2
\$\begingroup\$

Problem:

I am designing the electrical system for a vehicle that has 12 brushed dc motors / linear actuators, they are driven by 6 identical dual-channel motor driver boards, that provide a maximum output current of 10 A. Only six of the 12 motors are in use at any one time.

In order to save cost and weight, I would like to reduce the number of motor driver boards to three. In this configuration, each driver board channel would be connected to a SPDT switch that can be triggered by a high / low (3.3 V or 5 V logic level) electronic signal. This configuration is represented in the figure:

enter image description here

My attempts at a solution:

I have tried using a SPDT relay for this purpose. Due to the current draw of the coil of each relay, all six of the required relays could not be driven from a single output pin of the microprocessor board that is being used. Therefore, a lower power switch was needed to provide current to all of the boards. This resulted in a complicated design that was unreliable. Around 20% of the time, the relay armatures switched back and forth in high frequency oscillations when the coil was energised. This would only stop when the circuit was powered off. In addition, the circuit was heavier than a solid state solution.

I then tried building the same circuit using n-MOSFETs. In this circuit, the output of each motor driver board channel was in series with the source pins of two MOSFETs, whose gate were triggered by a 5 V signal from the microcontroller. In order to switch the output, one mosfet was switched off while the other was switched on (see Figure below).

enter image description here

The MOSFET had a nominal internal drain to source resistance of 0.050 Ω in the ON state. At a current draw of 10 A, the MOSFETs should dissipate ~ 5% of the power supplied to the circuit due to the drain-to-source resistance. However, when the MOSFETs are switched ON, the no-load speed of the motor decreases significantly more than 5%. I am uncertain why this is, though perhaps it is as the motors are a highly inductive load and this changes things. The data sheet for the n-MOSFET can be found here: https://docs-emea.rs-online.com/webdocs/0025/0900766b80025d5e.pdf

Questions:

Is there an appropriate component or circuit for this application? Also, are my approaches reasonable, and if not, how could they be fixed? I have researched triacs and think that these could be the basis for a solution to my problem, is this the case?

\$\endgroup\$
  • \$\begingroup\$ (1) Forget about triacs for a DC application. Once turned on they don't turn off. (On AC there is a zero-cross every mains cycle and they turn off then.) (2) At 10 A you'll lose 0.5 V on the MOSFET. Is that enough to cause the speed droop? (3) "At the max draw of 10 A, this should result in a power dissipation of 5 W inside the transistor. This represents about 5% of the maximum power draw of each motor." You might need to edit those two sentences. I think it has come out a bit garbled. \$\endgroup\$ – Transistor Apr 12 at 20:52
  • \$\begingroup\$ Thanks for the advice. A voltage drop of 0.5 V on the MOSFET would have been 3% of the supply voltage during my experiment (16.8 V), so I don't think this would have been the reason for the large drop in no-load speed. I am happy to be corrected on this, though. I have edited those sentences, which I hope makes the meaning clearer. \$\endgroup\$ – jholzer Apr 12 at 21:05
  • \$\begingroup\$ "... perhaps it is as the motors are a highly inductive load ..." On a DC circuit inductance only matters on switch-on or off but not on steady state. I note that you have no snubber diodes on the motors. \$\endgroup\$ – Transistor Apr 12 at 21:19
  • \$\begingroup\$ At what voltage do you drive these mosfets? If you drive them on 5V that probably is too low. Better use a (dual) gate driver (which replaces te drawn logic port as well). And mosfets with lower \$R_{DSon}\$. \$\endgroup\$ – Huisman Apr 12 at 21:27
  • 1
    \$\begingroup\$ My Rule of Thumb is the FET Ron ought to be < 10% DCR of motor and rate of deacceleration = current so if you change directions at full speed you may exceed the motor temp rating if too fast or often as reverse at full speed causes 2*V/DCR for some time to change speeds with inertial mass + heat. loss, so I start with <10% load DCR for RdsOn then do an energy loss budget. from the charts below in the answer the heat loss ranges from 0.5W to 300W depending on current which rises 300% when hot @ 175'C (ouch) \$\endgroup\$ – Sunnyskyguy EE75 Apr 13 at 3:38
2
\$\begingroup\$

Look at the curves on page 3 of your MOSFET's data sheet:

enter image description here

You are not fully turning on your MOSFETs. This means the drain to source resistance is higher than you think. You need to look at more than just the Vgs(th).

If you don't see the phrase "logic level" on the first page of the data sheet, you should scrutinize the conduction curves before trying to drive them with 5V or less.

Possible solutions are to use a gate driver, or logic level MOSFETs to drive the power FETs.

\$\endgroup\$
  • \$\begingroup\$ Thanks for this very clear answer. Looking at the graph for the 25 °C case, if I draw a line from a drain-to-source voltage of ~ 16 V, the corresponding drain-to-source current on the Vgs = 5.0 V curve is ~ 10 A. Am I right in thinking this means a drain-to-source resistance of 1.6 Ω? I think this would explain the observed decrease in motor power. I will look into the alternatives you have suggested. \$\endgroup\$ – jholzer Apr 12 at 22:30
  • 1
    \$\begingroup\$ That is one way to look at it. Note that that resistance will be highly voltage dependent. Another way to look at it is to notice how the current doesn't increase after about a volt of drain-source. This voltage drop is the real killer. That power is dissipated as heat in the MOSFET which further lowers the amount of current it can pass (as you can see by looking at second graph). \$\endgroup\$ – evildemonic Apr 12 at 23:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.