I still have work to do on my Kelvin (4-wire) Milliohm Meter project, one item of which is to improve the accuracy of the meter when measuring below 3 milliohms. I suspect the problem may be that the instrument amp is too close to the ground rail. To correct this I ordered a LM2662 which is a CMOS charge pump that can act as a voltage doubler or as a voltage inverter. This will allow me to create a negative rail which hopefully will improve accuracy.
I was brushing up on how charge pumps / voltage inverters work and decided to do a little mini project this evening with a 555 timer (newer version which can operate down to 2V and frequencies to 2MHz) to create my own voltage inverter. Here is the circuit:
Note: Although I am not using polarized capacitors they are being indicated on the circuit diagram for clarification should they be used. I connected it up on a breadboard using parts I had on hand and powered with 3V3 from my bench power supply.
- R1 = 1k
- R2 = 6.8k
- C1 = 100p
- C2 = 0.1u
- C3,C4 = 2.2u
- D1,D2 = IN4001
I need to order some Schottky diodes :-). Lo, and behold, it worked (at least with no load). Here is the output with the oscilloscope probing pin 3 of the 555 Timer and the -3V3 output.
The 555 timer is providing a 3V3 507KHz square wave with roughly 54% duty as seen on the yellow trace. The calculated frequency for the 555 Timer output was much higher but I expect my breadboard is adding capacitance and throwing off the calculation. The blue trace is the negative output measured to be -2.72V. The DC output with 20 mV divisions is shown in the next oscilloscope screen shot.
The noise is all within a 20mV band. And here is what the breadboard looks like:
The pencil is pointing at the LM2662 that I bought for the project.
It should be noted that the circuit with the listed components gives poor response unless the load is very light. See the simulation by Jon Clift in the comments as well as new experiments by me. The following table gives experimental results for different loads.
|Load (Ohms)||Pin 3 Square Wave Ouput (V)||Circuit Output (V)|
|No Load||3.3 V||-2.7 V|
|33k Ohms||3 V||-2.1 V|
|3.3k Ohms||2 V||-0.6 V|
|1k Ohms||1.8 V||-0.2 V|
Charge pumps and voltage inverters seemed like magic at first and it took me a while to catch on to what was happening (mechanical engineer here). I may order some Schottky diodes and inductors to explore other topologies when I have a bit more time. Comments always welcome.
EDIT: 2018-11-28 Realized I did not label all the pins on the schematic and updated it. Showed polarity in case electrolytic or tantalum capacitors are used. Added the load table from the Comments below. Clarified 555 Timer version and Diodes used.