Need a stable 10 uA current source

Thread Starter

Wolframore

Joined Jan 21, 2019
2,273
I need a stable 10 uA source for a regulation circuit. I’ve played with self biased JFET current source but concerned about stability over a range of temperature. I don’t mind calibrating for manufacturing variations so mostly concerned about temperature stability. It will likely be used in stable environments and +/- 10 % over 20-40C should be ok. Is it possible to run thermal analysis in LTSpice?
 

crutschow

Joined Mar 14, 2008
27,218
I would use an op amp/MOSFET current-source with a precision voltage reference.

Can the load be floating or does it have to be grounded?

Yes you can change the simulation temperature in LTspice to see the effects of that on the semiconductors.
You can also assign a temperature coefficient to the resistors (search for temperature in the LTspice help file).
 

Thread Starter

Wolframore

Joined Jan 21, 2019
2,273
I've tried this... the global temperature appears to have no effect on output current. I tried .TEMP 20 and .TEMP 80. This is the circuit I'm using.

1610488655807.png
 

Thread Starter

Wolframore

Joined Jan 21, 2019
2,273
I've also tried stepping - it looks like a rock solid output although the datasheet shows 250ppm/C temp stability.
1610488936240.png
1610488961182.png
 

Thread Starter

Wolframore

Joined Jan 21, 2019
2,273
I have tested some of the circuits using a hot air wand to raise the temp and I am seeing a drop in current using a standard LDO.. I don't have the parts to try this JFET out yet.
 

Thread Starter

Wolframore

Joined Jan 21, 2019
2,273
Must be something up with that model, I was able to get this to work with another JFET model, but shouldn't it have affected the resistor even a little?

1610490858776.png

1610490875518.png
 

crutschow

Joined Mar 14, 2008
27,218
shouldn't it have affected the resistor even a little?
How are you determining the effect of the resistor?

It you are taking about you first simulation then you need to delay the transient simulation (Time to Start Saving Data) a few ms, since there is a large initial transient that's suppressing the small changes in the plot.
 
Last edited:

Thread Starter

Wolframore

Joined Jan 21, 2019
2,273
Do you have to do something special? I thought the global temp would affect everything including the resistor.

post #4 shows no change to output current with changes to global temp.
 

crutschow

Joined Mar 14, 2008
27,218
Do you have to do something special? I thought the global temp would affect everything including the resistor.
Yes.
The global temp affects resistors but the standard model is a perfect resistor.
You have to define the resistor change with temperature. LTspice doesn't know that unless you tell it.
Below is the help file listing for that:
1610501572680.png
post #4 shows no change to output current with changes to global temp.
Try delaying the transient analysis as I suggested in my previous post.
 

Thread Starter

Wolframore

Joined Jan 21, 2019
2,273
Thanks @crutschow i will try to determine the correct coefficient for 0805 SMD resistor.

I will try delaying the transisent analysis. Have you also seen strange transients as at the end of analysis?

Exactly how stable do you need the 10μA to be?
I‘m a perfectionist unfortunately drive myself crazy at times, but 10% between a reasonable indoor ambient should be fine for this. I can see future uses where I would like to offset the thermal effect and get somewhere under 1% between 20-80C.
 

Thread Starter

Wolframore

Joined Jan 21, 2019
2,273
I made a variable resistor that sweeps by making R=10k-(10k * time) at time = 1 there is a large transient that can’t be explained. Looks like a large inductive spike.
 
Last edited:
Top