago (289kB, 73 seconds, MP3). The graph shows the audio amplitude as measured at the microphone.

0:05 WTMJ   620 Milwaukee
0:15 WSCR   670 Chicago
0:23 WGN   720 Chicago
0:30 WBBM   780 Chicago
0:39 WOKY   920 Milwaukee
0:45 WMVP 1000 Chicago
0:52 WISN 1130 Milwaukee
0:58 WEMP 1250 Milwaukee
1:02 WMCS 1290 Milwaukee
1:05 WJYI 1340 Milwaukee

TEST 3 -- WOKY - 920 kHz, Milwaukee, 6:30 PM, 12/23/00 (212kB, 26 seconds, MP3). For this test, the earphone jack was directly patched into the laptop's sound card, so the fidelity is not affected by the microphone. As can be heard, the audio is clearer, and has better high frequency response. The volume control was set near minimum. On some stations, with certain program material, at certain volume levels, some distortion was noted. With the patch cord used, the TR-1's audio stage was probably not terminated in the same impedance as the earphone (would have presented) so this could affect the results. This, too, was performed prior to alignment. All in all, though, the sound was pleasing, especially considering the early, groundbreaking design. I'd say "congratulations" Regency! You started it all!

TEST 4 -- "Working some DX" (long-distance stations) (298kB, 37 seconds, MP3).
On 2/10/01, from 7:00-8:30 PM, I recorded signals from some distant stations. They are spliced together here, and show impressive sensitivity. This test was run after I had aligned my ivory TR-1. The alignment seemed to help the sensitivity, making it fairly "hot". The callsigns, frequencies, and approximate distances are:

Freq    Station    Location    Miles   
1020 KDKA Pittsburgh, PA   450
  650 WSM Nashville, TN   490
1530 WSAI Cincinnati, OH   330
1210 WPHT Philadelphia, PA   700
  700 WLW Cincinnati, OH   330
  840 WHAS Louisville, KY   360
  880 WCBS New York, NY   750
  770 WABC New York, NY   750

TEST 5 -- Lab Generated Signal (178kB, 2 seconds, WAV - in case you want to experiment with the data). A synthesized modulated signal generator (Hewlett-Packard 33120A) was used to apply a 740 kHz signal, with 400 Hz, 50% modulation, to a small wire loop positioned near the TR-1's loopstick antenna. No broadcast band signal was close to this frequency, the nearest being at 720 and 780 kHz, around 100 miles away. A moderate strength signal was applied, causing the AVC voltage to change from approximately 0.50 with no signal, to 0.15 with this signal applied. A strong signal drives the AVC voltage to nearly 0.00. There was some hiss in the resulting audio, due to the lack of signal strength.

Back To Top

The oscilloscope plot (left) shows the RF, the 262 kHz IF, and the audio signal (across the speaker, which was in the circuit). Even by eye, the audio appears distorted. A spectrum analyzer plot (right) shows significant components at 400, 800 and 1200 Hz with amplitudes of +1, -16, and -41 dB. This yields a total harmonic distortion (THD, calculated as a ratio of RMS distortion to RMS fundamental levels) of 13%. The level of quality of the signal generator would seem to imply that the distortion is truly caused internal to the TR-1. Further investigation showed that about half of the distortion is present before the audio amplifier stage (at the top of the volume control), with the remaining half being produced by the audio stage itself. The AVC voltage has about 1 mV of audio on it (on top of the 150 mV AVC level), but this slight unwanted feedback does not appear to account for the distortion being created (additional AVC filtering did not affect the distortion). The distortion at the speaker with 100% modulation was almost 27%.

 scope  spectrum 

The spectrum analyzer's tracking oscillator was then used to modulate the signal generator, with the spectrum analyzer plotting the response of the radio. The following frequency response plot shows the range from zero to 10 kHz, with a 10 dB/division vertical axis. The response is fairly flat from 250 Hz out to about 5 kHz.

spectrum analyzer


I had read-up on the repair process ahead of time, so it would go smoother. When I finally started the work, the whole process took under two hours. There's always a certain amount of both skill and luck involved in troubleshooting, and experience helps, to be sure. I really never thought I'd hear a TR-1 play, but when the ivory one turned up "defiled", I knew it was my chance. To hear music come from the radio, through those primitive transistors and Richard Koch's original circuit, is thrilling. It was the first. Thanks Regency! If you try the repairs too, I hope it goes well.

Back To Top

Dr. Steve Reyer These web pages on the Regency TR-1 were created and updated for many years by Dr. Steve Reyer, who passed away in 2018. He was a true enthusiast in the radio hobby and is greatly missed. Steve was Professor Emeritus (Electrical Engineering) at the Milwaukee School of Engineering. He earned his Ph.D. in electrical engineering at Marquette University. Steve had hobby interests in consumer electronics and in Milwaukee Architecture. We are honored to keep his extensive efforts in the hobby alive and available here at

Back To Top

Copyright © 2019-2021