Speed Controller - error in Circuit

The IRFZ44 is the bigger brother of the IRFZ24 with about three times the current rating but those are both ancient devices.

Nowadays you can get devices with Rds.on in the 1mR region. Looking through their selector, I see that Farnell stock the IRF2804 which is a 40V device in TO-220, 2.3mR, 280A (25C) for £2. That sounds like a reasonable candidate. You'd be doing well to blow that up, even with no flywheel diode fitted and no current limit....

Murray

I doubt if this board would benefit from Rdson much below 6mR bearing in mind it is designed for 10A operation. Lower Rdson devices not only put the price up but also have much higher value of gate charge requiring more power from the drive circuits. More than the 555 could deliver without slowing the switching speed. Better to save your pennies and stick with a lower spec device.

Thanks for the suggestions on devices. After losing the first board I fitted an IRF3205 Vdss 55V RDss(on) 8mR this did not fix it. As I understand it this device will generate more heat as the RDss is higher ? Anyway the output of the 7812 is 2.5 V and I can detect some heat coming off the board from somewhere with no motor connected. Does that mean the reg is dead or is something dragging the voltage down ? The PSU is sitting at its usual 22V or so volts. I checked the mosfet for a d-s short and it's ok. I didn't measure the 555 as the reg volts are out. I can put a scope on it if needed; I picked up an analogue 35meg dual trace Hitachi for next to nothing and can measure things even if I don't understand what the trace is telling me. I will try to find board 2 and measure that.

I had a suspicion that it was not the MOSFET that was at fault.

We have two possibilities:

1) Regulator is dead

2) The load on the regulator is so high that the regulator is going into thermal shutdown. Which means that the 555 is taking to much power (possibly dead).

First check that the there is not a Gate - Source short on the MOSFET. Just possible this is the problem.

If thats OK then remove the 555 and check the o/p of the regulator is now correct. Only problem is that the 555 is in such an awkward position that the easiest way to remove it will be to cut the legs off. But this means replacing the 555 even if it is not faulty. An alternative might be to lift the output leg on the regulator but this may have similar problems to removal of the 555. You will have to decide!

Until you have 12v across a working 555 there won't be anything for a scope to see.

Best of luck.

Trevor

The gate charge is about 160nC for the 2804 vs 20nC for the IRFZ24 ie a factor of 8. That's because modern devices have a much lower gate charge than those devices from The Ark.

The power required to drive the gate is 1/2 x Q x V x f (unless you know better?). With a 12V supply and switching at 1000Hz I reckon that's about 1mW - hardly enough to release the magic smoke. What speed do you think we should slow it down to so that the 555 is able to drive it?

If you are going to bother building some thing like this without any form of protection other than a fuse or a current limited PSU you might as well choose a robust FET.

Tony - the 3205 seems to be an 8mR device compared to the 70mR for the IRFZ24, so the losses should be correspondingly lower. The total gate charge is very similar to the 2804, so the 555 should be fine with it.

If you have got something dragging the output of the 7812 down to 2.5V, something must be getting warm (apart from the 7812!). Can you tell by using your finger? Or see where the current is flowing using a DVM to follow the track drops.

Try removing the 555 and see if the voltage recovers to 12V. If possible replace it in a socket. There's not a lot to go wrong really.

Murray

Gents

I'll recheck for the source of heat and probably remove the 7812 as its a little easier to access apart from the retaining screw on the heatsink - I had to shorten a 25mm hex bit and use a right angle bit driver when I changed the mosfet.

No great loss if I need to destroy the 555, at 10 for £1.70 they really are cheap as chips (sorry).

Murray please can you explain how to do this ?

"Or see where the current is flowing using a DVM to follow the track drops."

As one measures current in series ( and can't do that unless one lifts a device leg) do you mean measure the voltage drop various points to referenced to ground ?

Good news on the remaining board - it is also OK again in must have been the C/O switch.

Thanks for the info re RDss.

Tony

You need a sensitive voltmeter (200mV range?) and if you are lucky with the layout, you may be able to see the voltage gradually drop as you probe the track, following the fault current from the output of the regulator towards the return (0V ground) connection, with the other probe attached to ground. Most of that 2.5V drop will be across one device (most likely the FET, possibly the 555). If you see a track that has no measurable voltage drop across the length of it, the chances are that it isn't carrying much current.

It seems that the 7812 will fold back the short circuit current to about 1/4A in a dead short, so perhaps you are seeing only a bit more than that at 2.5V. That may not be enough current to allow you to follow track drops, depending on your meter. Given that there are so few compts in the circuit, it won't take long to remove them or cut / lift a few legs to locate the short.

Murray

The 7812 can protect itself in one of two ways. Overcurrent fold back or thermal shutdown. From experience thermal shutdown is more likely but in any case the output voltage would drop significantly which would explain the 2.5V. Easiest way to prove this is to lift the the Output leg on the regulator. When powered up the output leg should have 12V on it. If not then the regulator is dead.

Then it is hunt the failed component. If there is a short on the Gate-Source junction of the MOSFET this could cause the 555 to lock up and overload the regulator. Otherwise you are down to a failed electroytic capacitor ( the one we called C3) or more likely a failed 555.

Trevor

Gents,

I checked the MOSFET and it is OK so pulled the 7812 with the aid of the bit I described earlier

and soldered it to a bit of stripboard for testing. I put a led with resistor across it to pull a bit of current and the out put was around 2.9V so my conclusion is that the reg is fried. Before removal I tried looking for the source of heat, definitely not the 555 probaly the 7812 but the thing was only barely above ambient. I should have the new device in a couple of days & will put on in. I suppose I could power up the 555 with an external supply but its probably not worth the effort and I can change it if I need to. Anyway I'll report back whn I have some news

Murray,

Thanks for the explanation on the track drops I'll keep it in mind for the future.

Tony

Good work. At least its a cheap and easily sourced component.

The question is what caused it to blow. They are pretty rugged devices. Unlikely to be simple overcurrent as they are well protected.

The only serious vulnerability is the situation where o/p voltage is higher than input. You mentioned shorting the motor terminals this could cause this situation to occur and there is sufficient energy in C3 to do the damage. The good news is that this will only have damaged the regulator. A repetition of the fault can be avoided by putting a diode in series with the input of the 7812 say a 1N4001 or similar.

Trevor

Trevor,

New reg fitted and I'm a happy man - thank you and as you predicted the IRF3205 is fine. I will see how I can retro fit the protection diode.

I have learned a lot; not to jump straight to suspecting the output device and to check the simple things like voltages first.

I have had a quickl look at current limiting circuits but nothing stands out , any suggestions would be welcome.

Tony

It's stretching the thread topic somewhat, but I think this may be of interest:

Roger Amos [of Model Railway Electronics fame] has published his controller circuit here.

This is a clever combination of PWM and 'servo' Voltage control. ... Although intended for model railway use; it may prove very useful for small tools.

MichaelG.

Sorry to stretch the thread even further but having seen Tony's (very good) attempt at reverse engineering of a PCB assembly & his comment about a 3 layer board, I thought I'd offer some advice to anyone attempting to do something similar.

I hesitate to use the word 'never' but even so, as someone who designs PCB for a living (and has done so on & off for over 35 years), I should point out that PCBs never have an odd number of copper layers - it's always even (it's all to do with the manufacturing process).

So if you look at the edge of a PCB and think you can count an odd number of inner layers, you've either counted too many or missed one.

Brian

Single sided PCBs have an odd number of copper layers as far as I'm aware

Russell.

Very clever