This page last revised March 27, 1999.
If you've got one of these radios, you've got to provide power for it. Some owners power up a "stock" '348 with the original dynamotor, and if you go this route, then all you will need to do is provide 28 VDC (or 14 VDC for some models), and all will be fine.
I prefer to provide an AC power supply for the radio. The supply for my BC-348Q is built inside the radio case (or you could build it as a separate stand-alone unit; this offers the advantage of reducing the heat within the case.)
What makes this radio a bit peculiar is that the original radio has the B- (negative) from the stock dynamotor floating from the chassis. If you connect a conventional B+ supply to this radio with the negative side grounded directly to the chassis, you are likely to fry something! So the first thing to resolve before powering up one of these radios is to verify that the furnished B+ supply does not have the negative connected to ground. But if your power supply is intentionally grounded at the negative output, you can still use it; it just takes a bit of care (as discussed below.) It involves (as a first step) changing the biasing in the cathode circuit of the final AF amplifier stage.
These radios were provided with a filter inductor within the same case as the output audio transformer. Because the original output audio transformer provides for 4000 or 600 ohm ouput (depending on model), many owners choose to replace the audio output transformer with one having lower output impedance (like 8 ohms, to feed low-Z headphones, or a speaker). But if the existing audio transformer is removed, the inductor is removed with it, and herein lies the problem: Dynamotor output voltage (200-250 volts) in the original radio is connected with the negative going directly to the audio output stage cathode, and also to one lead of the inductor. The other end of the inductor is grounded. If the inductor is removed, and the B- from the dynamotor (or AC supply) is connected directly to ground along with the audio output tube cathode, then the audio output tube will draw excessive current (because there is insufficient grid bias).
So if you are planning on removing the original audio output transformer (or if it's already gone), you will need to modify the circuit somehow in order to provide proper bias for the final audio output tube. (Of course, if the transformer is already missing then a previous owner might have already made the necessary mods.)
Let's review this. You will probably have to provide alterations to the biasing of the audio output stage if either (or both) of the following is true:
A suggested method of accomplishing the biasing change is to break the original connection from the audio output stage cathode to ground, and install a cathode bias resistor (paralleled with an appropriate bypass capacitor). This was done in my radio by a previous owner; the installed resistor was 500 ohms, 1 watt, paralleled with a 20 microfarad, 150 volt electrolytic. (The 20 microfarad cap is overkill; I would recommend something like 2.2 microfarads for a 150 hz lower frequency rolloff.) The R-C cathode circuit enables the power supply to be connected with the negative directly grounded to the chassis. It improves over the original circuit design by eliminating any variation in B+ voltage to earlier stages which would otherwise occur due to variations in audio output stage current draw. (That is to say, I think it's an improvement; read on...)
When I first started to familiarize myself with this "new" radio of mine, I didn't give much thought to the power supply. But after I looked over the original schematic and compared it to the actual unit, I began to look at other aspects of biasing. It became apparent to me that the bias issues would apply not only to the audio output stage, but also to the first mixer stage; that stage was (in the original circuit) provided with a negative grid bias of about 1.8 volts (again, from the inductor cited above, in conjuction with 50 ohm resistor 108-2.) However, the "modified" radio (that is, with AC power supply) simply provided for zero applied bias voltage to the input grid of the mixer. Was this a problem? Could it be fixed?
After looking over the original schematic, I determined that the mixer bias could be "restored" to the original value with any of three different methods (listed in order of increasing complexity):I chose option 3 above. Three 1 amp diodes in series, bypassed by a 0.1 microfarad ceramic cap, resulted in cathode voltage to ground at the mixer of about 1.8 volts (and corresponding negative bias voltage of -1.8 volts, not considering the effects of grid leak current.) It seems to work fine, but I'm unable to detect whether the radio performs better than it did before the modification. If I had to do it over again, I would try option 1 first to determine if there was any detectable improvement. If the improvement was noticeable, then I would proceed with either option 2 or option 3 above.
Finally, it's probably appropriate to consider: is it possible that the original design was based on a need to somehow couple the bias voltages to current draw? In other words: was it a design intent to have the bias voltages vary according to B+ current draw, or B+ voltage? If so, then the modifications described above might actually worsen radio performance. I do not believe that this was a conscious design decision, but it might have been; implement these revisions at your own risk. In any event, the power supply in my radio is a commercial grade regulated supply (set at 245 volts), so I do not expect any significant variation in supply voltage as a function of load current. YMMV.