Custom pre-driver amplifier circuit.

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi everyone. I am a newbie on this forum. I am posting here to receive some feedback as to how my custom pre-driver circuit can be improved. This circuit works but it suffers from hum although this goes as soon as there is an input signal is present. Connecting a mobile phone as a sound source does not produce a mains hum. However, certain tracks play with significant noise.

I think, the solution lies in making a more complex input stage. I have seen some circuits that even employ positive feedback through a 10 Ohm resistor that is connected to ground and then to both the non-inverting and inverting inputs of the input differential pair but this is done indirectly.
 

Attachments

  • custom-pre-driver-circuit.jpg
    custom-pre-driver-circuit.jpg
    447.4 KB · Views: 461
Moderator
Joined 2011
I have seen some circuits that even employ positive feedback through a 10 Ohm resistor
that is connected to ground and then to both the non-inverting and inverting inputs of the input
differential pair but this is done indirectly.

The 10R isn't for positive feedback, it just breaks the ground loop that causes the hum.
That's just what you need. It goes from the input RCA ground shell (with no other connections to it)
to the audio circuit common. Avoiding Ground Loop Hum
 
Last edited:
Your hum might be oscillation. Your input resistor should be 10k (right first time) to match the feedback resistor.
Add a 100pF capacitor across it to keep impedance low at high frequencies.
Possibly include a470 ohm to 1k resistor in series with the input for the same reason, this creates an RC filter with even more HF suppression.
Your 10pF capacitor to suppress oscillation is not large enough without emitter degeneration in the input stages. Either add 100 ohms in series with each input transistor emitter or make the capacitor 100pF for starters.
 
Your hum might be oscillation. Your input resistor should be 10k (right first time) to match the feedback resistor.
Add a 100pF capacitor across it to keep impedance low at high frequencies.
Possibly include a470 ohm to 1k resistor in series with the input for the same reason, this creates an RC filter with even more HF suppression.
Your 10pF capacitor to suppress oscillation is not large enough without emitter degeneration in the input stages. Either add 100 ohms in series with each input transistor emitter or make the capacitor 100pF for starters.

Keep in mind very high gain here: 10K/100 +1 =near 100 ... Current gain in amp stage is aproximately 10. At the same time current gain in vas stage is near 10 to 11. That i consider on the low side.
 
Last edited:
The local feedback from the 90 ohm resistor helps to keep the frequency response high in the VAS, so there is still a lot of gain at HF. Needs a bigger compensation capacitor.
Not sure I follow the diagram correctly.
Are the resistors really 90 ohms and 69 ohms in the VAS and CCS - NPV's are 91 and 68.
Is the current source in the input stage really 480 ohms should this be 4.7k?
Why use a 10k across the base-emitter of the VAS?
 
Excuse me for not explaining how my circuit works. The following is an explanation. The differential pair are supplied current through a chain connected to the negative side of the split power supply. Current first goes through two series connected diodes which produce a pseudo-constant voltage drop of around 07V x 2 = 1.4V. This is then used to directly drive the base of the final stage's constant current source. Moving upwards along the same chain, current first flows through a 480 Ohm resistor and then it branches into a 5.1V Zener diode that has its other end connected to ground and a resistor that supplies current to the differential pair. Since the Zener diode is essentially in parallel with the Base-Emitter junction of the input transistor and the 2.2K resistor, the current in the current source should be (5.1 - 0.7)/2200. The voltage across the input resistor that is connected to ground is in the range of a few micro-amperes, so its voltage drop can be ignored for calculation purposes. The output from the differential pair is developed across a bias diode that provides a guaranteed drop of 0.7 volts and a 1K resistor in series. The next stage is the pre-driver stage. The output from the differential pair is connected directly to a PNP transistor with it an Emitter resistor of 90 Ohms. Since we have a 0.7 volt voltage drop across the Base-Emitter junction and the diode at the output from the differential pair, the voltages that vary with a signal are across resistors, the 1K and 90 Ohm resistors. This means, the current gain is defined solely by these resistors which is around 1000/90. The 10K resistor connected across the Base-Emitter junction is to guarantee that the transistor's capacitance discharges when dV/dt is very large. In such a situation, the transistor is alsmost a switching transistor. Needless to state, I added this resistor after I found it greatly improved performance. Without it, there was distortion that could be noticed. Regarding the 10pF capacitor at the output instead of the usual 47pF - 300pF capacitors, I tried to use the standard range of capacitance but the circuit misbehaved. The output from the PNP transistor is connected to a constant current source with a preset current of around 0.7/69. This arrangement is common, so there is no need to re-explain it here. The dangling middle transistor is to provide an accurate bias for the driver stage.
 
The 10K resistor connected across the Base-Emitter junction is to guarantee that the transistor's capacitance discharges when dV/dt is very large. In such a situation, the transistor is alsmost a switching transistor. Needless to state, I added this resistor after I found it greatly improved performance. Without it, there was distortion that could be noticed. Regarding the 10pF capacitor at the output instead of the usual 47pF - 300pF capacitors, I tried to use the standard range of capacitance but the circuit misbehaved. The output from the PNP transistor is connected to a constant current source with a preset current of around 0.7/69. This arrangement is common, so there is no need to re-explain it here. The dangling middle transistor is to provide an accurate bias for the driver stage.

Well if the 1K resistor could not do it togheter with the 90 Ohm in emitter. the 10K adjust bias and affect switching time by less than 10%
But great work Ua741
 
I was questioning the value of the 480 ohm resistor. The current flowing in it will be
(18-1.4-5.1)/480=24mA which seems high.
The tail current in the LTP is approx. 2mA. 5mA would be enough for the 5.1V zener diode, totalling 7mA which would only need a 1.5k resistor.
The distortion in such a circuit depends on the differential base-to-base voltage in the LTP.
Your 2.2k resistor is not a long enough "long tail" to keep the distortion low: there will be significant modulation of the tail current due to voltage modulation across this resistor.
You should use a proper CCS.
So my question is what loading have you got on the output to require the signal currents in the LTP to deviate from the quiescent values so much that there is high distortion.
If it is connected to an output stage it would be useful to see how/what.
 
...also the LTP currents are not balanced. Because of the high current in the 480 ohm resistor the diode voltages could be higher than 1.4V, so the current in the CCS could be 12mA. That causes a 1.1V drop across the 90 ohm resistor. If we assume that the PNP VAS transistor base -emitter voltage is about the same as the 1N4148 voltage in the base then the current in the left hand transistor is going to be 1.1mA and that leaves 0.9mA in the right hand side.

The 10k base resistor you mentioned makes this even worse, and is only going to make a minor difference, compared to 1k, so if you found it made a big difference there is something wrong, somewhere.
 
All these replies show that I should consider redesigning the circuit paying special attention to:
a) use more negative feedback in the differential pair. An emitter 100 Ohm resistor was suggested earlier in this thread
b) use a compensating capacitor of 100pF in the final stage
c) remove the 10K Ohm resistor shunting the BE junction of the final stage
d) increase the overall current gain of the amplifier by reducing the 90 Ohm resistor. I am thinking about using a 30 Ohm resistor instead. This multiplies the present current gain by a factor of 3
e) reduce the global voltage gain of the amplifier instead of keeping it at 100 (gain is NOT expressed in dB)
f) replace the 1K Ohm resistor at the input with a 10K Ohm resistor. As suggested, a 100pF capacitor will be connected in parallel of this resistor
g) use a 470 Ohm resistor in series of the input.
h) use a 10 Ohm resistor between the common ground the global feedback ground and signal ground.

While thanking everyone for their contribution, I will now redesign the circuit according to the above and post BEFORE I actually build the circuit.

P.S.1.
I learnt in another thread that differential pairs and current sources suffer from various problems. Since, 'my' circuit it built with these two, it makes sense to try to mitigate them. One such problem is current sources tend to be slow due to a high capacitance.

P.S.2.
I was asked what power stage this circuit will drive. The power stage is a large ouput stage of a pre-existing amplifier. It has 4 2SC5200 and another 4 2SA1943 transistors to drive the output. The emitter wire-wound resistors connected to the positive and negative terminals of the split power supply. The Class AB configuration is achieved in the driver stage which supplies current to two 22 Ohm resistors connected to the positive and negative terminals of the power supply. These resistors are connected to the bases of the 2SA1943 and 2SC5200 transistors. The emitters of the driver stage have a low resistance to further linearise the circuit.
 

PRR

Member
Joined 2003
Paid Member
All these replies show that I should consider redesigning the circuit paying special attention to:.....

I would strongly suggest the Douglas Self book on power amplifiers. It may not be the Last Word on fine audio power, but it touches all the basics (like unbalanced input pair).

With the Bob Cordell book for dessert. He touches many of the same topics but with different words and side-branches.

You can learn LOTS more from these (and similar) books than by random forum chatter.
 
I would strongly suggest the Douglas Self book on power amplifiers. It may not be the Last Word on fine audio power, but it touches all the basics (like unbalanced input pair).

With the Bob Cordell book for dessert. He touches many of the same topics but with different words and side-branches.

You can learn LOTS more from these (and similar) books than by random forum chatter.

You are telling to read and study. I will do just that and thank you for the reading resource. I remember, a long time ago, when I was still a youth, a old man told me 'a book is a teacher'.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.