Hi All,

I've been working on motor timers for a small Free Flight model aeroplanes (ie not radio controlled) for a local competition and would appreciate some guidance please. I've been learning my way through this project but am very much a hobbyist and would appreciate having any pitfalls pointed out. This thread details the motor timer I've mentioned before in a thread about the other half of the project - the 'Dethermalizer' function that prevents the model from flying away if it catches big thermal lift in the gliding phase: https://forum.allaboutcircuits.com/threads/mic1555-sequential-timer-triggers-on-power-up.182657/ As mentioned before the footprint and weight of the timer needs to be kept to an absolute minimum wherever possible.

In the climb phase the model uses a one cell Lipo at 3.7V running a DC coreless 8mm 'Pager style' motor. The motor is started by a momentary push button and runs for maximum of 8 seconds as a one shot before stopping. The model then glides for as long as possible. The motor timer has gone through a few stages of development as I will outline:

1) The simplest iteration of the timer is one we have used for some years in our model club, a simple resistor/capacitor and MOSFET as shown in the first circuit attachment. However as the rc voltage drops through the linear region of the FET there is a period of decaying motor power and some loss of climbing time.

2) An improved version was to use an LM321 Op Amp Comparator to provide a 'hard stop' by driving the FET gate low when the r/c reaches a reference voltage. However a further problem was that while gliding the coreless motors will freewheel in the airstream which creates a lot of drag and spoils the performance so some sort of propeller brake was required.

3) I had the idea of using a P channel FET to open a short across the motor once it had stopped, but hopefully avoiding any shorting of the battery. I had experimented with the simple timer by matching the gate thresholds of the two FETs so that they did not overlap and avoided shorting the battery. This worked to some degree but wasn't very repeatable. Once the hard stop OP Amp timer came along I initially used the same Op Amp output to drive both the N channel motor FET and the P channel braking FET. This worked in practice, as can be seen here:

With the hard stop, the gate thresholds didn't seem to matter but someone pointed out that even with fast switching there is potential for a brief short. I don't have equipment to detect a short in any case so my first question is: Would driving both FETs on one pin really be problematic or is the switching (both on and off) fast enough to avoid any practical battery short issues? Weight and footprint is everything so if the potential issues are trivial then the simpler circuit would be preferable.

4) In order to solve the 'two FETs on a single output' problem I came up with a double Op Amp version which used a three part voltage divider and a single r/c. This - in theory - was to separate the switching of the motor and the braking short by a few micro seconds. I need it to be as soon as possible to use the back EMF to stop the propeller - if the brake is applied too late the prop continues to spin and will re-gain momentum from the airflow. The resistors in the voltage divider are 470K/3.3K/220K ie with 3.3K separating the ON and Brake. This gives me a ratio 68%/0.47%/32% so I think the braking short is applying about 50 milliseconds after the motor is stopped.

My second question is: Have I solved the problem with this double OP Amp circuit or am I missing something that could still cause issues? I can't really measure the voltage accurately enough (multimeter) to be sure that the Op Amps are working as I hope they are. Will the OP Amps reliably switch in sequence as the r/c voltage falls or is there any source of inaccuracy here that I'm not aware of?


Thanks in a advance for any help and as always I'm open to hearing ideas about how I could achieve a hard motor stop and brake circuit in a simpler, lighter or a smaller footprint.

Thanks,
Jon

 

However a further problem was that while gliding the coreless motors will freewheel in the airstream which creates a lot of drag and spoils the performance so some sort of propeller brake was required.

I would think that a free-wheeling propeller would have *less* drag than a fixed one.

AND - in schematic #4, Q2 applies a dead short across the battery.

The nice thing about the three-resistor reference voltage string is that it guarantees a deadband when both transistors are off during a transition in either direction, so keep that part. For the dynamic brake, try this:

Keep the p-channel FET. Disconnect the drain from GND, and connect it to the bottom end of the motor (the node with the n-channel FET drain). It is *critical* that the output of the lower comparator (reference designators - !) gets very close to Vbat when in the high state. If Q2 is on just a tiny bit, it shunts current around the motor during the run phase.

ak

 

AND - in schematic #4, Q2 applies a dead short across the battery.

Ah. Yes... that would be because I've drawn it wrong. Circuits #3 and #4 now corrected and show what I have actually prototyped

 

The nice thing about the three-resistor reference voltage string is that it guarantees a deadband when both transistors are off during a transition in either direction, so keep that part.

That's good - I thought that was what I was doing but I'm aware I don't know what I don't know.

For the dynamic brake, try this:

Keep the p-channel FET. Disconnect the drain from GND, and connect it to the bottom end of the motor (the node with the n-channel FET drain).

That is what I was already doing actually - I just should have checked it properly when I drew it out.

It is *critical* that the output of the lower comparator (reference designators - !) gets very close to Vbat when in the high state. If Q2 is on just a tiny bit, it shunts current around the motor during the run phase.

Thanks - this makes sense. I had been using standard LM321s but I have ordered some rail-to-rail LMV321s which should help hopefully? I was also looking at the P FET gate threshold to make sure of some headroom. I'm using SI2347DS-T1-GE3 which has a moderate but not excessively low VGS(TH).

 

I would think that a free-wheeling propeller would have *less* drag than a fixed one.

Aerodynamics is the one subject I am fairly comfortable with as it happens. The answer to this one is "it depends."
A big low RPM/high pitch prop with low shaft loads may well be less drag if allowed to free wheel - as in the case of a classic rubber powered model plane. But in electric models with a small blade area/high RPM/low pitch which is turning over a tiny generator then the windmilling drag is surprisingly high and it is a significant glide performance advantage to stop it.

 

I'm aware I don't know what I don't know.

CHAP. XVII. The Master said, 'Yu, shall I teach you what knowledge is? When you know a thing, to hold that you know it; and when you do not know a thing, to allow that you do not know it;-- this is knowledge.

Confucius (September 28, 551 – 479 BC)

Or, in American . . .

Know what you know, and know that you don't know what you don't know.

ak

 

Thanks for the help AK.

Just one further question while I think of it:
How much of an issue is it to use a momentary switch to charge or discharge a capacitor?
In the simple circuit #1, R2 was added to limit current on the switch but is it really necessary?

As I understand it the initial charge of the empty C1 is basically a short for a fraction of a second but is this an issue when it's so brief?

The switch is rated for 50mA. The capacitors are usually around 33uF to 100uF, R1 around 1M and I was using a 100R for R2.

This question also applies to a version of #1 where a momentary switch across the cap is pressed to discharge it and so end the motor run.

Thanks.

 

Depends. I tend to dismiss transients in the microsecond range, and into the millisecond range if the contacts are rated for 100 mA and above. Electrolytic capacitors have a property called the equivalent series resistance (ESR) that is just what it sounds like. This can be used to estimate the peak charge/discharge current, giving you a number to work with when selecting the parts.

A common rule of thumb is that electronic components should be rated for twice their in-circuit conditions. Use a 25 V cap in a 12 V device, a 1 W resistor in a 1/2 W circuit, etc. But for repeated contact transients, I'd go with 5x to 10x

Wally (crutschow) is more sensitive about this than am I. More than once his only comment on one of my circuits has been that it needs a switch-contact-protection resistor.

ak

 

Ok thanks. I've been using small case Tantalums until recently but have moved to 0805 Ceramics in pursuit of miniaturisation. The ceramics seem to have much lower ESR (?) and so larger potential current (?) although for a briefer period.

I'm finding it difficult to find specific ESR figures but based on similar components I think the high current is over and done in a few microseconds.

An example using a 47uF Tantalum with 3 ohm ESR gives me a max current of 1.4A and the cap should have charged to 2.1V in 0.1 milliseconds.

So if I've understood the principles correctly I'm in the order of magnitude where I can probably ignore the R2. Wally may disagree

I know it makes sense to just chuck it in for safety's sake but as always I'm trying to simplify and lighten everything for flying purposes.

 

My latest effort - a timer for scale models with adjustable PWM output. 0.4g and half an inch square.
Basically a MIC1555 in monostable mode driving the CS(enable) pin of a MIC1557 in astable mode.

Thrust adjustment makes trimming out a flying scale model much easier.

 

Would driving both FETs on one pin really be problematic or is the switching (both on and off) fast enough to avoid any practical battery short issues?

I dont see how a short is possible. To me, its fine

 

I dont see how a short is possible. To me, its fine

If both FETs are allowed to be 'on' at the same time you would be shorting the battery.
The concern is that switching with the single pin means that this may be possible if the gate thresholds overlap (ie with low signal FETs) Transient but possible.