[SOLVED] Constant Current power Supply

Hi,

One post ago you decided swichig regulator, now linear.
One post before it was a 3.6V LiIon now it is a 12V lead acid.

And it seems to me you didn't calculate anything at all.With 12V and linear regulation now you have to dissipate 12W of power. This is much heat. Now you need a big meatal case with heatsink, and lifetime of lead acid battery is not longer than LiIon.

High quality, high reliability, low cost. You can't have all in one.

Klaus

- - - Updated - - -

Hi,

One post ago you decided swichig regulator, now linear.
One post before it was a 3.6V LiIon now it is a 12V lead acid.

And it seems to me you didn't calculate anything at all.With 12V and linear regulation now you have to dissipate 12W of power. This is much heat. Now you need a big meatal case with heatsink, and lifetime of lead acid battery is not longer than LiIon.

High quality, high reliability, low cost. You can't have all in one.

Klaus

To br honest,I am super confused on the source selection.
So I asked for your help on the selection criteria.
So could you please make a selection for me.
I need light weight instrument with good backup.
Now you suggest a effective way to solve the problem

Hi,

if you want to design such a measurement tool, then you should be able to calculate U, I, R, P and these basics.
The same is with battery backup time. A x h = Ah.

I gave you the selection criteria for switching regulators. But i can´t make decisions for you because i don´t know your needs, device availability, price and so on.

Klaus

Hey Klaus,
I will stick with the switching regulator with a 3.6v Lithium cell.
That is finalized for sure.
But Could we use a discrete switching circuit so as I can lower some cost.

- - - Updated - - -

But If we are using a switching regulator then there isn't a large heat disipation on the switching regulator.so a lead acid battry would work nicely.

Hi,

new simple trial:


you have a ucontroller.
use 8 bit PWM
? -> what max. pwm frequency can you generate?
feed the pwm to a halfbridge. either build your own, or buy a device. (VCC 3.3V, R-on < 200mOhms, ...)
L1 and C1 build a low pass to get a DC voltage to supply the DUT.
R1 and C1 are to suppress resonance frequency.
Connect DUT (0..200mOhm)
Slowly increase PWM from 0/256 starting, until ADC0 reads 2V.
Now DUT is supplied with 1A.
convert and average ADC0 (at least 32 samples)
convert and average ADC1 (at least 32 samples)
divide ADC1/ADC0
multiply the result by 200 and display the integer value "R" in mOhm

... for most of the missing values i need the PWM frequency.

use RR_in and RR_out low offset opamps.

for sure one can improve circuit and software... later...step by step.
For example a 1A polyfuse in the positive DUT supply to prevent form accidentely short to GND.

Klaus

Hello Klaus,
I am Working On the Current Section,The Half Bridge Method Seems a good type of Current controlled source.
A maximum Of 125Khz PWM signal can be generated.

Ashkar, I have serious suggestions about your design approach which is why you are no closer to a solution after 65 responses than the start.

1: Describe what you measuring in a meaningful way with a sense of purpose.
2. Define all the electronic, physical , mechanical and environmental variables for inputs and outputs.
3. Define scope of project. (bandwidth of experience, cost, quantity)
4. Learn from others is best way by Reverse Engineering to learn how then find how to define your own specs and design.

A bad way is say next to nothing and assume we know everything about your requirements. e.g. Some one kindly Suggest me a 5 Amperes Constant current power supply for Micro Ohm Meter.

for example;

I wish to test connection losses for a Large Solar Power Array and want to find where the micro ohm losses exist from DC to the SMPS charger rate of x kHz.

I need 10 uOhm resolution and 10 Ohm range, but would prefer 1uOhm resolution and 100 Ohm range with an accuracy of 0.5% full scale on 8 bits.

Then show photos of application and ask best way to solve measure in shortest time, cost or best accuracy or some other preference.

Click to expand...

Reconsider starting again.

View attachment k.bmp
Look at this current driver

Hi,

why don´t you use available schematics?
or use simulation tools?

and what should this circuit do?

Do you know how a PNP transistor works and a p-channel mosfet with it´s built in body diode?

SunnySkyguy gave a lot of helpful hints. use it.

We miss:
*a detailed description
*voltage, frequency, duty cycle of pwm signal
* values of all transistors.

Klaus

Hi Klaus,
I have build the circuit using a simulator but then I exported the sch to BMP for easy readiblity.
So this circuit is a single mosfet driven constant current power supply that provide a constant current to the DUT ,
To limit the Current I LL use ur method of reading the voltage from the 0.2 ohm shunt resistance and increase the pwm duty until a 1 amp current is approched on the driving mosfet.

Resolution- 8 bit pwm
Frequency -20 kHz

Transistor bd138 and bd140
Mosfet -irf540

Ashkar:

How much ripple voltage can you tolerate on your micro-ohm resistance measurements?
What accuracy do you expect?

How fast is measurement period? Does this slow down by averaging due to noise?

Does it heat up by conduction losses and therefore change resistance?

Get my point? you need to define a better question with a better design spec.

Sunny:
Acceptable ripple is 1to 100uV p-p
resolution -on a scale of 0-2000 milli Ohm range- an error of +_1digit only.
100 samples per second and their average is applied for a great accuracy using a 16 bit ADC.
I hope that I might rovide a good knowledge of my setup.

Thankyou for some of the answers to my questions

I think 2m to 20mV p-p ripple is reasonable, 1 to 100 uV is unreasonable. The achievement of this requires extremely low ESR cap and relative to Rs of choke and MOSFET switch.

To expect +/-1 digit of accuracy with +/-1 digit of resolution on 12 bits means error is < 1/2 digit decimal with 1/2 digit quantization estimate or 1/2 in 2000 or 1 in 4000 which is more than 72 dB SNR. Averaging 100 readings will improve only random noise by root (100) =10 or 20dB This requires extremely tight control of reference voltage, reference voltage noise , S&H noise.