2
\$\begingroup\$

I am participating in a student project where we are building small, portable vehicles, starting with a pocket bike and some materials. We are also allowed to buy parts on our own.

Our current plan is to make a motorized long-board (with 49cc gas engine, really dangerous plan, I know!).

What would be really cool, is to have load cells in the front and rear, to control throttle and brakes. The throttle and brakes are controlled by wires, and an attempt at measuring the force needed for braking gave a result around 20N.

So I thought "Maybe we can get some kind of proportional pull solenoids, and connect them to a control circuit and to the wires". And now I have sketched a schematic for one such circuit. The full thing is HERE!

schematic

simulate this circuit – Schematic created using CircuitLab

  • On the left I reduce (divide) the supply voltage with a potentiometer for calibration according to three uncertainties: 1. The gain of the op-amp. 2. The exact weight of the rider. 3. The bias of the cheap Chinese flat load cell.

  • Note that I have tried to model the sensor as a Wheatstone bridge, and the state of it with a small pot-meter in one of the lower arms of the bridge.

  • Then in the middle, I lead the tiny two-voltage signal into an op-amp without any feedback.

  • After the op-amp I have placed a diode, which is supposed to represent a forward reference diode that basically just subtracts a fixed voltage (ideally), representing about half of the weight of the rider, so that the board does not easily accelerate and brake simultaneously, and of course to avoid that it accelerates or brakes when you are standing neutrally.

  • Then there is a resistance, dimensioned to limit the base-emitter current of the transistor that switches on the current to the actuator that pulls the wire. I have, for now, represented the actuator by an inductor.

Practically, the calibration will be done after the rider mounts the board. He/she will bend down and turn the pot-meter (which, I assume, can be shared between front and rear pedals) slowly until the actuators start pulling the wires, and then just a bit back again, so the transistors don't leak too much current through the actuators, emptying the battery too quickly.

Note that strict linearity between load and acceleration is not a requirement, although some predictability and reasonable behavior would be nice. It is a learning project, most of all. (For the real thing we would be using electro-motors and a digital circuit, I suppose.)

So my question is simply: Does this make sense at all?

I do have specific doubts:

  • Some of the sensors I have read about specify "input impedance" and "output impedance" both around 1kOhm. Am I right to assume that means the resistance of the arms in the bridge are around that value, or is that a complete mix-up?

  • Is it OK to lower the voltage over the sensor instead of doing difficult (hard to find) feedback tricks with the op-amp, or will this incur unacceptable noise or something?

  • Does the gain of cheap op-amps vary too much that a pot-meter will suffice for calibration?

  • Can a forward reference diode be used this way?

  • Will running all parts of this circuit from the same battery be a problem?

  • Can a real proportional solenoid be modeled simply as an inductor, and if so, what would be a typical inductance value of one such device running on 5V, pulling about 10mm max with a force of 20N?

  • Will such an actuator be fast enough for emergency braking, yet slow enough that driving on uneven ground won't throw you off the board? If not, do I need to run it on a higher voltage than 5V, or (in the second not case) soften out the signal another way?

Now if this all makes sense: How, in what way, am I going to fail trying to implement it? (According to your experience, that is.)

I really welcome a recommendation on what kind of actuator I should look for. The trade websites I am searching do not help me well narrowing down the search.

Thanks in advance for helping me sort out some concerns in the planning process. I have tried to find answers on my own, but it seems the essentially relevant pieces of knowledge are a bit hard to locate and get a grip on.

Update: After reading this tutorial it turns out that it may be possible to use a simple pull solenoid, but it will require some tricks. I think making a biased, periodic control signal with proportional bias and amplitude could be the way to go. And it seems I need something called a "flywheel diode" to protect the BJT.

\$\endgroup\$
  • \$\begingroup\$ What did you sketch the schematic in? \$\endgroup\$ – ThreePhaseEel Feb 22 '16 at 23:24
  • \$\begingroup\$ This: falstad.com/circuit/circuitjs.html \$\endgroup\$ – Elias Hasle Feb 22 '16 at 23:28
  • \$\begingroup\$ Can you redraw it using the built-in schematic editor? \$\endgroup\$ – ThreePhaseEel Feb 23 '16 at 12:44
  • \$\begingroup\$ @ThreePhaseEel Done. :-) \$\endgroup\$ – Elias Hasle Feb 23 '16 at 13:17
2
\$\begingroup\$

This is quite a broad question and I can't hope to answer everything. I can share a few thoughts.

Firstly, as you have configured this op-amp, it is essentially a comparator. Op-amps have very high gain, high enough to be considered infinite, in most cases. If the non-inverting input is higher than the inverting input, the op-amp will drive the output as high as it can, nearly 5V. And if the non-inverting input is lower, it will drive the output as low as it can, nearly 0V.

There isn't really anything between these extremes. I think this will result in a very uncomfortable experience for the rider since there are only two levels of braking: all or nothing. These very sudden changes in acceleration (high jerk) won't give the rider much opportunity to adjust her balance.

Also, when the balance is such that braking is just turning on, small perturbations from the road surface, etc, will result in the brake rapidly turning on and off. Besides being uncomfortable, it's probably not the best thing for the mechanical components, either.

To solve this, you'll want to add some negative feedback to the op-amp to reduce its gain. This will make it possible to brake somewhere between full-on and full-off.

You probably also want to add a filter between the load cells and the op-amp. Bumps in the road will create a lot of noise, and unless you ignore this noise the brake will be rapidly switching between on and off with every little bump in the pavement. The balance of the rider is going to change much more slowly in relation to road vibration, so a low-pass filter would be appropriate.

With negative feedback for reduced gain and low-pass filtering, you'll probably end up with something like this at the core:

schematic

simulate this circuit – Schematic created using CircuitLab

Many details omitted of course: you'll need to do some more research to get the full picture. I'd suggest "op-amp low pass filter" and "differential amplifier" as search terms to start.


The thing you need is a flyback diode. Say you had this:

schematic

simulate this circuit

When you close the switch, the current in L1 increases until it becomes limited by the coil's resistance. (If L1 were an idea inductor, then current would just increase forever.) Now when the switch opens, the current through L1 can't change instantly. If you don't provide a place for it to go, it will find a place by creating a very high voltage, with the bottom of the inductor being a higher voltage than the top.

In this case, it will probably arc across the switch. In more complex circuits something else might happen, but in most cases, it's not good.

By adding the diode, you give a place for this current to go. The voltage across the inductor is limited by 0.6V and nothing gets broken.

\$\endgroup\$
  • \$\begingroup\$ Thanks for answering some of my questions. So: Could you say that the rated gain of the amplifier at 100000 is an understatement? Because my circuit, as drawn, is able to correct for gains in that range (by downsizing the input), possibly at the expense of something else. \$\endgroup\$ – Elias Hasle Feb 23 '16 at 14:42
  • \$\begingroup\$ @Anonymous TI's datasheet says LM358's typical open-loop gain is 100dB, or 10000000000. The output BJT adds even more gain, maybe another 20dB. That would mean to get a 0-5V swing on the output, the input swing needs to be on the order of 5pV. Not only is that impracticably small (thermal noise alone is three orders of magnitude greater), but the gain of op-amps isn't something that's tightly controlled. The LM358 datasheet doesn't even specify minimum or maximum values. That's not a parameter you want to rely on in your design. \$\endgroup\$ – Phil Frost Feb 23 '16 at 16:45

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.