Simple rc esc

[Andrew Cowan]

Just a thought - have you thought about a low PWM to make it quite? You proably wouldn't hear a 200Hz PWM signal, and if you did, it would sound much more like an engine that the current PWM signal you have.

A

 

[8man1320]

That's the whole idea. I can change it higher or lower, to tune the sound out.

I will add a reverse relay, once the forward is ironed out.

I am thinking about adding a li-po cutoff later.

Even if I have to spend a bit to build it. It will be WAY better (for me)
than anything that I can buy. And if not, I change the program.

 

[Andrew Cowan]

Making the frequency lower makes it lots easier - you should be able to drive the FET directly from the PICAXE.

A

 

[BeanieBots]

In your initial post, you talk of running at slow speeds.
In most cases, a low PWM frequency gives better speed regulation at low speeds. With a 'proper' driver, you will have scope to experiment with different frequencies to find which work best for your motor, application and ears

 

[boriz]

“Do I really need that many caps??”

Maybe, maybe not. It’s one of the compromises I mentioned earlier. 470uF electrolytic will be ok for low/medium frequency filtering, but effectiveness will roll off at higher frequencies. The 100nF in parallel captures the high frequencies that get past the electrolytics. It’s not unheard of to use three or even four parallel capacitors, each a couple of order-of-magnitude smaller than the last. And when you’re taking it really seriously, each with its own unique chemistry.

Also note that PCB tracks are effectively low value inductors/resistors, and proximal tracks are transformers/capacitors, so keeping the filtering caps close to what you are trying to filter helps to improve effectiveness and reduce possible ringing/oscillating effects. This becomes more important/significant with larger currents and higher frequencies. It’s down to good design technique really. It may work perfectly without the extra capacitors/precautions, but better safe than sorry. Especially when you’re talking just a few pence worth of components.

 

[rolfi@tisco.no]

Indeed , the MOSFET with dissipate power and get hot during switching, and with an inductive load , such as a motor, switching OFF is the critical part...
However, little has been said about ON-losses due to MOSFET resistance, which should be low. For a MOSFET, that implies a large maximum (drain) current spec. The data for IRF530, state a typical ON drain-source voltage of 1V at only 6A, driven by a 7V gate voltage, yielding 6W of dissipation. Without a sizeable heatsink, that will give you hot MOSFETs, and increased ON-resistance.. in fact thermal runaway may occur, leading to permanent failure! It will also steal power from your motor.
Considering the currents mentioned by 8man1320, the IRF530 is thus a bad choice.
Better with a really low ON-resistance type, like the Fairchild FDU8580 (or IRF3706). It is reasonably priced, housed in a really neat and small TO-251AA package, and will consume around 1W when delivering 10A(!) at 7V gate drive. A truly impressive litte fella! The gate charge at 7V will be around 12nC, much less than for 2 IRF530’s, and easily handled by the TC1411 driver. Even a simple self-made cicuit made from a push-pull emitter follower will do, with bases directly connected to the 8M output. (Try using non-darlington(!) 500mA transistors, like BC337/327 (and many others), and tie the emitters to the 7V supply (pos side) via a 1kohm resistor..).
Now, if the FDU8580 (or IRF3706) is hard to obtain, there are lot of other suitable types. Just search for low ON drain-source resistance (< 20 milliohm), low gate threshold voltage, and even low drain-source voltage ratings!

 

[8man1320]

I'm going for the TC4424 w/ IRF3706 FETs.

Now to figure out what I don't have here, so that I can can order.

 

[Goeytex]

My apologies for suggesting the IRF530. I did not realize the currents involved with these RC
motors.. Rolfi is correct it is inadequate for this application.

So digging deeper, I do not have a SPICE model for the IRF3706 but I do have one for the
RF3717 which is very close .

I ran a transient analysis at 10Khz and 50 percent duty cycle on a .3 ohm resistive load using
2 IRF 3717's in parallel. Here are the efficiency results.

--- Efficiency Report ---


Input: 85.8W
Output: 85.04W

Ref. Irms Ipeak Dissipation

FET 1 8418mA 11897mA 389mW IRF3717 1
FET 2 8418mA 11897mA 389mW IRF3717 1
R1 16836mA 23793mA 85040mW Load
R2 16mA 423mA 2mW 10 ohm Gate Resistor
U1 16mA 423mA 1mW FET driver

The IRF 530s had to dissipate well over 4 watts each given the same conditions. I realize that the load is actually inductive but the comparative results are still relevant.

I also looked up several popular manufacturers of ESC's for radio cars and noted that they parallel up to 5 FETs or possibly more to distribute the load and the heat dissipation. They also use substantial heat sinks, and 1 even had a small fan attached.

If the Spice calculations are close, the heat sink will be required to dissipate a total of about 800 mw of heat if using 2 IRF 3706's in parallel, given an 8.5 amp average (RMS really ) current. I would figure on about 3 watts to keep it on the safe side. A heat sink for an older Athalon CPU should work nicely.

I have had good results bonding SMD FETS and other parts directly to aluminum and copper heastinks using "Artic Silver" Thermal Epoxy.

Goeytex

 

[boriz]

“(RMS really )”

You sure about that? For a square wave?

I got a bunch of P55NFO6L MOSFETs from Rapid, ‘coz they were cheap. 55 Amps continuous (220 Amps pulsed), 60 Volts drain max, Low (logic) threshold, On resistance 0.016 Ohms with 5v Gate and 27 Amp load. Great specs for low frequency Picaxe stuff, but only discovered later the fairly significant 1.7nF gate capacitance. Guess that’s why they were cheap

This Gate, driven from a Picaxe output through a 220R resistor will get to 90% of 5v in about 800nS. Overestimating a little (always a good idea) let’s say 1uS. So for a good 99% fully-on-time during one cycle, you’re looking at about 10KHz at most. Could be boosted a little by using a small cap across the 220R.

You could probably double that, 20KHz or more using a driver, though there’s probably a law of diminishing returns with increasing frequency by virtue of the motor coil inductance. I’m with Andrew on this. A few hundred Hz would be my choice (for initial experimenting).

No reason why you should go with this MOSFET. Just happens to be the type I have lying around. Spend a little more and you’ll get better specs.

 

[boriz]

One of the beauties of Picaxe is that once you have a MOSFET and test rig, It&#8217;s dead easy to change the frequency during experiments. Could even use a Pot to dial the frequency &#8216;live&#8217; while running the motor. This would also enable comparisons of torque/thrust/efficiency to be made fairly easily.

 

[boriz]

Come to think of it. You may even discover that for low revs, one frequency is best, but another best for higher revs. Again, easy to implement in Picaxe code. I'd love to to see your results posted here.

 

[8man1320]

As for results.. when I get them, sure I'll post them up.

2 IRF3706 with a heatsink. Any reason not to use these?

I'm ordering today.


Wow, the athlon heatsinks are big.

 

[pyrogaz]

“(RMS really )”

I got a bunch of P55NFO6L MOSFETs from Rapid, ‘coz they were cheap. 55 Amps continuous (220 Amps pulsed), 60 Volts drain max, Low (logic) threshold, On resistance 0.016 Ohms with 5v Gate and 27 Amp load. Great specs for low frequency Picaxe stuff, but only discovered later the fairly significant 1.7nF gate capacitance. Guess that’s why they were cheap

I'm using four of these mosfets in a picaxe based 24v 130A speed controller, I use a TC4427 to drive the gate to 12v and a PWM frequency of 20khz. The 130A rating is a worst case stalled motor and mosfets held at 12v, they have survived numerous 20 second tests of this scenario as well as a constant 40A for 1 hour with no heating so very happy with them.

 

[rolfi@tisco.no]

I think it will be quite OK for you to use only one IRF3706, it's ON-resistance will be around 10milliohms. 12A of current makes 1.5W when fully on. Switching losses will increase with frequency. Let's say 10kHz, i.e. a 100microsec period. A 0.1microsec switching time should be fine, - that implies that the gate charge for the IRF3706 - ~30nC in this case - has to be charged/discharged during 0.1us, which yields an average charge/discharge gate current of 30nC/0.1us = 300mA. So, the driver should be able to source and sink (not the least!) ~500mA. (The average gate current, however will only be 0.3mA..)
Now, the IRF3706 comes in a TO220 housing, so a couple of watts might not require a heatsink, at least not a big one. A small piece of alu or Cu plate will probably be fine, may be 3 -4 sq inches? At least you could give it a try..

 

[Goeytex]

“(RMS really )”

You sure about that? For a square wave?

Hmm.

Do you really want to quibble over the use of "RMS" with a square wave ?

I don't.

But a quick Internet search will give the formula for RMS Voltage of a square wave. SPICE also calculates RMS vs "Average" . Besides a square wave is not really "square" and has other components involved. No ?

 

[boriz]

Well if you're sure, that's good enough for me