How would one make an AC Programmable load?

[staze]

All,

There's all the talk of DC programmable loads... largely stemming from Dave's quick video on making one. But I've been thinking, what would it take to make an AC one?

The main reason I was thinking of it was because I have a couple little enclosures with power resistors in them for dummy speakers/headphones for testing (audio) amplifiers. It would be kinda cool if you could just have one box that was a programmable load as far as AC signals were concerned. I thought MOSFETs were relatively bidirectional, but apparently not really (you can't really use them for switching AC signals without some work).

Am I being blind to the obvious (probably)?

[Jorpy]

I might be talking out of my ass now, but is it possible to simply have a rectifying stage before the usual circuitry?

[staze]

I might be talking out of my ass now, but is it possible to simply have a rectifying stage before the usual circuitry?

See, and that's what I thought... but according to someone else, it would make the load non-linear (drop across rectifier). So the impedance would vary with the input voltage...

There must be some way to do it... =P

[djacobow]

How about a transformer with a fat resistor on one side and multiple taps on the other side, with relays controlled by a uC? With a clever tap/winding arrangement of power of two ratios, you might even be able to get 8b or so of control of the impedance seen by the source.

[penfold]

It depends what kinds of voltages and frequencies you're dealing with.  For appreciably high voltages you could quite easily ignore a silicon diode drop or active rectifier arrangement.  If you get your control right and pick the right points to take feedback from you can compensate for rectifier drops etc

[staze]

It depends what kinds of voltages and frequencies you're dealing with.  For appreciably high voltages you could quite easily ignore a silicon diode drop or active rectifier arrangement.  If you get your control right and pick the right points to take feedback from you can compensate for rectifier drops etc

Well, I was figuring a rather wide range, but my initial thought was as a replacement for a dummy speaker/headphone load... so we're talking a few volts... Obviously the easy solution is just a box with a rotary switch in it to select 4R, 8R, 100R, etc. But there has to be some way to do this "smart". =)

[firewalker]

For just a simple resitive load, I guess complimentary MOSFET or BJTs could be used. Each will conduct for half the period.

Alexander.

[woodchips]

Have a look at the Marconi TF893A, all the work has been done. Power to 10W, impedances from low to high.

Been a while since looking at one but sure it is based on a transformer.

[staze]

Yup. A transformer with a crapton of taps... huh. Not exactly simple. =/

[tom66]

You could modify a variac by putting another winding around the core, maybe? Then connect that to a resistor. A typical variac would have near zero DC impedance and response up to maybe 400Hz before inductance became a major issue.

[ejeffrey]

DC loads are normally constant current.  That would be weird for AC.  You more likely want constant resistance.  To do that, make a constant current load but use an analog multiplier to multiply the set point by a scaled version of the input voltage.  That makes the current proportional to voltage, aka a resistive load.  A single device will be unipolar.  You can use a rectifier or build two separate loads for the positive and negative half waveforms.

If you only care about average power, a simple rheostat will do the trick.

[ConKbot]

Because even specifying what the load behavior is is difficult, let alone building it.  Even a simple load with a non distorted current waveform and a power factor that is anything other than 1 will have to source power during part of the AC cycle.   Youre going to need a 4 quadrant power supply (source and sink positive voltages, source and sink negative voltages)

A transformer hooked upto a class AB amplifier with provisions to clamp the supply rails would be a start for being able to do a programmable current waveform/power factor. 

[NiHaoMike]

A PFC front end will work nicely for higher powers and in theory can be scaled down.

[David Hess]

I might be talking out of my ass now, but is it possible to simply have a rectifying stage before the usual circuitry?

See, and that's what I thought... but according to someone else, it would make the load non-linear (drop across rectifier). So the impedance would vary with the input voltage...

The voltage drop across the rectifier would not be a problem so much as making a constant resistance load which may be adjusted unless the AC voltages involved are low.

I would use either an existing power factor correction front end like NiHaoMike suggests with a constant current load or a constant resistance active load like ejeffrey suggests.

On intriguing way to make a power MOSFET operate with a constant resistance is to use a matched pair where one controls the Vgs of the other.  This was commonly done with JFETs to make voltage variable resistors and I have done it myself with small signal MOSFETs in automatic gain control circuits.

You might even be able to do this with back to back power MOSFETs and avoid using a rectifier.

[Jeroen3]

Who uses diodes, you can get ideal rectifier bridged based on mosfets and an ic.
Even mosfets nowadays go above 500 Vds.

[David Hess]

Who uses diodes, you can get ideal rectifier bridged based on mosfets and an ic.
Even mosfets nowadays go above 500 Vds.

Which bridge rectifier controller did you have in mind?

The Linear Technology LT4320/LT4320-1 only operates up to 72 volts and does not operate below 9 volts where the forward voltage drop of the MOSFETs will still be that of the body diodes.

[Marco]

You could use MOSFETs but you're always going to get some cross over distortion when you fully turn on the one not being used (to get rid of the body diode voltage drop).

[richard.cs]

For low voltage loads you could use a transformer to step up to a higher voltage where you can ignore diode drops and then use mosfets inside a diode bridge. But really  for audio stuff switched resistors are the way to go.

[mikeselectricstuff]

For this application, dummy loads for audio amps , you only need a few specific values, so resistors are by far the easiest solution.
You could use 2 back-to-back MOSFETS as an AC variable resistor, but the drive & loop control would get somewhat tricky, more so as frequency increases

[staze]

Wouldn't you have to dynamically change the load for an AC signal since the resistance would change (for a mosfet) as the AC voltage varied? You'd have to limit frequency since that control loop would have some upper bounds.

[tom66]

For this application, dummy loads for audio amps , you only need a few specific values, so resistors are by far the easiest solution.
You could use 2 back-to-back MOSFETS as an AC variable resistor, but the drive & loop control would get somewhat tricky, more so as frequency increases

hmmm just out of curiosity, if frequency is not critical (say only up to 100kHz)
would the best selection criteria be Rds-on? lowest possible practically?

Even 100kHz would be tricky.

Say you need just 1 degree phase error at that frequency. Rough guess says that's 30MHz bandwidth for your control loop and to drive your MOSFET at.

Ever tried to drive a ~nF of MOSFET capacitance at 30MHz? Tricky... Sure, it's not full swing from 0 to max load, but it's still a difficult task.

[Marco]

Wouldn't you have to dynamically change the load for an AC signal since the resistance would change (for a mosfet) as the AC voltage varied? You'd have to limit frequency since that control loop would have some upper bounds.

You can use a variation on something like this (you also need some logic to turn the MOSFET in reverse conduction full on and take some care not to get oscillation, apparently this was not trivial for Linear with their ideal rectifier).

http://3.bp.blogspot.com/-Rlfjp5Q9B0A/UJrcNdjZgNI/AAAAAAAAANo/hqhb5Eso9_8/s1600/Fig%2B5%2BCR%2BFET%2Bload%2Bckt.bmp

[edavid]

Have a look at the Marconi TF893A, all the work has been done. Power to 10W, impedances from low to high.

Been a while since looking at one but sure it is based on a transformer.

The original US version of this is the General Radio 1840A.  They are usually not too expensive.

Another practical solution for audio amp testing is the Clarostat 240-C power decade box.  It uses wirewound resistors, but works fine for most purposes.

[Marco]

during that time before i saw those stash of 200w 6Rs, i had the idea of using a spare 2kw sub woofer coil as a test load, an inductive load could be a better load ? maybe? dip the whole shebang into a bucket of water to dissipate heat. yes? no? i remembered in another service shop (some test shop for RCFs), they dunked a smaller 100w R into a bucket of water for extended tests.

If you're going to use water you could just use a water resistor, high frequency capable, can handle lots of power ... just don't smoke near it if you have a large DC component.

[staze]

just don't smoke near it if you have a large DC component.

Lol!

For those that don't get it, it would be liberating a fair amount of hydrogen and oxygen.

Would think just some nichrome in an oil bath would be safer.