Blog Post Actions
Group Actions
Engagement
  • Author Author: peteroakes
  • Date Created: Date Created
  • Views 5425 views
  • Likes 7 likes
  • Comments 32 comments
Related
Recommended

Electronic DC Load - Design and Build to test PSU Project

peteroakes

Back tot he Main Modular Bench power System Project The Modular Bench Power Supply ++, The Essential DIY Build for Every EE Student and Old Timer alike...

 

I keep promising myself I will build an Electronic DC Load and have kept defering it, well no longer, I finally ot it designed and tested on a breadboard and will be following up with a completed unit in a project case

 

It can go way above 30V (60 to be precice and upto 5Amps), the MOSFET I used the IRFP064 is able to go to 70Amps but thats more than my wiring would stand and way more than i need so I designed the load to go upto 5Amps for now. All I need to do though is change the current sense resistor and it can easily become a 10Amp unit

 

For the heatsink I tested with a standard one but subsiquently found an old CPU heatsing with a Fan attached which looks like it will do a grand job of cooling the MOSFET etc so it will be the one I use in the build.

 

Testin showed I can load as little as 250mV and still control upto an Amp which is great and once im over 1V source voltage I can go all the way to 5Amps with ease.

 

Now you all know the math "Power = Volts * Amps) so therefor will know that at 30V an 5Amps theres 150W being dissipated in the MOSFET and thats a lot of heat. I am pretty sure the CPU heatsing even with the fan will still get pretty hot and I look forward to characturizing it once the bukd is done. till then I have no idea how well it will work but I am sure it will be better than the one I tested the circuit with

 

for now, here is the schamitic I came up with

image

 

It will use between 12 and 24V for the DC in, has a REF02 Voltage reference to provide the manual setting via a 10 Turn pot and has a trimmer to accuratly set the top output to precisly 5A if so desired.

I also included a BNC input as an alternate control allowing a signal generator to control the load enabling many more test scenarious to be executed like transient response testing, noise response, Waveform based profile, pulse response etc.

 

The design does have a frequency limit before the output falls off of only a few hundred Hertz but that should be more than enough for this. It may be way better once it is off the bread board and the wiring is significanly shorter too but I wont be able to test that till the buld is complete. I will add a video to this blog once that is done and tested

 

anyway, enough banter, here is the video (Yes Im back to my 1+ hour videos... sorry I tried)

 

Hope you enjoy

 

Thanks to Vishay Precision Group ( VPG) Bulk Metal Film division for their kind donation of the current sense resistors and precision metal film resistors used in the resistor divider part of the design. It takes the use of this project to the next level by bringing stability to the key parts of the design in the order of a few parts per million... oh yaaaa.

Parents Comment Children
  • in reply to Former Member

    Hi Monica,

    Su Englais esta bien, mejor que mi Espanol.

     

    Don't worry too much about R-1. You can build one that will be accurate enough for this application. Just get a regular 0.1 ohm resistor and tie extra wires close to the body that will be used to take the voltage reading back to the Op Amp. Solder the wires. I have mocked up a picture to show you what I mean. This is a poor man's Kelvin Resistor.

     

    image

     

    John

  • in reply to jw0752

    Perfect , that is exactly what I would have done if Vishay had not been so generous with some samples and is a perfect example of how to roll your own when the needs arise

     

    Thanks for sharing

  • in reply to jw0752

    I have done this with my system, but in all realism when there is a tolerance value to the resistor, it is not going to make much of a difference until the resistor gets really low 0.01ohms.

     

    No matter what, the whole set-up will need to be calibrated with a know source. I think those resistor have a 5-10% tolerance, so a couple of mm of wire attachment may not be an issue.

     

    Malcolm

  • in reply to waelect

    Ah, a common misconception

     

    The tolerance of the resistor will as you say affect the base value, so 10mOhm +-5% could be between 0.0105 or 0.0095 ohms, and as you say, to make the instrument accurate, you will need to either know what that value is precisely or you calibrate the error out using a good current meter under operation and adjust the reading in software to fix any actual tolerance errors. As I said, the tolerance is not as important as stability, you can easily fix the tolerance in software, you can't fix the stability

     

    If you have a resistor for the current shunt that is over a few ohms in value  then yes you get diminishing return on the use of a Kelvin Connection (4 Wire), but the tolerance of the resistor is irrelevant in this, the same occurs if the current you're measuring is very low as a low current means a low volt drop.

     

    Now with a current shunt where you may be measuring 10 - 20 Amps or more, having a kelvin connection can make a huge difference

     

    Lets assume the wire resistance is only 1mOhm and the resistor is 10mOhm. The wire represents 10% * 2 (there are 2 wires) so 20% of the reading, if the wire is a lower resistance it can still be significant compared to the base resistor value!!!

    10mOhm at 20A = 200mV

    2mOhm at 20A = 40mV

     

    Now if you connect the current through the wider part of the resistor leads (Further away from the resistor body), we still have the 20A and we still have the above voltage drops in the wires and the resistor element but if we can get the voltage measurement as close to the resistor element as possible we reduce the amount of error induced by the wires.

    As the load to the voltage measuring circuit is insignificant compared to the shunt resistance (100K - several Mohms or more) it will have no significant impact on the measurement (Voltage Measurement Resistance in parallel with shunt resistor, so maybe 10Mohm in parallel with 10mOhm)

    So at low shunt resistor values and high currents it makes a big difference and can bring your measurement accuracy upto fractions of a % compared to many %.

     

    Now back to the point about the tolerance of the resistor, once you know its value or have calibrated out its error and providing it is a stable resistor (Some of the vishay shunt resistors I have are 5% but only a few PPM/degC so super stable but potentially not too accurate to its nominal value). If we can find out its actual value and it does not matter if it is 0.0095 or 0.0105 or anything in between, we can now compensate for this and know that it will not drift with temperature due to the stability of the resistor.

     

    You could measure the resistor by using an accurate DMM on amps range to pass through a current, now use a separate DMM to measure the voltage as close to the body of the resistor as possible or if it is 4 terminal (DON'T TAKE OUT THE CURRENT METER TO MEASURE THE VOLTS, THIS WILL CHANGE THE CURRENT FLOWING), on the other terminals. then simple ohm's law allows you to determine the actual value. If your meters are say 0.05% accurate then you can be sure you have a resistance reading better than 0.1%

    If you have a good 4 terminal resistance meter you can also simply use that to measure the resistor if it goes down to that range, and if it is 4 terminal then it probably does.

  • in reply to jw0752

    Same solution I have used to calculate with two analog readings the power consumption with the Bitscope Micro. The only difference is the resistor I have found that is no 0.1 but 0.45 Ohm.

     

    Enrico