Resistive but reasonably wideband electronic load, anyone?

[n3mmr]

If you limit the power capability to a few watts, is it possible to build an electronically controlled resistor for frequencies from dc to say 40-50 kHz?

[exe]

I think so, if one can build a power supply with 1MHz bandwidth, then I see no reasons why it's impossible to build an electronic load.


Here is a design for, afaikm 100kHz bandwidth, which doesn't seem enough for your purpose: https://www.eevblog.com/forum/projects/dynamic-electronic-load-project/ . But by using faster opamps, smaller and more modern fets (=less input/miller capacitance), and and adjusting compensation network, it can reach higher frequencies. Or, may be, bjt would be better because they are designed for linear operation, unlike modern fets.

Update: ah, resistive load... I don't know. Seems to be a much harder objective. If it doesn't need to be precise, then going digital could be an option, not sure about fully analog.

[tszaboo]

It depends on your voltage, current etc. Most loads use FETs because Rds is lower, and they have higher power capabilities. If you can tolerate regular transistors, then you can go much higher in the frequency range.

[exe]

As an idea: a voltage-controlled amplifier sets current through the pass element. Making it stable can be a challenge.

[T3sl4co1l]

Synthesizing a resistor requires having a ratio of voltage to current.  If it must be electronically variable, then that ratio must be variable, and you must have a multiplier function block.

Your old school OTAs (LM13700) and multipliers (AD633) can do this, and offer enough bandwidth to do what you ask.  Don't expect a terrific noise figure.

Doing more than a few MHz would be challenging because you can't get power transistors with low enough capacitance or high enough speed to do it.  Depending on resistance setting of course -- you can always swamp capacitance with a lower resistance, or inductance with a higher resistance.  These are competing forces, so the effective range gets narrower as frequency goes up.  Eventually, range goes to zero, and it becomes a fixed resonant circuit.

Tim

[filssavi]

I don’t see why you couldn’t
Depending on your requirements and proficiency it can be done fairly easily with an FPGA and an inverter, you can either dump everything on a power resistor bank or do backfeed to the grid

[GerryR]

Motorized Pot anyone!?  What range of resistance?


https://micronor.com/products/motorized-potentiometers/


Note: Added link.

[exe]

What about this idea? Just mirror current of a real resistor, but with a gain.

A value of "e-resistor" can be set by varying gain and/or attenuating sensing resistor. Or simply use a pot .

PS opamp, bjt, and values of resistors are randomly chosen, I didn't think much about them.

[CatalinaWOW]

Two questions that determine feasibility of a "resistive" load. 

First, how much resolution does your load require?  To illustrate the point take two power resistors and two relays.  With this arrangement you can get four load states (only three of which are useful).

What do you mean by bandwidth?  The above example, with appropriate choice of resistors and wiring can provide a resistive load over many MHz.  But if you mean a load that can adapt to constant dissipation over a range of supply states the problem is far more difficult.  Or if you want to change resistance at MHz rates the problem is doable, but requires attention to a variety of details.

[shakalnokturn]

It looks like the requirements need better specifying.

[David Hess]

If an FET instead of a bipolar transistor is used for the output device, then the channel resistance can be fixed subject to channel modulation effects which can be mostly trimmed out.  Some resistive gain control circuits work this way and it used to be common in audio applications.  Some early oscilloscopes used this method up to 10s of MHz.  Gain control linearity can be improved by using matched output devices with the resistance on one controlling the open loop resistance of the output device.

Another way to do it is to drive a fixed resistor load with a mirror of a variable fraction of the input voltage.  Now the drive circuit is a voltage output so any class-AB or even class-A amplifier is suitable and there is none of this nonsense about output device capacitance.  Audio power amplifiers are sometimes tested this way and it can be done up to microwave frequencies to apply a variable load to an RF device.