HOW THEY SHOULD HAVE DONE IT Carrier Meter Retrofit By David Wise Based on work by Gary Gitzen September 2007 September 2014 Many R-390 and R-390A receivers lack meters, because of regulations concerning disposal of hazardous substances. (The meters contain a small amount of radium.) The 1mA CARRIER meter M101 is hard to replace, because its 17-ohm coil resistance is not commonly available. The circuit is tailored to work with this resistance, while most 1mA meters are 100 ohms. A 100-ohm meter inserted in an R-390(*) will read very low. Jan Skirrow published a modification that uses an op-amp to boost the voltage to a level that can drive an ordinary meter. Although it works, and its intent of not modifying the original circuitry is laudable, it strikes me as a kluge. Gary Gitzen proposed a mod that avoids this, instead changing a component value to increase the output. Pete Williams tried it in his radio and reported encouraging results. I experimented further, and I believe I have perfected it. OBJECTIVE Change the design to use a 100-ohm 1mA meter instead of a 17-ohm 1mA meter. HIGHLIGHTS - Works with a variety of replacement meters - Works with the real meter - Affects the IF deck only - No side effects - Compatible with all other mods (it subsumes my Carrier Zero mod) - Gives smooth linear zero adjust just like my Carrier Zero mod SUMMARY OF CHANGES R-390A 1. Change R537 (across R523 CARRIER ZERO pot) from 22 ohms to 68 ohms; 2. Rewire R523 per figure (remove jumper and move one wire); 3. Change R548 (V506A cathode) from 27 ohms to 150 ohms; 4. Change R549 (V506A plate) from 82k to 75k 1W; 5. Add a 1N914 from junction of R547 (220k), R544 (2.7M), and R546 (180k) (anode) to ground (cathode); 6. Change R524 (V504 cathode) from 680 ohms to 620 ohms. R-390 1. Change R537 (CARRIER ZERO pot) from 15 ohms to 100 ohms and rewire per figure; 2. Add an 82-ohm resistor across R537; 3. Change R559 (V511A cathode) from 27 ohms to 150 ohms; 4. Change R560 (V511A plate) from 56k to 51k 1W; 5. Add a 1N914 from junction of R557 (220k), R558 (1.2M), and R556 (180k) (anode) to ground (cathode); 6. Change R536 (V506 cathode) from 820 ohms to 750 ohms. As you can see, this mod would have cost Collins a resistor and a diode. Why didn't they do it? Either germanium was costly, or they didn't think of it. DISCUSSION Note: I use R-390A nomenclature below but the R-390 is the same. The R-390(*)/URR carrier meter circuit is tailored to the meter it uses. With no signal, V506A is saturated, pulling 2.2mA, which drags the plate down to 25V and raises the cathode to about 60mV. At max signal, V506A is near cutoff, pulling about 0.4mA . That plus the meter's 1mA current puts the cathode at 38mV. This is a 22mV swing to get the meter from 0mA to 1mA, which means the meter plus series resistance due to R523 and R537 must be 22 ohms. With a 17-ohm meter, R523 in parallel with R537 must be 5 ohms. V508's 12mA cathode current through this develops the no-signal 60mV idle point. If we increase R548, the voltage swing will be larger and we can use a higher-resistance meter. R523||R537 has to increase too, to supply the higher reference to match the higher idle voltage on R548. I was skeptical about this idea. I worried that the cathode degeneration would reduce V506A's gain, affecting the SLOW integrator time constant. I breadboarded the carrier meter circuit with SLOW AGC, wiring one side of my 12AU7 per the R-390A and the other with the mod. I applied a variety of step-change voltages to both AGC circuits, and found that even with R548 at 180 ohms, the highest value I tried, the two AGC lines tracked to within 0.1V at all times. In addition, once I adjusted the scale factor so both sides read 100 at the same input, the meterss tracked too. I tried this with several different 12AU7's. All did fine. Therefore, practical values of R548 don't add enough degeneration to degrade the time constant action. I believe the R-390 designer had only one reason for the 27-ohm R548. It is not explained in the manual. In fact, the manual obfuscates the issue. I only discovered it through experiment. With no signal, the R549 AGC delay current from B+ tries to drive the AGC line positive. According to the manual, connecting the V504 and V508 suppressor grids to the delay node (bottom of R544) prevents this. But suppressor grids are very poor diodes. You can check this easily on your own radio. With no signal, pull out V506 and connect a microammeter from the AGC line to ground. It will measure about 15uA. If we assume that V506B's grid makes a high-perveance diode, then out of the 75uA (200V/2.7M) AGC delay current, about 50uA goes to the suppressor grids, and the delay node will be about 3V (15uA * (220k||180k)). It is actually V506A's control grid that clamps the AGC line. I discovered this when I increased R548. Whatever the idle cathode voltage on V506A, the no-signal AGC voltage was the same. It must have seemed obvious to the designer that to keep the line close to ground, V506A's cathode must also be kept as close to ground. With the original circuit, it's about 60mV, which probably represents the minimum you can achieve with a reasonably-priced meter. While I was taking these measurements on the breadboard, I discovered that the V506 grid puts out a significant contact current. I invented this term from my previous reading about "contact potential bias", which is used in every All-American Five radio design. Even though the grid wires are small, widely-spaced, and negative, a few of the zillions of electrons streaming through them toward the plate fail to dodge around the wires. This current can develop a voltage across a suitable resistance, giving bias without a negative supply, and you can leave the cathode grounded where it can't pick up hum. To use a meter with more than 17 ohms, you need to increase the other circuit resistances to get enough voltage to push 1mA through it. When I increased R548 to the 150 ohms Gary Gitzen recommended, I saw about +0.35V on the AGC line. The AGC delay point was about 6V positive, but I hadn't breadboarded suppressor grids. Knowing they don't make much of a clamp, and remembering that the R-390A already has a semiconductor diode involved with AGC (CR101, used for diversity operation), I put a 1N34 from the delay point to ground. The voltage went right down, but I was nonplussed (pun intended) when I saw that it was 0.3V negative. Then I remembered about contact current and tried a silicon diode instead of germanium. A 1N914 gave 0V. It was perfect; at -.35V bias, V506A's 2uA contact current through R547 develops -0.5V, exactly cancelling the 1N914's 0.5V forward bias. On the other hand, I found that with R548 at 150 ohms and M101 at 100 ohms, 1mA was not reached. With R548 at 180, it still didn't make it. Values higher than this actually yielded smaller readings due to the increase in R523||R537 (which appears in series with M101) decreasing the sensivity faster than the increase in R548 increased it. There is no convenient way to eliminate this resistance while preserving the feature where overload in MGC mode causes a meter deflection. It looks like 150 ohms is optimum, and something else has to be done to regain sensitivity. I found that R549, the V506A plate resistor, has a strong effect on meter gain. How does this happen? R549, being much larger than a saturated 12AU7's Rp, determines the idle current. When the tube is near cutoff, however, the tube, not R549, determines the current. Therefore, lowering R549 gives a bigger swing between maximum and minimum current. Or, gets the same swing for a smaller change in AGC. This is the reason R549 is 56k in the R-390 and 82k in the R-390A. Not because of the change in B+, but rather because with fewer controlled stages, the R-390A develops more AGC for a given input; its larger R549 decreases the swing so it still reads only 100dB. Now I can get full scale on a 100-ohm meter. I played around with R548 and R549, looking for combinations that use standard resistor values while achieving the same meter gain as the original circuit. Here are my results for the R-390A, using R523||R537 = 40 ohms for the modified cases and 5 ohms (6 for the R-390) for the original design. (Why 40? Read on.) R548 R549 AGC for 100dB dB at original AGC 27 82k -12.4 100 91 56k 1W -12.6 99 110 62k 1W -12.2 101 120 68k 1W -12.7 98 150 75k .5W -12.4 100 And for the R-390: 27 56k 1W -8.8 100 150 51k 1W -8.5 103 These R549 values can also be approximated by tacking a standard resistor in parallel with the original. 51k is 56k||560k, 56k is 82k||180k, 62k is 82k||270k, 68k is 82k||390k, 75k is 82k||820k. In the same email that recommended 150 ohms for R548, Gary rewired R523 as a pot and eliminated R537. This not only adds 100 ohms to R524, it presents a series resistance to the meter that varies as you turn R523. I wired R523 as a current divider as described in my Carrier Zero mod, and put R537 back, calculating a value that gives some adjustment headroom, but also delivers the reference voltage near the point where resistance seen by V504's cathode goes through its maximum, so its rate of change is low. It turns out that for 0.35V, (the Vref required when working against a 150-ohm R548), 68 ohms across R523 fills the bill. This combination adds about 40 ohms to the meter and to R524. V504's bias only changes 10% across the adjustment range. This setup will handle a 1mA 100-ohm meter, which also turns out to be the toughest case. Here is a table of meters that will work in the proposed circuit with no changes. If your meter is lower resistance, just add a series resistor to make the total shown here. That includes the correct 17-ohm 1mA meter; just add 82 ohms. 1mA 100 ohm 500uA 390 ohm 200uA 1260 ohm 100uA 2710 ohm 50uA 5610 ohm I looked at panel meter specs by Simpson, Triplett, and Weston. Most models meet or exceed this requirement. How did I get these values? The meter plus 40 ohms is bridged between 350mV and V506A with its 150-ohm resistor to ground. Idle current is 2-1/3 mA (350/150). 1mA meter current is a 140mV drop from 350mV through the 40 ohms and the meter's 100 ohms, so V506A K is at 210mV, so 1.4mA is going through the 150 ohm resistor, 1mA of which belongs to the meter, so V506A is conducting 0.4mA at 100dB. 0.4mA through 150 ohms is 60mV, so it's equivalent to 150 ohms terminated in 60mV. The 150 and the meter and the 40, all in series, go across (350mV - 60mV) or 290mV, so for a given meter full-scale current "Ifs", the proper meter resistance is (.29/Ifs)-190.