fisher: Biasing Circuit in Amplifier

The Fisher 600T has two different circuits for the amplifier section, one with an adjustable bias one without

The adjustable bias was removed in the last production runs

Which circuit is best suited for the amplifier section?

• amplifier (53 KB)
• amplifier2 (290 KB)

To thank the Author because you find the post helpful or well done.

Can a NTE5801 or 1N5801 relace the 1N2326 diodes in the biasing circuit?

To thank the Author because you find the post helpful or well done.

Hi Omer:

It makes sense that the diode was introduced in the last version of the circuit. Is the diode mounted on the heat sink of the the main power transistors? The main reason for the diode is to reduce the bias of the power transistors when they get hot. This provides thermal stabilization against heavy loads and, in this particular bias circuit, against higher power supply voltages. Bias circuits for bipolar transistors without stabilization diodes were common in early transistor designs, but became largely obsolete.

The bias circuit for the version with the adjustable bias and no diodes is largely a voltage divider from the power supply that drives the bases of the output transistors, so the standing current of the transistors is strongly dependent on the line voltage. You could verify this with a Variac adjusted between 110VAC and 125VAC. The version with bias diodes has much better rejection of supply voltage variations and the bias circuit is largely a function of the current dropped across the bias resistors. You can think of the diodes as little voltage regulators to bias the transistors. 

The early version without the diodes depends entirely on the 0.33R emitter resistors to mitigate variations on the supply voltage and thermal variations. These 0.33R resistors would best be made of metal, like copper and also mounted on the heat sink right under the power transistors, so that they go up in value with transistor temperature and reduce transistor bias. 

The other use for the adjustable bias is to get the ouput set to 0VDC at the speakers. This bias point is not adjustable with the diode version and depends on the resistor and transistor match between the upper -38V and lower +38V circuits. I don't see any other regulating feedback mechanism for the DC output voltage. The transformer coupling cuts the DC forward path.

On a different topic, the reason for the transistor cascode is to split the supply voltage evenly between the two transistors and allow for the use of transistors that have a Vceo rating that is less than the combined 76V supply voltage. The transistor pairs also cut the power dissipation per transistor in half, thus helping with thermal stability too. The transistor stack allows for a nominal doubling in output power at the same transistor current. The annotation of DC voltages in the newer schematic is not correct. For example 200mV DC at the base of Q7, should be around -20V as shown in the older schematic with the adjustable bias trim pots.

The location of the 2A protection fuses at the collectors with the 330R resistors in parallel is rather peculiar. What could the 330R resistor possibly do if the fuse pops? One possibility is that if the transistor stack is shorted out on one side and pops its fuse, then the 330R resistor provides some current shunting to prevent excessive DC current being coupled to the speakers by the circuit half that still works. 

I recommend using external fuses in series with the speakers to protect the speakers from failure in the amp. Alternatively, large back-to-back external coupling electrolytic caps could be used, but that defeats the virtues of a DC-coupled amp.

Replacing the diodes with a different type can be done, but the diodes must have a very similar drop at the operating current, say within 20mV. This will keep the standing current of output transistors conservatively from changing more than a factor two from the intended bias value. Keep in mind that the forward drop of the diodes is on the order of the 200mV, so they are probably germanium. Silicon diodes would drop around 0.7V. Modern schottkys of the correct voltage drop could be substituted for the originals. Vintage diode-connected germanium transistors may also work, if the forward drop is correct. The top and bottom diodes need to also match as well as possible to prevent DC voltages at the speaker output. Something like 1VDC at the output should be tolerable by most speakers. You can check to see if connecting the speakers causes the woofer to move excessively from the rest position with the residual DC voltage at the output. When experimenting with diodes, be sure to monitor the drop on the 0.33R emiter resistors and the DC output voltage before you connect any speakers.

Regards,

-Joe

To thank the Author because you find the post helpful or well done.

Thank you for the detailed explanation, many people that are restoring the 600T are removing the diodes and returning the 10 ohm pots for adjustable biasing  Your explanation, does not warrent this change!

Fortunatly, my 600T still has viable diodes in the biasing circuit  Since the germanium diodes are now "unobtanium", what whould be the appropriate Schotsky diode for replacement, if and when a replacement is needed?

In addtion, 2N1547 power transistors are still being manufactured in China ( I am not sure what their average gain is, and how leaky they are), could a Russian GT703G function in replacement of the 2N2148's that were originally used in this receiver?

I am not too thrilled with NTE replacements, nor their hefty price tag!

Thank you again for your detailed answer

Omer

To thank the Author because you find the post helpful or well done.

Hi Omer:

I would not give up the diodes, but it is possible to add a pot to tweak the standing bias current in the output transistors and the DC voltage at the output. 

Modern Schottky diodes have a forward drop around 200mV-400mV, depending on diode size and bias current. The Germanium diodes in the bias circuit seem to drop around 200mV. Perhaps a sufficiently large Schottky diode could be found that has about the same voltage drop as the original germanium diodes.

A diode connected germanium transistor, like the kind used in the audio output of pocket radios, should have the best chance of matching the original germanium voltage drop. You could even add a trim pot to the diode-connected germanium transistor to adjust the overall Collector-Emitter voltage drop. The trim pot would have its ends connected to the collector and emitter and the pot wiper would be at the base of the germanium transistor. The pot value should be in the 100Ω range. The pot can only increase the drop across the collector-emitter, so be sure that with the base tied directly to the collector the voltage drop is less than desired. The collector and emitter would be the new diode terminals. The anode and cathode of the diode are indicated by the arrow direction of the emitter symbol. A PNP/trimmer combination would have the cathode at the collector and the anode at the emitter. The reverse connection is true for a NPN/trimmer combination.

I am not familiar with the Russian germanium transistors you mention. The Vceo rating should be at least 50V, the Vbe voltage at the 30mA bias current should be similar to the original transistor, but a new Vbe voltage might be accomodated with the diode-connected transistor with trim pot. The beta should be at least as high as the original. Half the beta may still work well enough with an adjustment to the bias circuits.

In any case, substitutions of components needs to be done equally for the positive and negative supply circuits and requires careful monitoring of bias currents and DC output voltage without any speaker connected. I recommend using a Variac to bring up the supply voltage slowly while monitoring the bias current as a drop acorss the 0.33Ω emitter resistors and DC output voltage. The diodes, or diode-connected transistor substitutes, need to be mounted on the heat sink in order to help with thermal stabilization.

Regards,

-Joe

To thank the Author because you find the post helpful or well done.

I am kind of shooting in the air on this!

Would a 1N5111 Schottky diode work?

To thank the Author because you find the post helpful or well done.

Hi Omer:

Silicon bipolar junction diodes, like the 1N5111 Zener diode you suggest, are universally useless as replacements of germanium diodes because the forward drop is radically different. No point in looking for a silicon equivalent to germanium. Silicon forward drop is ~0.7V, Germanium forward drop is ~150mV. The silicon Schottky forward drop is ~300mV. Even silicon Schottkys are too far off at 300mV to be useful as 1N2326 replacements.

Converting this amplifier design to silicon is very doable, but does require changing resistor values and replacing both germanium diode and transistors with silicon. Mixing germanium with silicon does not work with this design without major modifications.

If I had to replace a 1N2326, I would use a PNP germanium low power audio transistor like an AC125 in a Vbe multiplier configuration as shown in the following diagram.

The schematic is only a suggestion derived from data sheet values and has not been verified with simulation or measurement. During all experimentation you should have the speaker disconnected until the output DC voltage has been adjusted to zero volts. Mounting the AC125 on the heat sink with the power transistors takes the most advantage of thermal stabilization.

The AC125 are cheap and common in auction web sites. I mention this particular type only because the AC125 data sheet on Radiomuseum.org is extensive enough to specify the forward Vbe drop. It is very likely that a comparable low power audio germanium transistor will have a low enough forward Vbe drop to work. The advantage of the Ge transistor is that the total voltage drop can be adjusted precisely with the trim pot circuit to get the amp perfectly biased.

The forward drops are specified in the data sheets as:

1N2326   135mV at 2mA
AC125    105mV at 2mA

In theory, the 25mV lower forward drop of the AC125 wired as a diode with base tied directly to the collector as the cathode and emitter as the anode, will run the output Germanium transistors at a little less than half the desired current, unless the Vbe multipler trim pot is added to increase the forward drop slightly as shown in the diagram.

The 1N2326 data sheet explains the advantages of using the 1N2326 to bias ouput power transistors as improving thermal stability and immunity from supply variations, as outlined in the earlier post.

Excessive germanium diode leakage is not a serious issue with the diode in forward conduction. It may however indicate that the diode is rusting out, which may eventually lead to a short. In this particular circuit, if one diode shorts out, it will turn off it's associated power transistor and the 30mA that is available from the other half of the circuit will appear at the output. This should not kill the speaker, but the sound will be very distorted. The max current that a speaker can handle is simply the Sqrt(power rating/impedance). For a speaker rated at 10Wrms max and 8 Ohms, the max speaker current rating is 1.1Amp. A slow acting 0.8A fuse would be recommended in series with the speaker to guard against catastrophic shorting failure of the power transistors. The built-in fuses in the amplifier limit the output current to 2A.

Regards,
-Joe

To thank the Author because you find the post helpful or well done.