Upper Limits of Biasing 2SC2922 2SA1216

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
look at Sanken datasheet.
fT is max @ 2A to 5A
hFE is max @ 500mA to 1500mA
These are clues that this device operates at it's optimum when currents are high.

In a ClassAB output stage the output device bias current is normally set to minimised crossover distortion. This bias setting is referred as optimal ClassAB bias. The optimum bias is when the total output Re times the bias currents generates a voltage of 26mV @ Tj=25degC.
The internal Re is quite small and for larger values of external Re is often ignored.
setting bias to 25mVre is common, particularly if the external Re=0r47.
If external Re=0r1 then Vre might be around 18 to 20mVre.

Taking those recommendations and creating a table similar to what D.Self shows in his papers/book gives the following:
Re=0r56 Vre=25mVre Ib = 45mA
Re=0r47 Vre=25mVre Ib = 53mA
Re=0r33 Vre=23mVre Ib = 70mA
Re=0r22 Vre=22mVre Ib = 100mA
Re=0r15 Vre=20mVre Ib = 133mA
Re=0r1 . Vre=18mVre Ib = 180mA

None of these optimal bias currents gets as low as 20mA.

But, take care with device Dissipation (Pq) and Thermal Stability.
High Vcc does not go well with low Re (see Cordell)
High bias current does not go well with small heatsinks, nor with high Vcc.
 
I would go for 0R22
At for example 3 Ampere peak output, this would create a
voltage drop of 3 x 0.22 = 0.66 Volt across emitter resistor.

For a 0R47 resistor drop would be 1.41V at peak current 3A

*** 3 Ampere peak output corresponds to 36 Watt RMS into 8 Ohm
 
Ok, the emitter resistors are .33ohm and 20mv measured across them, so 60mA or so bias current. I'm getting a 170 F temperature of the heat sinks. I took the bias current to about 90mA with obvious sound degeneration. So, are my heatsinks possibly too small? They got significantly hotter. For the four devices per side, I have an 8" x 2.5" extruded aluminum heat sink w/ 3 fins/inch.
 
90mA requiring Vre (external) of 29.7mVre is far too high.

The 20mVre seems much more likely as the optimum ClassAB bias. This will take account of 6mV (total = 26mVre) dropped across the internal RE of the Sankens which do have a higher value than some due to their construction.
 
Thank you for your replay :)

I know what the service manual adjustment idle is "12mV and after the cabinet is close 30 min after it will decrease to 10mV" I just looking for what the max and best setting will be for single 2SA1216/2SC2922 that is matches with an emitte of 0.1R 5W 2% Mundorf Resistor. Right now i have adjust the idle to 20mV "with cabinet close 18mV after 30 min" and fund it much better sounding... But i will be sure if this is okay... As told it is a single push pull with one 2SA1216 and one 2SC2922 for each channel.
 
High Ft power BJTs have internal emitter resistors to balance internally paralleled transistors and some of them have fairly high resistance, thus low gm , gentle Ic-Vbe slope -straight region in the plot- at high current. 2SA1186 2SC2837 are typical one, they have around 0.1ohm, read from Ic-Vbe plot.
 
The glaring omission from this thread so far (unless I have somehow missed it) is the output stage topology. CFP or Darlington? It makes a huge difference to optimum quiescent current. To a first approximation the best current does not vary with BJT type, but it varies greatly with topology and emitter resistor value.

Exactly. CFP optimum quiescent current is always lower than the one for EF configuration. Strange - nobody mentioned that earlier, the thread is long enough already :)

Also - a note to the OP - bias is voltage (Vbe). Sure, it's got Ic, associated to that voltage (and that Ic strongly depends on temperature), but still - when we talk about bias - we talk about voltage.
 
For the CFP the correct setting is not a voltage, although I can understand why many people think that it is. It is a current which varies with 'emitter' resistor, so at first glance (by someone who has not actually done the theory i.e. almost everyone, including Self) it can look like a voltage.

For the Darlington I am not certain that there is an optimim quiescent current, but I have not studied the Darlington.
 
If you consider CFP compound pair as a single "module" element, you set the voltage at its bases as bias anyway. Same kind of Vbe multiplier is used for biasing, which is a voltage reference (although, in case of CFP it tracks the temperature of the "driver" stage transistors).
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.