SC200 Power Amplifier Bias Voltage Variation Across Output Resistors—Is 4.4mV Realistic?

Hello everyone I have built the SC200 power amplifier. The article states that the bias voltage across the 0.1 ohm 2 watt output resistors (test points TP3 - TP7, TP4 - TP7, TP5 - TP7, TP6 - TP7) should be 4.4mV. On one amplifier board I have found that the closest voltages i can get to 4.4mV at the test points are  4.7mV, 5.2mV,  6.2mV accordingly. I have yet to measure the last test point.This is a variation of about 30%. i think the second amplifier is better but not much. i will check it tonight.i don't think it is possible to get 4.4mV across every test point. The article test procedure for the amplifier alludes that this is possible. Has anyone got any experience with this or built the amplifier and tested it????.Can anyone help.
RegardsHeath

Hi Heath,

I think this difference is due to the beta (hFE) of the transistors. For the FJA4313 they can vary from [55, 110} for the R version of the transistor, or [80,160] for the O version of the transistor.

So, for the same bias voltage at the base of the transistors, there will be a slightly different current gain for each transistor. This different current gain then either increases or decreases the current through the Collector-Emitter, slightly, and you will therefore see a slightly different voltage across the output resistors.

This is also true for Q11 and Q12 which drive the output transistors, so these are part of the cause too.

You probably do not have the relevant equipment (curve tracer), but you can match the transistors for their current gain, and this will reduce the voltage variation, but will not eliminate it totally. Same for the FJA4213 transistors (the negative swing components). Same for the driver transistors Q11 and Q12.

You will not be able to eliminate this variation completely, but, my approach would be to accept the variation unless you do match the transistors.

What does this mean - the TP which has 4.4mV indicates that 44mA is flowing through the resistor, and the TP's that have 4.7mV, 5.2mV and 6.2mV each have 47mA, 52mA and 62mA flowing through them respectively. The amplifier performance (THD) will be slightly degraded with regards to the matched transistor version - how much, i am not sure.

Hope this helps.

Regards,
Richard.

Hi Richard I have checked the resistor spec sheet and the tolerance is ±(5.0%).On top of this there is the Temperature coefficient for the resistor of ±400PPM/°C.So the resistor variation  is about (0.1ohms) +/-  (0.005 ohms + (400ppm* using 20 degrees C above ambient).= 0.1 ohms +/- 0.013 ohms ie 0.087 ohms to 0.113 ohms.This can also account for the variation in voltage drop.I am not sure how much this variation in voltage drop across the output resistors affects the distortion measurement.Does anyone else have any idea???
RegardsHeath

Hi Heath,Yes, any variation in values for same function components will affect values measured and distortion performance.
The transistor variation is greater than the resistor variation, as per your measured values you have posted.
To test the variation of resistor values (although they are small), you can use LTSpice XVII. This is a free simulation program. Once the circuit has been created in the simulation, you can then simulate the distortion, by changing the values of the resistors to 0.113Ohms and 0.087Ohms. So you can run the simulation with perfect resistors, then modified resistors to see the difference in THD.
Of note when using the simulation, is you must click the little hammer icon, which is the LTSpice XVII control panel. Select the Compression tab, and uncheck (Enable Second Order Compression, Enable First Order Compression, and Only Compress Transient Analysis). Once these have been unchecked, then the simulation THD figures will be more accurate.
Regards,Richard.

Hi Richard i have worked out what the dificulty in biasing the amplifier. Apart from the transistor variation and resistor variation that is. The Suntan 1kohm trim pots i used appear to be highly nonlinear near the end of travel around the 105 ohm point where the transistor bias point is. Adjusting the pot even a fraction at this point produced unpredictable result in the bias current.So I reduced the 12o ohm resistance in series with the pot to 83 ohm which meant i would using different point in the 1kohm trim pot. This worked producing more linear resistance of the pot. i can now adjust the bias reasonably well. I could have shorted out the 120 resistor completely I think.. Alternative used a better quality trim pot like a Bournes for 60 pence more instead of cheaping out. The Suntan trim pots are not very good.
RegardsHeath