A simple active DC dummy load for PSU testing

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hello folks,

here's one suggestion for a DC dummy load. The basic idea isn't mine, I merely adjusted it to meet my needs. Here's how it works:

-The bipolar power transistors are acting as resistors, according to the well-known equation R=U/I. When there is a certain dc voltage across the transistor (Vce) and simultaneously a certain current flows through it (Ic), the transistor will look like a resistance Req=Vce/Ic to the voltage supply.
-The equivalent resistance defined earlier will be controlled via op amp feedback, as the op amp tries to keep the voltage across the sensing resistor equal to the reference voltage formed with resitive voltage division circuit. Hence, increasing the reference voltage (op amp +pin voltage) increases the current drawn from the dc supply.
-The op amps are powered with dual 5 V supply regulated from the dc supply voltage. The reason for choosing 5 V is to be able to drive the bjt:s (to have a bit headroom after the base-emitter voltages are met).

Improvements:
-Diodes should be attached to protect the transistors
-Better supply bypassing for the op amps
-More clever component choices/ratings
-Maybe individual emitter current sensing for every power bjt
-At this moment the input dc voltage is limited by the regulator, approx. 7 V < Uin < 40 V. By using a small switch-mode supply one could operate the op amps even if the input voltage is below 5 V.

Any thoughts? :)
 

Attachments

  • dummyload.pdf
    12.1 KB · Views: 1,307
The most important reason for me is the adjustability of the load. I have built some SMPS and I like to make measurements with different loads - this way there will be unlimited options for a load resistance.

This method may be somewhat expensive, but at least it will look more professional than just a single power resistor ;)
 
Buy a rheostat and put it in a nice box. If looks matter in your lab, then this solves your problem.

Anyhow, if you insist on going active, there's some work to be done. The first thing that caught my eye, is that you supply the upper 741 from +5 and 0V, and then you connect the inverting input to a node that will go from 0 to +200 mV. This is way outside the input common-mode range, you are better off giving the opamps their own symmetrical power supply, independent from the U+ and U- terminals. This will introduce the need for a power supply, but hey, a power cord and on on/off button with pilot light will make your thing look even more professional! ;)

The second thing is that you simply parallel the power transistors. This is not done, load sharing resistors are mandatory!

The third thing that I found is that your schematic doesn't behave like a resistor at all. It is a current sink (looking at U+ only).
 
I didn't check your schematic for problems, but the design idea should work.

Another way is to use a PWM controlled transistor and a resistive load. By altering the duty cycle you can alter the load. You need an output inductor coupled to smooth out the load.

Using a PWM controlled load would take the heat off the transistors, but it may or may not be more advantange then your design. I suppose with the PWM you could use xformer gate drivers to isolate the controller from the load (only the power transistors would directly touch the load), and would not need both an Vee and Vcc to drive both halfs using an isolated PWM drive system. You could then use N-channel Mosfets\BJTs\IGBTs for both halfs, and there is probably less risk of something going bang if you accidentially exceed the transistor's power dissapation limit. You save money on the transistors but pay more for the power resistor you'll need to dump load into.

I have an old 1000W programmable load that uses a bank of BJTs that alter the base current to change the load resistance, but it uses a lot of of heat sink real estate for all of the BJT power transistors.
 
i'll try to find it if i have time, but in the 70's NASA published an active load using power MOSFETS and a bridge rectifier. it could act as a load for DC or AC. Mackie uses something similar in their stage monitor speakers as an active clamp. once the voltage across the speaker terminals reaches a predetermined level, the active load takes over, clamping the voltage, and diverting the current to keep the speakers from being fried by being overdriven.
 
A similar idea has once been published in the Elektor magazine. I recall having seen it once.

A bit of googling yielded the following links:

Self powered dummy load

200W active load

No doubt that you can find several ideas using NTRS.

If the only thing you want to do is draw current from the supply, then an active load would work. The PWM idea posted by TechGuy does sound appealing, as a resistor as the dissipating element is far cheaper than transistors + heatsinks, and a lot more robust too. However, there are numerous tests you can perform on a power supply which involve AC. For example output impedance vs. frequency, load transient recovery response and control loop stability to name a few. A simple resistor behaves very predictably with these measurements, but how about the active load?

It all depends what you want to do with it. In a production environment, I can understand the use of a test load that is adjustable over a very wide range, to quickly test the PS over its entire load range. In that case, I would consider developing an all singing, all dancing test box that runs as many tests as possible in one go. But in a lab environment, I think that any specialized box will quickly turn itself against you, because you'll want things that the box doesn't support.
 
IMO, the most useful thing is a constant current load, not constant resistance. Fortunately, that's the easiest thing to build. A simple MOSFET design was discussed in an earlier thread. This was from a Siliconix app note way back. Since the current is controlled by an input voltage, it's an easy matter to hook the circuit to to a DAC and put it under computer control. Then, using one of the inexpensive DVMs with an interface, you can automate all manner of tests. Probably overkill for most of us, but a lot of fun to play with.
 
basically, this is what was in the NASA tedh brief...
 

Attachments

  • mosload.jpg
    mosload.jpg
    22.9 KB · Views: 787
the "ground" can be a local isolated ground, as it only needs to provide a return for everything after the bridge rectifier. if you are using it as a test load where you need to monitor the voltage, the measurement is taken before the bridge. you use the 1 ohm resistor to measure current. i would use a 10 turn pot for the current control to get a fine adjustment. for higher current capability, use higher current devices, or parallel FETs. the real advantage of this circuit begins to pay off when you are testing very high currents, the device with heat sink takes up less space than an equivalent resistor. however if you look at digi-key or newark catalogs, you will find power resistors with high wattage ratings in to-220 and to-3p packages, but they're pricey.
 
Conrad Hoffman said:
IMO, the most useful thing is a constant current load, not constant resistance. Fortunately, that's the easiest thing to build. A simple MOSFET design was discussed in an earlier thread. This was from a Siliconix app note way back. Since the current is controlled by an input voltage, it's an easy matter to hook the circuit to to a DAC and put it under computer control. Then, using one of the inexpensive DVMs with an interface, you can automate all manner of tests. Probably overkill for most of us, but a lot of fun to play with.

But if you have a bjt with emitter resistor and a circuit which keeps the resistor voltage constant, wouldn't that look like a constant current sink/load as well? At least if the resistance doesn't change much with the temperature.

The reason why I'm interested in using power bjts is that I happen to have a bunch of them unused, as well as several larger heatsinks. This way it would become very cheap to make at least a smaller test version, say ~50-100 W load.

As can be seen from the first post showing the crude schematic, the load current is controlled via a voltage signal. I intended to add an external voltage input (selectable) through which one could the load "stepping" by inserting square wave signal at required frequency. This would allow transient response measurements.

Well, I think I'll be stubborn and desing a smaller non-isolated load (several tens of watts) first and see what I'll get out of it. Hopefully only small amount of smoke and noise :)
 
one advantage of MOSFETs over bipolars is linear voltage control of the current with little or no loading of the control voltage source. you could use a modulated dc source to test the transient characteristics of a power supply. it's easy to attach a function generator where the control voltage is in the schematic.
 
Bootstrapper said:
here's one suggestion for a DC dummy load. The basic idea isn't mine, I merely adjusted it to meet my needs. Here's how it works:
-The bipolar power transistors are acting as resistors, according to the well-known equation R=U/I.

I posted one such device,
I have used for testing transformers drop and voltage output level.
I did this back some years ago.

There were people of high intelligence posting.
But not anyone understood my concept.

thanks, Bootstrapper :)
you know the idea well

Here are 2 images of my little device.
If anyone is interested, I post my schematic used, in real life application.
Works like a charm ;)

The good alternative is of course passive adjustable power resistor.
I have those, too.
But just turning one trim pot
made my power supply/transformer testing very much more swift and easy.


Useful? Yes!!!
Difficult to build? No!!!
 

Attachments

  • lineup_active_dc-load_with_adjustable_dc-resistance_1.jpg
    lineup_active_dc-load_with_adjustable_dc-resistance_1.jpg
    22.4 KB · Views: 636
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.