Adding a known load to a DC motor

[awaiting]

Hi everyone,

I have implemented PI control of speed of a PMDC motor under no load condition. I now intend to implement the same with a load added to it.
I want to compare the results with simulation as well. Is there any way of adding a known load to a DC motor. In my set up the dc motor is mounted on a metallic box.

Thanks,

 

[zoki85]

Motor dynamometer

 

[Windadct]

If you have another motor - you can use as a generator and connect a resistor as load. Then as you change resistances you can change the load on the motor.

 

[zoki85]

If you have another motor - you can use as a generator and connect a resistor as load. Then as you change resistances you can change the load on the motor.

Good idea if he knows well enough characteristics of another motor

 

[Windadct]

I was just thinking to use the same type motor - by monitoring the input and output electrical power - you can then "see" the characteristics of what is between the measurements and divide by 2 - roughly speaking a poor man's dyno.

 

[zoki85]

I was just thinking to use the same type motor - by monitoring the input and output electrical power - you can then "see" the characteristics of what is between the measurements and divide by 2 - roughly speaking a poor man's dyno.

"Roughly" is a good word. If you have two identical machines operating like this, total loses (P_in-P_out) won't be shared in proportion 1:1 between them.To illustrate suppose measured input power is P_in=10 kW, and let the efficiency of the first machine η₁=0.9. Loss of the machine is
P₁=(1-η₁)P_in=1000 W, and power throughput P'=η₁⋅P_in=9 kW. Suppose now that efficency of the second machine is also η₂=0.9. Then, the loss of the second machine is P₂=(1-η₂)⋅P'= 900 W, and P_out=η₂⋅P'=8.1 kW.Thus, one makes about 5% error assuming that loss of the motor 1900/2 = 950 W. The error increases in the case of small power machines or/and if the operating point is far from nominal.

 

[Windadct]

Hmmm.. somehow I think that 5% error for the OP would be pretty good... esp if he does a few boundary cases : no load on the motor and not linked to the load Generator, not linked to the Generator and the motor in stall (full torque) for a second, linked to the load Generator w/no electrical load - etc...lastly he can swap the two motors and compare the datapoints. I'll bet with some "good practices" he can determine the characteristics to less than 1%. In the real world - that is actually very good.

 

[OmCheeto]

Hi everyone,

I have implemented PI control of speed of a PMDC motor under no load condition. I now intend to implement the same with a load added to it.
I want to compare the results with simulation as well. Is there any way of adding a known load to a DC motor. In my set up the dc motor is mounted on a metallic box.

Thanks,

What is the watt rating of your motor?

 

[zoki85]

Hmmm.. somehow I think that 5% error for the OP would be pretty good... esp if he does a few boundary cases : no load on the motor and not linked to the load Generator, not linked to the Generator and the motor in stall (full torque) for a second, linked to the load Generator w/no electrical load - etc...lastly he can swap the two motors and compare the datapoints. I'll bet with some "good practices" he can determine the characteristics to less than 1%. In the real world - that is actually very good.

In the real world in many cases 1% accuracy is acceptable even for methods with professional dyno equipment.
Simple P_out/P_in = η_tot = η₁⋅η₂ depends on assumption η₁=η₂.
This may be preceise enough for a bulky machine operating near rated load, but for "toys" with P_nom<1 kW operating far from their nominal point, 5% preceison would be quite success.

 

[Windadct]

Z85 -- I think we are on the same page -- but then in my "real world" there are kind of 2 real worlds... academic and industry... in an academic (lab) setting 1% is a good goal for real systems, in industry 5%... the challenge I face, that drives me crazy is academics (without much real lab experience ~1%) come into industry so used to "perfect" simulations and can not accept the 5% of industry ( paper engineers)...so to me this case is actually a very good exercise. However the question needs to asked at the outset - what are you trying to learn or achieve, including how accurate do you want to be.