Constant Current Source (Again?)

[theo92]

I saw all the previous threads regarding constant current sources. But none of them seems to fully help me.

I'm designing a constant current source to plot a V~I curve of an incandescent lamp. I did it previously using a constant voltage source using PIC's PWM and a BJT, but I'm not satisfied. Now I want a current source.

At first, let me clear my needs:

I need two current sources, first one would deliver low current, say 1mA to 15mA, to measure the cold resistance of the filament accurately. The next one would provide 100mA to 1.5A to let the bulb glow at it's full operating point. Accuracy for measuring the cold resistance is critical. I'll use PIC's onboard ADC to measure the voltage, and a MAX4372 or similar current sense amp for current measurement. I think I'd need two shunt resistor, one of higher resistance for measuring the cold resistance (as it'd be very low), and another of lower resistance to measure the higher current upto 1.5A. Will I need two different current sources for two different current range? Or can they be integrated into a single one?

I've searched a lot about the current source. Found multiple ways, and become confused. LM317 seems the easiest option, but it'll not the suitable for my needs. Next, is an opamp + MOSFET (or BJT)+ a sense resistor. Can you suggest me such a schematic with required equations? The current of the current source will be controlled through a digipot or PWM. I prefer digipot though.

And the analog side will powered from a 15V power supply, and the digital ones, from a 5V regulator of course.

 

[alec_t]

Will I need two different current sources for two different current range? Or can they be integrated into a single one?

I see you answered your own question!

The current of the current source will be controlled through a digipot or PWM. I prefer digipot though.

 

[ericgibbs]

hi theo,
I would say you would require two separate CC source for such a wide range of currents.

What is the resolution of you digi-pot.?, this could be controlled to increase the CC in steps.
I would consider it would be difficult to measure accurately the resistance of the lamp when using PWM

 

[theo92]

I would say you would require two separate CC source for such a wide range of currents.

Ok. I thought right!

What is the resolution of you digi-pot.?, this could be controlled to increase the CC in steps.
I would consider it would be difficult to measure accurately the resistance of the lamp when using PWM

I think an 8 bit digipot (256 steps) would be fine. And in my previous design, I used the PIC's PWM at 48KHz, filtered through a second order low pass, then a power transistor (I preferred a MOSFET, but a logic level MOSFET was unavailable here), and I had severe problems in measuring the resistance.

I need some CC basics, and an application schematic as per my requirements, I've written above.

 

[ericgibbs]

hi,
Which type of lamps are you trying to measure the cold and hot filament resistances, I see the two CC ranges.??

 

[theo92]

hi,
Which type of lamps are you trying to measure the cold and hot filament resistances, I see the two CC ranges.??

Yes, various kinds of incandescent lamps (vacuum or gas-filled or tungsten-halogen lamps) from a voltage range of 2.5V to automotive 12V.

 

[ericgibbs]

Yes, various kinds of incandescent lamps (vacuum or gas-filled or tungsten-halogen lamps) from a voltage range of 2.5V to automotive 12V.

The reason for asking about the types [ working voltages] is because of the compliance required in the CC circuit.
All this means really if a 12V lamp rated at 1Amp @ 12W was being tested, the CC source would to a higher supply voltage than 12V in order to maintain current control..
If it had been 230V lamps,, it would be difficult.!!

Are you able to acquire most types of PMOS fets and other OPA components OK, in WB.??

 

[theo92]

The reason for asking about the types [ working voltages] is because of the compliance required in the CC circuit.
All this means really if a 12V lamp rated at 1Amp @ 12W was being tested, the CC source would to a higher supply voltage than 12V in order to maintain current control..
If it had been 230V lamps,, it would be difficult.!!

Are you able to acquire most types of PMOS fets and other OPA components OK, in WB.??

Did you mean p-channel MOSFET by PMOS fets and OPA series opamps by OPA components? Yes, they might not be available in general shops, but I can purchase them through element14 (formerly farnell), they have a very wide range of stocklist.

No, I'm not going to control a 230V lamp anyway. I'll study the power laws (like R varies T^1.2 for vacuum lamps, and slightly changes for gas filled lamps due to convection of hot gases around the filament) between various parameters, like temp., voltage drop, current, power etc of low power lamps, 12V max.

 

[ericgibbs]

Yes, P Chan MOS and general purpose OPA's.

Are you asking for a fully working circuit diagram for these CC's.?

 

[theo92]

Yes, P Chan MOS and general purpose OPA's.

Are you asking for a fully working circuit diagram for these CC's.?

Well, my requirements are quite odd. So, an exact CC schematic might not exist. I need some hints, some sample circuits, and some relevant equations, so that I can customize it as per my requirements.

 

[ericgibbs]

Well, my requirements are quite odd. So, an exact CC schematic might not exist. I need some hints, some sample circuits, and some relevant equations, so that I can customize it as per my requirements.

hi,
These examples may suggest ideas. using LTspice.

The FDS PMOS is a dual, the only change is the MOS Source resistor 1K or 10R.

The MOSFET will require a 'very good' heat sink for the high current CC

I have assumed a 5K digit pot

 

[theo92]

hi,
These examples may suggest ideas. using LTspice.

The FDS PMOS is a dual, the only change is the MOS Source resistor 1K or 10R.

The MOSFET will require a 'very good' heat sink for the high current CC

I have assumed a 5K digit pot

Thanks a lot. Really a lot. I've seen the duo at a glance for now, they are just perfect for my purpose. I'll figure out deeply tonight. Meanwhile, can you provide me the equations which you used to calculate the value of the resistors?

You solved me, really

 

[theo92]

I've an idea. Both the two designs can be integrated into a single design. There'll be a relay connected to a PIC pin. The relay will connect one of the two resistors (by given signal from PIC) to the source pin of the FET.

 

[ericgibbs]

hi,
I would consider using a P JFET as a constant current source for the digipot drive, like these examples.

It would help to make the digpot response more linear

NOTE: for the high current CC, the source resistor is running at ~25W !!, so choose a high power resistor, also for the high current CC it will be necessary to increase the 15V supply.

For example at say 1.5A with a 8.7R thats a drop of 13V! so to have 12V across the Lamp and some across the FET you will require approx 30V.!
So choose a power P MOS FET Vds rating and OPA to suit.!!

These circuits are only a guide.

 

[theo92]

Ok. I'm checking these out. What should be the Vds rating of the pMOSFET for optimum performance? And what why do you suggest OPA?

 

[crutschow]

I would recommend a variation on Eric's circuit with feedback around the MOSFET for better current stability and accuracy as shown on page 2 of this reference. You can replace the JFETS with MOSFETs.

The constant current simply equals Vctrl / R for the left circuit and (Vsupply - Vctrl) / R for the right circuit. This minimizes any variation of the current due to transistor Vgs drift from temperature, etc.

 

[Claude Abraham]

This sounds spooky & familiar. I developed such a circuit in the 90's. Are you a pro? Are you working for an OEM or a contractor hired to develop a lamp driver? If so, why are you asking here? Mostly hobbyists here. I'd suggest hiring a well established analog development (hardware) EE.

My driver is proprietary but not patented, & I might be liable if I divulge it. But driving incandescent lamps involves knowing the I-V curves, life vs. temp curves, color temp curves, etc. Also turn-on when cold can be a problem so a constant warming current is employed. Even if I could help you, it wouldn't be for free. This is pro EE arena, & practitioners who do this are paid. Just thought I should mention it. Best of luck.

 

[theo92]

This sounds spooky & familiar. I developed such a circuit in the 90's. Are you a pro? Are you working for an OEM or a contractor hired to develop a lamp driver? If so, why are you asking here? Mostly hobbyists here. I'd suggest hiring a well established analog development (hardware) EE.

My driver is proprietary but not patented, & I might be liable if I divulge it. But driving incandescent lamps involves knowing the I-V curves, life vs. temp curves, color temp curves, etc. Also turn-on when cold can be a problem so a constant warming current is employed. Even if I could help you, it wouldn't be for free. This is pro EE arena, & practitioners who do this are paid. Just thought I should mention it. Best of luck.

No, I'm not a pro, not hired by anyone. This is my own project to experimentally arrive at the various exponent rules for various kind of lamps. Actually, I've made a model for the gas field lamp, taking Langmuir stealth and conduction loss into consideration. I'll see whether it agrees with experiment. My first prototype version was PWM controlled, I was not happy with it. So, I'm switching to a CC source. The device has a USB connectivity, and a host software, which allows to export the gathered data directy to mathematica notebook format, to plot the results just by a click!

I'm an undergraduate student of physics. There's semiconductor electronics in my course in this year. I've a little less basic sense about semiconductor basics, though I expect I'd know a lot about them at end of the year. I'm a peculiar hobbyist in nature! I've worked with PIC's a lot. Analog electronics is like a scattered card puzzle to me (for now, will not be after a year!), Eric solved a part of it, I'm exploring his circuit.

I'm thankful to electrotech, digital-diy and eevblog, I've learned so many things from them, and still.

 

[ericgibbs]

hi theo,
You realise that as you driving the lamp filament with a constant current that it would be possible to directly measure the voltage drop across in the filament in order to determine its resistance at that current.??

Depending upon the 'stability' of the required constant current, a negative feedback circuit from the output may be required.

As you are using a PIC/MCU for the calculations, it would also be possible to connect a low value precision resistor in the low end of the lamp.
By measuring the voltage drop across this resistor you could determine the 'actual' CC current, then measure the voltage across the filament and resistor, subtract the resistor Vdrop from that reading. Then calculate the filament resistance from those values.

 

[Reloadron]

Hi Theo. didn't you care for the ideas shared in this thread? If you look at the suggested circuit you will find striking similarities in the design. Additionally how Eric suggest measuring the voltage drop for a known current to derive the resistance.

Ron