amp transistor stops when speaker connected

Thread Starter

Robyn

Joined May 1, 2013
4
Hi everyone, I am trying to design an AB amp for my car. I have put together part of the circuit by trial / error / maths so far but i stumble upon a behaviour that i cant explain. If i probe the emitter of the transistor with an oscilloscope i can see my sine wave but as soon as i plug a speaker between Vout and gnd the signal stops. The op amp is a LM358N and the transistor is a TIP42 (ill add the 41 when i get a satisfying top half of the wave).
What am I not doing properly?

Thank you for your time!

Ps. The circuit is derived from a circuit that i found in "41 projects using 741 IC" and that gives me the same behavious even when complete.
2016-01-31 16.41.31.jpg tmp_3055-20160131_162125~21739793040.jpg
 

Dodgydave

Joined Jun 22, 2012
11,284
Your op amp gain is showing a reduction of a half, you need to swap the 10k and 5k resistors, then the gain will be X2, put the 10k feedback resistor from the Emmiters junction.
 
Last edited:

blocco a spirale

Joined Jun 18, 2008
1,546
It would be unwise to connect a loudspeaker to Vout because there will be around 5.3VDC on there which will destroy most small loudpeakers pretty quickly. I'm guessing that your power-supply cannot deliver the >1.325A required to heat your loudspeaker voice coil (assuming a 4 Ohm car loudspeaker) and is cutting out. You should connect a capacitor of at least 1000uF between Vout and the loudspeaker (+ to Vout) to block the DC.
 
Last edited:

crutschow

Joined Mar 14, 2008
34,281
Looks like you have the "error" part down pretty well.
You need to add the output capacitor as blocco a spirale suggested.

But note that then there's nothing to discharge the capacitor.
You need either a small value resistor from the transistor emitter to ground (which wastes power) or add a PNP transistor to form a push-pull output as shown in the other circuit you posted (but that needs proper AB bias to avoid crossover distortion).
 

Thread Starter

Robyn

Joined May 1, 2013
4
Thanks a bunch guys, i added the output cap and an input one as well, increased the gain and moved the feedback resistor from the output of the opamp to the emitters of the transistors. Everything works dandy now.

I still have a couple of questions though.
- Do i need a discharge resistor on my input cap as well?
- Also how do i pick the value range of the input and feedback resistors (say 10k/120k VS. 100K/1.2M)?

Thank you again for all your precious help.
 

Attachments

Dodgydave

Joined Jun 22, 2012
11,284
Your feedback resistor is in the wrong place, it should go to the Emitters, not to the supply line. As for the choice of resistors, it depends on what input resistance you want,i would increase the two 1k resistors to 10k or higher to reduce current.
 
Last edited:

blocco a spirale

Joined Jun 18, 2008
1,546
The ratio of the input series resistor to the feedback resistor sets the gain but the input resistor also determines the input impedance of the amplifier. 10k - 47k is reasonable for an inverting amplifier like this otherwise, in order to achieve a useful amount of gain, the feedback resistor value starts to become too high. You don't need a discharge resistor for the input cap.

There are a couple of easy enhancements you should make:
Put a 10uF capacitor across the lower 1k resistor to reduce the amount of power-supply noise entering the amplifier via the non-inverting input.
Add a means to bias the output transistors in order to reduce crossover distortion and improve sound quality. As it stands, this isn't a class A/B amplifier.

It's a good idea to simulate your designs first using one of the free electronic simulation programs.
 

ian field

Joined Oct 27, 2012
6,536
It would be unwise to connect a loudspeaker to Vout because there will be around 5.3VDC on there which will destroy most small loudpeakers pretty quickly. I'm guessing that your power-supply cannot deliver the >1.325A required to heat your loudspeaker voice coil (assuming a 4 Ohm car loudspeaker) and is cutting out. You should connect a capacitor of at least 1000uF between Vout and the loudspeaker (+ to Vout) to block the DC.
As it is; the cone will be slammed against the end stop - so it probably wont give much sound for however long it takes the coil to burn out.

The ratio of the 2 1k divider resistors could be changed to present a more acceptable voltage across the coil, that in itself would limit the output swing - as would the fact that DC through the coil would deflect the cone, then it would have less travel in one direction.

A coupling/matching transformer on the output is the way to go.
 

blocco a spirale

Joined Jun 18, 2008
1,546
As it is; the cone will be slammed against the end stop - so it probably wont give much sound for however long it takes the coil to burn out.

The ratio of the 2 1k divider resistors could be changed to present a more acceptable voltage across the coil, that in itself would limit the output swing - as would the fact that DC through the coil would deflect the cone, then it would have less travel in one direction.

A coupling/matching transformer on the output is the way to go.
Which bit are you replying to? In his last schematic he has moved to a PP output stage and added a capacitor at the output to block the DC.
 

ian field

Joined Oct 27, 2012
6,536
Which bit are you replying to? In his last schematic he has moved to a PP output stage and added a capacitor at the output to block the DC.
I just ignored the stuff about adding a capacitor - that way round gets more complicated than any advantage it gives.

If you want capacitor coupling - make it a full complementary pair emitter follower complete with automatic bias etc.
 

blocco a spirale

Joined Jun 18, 2008
1,546
I just ignored the stuff about adding a capacitor - that way round gets more complicated than any advantage it gives.

If you want capacitor coupling - make it a full complementary pair emitter follower complete with automatic bias etc.
He has a full complementary pair emitter follower, but no biasing, as yet.

We all know this is a poor amplifier but... one step at a time.:)
 

ian field

Joined Oct 27, 2012
6,536
He has a full complementary pair emitter follower, but no biasing, as yet.

We all know this is a poor amplifier but... one step at a time.:)
I've seen circuits online for the complementary pair having the bases strapped together with no bias diodes. The follower is included in the op-amp nfb loop, and its claimed that this eliminates the crossover distortion.

As I haven't tried it myself - I can't comment on whether that actually works or not.
 

RichardO

Joined May 4, 2013
2,270
I've seen circuits online for the complementary pair having the bases strapped together with no bias diodes. The follower is included in the op-amp nfb loop, and its claimed that this eliminates the crossover distortion.

As I haven't tried it myself - I can't comment on whether that actually works or not.
This does work -- with 3 stipulations.

First, the op-amp must be fast enough to transition the dead band (the voltage range where both transistors are off) in a period of time that is small compared to the period of the highest frequency signal to be amplified.

Second, the op-amp must have a high open loop gain to "swamp out" the loss of gain caused by the lack of bias.

Third, the op-amp needs to drive the load even when both of the transistors are off. This is done by putting a resistor from the bases to the emitters of the otherwise unbiased transistors. This requires an op-amp with more than a few milliamps of output drive.

I have built an output amp using a current feedback op-amp similar to this simulation and it works quite well:

HW_power_amp.png
 

Attachments

Top