Negative charge pump

[Kevin Weddle]

Here is something a drew up for those interested in a negative voltage supply made with a charge pump.

[audioguru]

Hi Kevin,
Like I said in your other post about a simple amplifier, your 3 left side transistors will work well making -12V pulses, when it is driven with a square-wave oscillator like a 555.
I don't know what the right side transistors will do. I think that the PNP transistor at the extreme right side will do nothing but breakdown since its base is much more positive than its emitter!
Did you try it?

It might be best to make a simple 2-transistor 12V p-p push-pull high current amplifier that drives a voltage doubler that is made with rectifiers.

[Kevin Weddle]

Audioguru I have great news. I tested this circuit using 5 volts and it works. I could not hook up the two circuits together because of the lack of breadboard space. However, I tested them independently and found the first circuit to give me -3 volts and the second circuit to give me -9 volts. All from a +5 volt supply. I'll post the resistor values. I had to use the square wave generator as opposed to the clock for current reasons I believe. The -9 volts from the second circuit was made with -5volt supply. As I said I had to run the two circuits separately.

[Kevin Weddle]

Here is the new circuit. I hope to build it soon. You might experiment with different resistor values and bigger transistors and larger capacitors for maximum current capabilty. Please disregard the opamp biasing.

[audioguru]

Hi Kevin,
I am glad that your circuit worked with a +5V supply.
However, I have analysed the 1st half of your last circuit and found 2 serious problems if it is used with a higher supply voltage:

1) When the capacitor is charging, nothing limits the current. The transistors might be destroyed.
2) When transistor T3 conducts to transfer the capacitor's charge to a negative pulse, the absolute maximum negative bias voltage of the B-E junction of transistor T2 is exceeded, without current limiting, and the transistors might be destroyed by both the excessive voltage and the unlimited current.
Please see my sketch:

[audioguru]

Hi again,
In your more detailed schematic, the opamp used to feed CL1 is driving two 100 ohm resistors with a total output current of about 226mA, much too high a current for an ordinary opamp.
If you use a power audio amp to provide the required current, its output swing certainly wouldn't be anywhere near 0V and +12V.

[Kevin Weddle]

Audioguru, I am having trouble transfering the charge to ground. I can only get slightly over -3 volts from the first circuit. What could the problem be?
I added a PNP to the first circuit during the charge transfer to help get it at ground potential. If I get it working I'll repost the circuit.

[audioguru]

Hi Kevin,
Which opamps are you using to provide such a high output current?
Their outputs certainly won't swing from 0V to 5V unless they are rail-to-rail types and are driving a low-current load.

You are losing a B-E voltage drop on emitter-follower transistor T1, and maybe more voltage everywhere else, because the transistors don't turn-off properly with a "0V" voltage of 1V when the outputs of the opamps go low.
Please see my attached sketch:

[Kevin Weddle]

I have tried everything it seems. I used emitter follower transistors on the opamp outputs. I have put diodes in to get rid of base current when the output goes low. I even put that other transistor as I was wondering if it needs it to put the capacitor at ground. I think when you disconnect the other end the voltage does not do anything. But I have now two transistors that put the capacitor at ground.

[audioguru]

I have tried everything it seems. I used emitter follower transistors on the opamp outputs.

Then you have reduced the opamp's output voltage swing from 1V to 4V to only 1.7V to 3.3V.

[Kevin Weddle]

okay, I fixed the first part of the circuit for the negative charge pump. I was missing a diode in there. Here is the circuit without the resistor values shown. You can play with the resistor values, but I will expand the circuit as I promised and post it.

[audioguru]

Hi Kevin,
Your "fixes" to your circuit will cause big problems:
1) You have shorted the bases of the first transistors together so now the upper transistor cannot be an emitter-follower. Each transistor should have its own base resistor.
2) The diode that you added to the first transistor to prevent it from reverse breakdown causes another voltage loss for the charge-pump cap:
a) 0.7V for this diode.
b) 0.7V for the upper transistor's B-E.
c) Maybe 1V for the upper transistor's base resistor.
d) About 0.1V (if you are lucky) for the lower transistor's saturation voltage.
e) About 0.1V for the right transistor's saturation voltage.
f) 0.7V for the opamp's buffer transistor's B-E.
g) About 1V for the opamp's saturation voltage.
Therefore the total voltage losses are at least 4.3V. So your "-12V" will be only -7.7V or less and will have a very high ripple. Reducing the ripple with another diode and big filter cap will reduce the voltage to -7V or less.

Your charge-pump transistors don't have a resistor from their base to ground to turn them off quickly.

[Kevin Weddle]

Okay, here is the tested circuit. It delivers about -8volts. The frequency of the square wave is undetermined.

[audioguru]

Good, now you have a working circuit.
Can you get enough current out of it to supply an amplifier?
When it is fully loaded by the amplifier, how low does its output voltage drop?
What is its input current at full load?
How much ripple on its output?

[Kevin Weddle]

Hi audioguru. We have circuit building problems. I have transistors and opamps that go out. You know how it is. I really get tired of these problems. But I have some interesting results. I did manage to double the circuit, basically, and get -10volts. The output of the first stage was -3volts into 10 ohms. That is 300mA without much ripple. The current is way too low for my ultimate goal of producing an inline amplfier for the speakers. I wanted a simple solution. Would it be possible to bump this circuit up a bit to provide a solution to amplifying beyond the capabilty of the radio. I think this is something that everyone would be interested in.

[audioguru]

Hi Kevin,
I have always used everything within their ratings over my fairly long career and have never burnt-out a transistor, opamp nor LED.
Perhaps your problems are caused by neglecting the ratings.

I don't think it is possible to build a simple and efficient high-current DC voltage doubler, that's why car radios have been using bridged amps for years. When the amps are bridged, the signal voltage across the speaker is doubled. A bridged amp is simple, they come in a single IC.
Check around and you'll find that most car radios have bridged amps. They don't sound loud and powerful because the cars come with cheap 20 ohm speakers. Driving half-decent 4 ohm speakers, a bridged amp sounds like it has 10 times more power than with the original speakers.
Many bridged amps can drive a 2 ohm load, for double the power again.

For those people whose car audio systems rattle windows a kilometer away, their very high power amps have a Switched-Mode power supply where a high-frequency high-current square-wave is stepped-up with a transformer to a much higher voltage. Then it is rectified and filtered and used to supply high-voltage high-current amps.

[Kevin Weddle]

Hey audioguru, everything is copasetic. Here is my lower current negative charge pump. The first part is very similar to the second delivering a total of -10.3 volts from 0 and 12. Take a look. The NPN's are 3 A becuase that is all I have left. The PNP's are about 500mA. The capacitors should be fairly large for maximum current.

[audioguru]

Hi Kevin,
I don't know what your transistors are supposed to do.
1) The transistors at the bottom are fighting each other because the NPN one is pulling to -7.3V while the PNP one is pulling to ground at the same time. They should be push then pull like the input stage.
2) That NPN transistor at the bottom gets only 1.1mA of base current so it won't do much. It also doesn't turn-off very well with 0.66V of forward bias.
3) The PNP transistor at the upper right can be replaced by a diode.

[Kevin Weddle]

Audioguru, I simply removed the transistor and increased the voltage output from -10.3volts to -13.3volts. I know what you mean by the NPN that transfers -7.3. I think that the resistors could need fine tunning. What is interesting is that in spite of the fact I got
-13.3volts, there was a voltage drop from -7.3volts to -5.5volts. So I think the NPN that transfers the -7.3volts is doing it's job somewhat efficiently.

[audioguru]

Hi Kevin,
What are you going to use the negative 13.3V for? ???
Car radio audio amp ICs don't use a negative voltage. ::)

[Kevin Weddle]

Yeah. I think the amplifier is a pushpull. The volume is adjusted internal to the chip. The amplitude of the input is taken into account and it drives the pushpull.

[audioguru]

If the car radio is new, maybe it has an all-digital Class-D switching audio amplifier. They are so efficient that a small surface-mount package can provide 4 bridged amps without requiring a heatsink.
I don't want to touch those amplifier ICs because they have 56 tiny legs and their "power-pad" bottom is also soldered to the PCB. Try doing that by hand!

My car's factory radio/CD player must have 4 bridged amps because both wires for each speaker carry the signal. After replacing the crappy high-impedance tiny-magnet factory-installed speakers with half-decent 4 ohm big-magnet ones, I get more than enough power/volume.
Yeah, its volume control must be digital too, because you can turn it around and around. It must provide "up-volume" or "down-volume"
pulses using an optical or hall-effect sensor then stores the setting in memory.