Monday, September 14, 2009

PRC-47 Modifications

[2 December 2010: Added an Addendum to the "Replacement of the Switching Supply Power Transistors" section, below.

8 February 2011: Added another Addendum at the very end of this post.]

The PRC-47 is a Vietnam War era SSB transceiver designed to operate from 2.000 to 11.999 MHz in 1 KHz steps. It is USB only, but can also operate CW or FSK, and it's designed to be powered by either a DC supply (from 24 to 28 VDC) or a 115 VAC, 400 Hz supply.

Transmit power (into a 50 ohm load) is rated at 100 watts PEP (High Power Position) or 20 watts PEP (Low Power Position).

Here are some pictures of my PRC-47:

(Click on Image to Enlarge)

Note the modular construction with plug-in modules...

...And an easily accessible chassis:

Replacement of the Switching Supply Power Transistors:

I've had several PRC-47 transceivers in which the two 2N1653 transistors (used to convert the 24 VDC to 24 V "square-wave" AC) were bad. I replace these two transistors (Q1 and Q2, on the chassis, just under the faceplate) with more modern 2N5884 transistors, which seem to work just fine.

[Addendum, 2 December 2010]

The original transistors used for Q1 and Q2 in the Switching Power Supply are Germanium, and are either 2N1166, 2N1653, or 2N2287 transistors (I have one Tech Manual that specs the 2N1653, and another which specs the 2N2287, and others have told me that the 2N1166 is also used in some units).

Germanium transistors are difficult to find, so, to replace the original "failed" transistors, I chose a more common Silicon PNP transistor. The 2N5884 which I use is not a perfect match for the original transistors, but it's close. Here's a comparison of the specs of the three original transistors (from
the fourth edition (1969) of Motorola's "The Semiconductor Data Book") versus the 2N5884 (from ON Semiconductor's website):

It's worth noting that diodes CR1 and CR2 in the Power Oscillator are there to limit the collector-emitter voltage of transistors Q1 and Q2 (the 2N1653 transistors) to 26.5 volts, so, assuming the diodes are still OK, the 2N5884's max rating of 80 volts should provide plenty of headroom. (I haven't made any measurements to verify this, though).

If anyone is concerned about breakdown voltage, you might try experimenting with the MJ15004 transistor -- it's rated to 140 VDC. However, its Ic (continuous) rating is only 20A, versus 25A of the 2N5884, so there's a tradeoff. Personally, I'd go with the 2N5884.

By the way -- the higher Vce(sat) of the 2N5884 (or MJ15004) might result in a lower plate-voltage to the PA tube (because the voltage swing at the primary of T1 will be lower (26.5 VDC - Vce(sat)), and thus lower output power. I haven't verified this, but if you find it to be the case, you can try bumping up your DC input voltage to, say, 28V, to help counteract the swing limitation due to the higher Vce(sat) of the Silicon transistor.

(And whichever transistor you use, please report back with your results!)

PTT Not Working?

If PTT doesn't work on your radio, check the following:
  1. Ensure that CR18 is installed in the AF Amplifier Module (I use a 1N4148).
  2. Ensure that the "PTT" wire (green, in my set) is connected to J2.11 on the chassis (this is the DB-25 jack into which the AF Amplifier Module plugs). If not connected, you might find it tucked to the side.
(Note: It's possible that very early versions of the PRC-47 don't have a PTT function, but are VOX only. I have an "Advance Copy" of TM 11-5820-509-35, dated November 1963, which shows no connection to pin 11, nor the existance of CR18 in the schematics (I presume the set, at that time, was VOX only). However, my later (electronic) version of the manual, dated July 1974, shows these connections.)

LSB Modification:

A radio which only operates USB in the range of 2 to 12 MHz is of limited use to a radio amateur. However, the PRC-47 can be easily converted to LSB operation. To do this, you need to replace the mechanical filter in the IF module (which happens to be a 500 KHz, LSB filter) with a 500 KHz, USB filter. And voila, you'll have LSB.

Here's how I make this modification:
  1. Remove the Amplifier-Modulator module from the radio and remove its covers.
  2. Remove the mechanical filter and replace with a Collins F500-Z4 (or equivalent) filter.
  3. Apply +20V to P4 pin 3 (this will supply power to the module) and the 20V return to the module case. (I find it easiest to attach +20V to the far left side of L9.)
  4. Set a signal generator to 501.5 KHz and apply the signal to J3.
  5. Measuring at J1 (with either a scope or spectrum analyzer), adjust the generator's level so that you see a signal (but don't overdrive the module), then...
  6. peak the measured signal by adjusting the two variable caps, C15 and C17.

AGC Modifications:

In my opinion, the "stock" PRC-47 AGC design results in severe and unnecessary distortion of the receive audio signal. One of my first goals was to attempt to improve the quality of the receive audio. In trying to fix the AGC I'd find a solution to one problem, only to then have another problem (previously hidden) reveal itself to me. And so the modifications, like coral, grew by accretion. Thus, they may not all be necessary (because a later mod may actually cancel the need for an earlier mod), but it would take much more time to determine which mod is irrelevent and which is not, so I've left them as I've implemented them.

First, a bit of background...the PRC-47 AGC is audio derived and results in two AGC signals: -AGC (which controls the gains of the the input RF preamp tubes) and +AGC, which controls the gain of the IF stage.

For SSB operation the PRC-47 has, in my opinion, an AGC decay time which is much too fast. The +AGC line is the dominant actor for normal SSB signals, and thus, to increase the decay time of C42, I changed R74 from 22K to 220K, and then added an emitter-follower (2N2222) to keep C42 from being loaded by successive stages.

I found, though, that when I did this, signals at normal "everyday" signal strengths sounded good, but very strong signals still distorted. A bit more investigation revealed that, with very strong signals, the +AGC signal was being driven so high that it was driving the IF stage transistors (Q2 and Q3 in the Amplifier-Modulator module) into cutoff!

I fixed this by limiting the the level to which the +AGC signal can go with a 13V zener (which, in my radio, results in a clamp voltage around 12.6 volts). Not elegant, but it keeps the IF transistors out of cutoff, and the -AGC signal (which kicks in at higher levels, and which is not limited) performs the AGC function for those very loud signals. (By the way, the 390 ohm resistor that I added between the emitter of the Emitter-Follower and the +AGC line limits current when the zener is driven into conduction, and the 22K provides the original resistance-to-ground as seen by the Amplifier-Modulator module and provides a necessary bias path for the transistors in the Amplifier-Modulator module.)

(I actually use the signal at the emitter of the emitter-follower to serve an "S-Meter" function (described later) because, at the emitter, the AGC voltage isn't clamped by the zener. Thus the S-Meter covers a wider voltage range than it would have otherwise if I'd used the +AGC signal. This signal (from the emitter-follower's emitter) I call "Buffered AGC+", and I connected it to a spare pin on the module's DB-25 plug (pin 12) so that I could then route it over to the meter circuitry located elsewhere in the chassis.)

Another problem I encountered: after releasing PTT, there was increased noise from the speaker (lasting for about a second) until the AGC stabilized the signal level. When I monitored the voltage across C42, I'd see its voltage actually drop momentarily (thus increasing receiver gain) when I transitioned from Xmit to Receive, and then it would recover. Removing C35 eliminated this noise burst.

After I had removed C35, I discovered that sometimes I'd lose output power during xmit. When this occurred, I noticed that the +AGC voltage was rising (and thus cutting off IF amplifier gain). I suspected that noise on the +26VDC line might have been affecting AGC during transmit (because C35 was removed), and I modified the circuit to use the +20 VDC line instead. Note that this required paralleling R59 with a 33K resistor to keep the junction of R59/R55 at around 8 - 9 V during transmit.

Essentially, during transmit relay K1 applies 26 VDC to the VOX line. This turns on a 2N3904, which in turn switches on a 2N3905 which connects +20 VDC to R59 (less, of course, a Vce(sat) voltage drop), rather than the original 26 VDC.

Another issue: during transmit, because of the newly-added 2N2222 emitter follower, the voltage across C42 can drop to near 0 volts. Then, when switching back to receive, it's possible to have a "pop" on the attack of loud signals because this low voltage causes the gain to be too high.

I fixed this by clamping the voltage across C42 during transmit: a 2N3904 transistor turns on (during transmit), which forces the voltage across C42 to be about 3.8V (the 3.8V comes from a 3.9V zener). Then, when we transition back to receive, there's less of a difference between initial receive gain and the required receive gain. (We could actually have used a zener with a bit higher voltage (nearer, say, 4.8 to 5.4 volts), but 3.9 volts seems to work fine.)

Here are some voltage measurements on my PRC-47 with these mods:
  • Receive, no antenna connected: V(C42) = 0.4Vdc V(Buffered AGC+) = 5.4Vdc
  • Receive, 80 meters, w/antenna and atmospheric noise: V(C42) = 9.4 Vdc V(Buffered AGC+) = 8.9Vdc
  • Transmit: V(C42) = 3.8 Vdc V(Buffered AGC+) = 5.4Vdc
Here's a schematic showing these AGC Mods.

(Click on Image to Enlarge Schematic)

Here's how the implementation looks:

Adding an S-Meter:

Normally, the PRC-47's meter doesn't move when the radio is in Receive mode. I thought it might be nice to have some sort of indication of relative signal strength (having a "dead" meter always seems a bit unnatural to me). If you don't mind a meter that's uncalibrated and non-linear, then here's a simple mod you can make.

It does require using a spare pin on the AF Amplifier module to run a "Buffered AGC+" signal to the outside world (and a wire added from the DB-25 jack (J2 on the chassis) to the new components mounted elsewhere on the chassis (see photo below for component location and mounting -- I used pin 12 of the DB-25 (J2) for this new signal). You can find a description of this "Buffered AGC+" signal above, in my AGC mods. (A "buffered" agc signal is used to drive the meter in order to keep the meter from loading the AGC cap: additional loading would worsen agc performance by shortening the agc decay time.)

I simply "diode-OR'd" this new AGC voltage with the existing "Xmit Signal Strength" signal. "Diode-ORing" simply means that whichever of these two signals has the highest voltage level will be the signal which controls the meter reading.

During Receive, the "Buffered AGC+" signal runs from about 5.4V (no antenna attached) to around 17.7 volts max (for strong signals). And during xmit this signal is at 5.4 volts. The 7.5 volt zener keeps the meter at 0 when there's no antenna attached (because 7.5 volts is greater than 5.4 volts, the zener doesn't conduct), and there's a couple of extra volts of head-room to keep the needle at a reasonable (left-side of meter) deflection when receiving normal atmospheric noise. And because this zener doesn't conduct during TX either, only the Xmit Signal Strength signal feeds the meter during TX.

Xmit Signal Strength is actually a fairly low level signal (if I recall, from my measurements it's about 250 mV max), and thus I used a Germanium diode (1N34) for this side of the diode-OR because of this diode's low turn-on voltage.

The 200K resistor limits the current to the 50 uA meter so that, at the strongest receive signal levels, the meter is just at its maximum deflection.

Here's the schematic:

(Click on Image to Enlarge Schematic)

And here's its implementation. I added a terminal strip for the additional components.

Transmit Audio issues:

The PRC-47 transmit audio leaves a LOT to be desired. It has an audio "compressor" which will distort the signal. Carbon mics can sound crummy. An inordinate amount of crud from the switching supply is coupled into the signal, manifesting itself as an underlying "whine" sound.

The conversion of the 24 - 28 Volt DC source to AC (actually, a square wave), which is required to generate the other voltages that the radio requires, unfortunately results in nasty switching artifacts that appear everywhere throughout the radio (on the +20VDC line (and other signals)). The ultimate result? Whine on the TX audio.

I've worked for many frustrating hours trying to minimize this whine. I've had some success, but I've never been satisfied. Because I use an "amplified"D-104 as my mic, I finally resorted to a simple solution: apply even more gain to the D-104 signal, EXTERNALLY (so that it's much louder than the internally-coupled noise), and bypass the radio's internal gain stages (into which the noise was being injected and amplified). This doesn't completely eliminate the noise, but the noise does seem to be much less and just about tolerable.

During my attempts to reduce this whine, I tried any number of fixes. Some seemed to improve things, such as the addition of 0.01 uF from the "CW Key" signal to ground (I mounted this on the back side of the middle board in the AF Amplifier module). But many of the mods I tried had little effect.

(One likely area of coupling might be the "Mike Input" wire. This wire runs from the Mic Connectors on the front panel to J2 pin 25 on the chassis (the AF Amplifier module's connector), and during its run from the front panel to J2.25 it's bundled together with quite a few noisy wires in a wiring harness. Thus, noise coupling onto this wire is certainly a very real possibility. One mod I'd like to try to reduce this possible coupling is to replace the wire with a shielded wire (e.g. coax). Unfortunately, on my radio, the Mic Connector pins to which this wire connects are difficult to get to. So, I'm leaving this for another day...)

Here's the External Mic Preamp which I built to drive the PRC-47 (I needed more gain, even though I'm using an "amplified" version of the D-104 mic):

And here are the mods I made to the mic amplifier side of the AF Amplifier module:

(Click on Image to Enlarge Schematic)

(I recognize that this solution really only suits my particular situation, and it is of little use if one wants to use a carbon handset or mic with the radio. So I encourage readers to experiment and discover what works for them. One tip I can give -- a friend, Dick (W1QG), replaced R2 (47 ohms) with a current source (about 600 uA), and also added emitter degeneration to Q1 and Q3, to lower the gain of both of these stages. This apparently helped out when using a carbon mic.)

A note on how I adjust the "MIC AMPL GAIN" pot (R27) on the AF Amplifier module...

Adjustment of this pot is described in section 3-26 of the Tech Manual (TM 11-5820-509-35). I'm lazy, so rather than attach a generator to the mic input, I adjust R27 using the TUNE signal (when in Tune Mode) so that it measures 3.5 vpp at the measurement point (see section 3-26.b). Then, once I have R27 set to 3.5vpp for the TUNE signal, I adjust the gain of my EXTERNAL mic preamp to also give 3.5 vpp at the same measuring point (for voice peaks). (Note, diodes CR5 and CR6 act as limiters, and they limit both the TUNE and the mic's audio signals -- if you're using an external mic preamp and you set its gain too high, you'll grossly clip your audio signal, which can add distortion. The TUNE signal, by the way, isn't a nice sine-wave, but is clipped by CR5 and CR6.)

If the radio is already buttoned up, I adjust the gain of the External Mic Preamp to give the same envelope peak voltage on the RF output (monitored with a 'scope) that I get in the TUNE position.

Curing that "Donald Duck" sound on Sidetone:

During Transmit the PRC-47 can insert a small amount of the TX audio back into the receive path so that operators, while transmitting, can hear themselves talking. This audio is called "sidetone", and its usage comes from the telephone world, where it was found (allegedly), that if people heard themselves in their telephone handset's earpiece while they talked, there was less of a tendency for them to yell into the mouthpiece.

Unfortunately, the PRC-47's sidetone signal can sound very distorted (I liken it to "Donald Duck"). The cause seems to be an audio envelope that appears on the "Sidetone Gate" signal during transmit. I improved this by adding a 4.7 uF cap from the "Sidetone Gate" line to ground in the AF Amplifier module (cap '+' goes to ground, cap '-' goes to Sidetone Gate).

If using an external speaker in lieu of a handset (which I do), sidetone can be quite annoying, so I also turn the sidetone gain potentiometer (R46 on the AF Amplifier module) all the way down.

Other Notes:

1. Extender Cables.

When working on the PRC-47, it's nice to have a set of extender cables that will let you remove modules from the chassis for debugging/experimentation, yet allow them to remain attached. Cables for this purpose actually exist, but are difficult to find. However, it's easy enough to make one for the AF Amplifier module (which uses a DB-25 connector). Here's the one I made (shown in use):

2. Documentation:

Get the latest version of the Tech Manual (TM 11-5820-509-35). The one I have is in electronic format (PDF), dated July 1974. Various modules have been modified over the years (from the initial design), and it's worthwhile having the schematics for the latest designs (Chapter 1 lists the various changes that have been made to the radio since is inception).

However, there is one caveat regarding this manual: there ARE errors! The text was (apparently) read in via OCR, and the usual goofy OCR mistakes are the result (I guess proof-reading was either sloppy or non-existant). Also, the schematics can have errors. I've run across a few, but, fortunately, not many.

(Also -- I always keep a lab notebook nearby in which I jot down notes, modifications, and measurements as I'm going along. It's a habit I got into as an engineer and it helps tremendously when, years later, you're looking over a schematic and wondering, "Why the hell did I do that?" Which is exactly what happened to me when I started writing up this post on the PRC-47 -- many of these mods were made six years ago, and it was no longer clear to me why I'd made some of them. Fortunately, a quick glance through my lab notebook quickly resolved any questions that I had.)

3. Pot Settings.

I usually set the following pots as follows:

On the AF Module:
  • R46 (Sidetone Gain): Full CCW
  • R52 (AGC Gain): Full CW
  • R54 (Rcvr Gain): Set per manual (section 3-22). Or use your ear.
  • R27 (Mic Ampl Gain): See Note above (in the "Transmit Audio Issues" section).

4. Setting TX Gain:
  • See sections 3-27 and 5-4 in the Tech Manual.

5. Power Amplifier Bias Adjustments:

Per Dick, W1QG, the PA bias should be set so that the PRC-47 "idle" current (from the 24 volt DC supply) during transmit is 6 amps ("idle" means that there is no TX audio). Using the instructions in section 3-27.c of the Tech Manual, adjusting the bias so that the voltage at A5J2 is -140 volts, instead of -110 volts, seems to achieve this goal.

6. Replacing bad transistors in modules...

Many (if not all) of the transistors used in the PRC-47 are germanium, not silicon, devices, and, if they go bad, identical replacements can be difficult to find. Whenever I come across a bad transistor, I've had good success simply replacing it with a silicon device. For example, I might replace a bad PNP with a 2N3905, and a bad NPN with a 2N3904 or 2N2222 (because I have lots of these devices lying around my "lab"). Silicon devices have a higher Vbe, but not significantly higher considering that much of the PRC-47's circuitry is biased from 20 volts. Thus, I don't believe that use of silicon devices will significantly alter the bias point (and therefore, potentially, the gain) in much of the PRC-47 circuitry.

Also -- pay attention to the transistor's application. You want to ensure that whatever replacement transistor you choose won't get smoked! Fortunately, you can still locate specs for many of the original 'Ge' transistors on the web, and a quick comparison with a replacement 'Si' transistor's specs should tell you if your choice is adequate.

A Final Note!

It's always possible that I've made a mistake in these notes (or during the implementation of these mods). If there's something that looks wrong or suspicious to you, please feel free to contact me and let me know.

Many thanks!!!

- Jeff, K6JCA

Addendum, 8 February 2011 --

I've just received a note from Ron Boltz, K3TZJ, who writes:

Finished my PRC-47 sets that were here with two more arriving this week. I sent you some other info a month or so ago. Two sets had 2N2638 transistors in them so another number to add to the list.

I got to do some testing on the last set with 2N5884s installed. The frequency of the inverter goes up to 525Hz and the high voltage goes up to 1,710. The bias also increases to -132 volts. I discovered that the bias needs to be adjusted on some sets as the PA tube overloads if not set correctly. Several sets had the bias way low for some reason.

Three of the sets I worked on had the key lines cut at two places. Drove me nuts till I discovered them. One cut prevented the 800Hz oscillator from running during the tune cycle and the other place was on the PA over temp switch which prevented keying. All three sets were Marine Corp sets and had the same depot stickers on them. I wonder if this was a method of de-mil?

Ron Boltz