Is there an error in this Class B amp schematic?

[Chris Wilson]

I have built and  triple checked the section of schematic ringed in red. The amp gives odd waveforms when fed a pure 136kHz sine wave. I am looking at the output on my scope with a 50 ohm termination on a BNC T. I have been looking at every Class B amp with diode biasing I can find and they all seem to feed the signal to the junction of the 2 diodes. Thanks.

[w2aew]

Looks to me that this stage is setup as Class A, although the the two output BJTs operate in complementary class B. The feedback around the last stage should ensure the overall performance is class A, provided the output amplitude is low enough to keep the op amp output from hitting its output voltage limits.

[IconicPCB]

HI..

typically the diodes are shunted by a capacitor ( signal short )and the output stage is driven such that the bases of the two transistors are biased into some forward conduction to minimise distortion.

In this case the base bias is derived from the final OP AMP hence the connection to the base of the top transistor.

You might try a capacitor across the two biasing diodes.

Do make sure you have sufficient powersupply overhead for the opamps not to go into rail drive limits.You have a bit of gain in those two final stages.

[T3sl4co1l]

Uh...

Care to elaborate?  What is "odd"?

Driving one end of the pair of diodes is fine as long as they remain conducting.  Which they should, if the bias resistor to GND does its job.  I did something similar on this circuit,
http://seventransistorlabs.com/Images/WidebandAmp.png
although the bias resistor in series with the diodes is unmarked for some reason (looks like I used 27 ohms).

Tim

[wiss]

I have built and  triple checked the section of schematic ringed in red. The amp gives odd waveforms when fed a pure 136kHz sine wave. I am looking at the output on my scope with a 50 ohm termination on a BNC T. I have been looking at every Class B amp with diode biasing I can find and they all seem to feed the signal to the junction of the 2 diodes. Thanks.

Isn't 136 kHz a bit high for tl072, especially when it have to do something more than just drive a resistive feedback divider?
Feeding the output stage from either side is the normal way to do it, nothing odd about that.

[Circlotron]

I have been looking at every Class B amp with diode biasing I can find and they all seem to feed the signal to the junction of the 2 diodes. Thanks.

You could feed the signal to the junction of the two diodes on this circuit too if you add a 680 ohm resistor from the upper base to collector, same as the lower one. And yeah, caps across the diodes for sure. Say 100nF - 1uF.

[dannyf]

they all seem to feed the signal to the junction of the 2 diodes.

Most of them don't do that.

As it doesn't matter - the opamp will just bias its output signal accordingly.

[GK]

I have built and  triple checked the section of schematic ringed in red. The amp gives odd waveforms when fed a pure 136kHz sine wave. I am looking at the output on my scope with a 50 ohm termination on a BNC T. I have been looking at every Class B amp with diode biasing I can find and they all seem to feed the signal to the junction of the 2 diodes. Thanks.

Isn't 136 kHz a bit high for tl072, especially when it have to do something more than just drive a resistive feedback divider?
Feeding the output stage from either side is the normal way to do it, nothing odd about that.

The TL072 has a 3MHz GBWP. In that circuit it is configured with a noise gain of 11, yielding a closed loop bandwidth of  3MHz/11 = 273 kHz. At this unity loop gain frequency intercept the open loop gain will be at a 20dB/decade roll-off, returning a paltry 6dB of loop gain at the operating frequency of 136 kHz. However, that is not what matters much, as it is the amount of loop gain present at the harmonic frequencies that effects the degree of distortion reduction. Since the second harmonic of 136 kHz is actually at the unity loop gain frequency (the higher order harmonics don't even need to be considered now), there will be zero effective negative feedback for the purpose of distortion reduction. So, basically, in this specific example, it is rather pointless to include the complementary emitter follower into the feedback loop.

[Chris Wilson]

OK, some great and very helpful answers here. I am under manic presure from work, I will post in detail with some scope shots and a full run down of why I need it and what i want it to do, but VERY briefly I think it has more gain than I expected and / or was led to believe. I was seeing wedge shaped waveforms, almost square, out, when fed with 0 dBm sine waves from my 9good quality) sig gen. i was expecting a higher amplitude sine wave. Dropping the input level to about -19dBm results in sine waves, so i guess it's "clipping" very badly? Now whilst the exciter (my Kenwood TS-590 Drive output, meant for either amplifying for 136kHz and for feeding a transverter for other amateur bands higher than it's 6 metre maximum, is supposed to be 0 dBm, I can, by adjusting the modulation level via the PC sound card, which is how it's fed in digital modes, to a lower level, get sine waves out. So basically I am not sure how Kenwood rate this "0 dBm exciter output level". It may be i can adjust audio modulation to suit the gain of the amp. the other thing i don't undertstand is why a symmetrical square wave, when the amp feeds the x2 frequency converter it has to drive, and which is designed to go from sine to square output, shows the square waveform go to rectangular with longer top and bottoms, if you see what I mean, as the gain is changed via the amps input level. Only over a narrow input signal level do i get true square waves out.


I attach below the circuit this amp feeds, to double the 136kHz tpo 272kHz, and convert to square wave. This is required as the Class D main amp drives power FETS, and the driver chip needs double the frequency as it itself halves it..... 

Thanks for all the help!

[T3sl4co1l]

Well... what are you expecting?  The circuit is built with two gain of 10 stages at the end.  Surely you are using this circuit because you intended that gain?!

Tim

[dannyf]

I think it has more gain than I expected

Maybe your expectation isn't right? The two stages have a combined gain of 100x.