R-390/R-390A Carrier Adjust David Wise September 2007 The R-390A Carrier Zero pot, R523, is very touchy. A popular mod is to replace it with a ten-turn unit. This works, but it doesn't solve the underlying problem, it only covers up the symptom. I have developed a simple, reversible mod that makes the original pot work better than new. It spreads out the adjustment range and makes it linear. Only the IF deck changes, and you may not have to remove it from the radio. You don't even have to demount the pot. This mod is very effective on the R-390A. It can be applied to the R-390 with more effort and lower expectations. PREREQUISITES 1. This mod is for the original 17-ohm meter only. 2. If, like mine, your meter reads high (more than 100dB with 100mV of signal on the balanced antenna input), you can do this mod without removing your IF deck. Otherwise you will have to change a resistor inside. There is another, more involved mod that also allows you to use a nonstandard meter, which I will write up later. THEORY The R-390A Carrier Meter circuit places a milliammeter between an adjustable voltage and the variable voltage at the cathode of the AGC time constant tube V506A, whose grid is on the AGC bus. At zero signal, V506A is saturated and conducting a plate current of about 2.2mA . This current develops about 60mV across the 27 ohm cathode resistor R548. If the adjustable voltage on the other side of the meter is also 60mV, the meter indicates zero. As the signal level increases, the AGC bus goes negative, reducing V506A's current and hence the voltage on the minus end of the meter, and the meter goes upscale. The adjustable voltage comes from a variable resistor in the 4th IF V504 cathode. This tube is not controlled by the AGC bus; instead it uses the "cathode bias" technique, where the cathode current goes through 680 ohm R524, pushing the cathode positive with respect to the grid in a way that tends to minimize change. The operating point is about 13mA and 9V*. The designers inserted a small variable resistor at the bottom of R524 and anchored the meter at the junction. * Says the manual. In my radio, four different 6AK6 tubes yielded from 7.5V to 8.5V, so the average current is 11mA not 13mA. All my calculations below are based on 11mA. Getting 60mV from 11mA takes 5.5 ohms*. The smallest pot available at the time of the R-390 was the 15 ohm wirewound unit that thay used. As the radios went into service it became apparent that the change as the slider moved from one turn of resistance wire to the next was objectionable, and in the R-390A they replaced it with a carbon pot (Cost Reduction Report, section 4.3.3, page 15). (There is new speculation that low-ohm pots were also regarded as unreliable and their use became discouraged.) The smallest unit available was 100 ohms, and the engineer unimaginatively tacked a 22-ohm fixed resistor, R537, across it to bring it down to approximately the same maximum value as before. This new pot did not suffer from the stepping effect of its predecessor, but the adjustment range is compressed into the last few ohms of R523. Since the nominal setpoint is 7.3 ohms (7.3 in parallel with 22 is 5.5), the pot is normally at about 7% rotation, and it's really touchy. As the pot ages and gets scratchy, it becomes impossible to keep it stable. * These values are for the R-390A. The R-390 needs about 70mV from 10mA. This is discussed at the end of the article. As the pot goes from 0 to 100, the total cathode resistance seen by V504 goes from 680 to 698, essentially no change, so I simplified the calculations to assume constant current. Table 1. Voltage vs R523 Ohms (stock, 11mA, R537=22) V R ------------ .01 0.9 .02 2.0 .03 3.1 .04 4.4 .05 5.7 .06 7.3 <- nominal .07 9.0 .08 10.9 .09 13.0 .10 15.5 .20 100 You can see how nonlinear this is. 35% of the output range is crammed into the first 10% of rotation. At nominal, 1% adjustment equals 10% change. If we are allowed to increase the effective resistance in series with the meter (lowering its full-scale reading), we can improve this dramatically with a small wiring change. The stock design sets up R523 as a rheostat; the slider and the CW end are connected together. The CCW end is grounded, and R537 goes across. These parallel resistors form the lower leg of a voltage divider. The upper leg is R524 inside the IF deck. See Figure 1. If we modify this as in Figure 2, we get a current divider, analogous to the voltage divider you're all familiar with. When the pot is CCW, all the current flows straight to ground and the meter reference voltage is zero. When the pot is CW, almost all the current flows through R537, supplying 100mV. In the middle, some current goes one way, some the other, and we get an intermediate voltage. Let's derive the expression for reference voltage as a function of pot rotation x. Say I is the current in R524. It splits in R523. I1 goes to the left, and I2 goes to the right and down R523, which we set to arbitrary resistance R. Then I2/I = x/((100-x+R)+x) = x/(100+R), I2 = xI/(100+R), and V = I*I2*R or xIR/(100+R). It's linear! Table 2. Voltage vs R523 Ohms (modified, 11mA, R537 = 10) V R ------------ .01 10 .02 20 .03 30 .04 40 .05 50 .06 60 <- nominal .07 70 .08 80 .10 100 Notice that 10% rotation yields 10% voltage change, compared to 1% in Table 1. It is ten times easier to adjust. Scratchiness is ten times less obvious, and making replacement unnecessary in most cases. This is how Collins should have done it. All I can figure is, the engineer assigned this task was new and naive, distracted by other matters, or (the Collins Collectors Association will burn me at the stake) just plain incompetent. HOW TO DO IT My '54 Motorola was wired like Figure 3. Your mileage may vary. At R523, the slider and CW terminals are jumpered together; then there are two wires and R537. One wire goes to V504, the other to the meter. In my radio, the meter wire is white with green and blue tracers, and the V504 wire is white with orange and blue tracers. Both were on the slider. Remove the jumper and R537, move the meter wire to the CW terminal, and install the new R537 between the CW and CCW pins. That's it! Fire it up and revel in the smooth, easy adjustment. It should zero around 60%; if your 6AK6 is flat you'll have to turn it higher. The soldering heat may change the old R537. You might as well replace it with a new carbon- or metal-film unit. The value, which can range from 5.6 ohms to 22 ohms, determines the full-scale reading, and depends on many factors. To approach the original reading, use 5.6 . To meet or exceed the original reading, you will also have to change R549, see the end of the next section. I'm using 10 ohms for R537 and the original R549. COMMENTARY AND EXPERIMENTS There are three side effects, one innocuous, one beneficial, and one possibly detrimental. The latter can be neutralized by changing one resistor inside the deck. First, the cathode resistance seen by V504 increases slightly. At the nominal operating point, the stock design was about 685 ohms. Now it is 705. This is less than 4% change, well within tolerance. I could not detect any change in IF gain. Second, the meter full-scale reading is now independent of the zero adjust. Third, the full-scale reading may be less than before. It depends on the cathode current of 6AK6 V504. Strong tubes make it worse, because you have to turn down R523 to get a zero. You can increase the full-scale reading by decreasing R537, until you run out of zero-adjust range. At worst case the stock circuit will show the meter just 4.5 ohms, which is unreachable with the mod as documented so far. But you may not want the original reading; it may have been too high. Inject 100mV of signal into the balanced antenna connector and note the resulting AGC voltage. At this voltage, you want to read 100dB. I'm using 10 ohms, see Table 3 below. Note that 10 ohms yields a Thevenin resistance of 9.1 ohms, somewhat larger than the stock value of 5.5 . At 5.5, my meter reads high; this is what enables me to use the simple version of the mod. If your radio needs 5.5 ohms, then you will have to change R549; see the end of this section. The AGC voltage developed at a given input depends on the 6DC6's gain and its cutoff characteristic. Hot ones require more AGC. So do "wide" ones. I tried five different tubes. If you try this, it's essential that you measure the antenna voltage, because trim and choice of tube affects the input impedance. The AGC also depends on the cutoff characteristics of 6BA6 IF amps V501, V502, and V503. Wide ones develop more AGC. It also depends on the gain of 6BA6 AGC amp V508. A hot one will develop more AGC. I played with 6DC6 substitutes. I tried 6DK6, 6CB6, 6DE6, and 6BZ6. As you can see, they're either too sharp or too wide, but they're good enough to keep it glowing until you get the real thing. Table 3. AGC voltage with 0.1V on balanced antenna input, and R537 setting to get 100dB reading (1450KC, trimmed, IF gain set for best S/N) TUBE AGC R537 Total R --------------------------------------- Tung-Sol -13.3 9.6 8.8 RCA #1 -12.7 8.8 8.1 RCA #2 -12.1 8.2 7.6 RCA #3 '68 JAN -13.3 RCA #4 '62 JAN -13.6 12.2 10.9 6DK6 -8 6CB6 #1 -9.1 6CB6 #2 -9.6 6DE6 #1 -11.1 6DE6 #2 -12.3 6BZ6 #1 -14.1 6BZ6 #2 -15.4 Finally, the AGC depends on IF gain. Table 3 was compiled with IF gain set for best S/N ratio. The gain was 600uV, rather than 150, for -7V on the diode load. The Tung-Sol 6DC6 that got -13.3V in Table 3 gets -13.8V when the IF gain is reset to the standard value. In that situation, the perfect R537 would be 11 ohms. IF gain also affects the 10dB point. At 600uV/-7V, 10dB input does not cause much deflection. At 150, the meter reads above 10dB. Somewhere in between, it's right on. If you don't feel like searching for the S/N sweet spot, this trick will probably get you closer than the stock setting. The Carrier Meter reading for a given AGC voltage also depends on the no-signal V506A plate current, which in turn depends on B+ and the exact value of the load resistor R549. Raising B+, or lowering R549, increases the reading. The same 12AU7 that reads 100dB at -13V with an 82K pullup to 205V only needs -9V with a 56K pullup to 180V, the setup that obtains in the R-390. This opens the possibility of tweaking R549 to control the full-scale sensitivity. If you can't get a full-scale reading with my recommended R537 value, lowering R549 will get you there. Of course you have to remove the IF deck to do that, which I was trying to avoid. THE R-390 TM11-5820-357-35 is inconsistent regarding the 6AK6 final IF (V506) cathode resistor R536. All the schematics show it as 820 ohms, while the voltage/resistance diagram says 700 ohms, 8.2V. One owner reports 900 ohms and 9.4V, so the V/R diagram is probably wrong. In any case, the reference current is less. At the same time, the AGC time constant tube is driven harder (56K from 180V instead of 82K from 205) and therefore runs a higher cathode voltage, about 70mV nominal instead of 57. If you applied my technique to the existing 15-ohm pot, R537 would be 15 ohms, and the maximum reference voltage would be 90mV if R536 is 680, around 75 if R536 is 820. In the latter case, you'd find yourself setting the zero adjust around 95%; kind of close to the edge. This mod reduces the turn-to-turn resistance jump by a factor of two. Not having any direct experience with the R-390, I can't say if this is enough of an improvement to be worth doing. You could always replace the pot with a 100-ohm carbon like the R-390A. The benefits are not so great and the effort and risk of side effect are larger. Still, it's easy to do and undo; I'd try it.