Power Amplifier
Changes described here can be seen on page 3 of the revised schematic, published earlier.
This unit is from the 5th version, s/n 41001 and up, and the bias adjustment pots were replaced with free standing diodes starting with the 4th version, s/n 39000 - 41000.
The adjustment pots were open wirewound units subject to getting dirty and intermittent. Likely the cause of some of the early failures. There were
resistors in parallel but they may or may not protect against open wipers on the pots. Bias with the diode modification is dependent on component
tolerances. One channel appeared to be ok but the other channel was at about half the specified bias. Both channels had a small DC output at the speaker terminals.
Service manuals for earlier versions with the bias pots specified setting the bias using an intermodulation distortion analyzer. Something we all have
sitting right next to the voltmeter. Oh, you don't have one either? What a coincidence. Guess we'll have to improvise. After the change to non adjustable
bias, this was changed and the schematic indicates that there should be 19 mV across the 0.33 ohm emitter resistors <R97 through R100>. Be careful as this
does not directly represent emitter current. These resistors are also carrying part of the current for the bias network. My revised circuit has the emitter
resistors [R31 R32 R47 R48] handling only the transistor emitter current. I determined that the idle current should be approximately 30 mA. The replacement
emitter resistors are 1% tolerance. While this isn't necessary for performance of the circuit, it does allow for greater accuracy in bias current measurements.
I determined that the best way to correct this circuit was a return to the bias adjustments, but make them reliable. (transistor version of IBAM?) The diodes
<CR6 through CR9> of the later circuit were not attached to the heat sink so none of the versions ever had any direct thermal feedback. My revised bias
circuit uses a 10 ohm Bourns trimpot [R27 R28 R43 R44] in series with a 3.3 ohm resistor [R25 R26 R41 R42]. That combination is in parallel with a 6.8 ohm
resistor [R29 R30 R45 R46], so the entire combination can only be adjusted between 2.5 and 4.5 ohms and even if the wiper on the pot were to lose contact,
it can't go over 4.5 ohms, raising the bias to 60 mA, double the specified value, and probably not harmful. That would be 1.14 watts per transistor at
idle. The bias circuit was determined first by calculation, then by a lot of testing. The bias can be adjusted from half to double the specified value. The
pot is a sealed unit, so should be reliable over the long term.
Bias adjustment will be described later. There is still a test for the 4th and 5th revisions using the IM distortion analyzer and what the results should be
but nothing stating what to do if the distortion is out of spec.
1424
There were a couple of wiring errors, apparently from the factory, because the solder joints appeared to have never been disturbed. The 5th revision, s/n 41001
and up, has an error in the schematic diagram. There are 2.2 ohm resistors <R112 through R115> added to the collector circuits of output transistors <Q1
Q3 Q5 Q7> and the bias resistors <R73 through R76 and R83 through R86> are shown connected to the top side of these resistors. A small resistor in a
collector circuit with nothing connected between that resistor and the collector will do nothing. The unit actually had the bias resistor connection
made to the lower end of these resistors.
There are 4 electrolytic capacitors <C39 through C42> that are buried under a lot of resistors that make up the bias network. Since I was doing some
redesign, resistors are cheap, and they were in the way, I ended up completely stripping the power amplifier and rewiring. One advantage of this is that
todays 2 watt resistors are about the size of yesterdays 1/2 watt units so the circuit looks much neater after rebuilding. Before stripping, ALL wires on the
driver transformers <T2 T3> must be carefully labeled. If they ever got connected with the upper and lower drives in phase, the first signal applied
would turn on all 4 transistors and blow the entire thing sky high.
All power transistors were removed, cleaned, the heat sink cleaned, and transistors remounted with new insulating washers and fresh thermal grease.
After this, all were checked for possible case to chassis shorts. Soldering to the transistor sockets was done with the transistors removed to reduce the
possibility of heat damage. Some minimal rework was done later with the transistors installed.
1406
Since the actual output stage is isolated via the driver transformer, it can be tested independently of everything else. An external power supply was used, one
that I built many years ago. It is a dual supply with voltage adjusted by a variac on the AC line side. This was connected to the output stage (don't
forget the ground connection) and voltage brought up to approximately +/- 10 volts. All voltages were measured and found to be at acceptable levels. The
process was repeated for increasing voltage until the amplifier was up to the full +/- 38 volt supply.
Adjusting Bias
A voltmeter was connected across one of the emitter resistors and the associated trimpot adjusted for 10 mV, corresponding to 30 mA of emitter
current. The other trimpot in that channel was then turned to approximately the same physical position because the adjustments for the upper and lower
transistor pairs interact and this would get the adjustments closer to start. Then the voltage at the other emitter resistor was checked. Adjustments were
repeated several times until correct. Next, the DC output offset voltage should be measured. If it is positive, then the upper transistor pair needs slightly
less bias and the lower slightly more. If the voltage is negative, then the opposite. This is a very small adjustment. Before you think this is backwards,
remember that we are dealing with PNP transistors and the upper pair leads to the negative supply.
Bias was actually set approximately at first, then the amplifier was allowed to cook at about 10 watts, then the bias set with the output warmed up. Why 10
watts rather than full power? In normal operation, the average power delivered is going to be low enough such that peak signal levels are still within power
limits. Probably much lower than even 10 watts.
After several months of use, it was noted that the bias adjustments needed trimming. The bias had drifted up to around 12 mV (36 mA idle current).
Output Transistors
As mentioned earlier, the output transistors are house numbered versions of either 2N2147 or 2N2148. They are the same except the 2N2147 has slightly
higher voltage ratings. Be careful of replacement cross references. One of these 2 (don't remember which) crosses to a small signal transistor and more
recently, I've heard that the NTE cross reference is also incorrect for at least one of them. Actually, I'd probably redesign and rebuild before replacing
expensive transistors in this circuit. The TX-300 and early versions of the 600-T and some consoles show 2 different transistor numbers for the 4
transistors, but they were changed to the same part in later versions of the 600-T. I don't think it matters as long as the beta of the 4 transistors within
a channel are close. Since the power amplifier is independent (due to the driver transformers) it would be possible to redesign using more rugged and
readily available NPN (or PNP) silicon devices but I'd just design a completely new circuit first. Good as the present circuit is, unfortunately, it's simply
obsolete. 38215G will eventually have a redesigned circuit.
I should probably add catch diodes and a Zobel network. I'm guessing that the problems solved by those were not discovered in the early transistor era.
next: Driver and Control Preamp