Switch mode programable electronic load?

[zapta]

We had programmable linear power supplies and then came the programmable switching power supplies with better efficiency. 

Can a similar concept be applied for programable electronic load? That is, instead of disputing the consumed power on some component with a heatsink, convert it with a switch mode power supply and push the power back the the mains. This will allow smaller and quieter electronic loads with higher power capacity.

Does this make any sense?

[oldway]

Yes, this concept is applicable to electronic load, but with serious limitations.
I already used it to recover energy from the discharge of industrial batteries but with thyristor converters.
As you need to smooth the current, you will need an electrolytic capacitor in the input circuit of such an electronic load, and this prevent to make high slew rate current variations.
It can also be a challenge if you want your electronic load to work at low voltages <5V for example.

[nctnico]

We had programmable linear power supplies and then came the programmable switching power supplies with better efficiency. 

Can a similar concept be applied for programable electronic load? That is, instead of disputing the consumed power on some component with a heatsink, convert it with a switch mode power supply and push the power back the the mains. This will allow smaller and quieter electronic loads with higher power capacity.

Does this make any sense?

These are commercially available. I looked at them for a battery pack cycle tester but it turned out the extra expense wasn't worth the savings in the electricity bill even though the tester runs almost 24/7.

[tron9000]

I think this guy does something similar:
https://youtu.be/k5684mQJQRU?list=PLxaGnte1Dq0ly6tsTkCaqK3cf_viyqyEq

he's using PWM through a load changing its effective resistance.

[Whales]

This idea sounds perfectly fine.  Forseeable design challenges:

• Input range.  Unless you are using mains on the load in the first place, you will need to boost/buck and re-oscillate to match mains waveform.  This will probably lead to a very limited input range, but it might be better than I assume.
• Bulk and cost.  Up to a certain size you would probably find that traditional resistors are cheaper and better.

[oldway]

Up to a certain size you would probably find that traditional resistors are cheaper and better.

For limited input range, as for example 12V constant current battery discharge, you can make a stepless resistive load as schematic I have already posted in another thread.

[Kleinstein]

In many areas you need special permit to allow feeding power back to the grid. There also need to be safty measures as not to power part of you instllation if the fuses are blow. Just for this case you may still need a heat sink as a backup anyway.

As from the switching mode the load will not be as smooth as a linear one.

[oldway]

One can also wonder what would be the interest of wanting to recover so little energy, especially if it is amateur use.
To discharge a 12V car battery or to test a power supply for a half hour, it does not represent any interest.

[joeqsmith]

We had programmable linear power supplies and then came the programmable switching power supplies with better efficiency. 

Can a similar concept be applied for programable electronic load? That is, instead of disputing the consumed power on some component with a heatsink, convert it with a switch mode power supply and push the power back the the mains. This will allow smaller and quieter electronic loads with higher power capacity.

Does this make any sense?

https://www.eevblog.com/forum/projects/ugly-twist-to-a-programmable-load/msg666730/#msg666730

[albert22]

Here is a switching electronic load that uses light bulbs and pwm:
http://www.edn.com/design/analog/4345737/Versatile-power-supply-load-uses-light-bulbs

[oldway]

Here is a switching electronic load that uses light bulbs and pwm:
http://www.edn.com/design/analog/4345737/Versatile-power-supply-load-uses-light-bulbs

An important function of a dynamic load is to test power supplies with a current step to check the stability and dynamic response.

A load using the switching mode principle requires a filter of the current, which makes it impossible to make a current step with a short rise time.

It is therefore unusable in general.

[TopLoser]

http://www.farnell.com/datasheets/1907679.pdf

Not cheap though...

[zapta]

http://www.farnell.com/datasheets/1907679.pdf

Not cheap though...

They claim efficiency > 90%. Very good.

[joeqsmith]

An important function of a dynamic load is to test power supplies with a current step to check the stability and dynamic response.

A load using the switching mode principle requires a filter of the current, which makes it impossible to make a current step with a short rise time.

It is therefore unusable in general.

Consider much of the supplies made are switching and even some audio amplifiers are switching as well, to be honest I am not sure what would be considered a usable short rise time. 

Looking at the slew rates on the BK loads, they range from 1A/us to 0.5A/us (10/90).    Based on rise time,  I will assume we are interested in the turn on time (10% load to 90% load).   

Load1 shows the home made switching load as it was during the time I made the video.   I am using the Tektronix P6042 I had modified to increase the BW to look at the current.  It was set to 1A.      So 0.8V / 50mV = 16    16 * 1 or 16 Amps peak.  12.8A in 186uS or 69mA/us.   

However, the analog control signal is heavily filtered (not to mention it's based on an old Unitrode part running fairly slow).   I removed  the input filter and made the same measurement in load 2 which shows a rise time of 51us or 251mA/us.    About half the slower  BK units. 

But let's twist the throttle a little further on the home made switching load.   Load3 is now at 1.1V peak / 50mv or 22Amps peak.  So 17.6A/22uS or 800mA/uS.    Consider again this is not at all based on a modern switching design. 

There are many reasons why I see no point in using this sort of design for a load but the rise time is not one of them. 

Here is a switching electronic load that uses light bulbs and pwm:
http://www.edn.com/design/analog/4345737/Versatile-power-supply-load-uses-light-bulbs

Thanks for the link. 

[joeqsmith]

http://www.farnell.com/datasheets/1907679.pdf

Not cheap though...

- Response time 10-90% load step <1.5 ms

Very interesting setup.   I wonder how many they have sold and to who.   

[zapta]

Very interesting setup.   I wonder how many they have sold and to who.

And also why people are buying them, to reduce size and heat or to preserve energy.

[oldway]

- Response time 10-90% load step <1.5 ms

Very interesting setup.   I wonder how many they have sold and to who.

This is useless to make dynamic tests of power supply, that's far too slow.
As I said, that's the problem with this kind of technology...
For example, an Hameg HMP4040 power supply has a recovery time <1 ms after a load step from 10 to 90% for return within a +/- 10 mV window.
How could you mesure this if you electronic dc load has a responsive time of 1.5 ms ?

Another issue to consider is the cost/benefit.
This technology is expensive, is it worth the investment?
Except special cases of continuous uses, such as formating of industrial battery plates, most of the time, the electronic load is used only rarely, with low power and during reduced time. (half an hour is more than sufficient to reach thermal equilibrium of a power supply).

Even for testing underground trains choppers of 500 KW or more, a study had shown that it was not worth the investment and we used forced cooling resistors as load.
In the winter, warm air is used to heat the factory.

[BravoV]

- Response time 10-90% load step <1.5 ms

Very interesting setup.   I wonder how many they have sold and to who.

This is useless to make dynamic tests of power supply, that's far too slow.
As I said, that's the problem with this kind of technology...
For example, an Hameg HMP4040 power supply has a recovery time <1 ms after a load step from 10 to 90% for return within a +/- 10 mV window.
How could you mesure this if you electronic dc load has a responsive time of 1.5 ms ?

Another issue to consider is the cost/benefit.
This technology is expensive, is it worth the investment?
Except special cases of continuous uses, such as formating of industrial battery plates, most of the time, the electronic load is used only rarely, with low power and during reduced time. (half an hour is more than sufficient to reach thermal equilibrium of a power supply).

Even for testing underground trains choppers of 500 KW or more, a study had shown that it was not worth the investment and we used forced cooling resistors as load.
In the winter, warm air is used to heat the factory.

Although I'm not an expert, have to agree with Oldway's point of view here, especially on the response time.

The reason and the "main purpose" of an electronic load is to test and verify the power supply capabilities. Saving or recycling the loaded power imo is a much lower priority "feature", and it should not made the whole main task much more complicated, let alone significant cost increased in building it.

Loading the PSU doesn't always mean in a continuous constant current, voltage or wattage, a dynamic loading is actually favorable as it reflects the real world condition.

The load that is constantly "switching" on it's own is quite hard to tame, for example, head to this excellent project designed by one of our forum respected member -> Dynamic Electronic Load Project by Jay_Diddy_B , and look how a simple and a dirt cheap circuit capable of "torturing" at power supply at very fast period in uS (micro second), say like testing it's transient response.

Again, cmiiw, not an expert here, just take it with a grain of salt. 

[nctnico]

- Response time 10-90% load step <1.5 ms

Very interesting setup.   I wonder how many they have sold and to who.

This is useless to make dynamic tests of power supply, that's far too slow.
As I said, that's the problem with this kind of technology...
For example, an Hameg HMP4040 power supply has a recovery time <1 ms after a load step from 10 to 90% for return within a +/- 10 mV window.
How could you mesure this if you electronic dc load has a responsive time of 1.5 ms ?

Another issue to consider is the cost/benefit.
This technology is expensive, is it worth the investment?
Except special cases of continuous uses

Even with continuous use these loads are not cost effective. I have looked at these exact models for a battery pack tester in a production line.

[joeqsmith]

Although I'm not an expert, have to agree with Oldway's point of view here, especially on the response time.

The load that is constantly "switching" on it's own is quite hard to tame, for example, head to this excellent project designed by one of our forum respected member -> Dynamic Electronic Load Project by Jay_Diddy_B , and look how a simple and a dirt cheap circuit capable of "torturing" at power supply at very fast period in uS (micro second), say like testing it's transient response.

Again, cmiiw, not an expert here, just take it with a grain of salt. 

For period he talks about 3ms, not micro seconds. 

0-5A pulsed current at 330Hz

I did not see where he defined the DSOs numbers so lets look at his SPICE.   Risetime:


Looking at the above, seems like 10/90 is about 2A/7us or 0.286A/us.   Compare this with the BK loads at 0.5A/us for many models up to 1A/us for others and my switching load at 0.8A/us.  I guess I don't understand the response time comments. 

Would guess most designers would use a network analyzer to verify stability.  I use a screw driver but to each their own.       
http://www.ti.com.cn/cn/lit/an/snva364a/snva364a.pdf
http://www.ridleyengineering.com/ridley-universal-injection-isolator.html?showall=1&limitstart=

The larger loads may be for something more like burn-in.   If they were running supplies non-stop, that could be a lot of heat.  Need to consider the cost of removing the heat.  Air conditioner, heat exchangers....  Would like to see how it is being used. 

[oldway]

The larger loads may be for something more like burn-in.   If they were running supplies non-stop, that could be a lot of heat.

This is a very specialized application only for switch mode power supplies factories.
Not even 1% of the electronic load market.

Need to consider the cost of removing the heat.  Air conditioner, heat exchangers.... 

In winter and in cold weather, the hot air is used to help plant heating .

When the heating is not useful, we reject the hot air outside.

[joeqsmith]

The larger loads may be for something more like burn-in.   If they were running supplies non-stop, that could be a lot of heat.

This is a very specialized application only for switch mode power supplies factories.
Not even 1% of the electronic load market.

I don't know what the market breakdown is for electronic loads.   Other applications may be for testing generators or part of a load for engine testing.    For home use the only need I have ever had for one was to show an example of a switching design.   

Need to consider the cost of removing the heat.  Air conditioner, heat exchangers.... 

In winter and in cold weather, the hot air is used to help plant heating .

When the heating is not useful, we reject the hot air outside.

Right, reject the hot air outside causing global warming.     I can see where there would be cases that recovering this energy would make sense.   I would like to see some of these installations.   

They have a little video that gives a little background on them.   
http://www.elektroautomatik.de/en/aboutus.html

[C]

Might want to Google "jim williams load" and read the following
 
you get
Linear's Application Note 104
 Load Transient Response Testing for Voltage Regulators
http://cds.linear.com/docs/en/application-note/an104f.pdf

Design a 100A Active Load to Test Power Supplies
http://cds.linear.com/docs/en/article/WMSstory_From%20EDN%2009-22%20NoAds.pdf

Linear's Application Note 133
A Closed-Loop, Wideband, 100A Active Load
http://cds.linear.com/docs/en/application-note/an133f.pdf

[oldway]

Other applications may be for testing generators or part of a load for engine testing.

We are speaking about electronic loads for DC, not of grid inverters.
DC generators and motors (not alternators + rectifier) are tested by the energy recovery method using 2 identical generators mechanicaly and electricaly coupled so one works as a motor, the other works as a generator.
By control of field of one of the 2 generators, you can control the circulating current.
No need of a DC load.
The only wasted energy are the electrical and mechanical losses.

Right, reject the hot air outside causing global warming.

Problem of global warming is so big that power losses of a 2 or 4KW electronic load used once a week to test a 500W power supply you just repaired does not matter at all.

[joeqsmith]

Other applications may be for testing generators or part of a load for engine testing.

We are speaking about electronic loads for DC, not of grid inverters.
DC generators and motors (not alternators + rectifier) are tested by the energy recovery method using 2 identical generators mechanicaly and electricaly coupled so one works as a motor, the other works as a generator.
By control of field of one of the 2 generators, you can control the circulating current.
No need of a DC load.
The only wasted energy are the electrical and mechanical losses.

Right, reject the hot air outside causing global warming.

Problem of global warming is so big that power losses of a 2 or 4KW electronic load, used 1 time per week to test a 500W power supply you just repaired, does not matter at all.

I was thinking an engine (gasoline/diesel) being attached to a dyno and using the load to dump the energy from the absorber.  Basically what you describe but I have been at locations where they dumped the energy from the absorbers into large resistive banks.    The last place I worked at we just absorbed the heat in the windings then removed it with water.   

For home use the only need I have ever had for one was to show an example of a switching design.   

Now that I think about it.... I have actually needed to build a variable electric load once.   In the following video, I am taking the voltage off a 3 phase generator, rectifying it, then feed that into my DC linear load.   This is being used to test the torque/power output of a small brush electric motor. 



I posted about this toy dyno here if anyone is interested. 
http://www.msuk-forum.co.uk/topic/81512-self-built-rc-dyno/page-11