Precision Constant Current Sink

[borg]

Do you guys now a place where i can buy a constant current sink?

I am thinking of a simple box with two connectors that draws something around 1A (at something around 10V) with a high precision (100ppm or better).

There are of course programmable electronic loads that can reach this precision, but they are quite expensive.

[Rerouter]

for 1A, that is only an error of 0.1mA at 100ppm, most of the generic designs are more than capable of this if a low offset op amp was used, (probably use a larger resistance value to decrease noise) say a 2 ohm resistor, that would correspont to a 200uV offset, which is well within most cheap low offset op amps, note however it would still need calibration,

[borg]

I think i wasn't clear enough, the voltage source will be around 10V, not exactly (i.e. it can change over time). If it works between 9 and 11V would be good enough. But the current sink should stay at 1A. The electronic load that we have here does exactly that (it keeps the current constant, while the voltage is changing), but it is not precise enough.

This electronic load would probably work: http://www.home.agilent.com/en/pd-1000001519%3Aepsg%3Apro-pn-6060B/300-watt-dc-electronic-load?nid=-536902252.536880312&cc=DE&lc=ger

But i had hopes that there is a cheaper method for a constant current.

[ejeffrey]

He said you can make a DIY constant current sink that meets those specs fairly cheaply, but you will need a reference to calibrate it.

[robrenz]

Check out these two threads:

https://www.eevblog.com/forum/product-reviews-photos-and-discussion/array-electronic-loads/msg88995/#msg88995

https://www.eevblog.com/forum/product-reviews-photos-and-discussion/why-is-bk-precision-still-in-business-bk8500-programmable-load-oem-found/msg139470/#msg139470

[PA4TIM]

1A is not much, with a mosfet, a precision shunt with low tempco and a precision opamp that must be possible.
Problem will be that if you have a small shunt the shunt will stay most stable but the voltdrop will be very low. And if you use a larger shunt the tempco of the shunt will be a problem.

I'm not in to micro controllers but if you use a somewhat bigger shunt for good resolution and measure before its resistance change per Degree. Keep it as stable as posible with intelligent cooling, and during use you measure its temperature constant to control the cooler then your microcontroller can also correct the current caused by changing resistance by controlling the mosfet.

You need a good temp sensor, a uP, a nice zero drift opamp ( ltc1052 or so ?) and a big mosfet so that stays cool too.

Or shop at Keithley or Agilent.

[saturation]

This is an interesting problem, but I don't think there is a commercial pre-made solution out there if you need precision <= 100ppm, and I presume you mean ppm/^oC?  You'd very likely have to build it yourself as PA4TIM describes.  This was a question raised in a TI chip some years ago:

http://www.ti.com/analog/docs/litabsmultiplefilelist.tsp?literatureNumber=sbva001&docCategoryId=1&familyId=401

Eloads are constructed more for stress testing DUT.  When reading the specs, factor in the LSD drift, the actual accuracy is much worse.  The thermostat hysteresis provides a wider upper and lower limit of internal temperature drift which is often not allowed in precision devices.

Other options are 2 quadrant power supplies or 'source measuring units', and use it primarily to sink current.  They have better precision, but they cost far more than ordinary eloads.   But they are also limited by their cooling capacity, I do not know if their current sink specification is truly as accurate as the source specifications.

I think i wasn't clear enough, the voltage source will be around 10V, not exactly (i.e. it can change over time). If it works between 9 and 11V would be good enough. But the current sink should stay at 1A. The electronic load that we have here does exactly that (it keeps the current constant, while the voltage is changing), but it is not precise enough.

This electronic load would probably work: http://www.home.agilent.com/en/pd-1000001519%3Aepsg%3Apro-pn-6060B/300-watt-dc-electronic-load?nid=-536902252.536880312&cc=DE&lc=ger

But i had hopes that there is a cheaper method for a constant current.