Have a go yourself.
Look for a diode that can carry 5A (or more) continuously with a pulse current rating of 50A or so. The former can be searched for directly, the latter will probably require you to look into the datasheets. This diode only ever carries brief pulses of current, so you can go closer to the maximum rating of the device.
Once you're skilled doing that, try for a mosfet. Look for 10A or more continuous (or whatever is double the maximum expected continuous current.
Look then for a continuous current (for t <= 10 secs) exceeding the stall current. Sometimes this is not given near the top of the datasheet and you need to go looking for a graph showing maximum drain current against pulse length. Typically several curves are shown, each for a different duty cycle. For your situation, you might need to look at the 10 second pulse time and the duty of 0.5. (and then for safety, halve it!)
These figures will assume some heatsinking, the SO8 package provides limited opportunities for this. The larger TO-220 package allows you to bolt it to something.
4. If we do change the mosfet type, should the diode and resistors type, change as well?
I've mentioned the diode earlier. The short answer is yes.
The resistor is only a belt and braces safety measure. It need not change. And it can be pretty much the smallest component you can find
5. Is it more complicated to solder wires to a surface mount mosfet, than on to a leads one?
Yes. the leads are short and stumpy and designed to sit flat on a PCB. Presumably you're not using a PCB. This means "dead bug" wiring (google it). You'll need a steady hand, fine wire (wire wrap wire is ideal) and some patience.
Otherwise, what other considerations should one have
One of the main considerations is that the leads are used to conduct heat away from the component onto an area of board which allow it to radiate (or conduct, or convect) away. You won't have that, so your power dissipation will need to be kept even lower.
As an example, consider 20A through a resistance of 0.02 ohms. The power dissipated is 20*20*0.02 = 8W -- This could be what happens during startup and even worse could happen if the motor is stalled. The SO-8 package (well, that device we looked at first) could only handle 1.5W max continuous dissipation. If we read further into the datasheet, this may have required connection to some particular area of PCB, without it the max dissipation would be much lower.
if he is choosing a surface mount or leads, if he knows that he is going to solder the ends with wires? Is the leads mosfet much heavier?
The main consideration is how much of the package's specification requires the leads to be connected to a large piece of PCB. The same thing applies to larger through-hole devices, but their larger size and mass normally allows them to have a higher dissipation without being connected to anything.
I tried to recalculate the spec filters on DigiKey myself, but got lost in understanding and choosing all the spec options.
The only real change is to increase the range of currents (so for the mosfet change from 6A to 20, to 10A to 50A or similar). You might consider reducing the minimum voltage from 35V down to 20V just to give you more low voltage (which tend to be high current) options.
6. Just as one example, of the problems of understanding the specs on my own:
Why is it, that if the said current was 3 Amps, you asked me to filter, on 'id' field, 6-20 Amps?
I may have covered this above a little, however there are several reasons:
1) you never want to design a circuit to use a device right up to its maximum ratings (50% of them is probably safer)
2) the maximum ratings are often not achievable easily (and almost certainly not without heatsinking)
3) there may be better (and even cheaper) devices available that can carry more current than you require.
4) there is no point in looking at devices many times larger in capacity because at some point they're going to just be bigger, heavier, more expensive, and have other limitations.
So I chose a range of between double (at the low end) to around 5 to 10 times (at the high end) the maximum typical current.
Another example, out of many, about my difficulty of understanding the specs would be:
7. I want to switch the motor from the Arduino via the mosfet, using the Arduino 5V PWM or analog signal. What spec filter should i apply, to be sure that the mosfet can react to 5V and that 5V is not too high or too low for it?
OK, to know that 5V is OK, look for a device labelled as a "logic level" device. In more detail, this means:
1) That Vgs(th) (the voltage that the device starts to turn on) is well below 5V (in the current case it is given as typically 1.85V -- see page 2 under "On characteristics").
2) You will also find that 5V is well within the range of Vgs(max) (The second parameter listed on page 1 under "Maximum ratings" is a Vgs of +/-20V).
3) Thirdly, that at 5V (often specified at 4.5V) the device is substantially turned on. In this datasheet we see on page 1 the headline figure of 14milliohm Rds(on) at Vgs of 4.5V. This means the resistance across the device is 14/1000 of an ohm when you apply 4.5V to the gate.
4) For more details you can also look at the Vds graph (Figure 1 in this datasheet) You'll notice that it shows voltage vs current for many gate voltages. You can see that for low gate voltages the graph rises then hits a knee and starts to ho substantially horizontally. You want to keep well away from the knee! You will note that the knee does not even appear on the graph for Vgs of 4.5V, and that is for currents well above what would destroy the device. This tells you that the device will drop about 0.7V at 50A with a Vgs of 4.5V leading to a power dissipation of 35W which is enough to destroy the component VERY quickly. (the spec tells us that we can do this for 1/100,000th of a second)
Now to the analog Vs PWM.
Let me tell you that analog is a loser out of the box.
PWM is better, but remember I asked you about how often the motor would be switched?
One of the issues with PWM is that you need to turn the device fully on to fully off and fully off to fully on lots and lots. And a mosfet doesn't do this instantaneously. Effectively it transitions from on to off (or off to on) and we can assume that the power dissipated during this time averages V * I / 2 (so if you're PWMing a load taking 5A at 12V, the average dissipation during switching is 5 * 12 / 2 = 30W) This might be compared to dissipation of 0 when the device is off and a dissipation of 5 * 5 * 0.02 = 0.5W when turned on.
So now we need to determine how fast the device turns on and off, multiply that by the frequency, then by 2, and then by the switching dissipation to calculate the average dissipation over a second.
I'll do this quickly...
From the datasheet (page 2 "Charges, capacitance and gate resistance") we see the total gate charge is up to 50nC. The arduino can supply 20mA, so the time to charge or discharge the gate capacitance is C/I = 50E-9/20E-3 = 2.5E-6. SO it will turn on or off in about 2.5uS. We check the datasheet to make sure that this isn't less than the device's rise or fall time (we would use these if it did). Note that these are specified with a gate current of 10A! we're using 20mA.
Now let's assume we're doing PWM at a frequency of 4000Hz (not real fast). The switching losses are 4000 * 2.5E-6 * 2 * 30 = 0.6W
Our total dissipation is 0.6 + 0.5 = 1.1W. Note that the switching losses are larger than the other losses. Also note that the switching losses depend on frequency and the static loss on duty cycle. If the duty cycle were 50% the switching losses would still be 0.6W, but the static loss would be 0.25W. If we reduced the frequency to 1000Hz, the switching losses would fall to 0.15W. If we could increase the gate current (perhaps by paralleling several arduino outputs or by using a gate driver) we could reduce it further (note that paralleling outputs requires that you can switch them all simultaneously or you may make things *MUCH* worse)
Another question i have, would be:
8. If (with your help, i hope), a mosfet is identified, that is rated continuous 5 Amps and stall 50 Amps, would such a mosfet also be good, for a much less stronger motor, for example a 1 Amps continuous and 5 Amps stall, or does a weaker motor, also require a new type or mosfet?
The stall current is the motor's rating, not the mosfet's but I understand what you're asking.
Yes, you could replace the motor with a smaller load of the same type with no further thoughts.
Mosfets do leak a small amount of current, so something rated to switch several hundred amps might leak a tiny current which would make it less suitable to switch a 20mA LED, but it's unlikely you'll try something so extreme.
9. I think that perhaps i should indicate again, that i will be using a LiPo 3S 25-30C 1-1.5 Amps 11.1V (12.5 V charged) battery.
Yeah, you'll need to be careful because if the motor draws 50A surge currents you'll get really odd things happening to your arduino's supply rail and something called "ground bounce" that may make switching the mosfet harder than you imagine.
As you may have noticed, there are many things to be aware of and (maybe you haven't noticed yet) but tradeoffs you need to make.
oh, and good questions by the way.