k8gu.com
repair notes
Power Amplifier

When I purchased my first TS-930S, it had a number of "improvements" that the owner had added. One of his selling points was that he'd pushed the power output to 150W because the "28-volt transistors could handle it." [I bought the radio because it had three Inrad and one Kenwood filters plus the PIEXX board and the service manual, figuring that for $650 I could remove the PIEXX board and the filters even if the radio was junk.] Although I removed almost all of the other modifications, I naively failed to spend 30 seconds to reduce the output power. It wasn't until my PA failed that I learned how fragile the PA in the '930 really was.

When the 28.5-volt power supply fails, it is invariably related the failure of the MRF-485 driver transistors. However, if you have low or no output and a good 28.5-volt power supply, check the mechanical and electrical integrity of the coaxial cables that connect the signal unit to the PA. I got an outstanding deal on a "low output" '930 on eBay only to find it had dirty contacts on the input connector for the PA. Unfortunately, this is rarely the case. Further, the MRF-485 Kenwood used in the TS-930S and TS-940S is a low-hFE grade of the part that often difficult and expensive to obtain. Higher gain parts are readily available although still somewhat more expensive than the NTE-listed substitute, the NTE236.

W6NL claims that he made a successful substitution of the NTE236 for the MRF-485 as well. However, ON7WP/AA9HX makes a rather scathing assessment of the choice suggesting a much cheaper FET. Although I agree that the MRF-485 [and the NTE236] is being pushed very hard, his sounds like a convoluted solution which I believe is neither justified nor necessary. He offers no indication of whether he checked the hFE of the NTE236's he burnt-up before trying the FET. This is a subtle, yet crucial, consideration because I believe that high gain in the driver stage leads to instability, inciting both the power supply and driver failure. Of a handful of NTE236 parts that I ordered from Jameco Electronics, all but one were clustered around a DC hFE value of approximately 80, which is substantially higher than the YELLOW and ORANGE coded MRF-485's [specified by Kenwood] corresponding to a maximum hFE of 52. The remaining part measured at 52, suggesting that Leeson may have been fortunate to have used a lower-gain pair.

KB2LJJ/CT1APV offers some help in his rather complete discussion of mods for the TS-940S, which contains a similar, if not identical, PA to the '930. Apparently a design change to the bias network in the '940 production run allowed the radio to use the higher gain drivers. This change involves increasing the value of resistors in the part of the bias circuit where diodes are thermally-bonded to one of the drivers and the PA heat sink to mitigate thermal run-away. According to this suggestion, I increased the value of R16 to 2.2k Ohms.

After installing the new drivers, replaced the regulator transistor Q6. The bias on the emitters of Q2 and Q3 remained near 0.1 volts. Eventually, I tracked this down to zener diode D5, which I replaced with an NTE5078A. In the process, I also replaced Q9 with an NTE377, which turned-out to be unnecessary. But, I include it for completeness.

According to W6NL and the service manual, I reset the bias on the drivers to 60 mA. I conservatively chose to shoot for the low end of the bias tolerance suggested by the service manual to avoid provoking unnecessary instability. At the suggestion of KB2LJJ, I let the radio transmit in USB mode with no audio input for well over an hour, monitoring the bias current through L7 carefully for changes. I then reset the CW carrier and TUNE carrier to 110 W and 55 W, respectively. Finally, I listened to myself on a second receiver as I sent some CW at 35 wpm into a dummy load. Sounds good. I was able to test the radio in the January 2006 North American QSO Party CW contest for about 8 hours, where it was subjected to high duty-cycle CQing and high SWR on 160 meters. The radio has continued to perform well [as of February 15, 2006] under occaisional contesting and casual operating.

Parts and Suppliers

A list of all the parts I replaced in the PA are in the table below. I purchased my parts from Mouser Electronics and Jameco Electronics. Both of these companies carry nearly the complete NTE line, although Jameco tends to be slightly cheaper. Many larger cities have a local electronics emporium [not the one with the Answers, although they probably can order them] or two that carries NTE/ECG replacement semiconductors or even original replacements. This repair should cost around $35 dollars as I described. You may also elect to obtain original parts at higher cost from RF Parts or East Coast Transistor.

idpartreplacement
Q2,Q3MRF-485LBNTE236
Q62SC496(Y)NTE295
D5BZ-192NTE5078A
R161.2 k-ohm2.2 k-ohm

Good luck. I appreciate your feedback, particularly from readers who have actually attempted or completed this repair, either on their own or using my comments.

Update [18 June 2006]: the NTE236 has been discontinued. You may be able to find vendors who still have it; but, prospects are not good.