View Full Version : controller Charge Pump?

- 17th May 2005, 20:34
Hey guys I am looking for something that can act as a charge pump. Bascially I need an IC/Design that can pump out a definite quanta of charge each time it is activated. Activation would come from a voltage source.

The maxium continous rating would be less than 10 mA. Hopefully you guys know an IC that can be of use.

Ron Marcus
- 20th May 2005, 03:11
I need more info. Why a charge pump? What is your initial voltage? What is your needed voltage? I have run a 5 volt LCD screen from a 3 volt PIC using the PWM output. There are app notes for pushing out over 200 volts using a step up coil and diodes from a lowly PIC!

I like coils for step up, but the EMI can be a problem. What is the application?


- 20th May 2005, 16:43
the application is a bit odd. it is to control an oxygen sensor is a car, basically i need to source enough charge such that the catylitic reaction in the oxygen sensor is at steady state.

The amount of charge i need to pump in to reach that steady state is proportional to the amount of oxygen left over from internal combustion. it is this amount of oxygen that i am intersted in measureing. Most designs use voltage source pushing current through a small resistor to supply the current needed, and then use A/D to find to voltage drop across the resistor to find the steady state current.

But I would like to do away with using A/Ds, if i had a way to use PWM to control a charge pump or current source that would make things alot simplier.

I need something that I can control percisely to do the following:

supply: -4ma to +4ma
the voltage range of operation would be between -5V to 5V.

I am a bit stuck in finding something, I can explain things in more detail if needed but basically I need a PWM controlled current source/charge pump, even a voltage controlled current source/chrage pump is okay, My only requirement is that the resolution and percision need to be high and it should not be very complex.


Alan To

- 21st May 2005, 09:25
What about using a standard zirconium lambda sensor?

Sensor output ranges from 0.2 Volts (lean) to 0.8 Volts (rich).
A perfectly balanced or "stoichiometric" fuel mixture of 14.7 parts
of air to 1 part of fuel gives an average reading of around 0.45 Volts.


- 21st May 2005, 14:47
I am wokring with a wideband sensor and not the narrow band sensors.

- 21st May 2005, 15:19
See this link:

- 24th May 2005, 02:52
that is the techedge one, I am not looking to "bite" his design. His design is overly complicated in my opinion.

- 24th May 2005, 18:52

Feel free to "bite" the BOSCH design.
(See the block diagram of the IC CJ110).

But, doesn't it look familiar somehow?


* * *

CJ110 - Lambda Probe Sensing IC for LSU4.

CJ120 - Lambda Probe Sensing IC for LSU4.x
Ri Measurement; SPI-Interface; Diagnostics


* * *
Bosch LSU 4.2 Planar Wide Band Lambda Sensor:

* * *

- 24th May 2005, 19:34
humm, I was gonna do the design myself without the BOSCH IC, I was looking around for vendors of the BOSCH IC and never found any.

- 24th May 2005, 20:43


The CJ110 and CJ120 are on this list:



Robert Bosch Corp.
38000 Hills Tech Drive
Farmington Hills, MI 48331
By telephone: 248-553-9000
By fax: 248-553-1426

- 24th May 2005, 22:21
TY for the info.

- 25th May 2005, 18:31

Print the page 3 and 4 of the CJ120 PDF.
(The Block diagram page and the Application circuit page).


Add to the application circuit the two op-amp's and connect them
as they are depicted on the block diagram page.
(Draw that with a pencil inside the CJ120 on page 4).

The two op-amp's:
Pumpstrom-Regler = Pump current control
Pumpstrom-Meßverstärker = Pump current sense amplifier


PIN Vcc = +5 V
PIN VM = +2.5 V (Virtual ground voltage source)
(0.5 x VCC)
PIN US = +2.95 V (Nernst cell reference voltage source)
(450mV, reference to virtual ground)

The two op-amp's have a single polarity supply.
(+5 and GND).

On the output of the op-amp "Pump current control"
you will need a 100 ohm resistor.

* * *

The idea is to replicate what Bosch is doing inside the CJ120.

You will need a way to simulate the LSU4 during the testing of the circuit.
(DC +/- milliampere meter and a voltage that goes from 4.0V down to 2.5).

Do you understand how it works?


- 25th May 2005, 20:51
I have not seen the additions you have made on the PDF because the current computer I use gives an error with the file, I will look at it later on a nother computer.

I pretty much understand how the circuits works. I have made a citcuit long time ago that was very similar to what the Bosch ICs are doing. That was with analog circuits right now I am trying to do most of the Analog stuff using a MCU so to minimize tolerance problems and to cheapen things up and also to add dataloggin capabilities to it.

I built my first circuits based on the Bosch Datasheet for the Sensor itself because I never knew the Datasheets for the ICs existed publicly. The circuit I built before knowledge of the Cj120 datsheet was similar in many ways to the CJ120 but more simplier. Now that I seen the CJ120 the only thing I can not understand is on Page 3 or 4 of the CJ120 PDF, the use of Rical (82 ohms). The fact that I can not speak German only adds to the problem. But I am sure somewhere in my subconcious my mind is solving the issue, so it will only be a matter of time before I know what the RiCal does. I suspect it maybe a reference restance that the resistor in the pump Cell (Ri) should be at during normal operations. The reason I say this is becasue I know pretty sure that the Pump resistance should be close to 80 ohms during normal operation and not the 100 ohms that is shown on the diagram.

There is no need to simulate the BOSCH LSU sensor in my testing since I have about 4 of them lying around. Also the Pump current goes through an external 61.9 resistor, I suspect the 100 ohms you are talking bout is the internal resistance of the pump cell.

I would love to by the bosch ICs from rutronik but they require huge quantity purchases something in the thousands per order.

If I can digitize most of the analog circutry I can make a very cheap, fast and accurate Wideband reader.

- 25th May 2005, 20:58
BTW Luciano TY very very much for posting up the documents, because of the new information I found in there I am going to make a few design adjustments to my circuits and MCU program.

- 25th May 2005, 22:18
================================================== ======
My previous post:

The PDF was not modified. You will have to draw the two op-amp's
and their connections on page 4. The 100 ohm resistor on the output
of the op-amp "Pump current control" is present in the Tech Edge circuit.
================================================== ======

The CJ120 also measures the Nernst-cell internal resistance (Ri) which
is used to determine the temperature of the sensor. The internal resistance
of the Nernst-cell is 80 ohm when the sensor ceramic temperature is approx. 750°C.
By knowing the (Ri) you can control the heater supply voltage so that the
temperature of the sensor is kept at a nominal temperature of approx. 750°C
which is needed for *precise* measurements.

The Nernst-cell internal resistance is measured by pulsing the Nernst-cell via a
capacitor and a resistor.The pulsing must be done between 1kHz and 4kHz.(CJ120 3kHz).

On the Tech Edge Pty. Ltd. schematic you can see that they measure the Nernst-cell
internal resistance and also measure the effective heater voltage and current.

On the CJ120 block diagram you can see that the effective heater voltage and
current are not measured. The document http://wbo2.com/lsu/Y258K01005e03mar21eng.pdf
describes the maximum permissible heat up rate when the heater is switched on.
When you know the Nernst-cell internal resistance then you can control the heater
without measuring its voltage and current.


- 25th May 2005, 23:38
I have read the datasheet on the sensors many times before.

My original circuit had PID control of heater to maintain the 80 ohm resistence on the nermest Cell with a 3 khz ac signal, I made a mistake in my previous post about the 80 ohms in the pump cell when it should of been the nermest cell.

- 26th May 2005, 03:42
I just looked at the tech edge diagrams in detail, everythin looks fine a dandy except I did not notice a coupling capacitor to couple the 3 khz AC signal to sense the Nermest resistance, the "VsDrive" connection connects through a resistance (910 ohms) to pin 11 of the atmel8 AVR MCU, if you are coupling an AC signal of 0v to 5v then for the bosch lsu you would need something in the 10k range and also a capacitor. Also Pin 11 of the atmel8 AVR is not a PWM pin, if he is ouputing an AC signal he must be doing it via software.

I am thinking that the TE unit actually powers off the sensor and does DC sensing of the the resistence in the nermest cell, which would make the unit super slow.

I am jsut puzzled because no where on the schematic are any hallmarks of generating an ac signal, coupling an ac signal into the nermest cell, or low pass filtering to target a signal in the 3khz range. The only filtering around the VS port are low pass filters with a 3bd freq of about 10khz.

- 26th May 2005, 21:23

See attached BMP file.

Nernst-cell internal resistance (Ri) measurement.

The circuit measures the resistance of the Nernst-cell and
also the resistance of the RiCall 82.5 ohm. Both measurements
are done with the same I/O, so we measure with the same voltage
when we source (+4.xx V) and when we sink (0.xx V).
The PWM is not used.

The AC current is ≤ 250 µA.
(See 250 µA current direction when the I/O sink or source).

The differential amplifier will output a voltage when
Ri and RiCall have different values.The output voltage is
measured with the A/D of the MCU. With the resistors you
can adjust the gain of the amplifier.

* * *

The Tech Edge does almost the same with the I/O but
they use a 910 ohm resistor so the current is more
than 250 µA. (???). The capacitor is the 100nF C201.
They use 2V as the virtual ground.

(See page 6 of the PDF for Max. current load of the Nernst-cell).

* * *

This is just a theoretical circuit.
Never been tested.


- 26th May 2005, 22:26
910 ohms would mean they are violating 250ua parameter by more than 10 fold with a 0-5v signal, c210 looks like it is just part of a lowpass filter to remove noise rather than to couple an ac signal into the nermest cell.

It looks to me that they really are powering off the sensor and doing a DC measurement of Ri. I also have no idea why they use an external DAC rather than the avaible PWM ports on the Atmega8 instead.

Luciano are you affiliated with Tech Edge? are you also building you own Wideband reader? Why are you using RiCal of 82.5? I believe it is a reference resistance for the nermest cell, shouldn't it be 80 ohms?

The way I am doing the Ri measurement is different as I try to use as little OP amps and resistors as possible and use a bit of digital signal processing instead.

- 26th May 2005, 22:56
I see a problem with the circuit of my last post. (BMP file).

When we are not measuring the Ri of the Nernst-cell, the output voltage of the cell +2.95V (+2.5V +0.45V) goes through the two 10K resistors and the 82.5 ohm resistor back to the 2.5V of the virtual ground. (We have 20082.5 ohm connected in parallel with the cell).
The circuit will need analog switches (like the 4066) in order to insulate the Nernst-cell.

Two years ago I helped my cousin on a similar project.
He married, his wife sold the car, the project never started.


- 26th May 2005, 23:19
oh yeah I was gonna point out the 2.95V thing but you beat me to it.

I recommend putting a 1uf or 100nf capacitor in series with the 10k resistors. The problem I see with the circuit without proper capacitors in place is that you require the Nermest cell to be perfectly .45V above your virtual ground. If the nermest cell is slightly below or above the .45V then your differential amp will output a difference. If you filter out the AC signal itself without the 0.45 required DC offset and use the diff amp on the ac signal itself then the circuit should be cool.

I made a mistake when I said that the TE unit powers off the sensor, It is jsut that I use coupling capacitors and when I did not see any I jumped to that conclusion.

- 27th May 2005, 16:11
Interesting products:


Page 1-2 Lambda Sensor Mini-LSU 4.9
Page 3 the AWS_LSU 4.9 (Sampling rate 100 Hz per channel).


- 27th May 2005, 18:49
interesting, looks like the old LSU just in mini form.

- 28th February 2006, 23:37
So basically, if I understand this entire deal...

The Bosch chip outputs a voltage called Vcj and this voltage allows you to calculate IP.

IP = (Vcj -1.5) / 1.053 This works fine!

Now that you know the IP, you must do a lookup using the values that Bosch provided in the lookup table for IP vrs. 02%

Cool, now how in the world do you make their Math work on Xo2 to Lambda?

They have a formula stating...

Lambda = (Xo2 / 3 + 1) / ( 1 - 4.76 * X02)

Then they show the following as references....

IP = 1.40 = 12.0% Xo2 = 2.42 Lambda

IP of 1.40 = 12.0% Xo2 (This works fine)
IP of 1.40 = 2.42 Lambda <- Does not work

This does not work given the above formula using 12.0% Xo2 or I'm doing Math Wrong.

Lambda = (Xo2 / 3 + 1) / ( 1 - 4.76 * X02)

X02 = 12.0
12.0 / 3 = 4 + 1 = 5
12.0 * 4.76 = 57.12
5 / (1 - 57.12) = 5 / -56.12 = -0.089

Where am I going wrong?

- 1st March 2006, 13:17
I need more info. Why a charge pump? What is your initial voltage? What is your needed voltage? I have run a 5 volt LCD screen from a 3 volt PIC using the PWM output. There are app notes for pushing out over 200 volts using a step up coil and diodes from a lowly PIC!

I like coils for step up, but the EMI can be a problem. What is the application?


Ron, how do you run a 5V LCD from a 3V PIC using PWM ? i'd love to try it out.

- 2nd April 2007, 13:39
April 02, 2007


I have completely abandoned this project.
Please do not send me private messages about this thread.

Best regards,