rebuilt it in ltspice and made a few slight changes, it seems i can push it way above the 4 watt limit
although the things that i see at the base of q5 look concerning so concerning that an the crossover distortion of an op amp class b design is nothing
Ltspice does not list "rms" power, peak power 12 W is roughly 6 W effective power.
An op amp will not increase power.
Max power is quite easy to calculate:
24 V single supply ->
12 V max theoretical peak Voltage
Minus about 1.5 V for each output transistor + driver ->
10.5 V peak output Voltage ->
7.5 V rms ->
7.03 W at 8 Ohm
Looking at the original dual rail model. The negative power supply is wired incorrect.
Im not sure why LTspice even showed the model as valid. Or even showed a sinewave at the output.
Unless corrections were made after the screenshot.
This model would not even operate, output would be a = Flatline.
DC offset as expected was way up, wired correctly. The sinewave wouldn't even be symmetrical.
It is, but with single supply 24 V you only get max 12 V minus 1-2 V swing.
If you use a bridge circuit you can double it...
DC analysis with power supply wired correct. All transistors are 22c Temp.
Issues will be listed with quick explanations.
1) VR1 sets differential current at Current meter (AM8) to 6ma. Way to high.
For stability and enough current to drive T3, Typical values would be no more than 1ma to 2ma.
2) Differential transistors current T1 and T2 should be balanced.
Current meters AM4/AM5 show large unbalance 4ma and 2ma.
As expected would result in high DC on amplifier output. VF1 shows
212mV of DC offset. Something acceaptable be no higher than 30mV
and even with simplified circuits would be possible to achieve 3 to 5mV
over a wide temperature range. R2 value is too low and VR1 value should be higher
to set current to 2ma
3) Bootstrap current source R10 and R11 only provides 1.23mA of current for second gain stage T3 shown by current meter AM3
To effectively maintain stable bias current across the diodes. And provide enough AC swing for the whole
amplifier. DC current for T3 to drive T5 and T6 pre drivers and effectively swing to the full 24 volt rails
would need to be at least 6ma. And often in simplified design can be as high as 8ma
R11 and R10 bootstrap resistors should be set lower to achieve at least 6ma of current.
4) pre drivers T5 and T6. Dc current displayed by AM6 and AM7 are not only imbalanced.
At 641uA and 2ma. As emitter follower T5 and CFP T6 to operate as so would require at least
3ma of current. CFP has slight feedback so could likely drive the 2N3055 with less current.
All in All current needs to be higher. R12 and R13 should be set lower for 3ma of current.
5) Power transistors are fortunately not under biased. And surprising Over biased.
At roughly 200ma shown by AM1 and AM2 current meters.
Even though diode string only has 1ma of current. Resistor R5 is too high.
DC bias for quasi class A/B should be around 30ma to 45ma. When the amp has stabilized thermally.
Since we are starting at 22c or roughly cold start. Bias could be as low as 15 to 20ma starting.
And when output transistors get hotter amp will stabilize to 30ma to 45ma.
Since we are at 22c and already high bias. Likely the amp would thermal runaway or get to hot.
Also for more thermal stability R17 and R3 should be higher values than .1 ohms.
.22 to .27 more common. And since thermal tracking with diodes is often debated.
It works fine, but advisable to keep power transistors resistors at higher .27 ohms for thermal
stability and likewise DC stability.
6) R4 is simply not needed and reduces the amplifier bandwidth incredibly low.
Also compensation or Cdom capacitor C2 might need to remain at 100p to stabilize the amplifier.
but would be considered somewhat high. Again often needed to stabilize simplified amplifiers.
without going through a even longer explanation how that value or miller compensation is achieved
and calculated. R9 and R15 degen resistors and C2 would be very effective at stabilizing the amplifier.
and degen might even be raised after amplifier basic DC operating points are fixed. And open loop/closed loop
phase analysis is even possible. R4 lead compensation can be = abandoned
Last edited:
If you want simple, not to many components, why not build a chip amp , there are many to chose from.
This is AC ANALYSIS
Tested for full power in 4 ohm load.
This shows typical behavior of the 2N3055 power transitors when Baxadall diode not used.
As you can see AC current not DC current is unbalanced.
T8 is conducting almost all the current 1.72 amps shown by AM2
T7 is conducting less than half that current. 700mA
Not only will T8 typically clip earlier and rail stick, it also conducts most the power.
Adding the diode will distribute AC current more evenly. The non linear clipping
also be fixed.
Gonna eat dinner, will show new schematic in few hours with corrections and DC and AC analysis
This is new model with adjusted components, addressing problem list in post #34
DC Analysis shown above^
1) 2) VR1 now sets current at 2ma shown AM8
Differential circuit is as balanced as possible without a mirror roughly 1ma each AM4/AM5
Easily achieved with 1% to 5% resistor R2 910 ohms or E24 to E48 series tolerance resistor.
Differential now balanced shows respectable DC offset at output or VF1 9.77mV
3) Bootstrap current source R11 / R10 changed to 5.5mA
4) Pre Drivers current has been balanced to roughly 4mA and diode added to CFP driver T6
5) Output transistors biased to 30ma set by R5 for better thermal tracking 3 diodes used in the string.
Diodes D1 thru D3 to be mounted close to power transistor heatsink
6) Cdom capacitor C2 lowered to 33p and lead comp resistor removed.
T1/T2 when placed on circuit board should be mounted flat to flat close together for differential thermal matching
Amplifier input AC coupled or DC blocking capacitor C1 added. R1 and C5 form input filter for blocking radio interference.
AC ANALYSIS shown above ^
As noted before, these are expected currents at full clipping 4 ohms
With added diode to T6 CFP driver. AC current between both 2N3055 is now equal AM1/AM2
Unlike before when most AC current went to T8 AM2
Clipping of lower rail symmetrical with no rail sticking. Sorry to lazy to show scope views.
The behavior otherwise well known regardless.
When testing amplifier R5 could be lowered to 33 to 47 ohms to lower bias
and after extended warmup could be raised depending on needed bias.
Model shows 22c for all transistors. Diodes 20c and power transistors 27c for cold start.
Cold start could be set low 10 to 15mA and will reach 30 to 35mA when warm.
I used 30mA in the examples so people dont cry and whine bias is too low
which is why R5 set to a high setting to acheive 30mA at around 22 to 27c
Last edited: