I am trying to build a controllable resistive load for a test system and I was hoping that someone has solved this problem before.
The over all design is meant to drive four parallel LEDs & resistor pairs and to detect if any of them are open or shorted. The driving circuit has a 12v supply with a current sense and the low side is an FET that is PWM controlled to Gnd.
I need to design a variable load that can sink between 20mA and 300mA and will emulate the LED/Res load. This means that the resistance would be 60 to 600 Ohms.
I have looked at the programmable resistance chips like the MCP4012 or the AD5246 but the operating current range is too small.
I have also looked at commercially available ones but they are very expensive and meant for carrying 15Amps.
Cost is not a significant factor since I only need to build about three of these but I would hate to waste couple of thousand dollars when a couple of hundreds (or less) would do.
> I am trying to build a controllable resistive load for a test system
>and I was hoping that someone has solved this problem before.
>
> The over all design is meant to drive four parallel LEDs & resistor
>pairs and to detect if any of them are open or shorted. The driving
>circuit has a 12v supply with a current sense and the low side is an FET
>that is PWM controlled to Gnd.
>
> I need to design a variable load that can sink between 20mA and 300mA
>and will emulate the LED/Res load. This means that the resistance would
>be 60 to 600 Ohms.
>
> I have looked at the programmable resistance chips like the MCP4012
>or the AD5246 but the operating current range is too small.
>
> I have also looked at commercially available ones but they are very
>expensive and meant for carrying 15Amps.
>
> Cost is not a significant factor since I only need to build about
>three of these but I would hate to waste couple of thousand dollars when
>a couple of hundreds (or less) would do.
>
> Any suggestions would be appreciated.
>
>-Phil-
What are your specifications for accuracy? The LED + resistor pairs
do not behave quite like a resistor. Are there multiple LEDs in series
in each pair (seems like you'd want to do that with 12V available
and the max single junction voltage is only a few volts.
Could you use signal relays (either electromechanical or MOSFET) and fixed
LEDs and/or resistors to test at your limit values?
> the low side is an FET that is PWM controlled to Gnd
Varying the duty cycle would make a variable power resistor. For
example, if FET Ron is 1 ohm, then a 1:59 duty cycle would average
to 60 ohms. 1:599 would average to 600 ohms. To avoid very low
on:off ratios, use a series fixed power resistor. eg 60 ohms (including
FET Ron), 1:1 for 60 ohms, and 1:9 for 600 ohms. 300mA is nothing
to a $2 power FET wih low Ron and sharp drive (Russell has a cheap
3-transistor driver)
What I need to test & verify is the current sense on the power supply. Does it set a flag if the current is outside of the limits? If it is outside, what is the actual trip point?
The problem is that I need to be able to vary the current through a load that is not referenced to ground. It is reference to the +12v with the low side varying between 0 and 12v.
My specifications for accuracy are not the real problem. I can measure the real current with enough precision but I need a way to vary the load repeatability.
-Phil-
On 3/3/2012 1:50 PM, Spehro Pefhany wrote: {Quote hidden}
> At 04:17 PM 3/3/2012, you wrote:
>> I am trying to build a controllable resistive load for a test system
>> and I was hoping that someone has solved this problem before.
>>
>> The over all design is meant to drive four parallel LEDs& resistor
>> pairs and to detect if any of them are open or shorted. The driving
>> circuit has a 12v supply with a current sense and the low side is an FET
>> that is PWM controlled to Gnd.
>>
>> I need to design a variable load that can sink between 20mA and 300mA
>> and will emulate the LED/Res load. This means that the resistance would
>> be 60 to 600 Ohms.
>>
>> I have looked at the programmable resistance chips like the MCP4012
>> or the AD5246 but the operating current range is too small.
>>
>> I have also looked at commercially available ones but they are very
>> expensive and meant for carrying 15Amps.
>>
>> Cost is not a significant factor since I only need to build about
>> three of these but I would hate to waste couple of thousand dollars when
>> a couple of hundreds (or less) would do.
>>
>> Any suggestions would be appreciated.
>>
>> -Phil-
> What are your specifications for accuracy? The LED + resistor pairs
> do not behave quite like a resistor. Are there multiple LEDs in series
> in each pair (seems like you'd want to do that with 12V available
> and the max single junction voltage is only a few volts.
>
> Could you use signal relays (either electromechanical or MOSFET) and fixed
> LEDs and/or resistors to test at your limit values?
>
> --sp
>
>
>
>
>
>
At 06:07 PM 3/3/2012, you wrote:
>What I need to test & verify is the current sense on the power supply.
>Does it set a flag if the current is outside of the limits? If it is
>outside, what is the actual trip point?
>
>The problem is that I need to be able to vary the current through a load
>that is not referenced to ground. It is reference to the +12v with the
>low side varying between 0 and 12v.
An active circuit that simulates a resistor will have a settling time
as well. It won't do you much good if the PWM rate is 1kHz and it
take 10msec to settle.
The idea is trivial (draw a current that has a fixed
proportionality to the voltage, where the proportionality is determined
from the reciprocal of the simulated load resistance). I'll let you
decide whether you could pull this off cheaper than buying an off the shelf
solution (hint- multiplying DAC). The control loop will probably have
to be analog. That it is not referenced to ground is only a minor
inconvenience.
>My specifications for accuracy are not the real problem. I can measure
>the real current with enough precision but I need a way to vary the load
>repeatability.
>
>-Phil-
If discrete steps are acceptable, relay contacts switching resistors is a
very easy approach. 6 contacts & 6 resistors can give you 64 steps. It
should be very repeatable if done right.
>On 3/3/2012 1:50 PM, Spehro Pefhany wrote:
> > At 04:17 PM 3/3/2012, you wrote:
> >> I am trying to build a controllable resistive load for a test system
> >> and I was hoping that someone has solved this problem before.
> >>
> >> The over all design is meant to drive four parallel LEDs& resistor
> >> pairs and to detect if any of them are open or shorted. The driving
> >> circuit has a 12v supply with a current sense and the low side is an FET
> >> that is PWM controlled to Gnd.
> >>
> >> I need to design a variable load that can sink between 20mA and 300mA
> >> and will emulate the LED/Res load. This means that the resistance would
> >> be 60 to 600 Ohms.
> >>
> >> I have looked at the programmable resistance chips like the MCP4012
> >> or the AD5246 but the operating current range is too small.
> >>
> >> I have also looked at commercially available ones but they are very
> >> expensive and meant for carrying 15Amps.
> >>
> >> Cost is not a significant factor since I only need to build about
> >> three of these but I would hate to waste couple of thousand dollars when
> >> a couple of hundreds (or less) would do.
> >>
> >> Any suggestions would be appreciated.
> >>
> >> -Phil-
> > What are your specifications for accuracy? The LED + resistor pairs
> > do not behave quite like a resistor. Are there multiple LEDs in series
> > in each pair (seems like you'd want to do that with 12V available
> > and the max single junction voltage is only a few volts.
> >
> > Could you use signal relays (either electromechanical or MOSFET) and fixed
> > LEDs and/or resistors to test at your limit values?
> >
> > --sp
> >
> >
> >
> >
> >
> >
> > -
Trimmer on a power transistor base, then VCC to collector and 0v to emitter !?! Put some DMMs in line to see the load and adjust the trimmer as needed ?
Regards,
Steve Maroney
Business Computer Support, LLC
504.904.0266 Office
504.914.4704 Mobile
866.871.7797 Fax
You can easily produce a variable power resistor using a MOSFET. You
can easily produce a floating current source using a MOSFET.
If you are happy with manual adjustment and using a separate meter for
feedback the solution is trivial - apply variable voltage to MOSFT
gate.
If you want it to be open loop and calibrated the voltage + MOSFET
may would but odds are that the accuracy would not suffice. So you can
add intelligent monitoring and control. You feed the required state to
the circuit concerned and it provides the desired eg load resistance
by measuring V & I and keeping the ratio constant. dynamically.
In the most general case - floating current sense (small resistor or
Hall sensor or ...), differential amp to read current sense,
differential amp to read FET voltage drop.
Adjust FET gate drive and calculate V/I until desired resistance is achieved.
Odds are that a FET and manual gate voltage control will meet your need.
