ULN2003A driving IRF540N mosfet

Like this

"Here's one I prepared earlier"

http://www.element14.com/community/groups/internet-of-things/blog/2014/08/05/drive-big-things-with-added-safety-opto-isolators

Regards

Peter

Trust you to have polished your crystal ball. 

I always cringe when expensive hardware is not protected ...

The advantage of the opto-couplers is they can be remote from the processor, and even duplicated.

Mark

Wasn't the optoisolator the preferred method last time this was raised BTW?

No.  This is the first time I have heard of a optoisolator.  In the last discussion no-one even suggested them, which is why I was a bit confused when they came up here.

Link to previous discussion:  http://www.element14.com/community/message/113013/l/arduino-and-transistors-question--re-school-musical-production#11301…

I will take a detailed look at everything that everyone has posted on this thread and consider what I might do.

>> I see he's using an unlicensed copy of Johnny Cad and added his own logo.....This is simply terrible!

I don't think any version of Johnny Cad is going to cut it for the project I have in mind.  Is there an more digital alternative to Johnny Cad?

Actually the opto-coupler did come up in reply #105

http://www.element14.com/community/message/114536/l/re-arduino-and-transistors-question--re-school-musical-production#114536

To be fair optoisolator and optocoupler are basically the same device.

Sometimes the 'isolator' can have a mechanical interrupter (ie mouse wheel, door switch, end stop) that breaks the link between the LED and detector.

You have obviously gotten to the point where your skills are being used more and more.

This is the opportunity to plan for the future and make the board/boards able to be multiversal.

If you make the board with an optocoupler and bring out both the Anode and Cathode along with the ability to change the series resistance, then it could be driven by almost anything you want.

Several could even be run in parallel, and at some distance from the controller, or switch bank or whatever.

I did this once with a Jackpot system controller which switched 24vAC and was perfect when one establishement wanted two signs linked together.

Mark

Optoisolator/optocouplers are are used to describe what is basically the same thing.  After a while people may use the wrong exact term but every one knows what they mean (fortunately) I'm sure this goes for many fields of knowledge.

If you go the E14 route I'm sure you can pic up some of the quad packages like this one here:

http://uk.farnell.com/toshiba/tlp620-4-gb/optocoupler-quad-5kv-transtr-o/dp/1225838

This is the UK one but I'm sure you can find the part in the .Au  E14or Farnell

For the cad package have you considered Eagle which has a free version for low complexity boards (this one should be OK).

Or go for completely opensource and go for Kicad

http://www.cadsoftusa.com/

http://www.kicad-pcb.org/display/KICAD/KiCad+EDA+Software+Suite

For faster response time you can use the Darlington output versions, this provides faster rise time on the outputs, these are what I used for the Electronic Referee project

mcb1 wrote:

 

Actually the opto-coupler did come up in reply #105

http://www.element14.com/community/message/114536/l/re-arduino-and-transistors-question--re-school-musical-production#114536

 

Aha!  So they did...

Must have skipped over those responses at the time (10 days before the musical started - lighting order due in 3 days).

Some of them are very interesting, and sort of complicated for what I want to do.

What are the chances of the MOSFET breaking anyway?

I was doing a bit of research last night on the different ways and reasons that MOSFETs fail.  I found this website particularly interesting:  4QD Information Section

In summary:

• Too much voltage supplied (either to the gate or source/drain) 12 VDC is well inside the allowable limits of IRF540 (which is max 100V)
• Too much current supplied (source/drain) only going to be supplying 0.5A on average maybe up to 1.0A at times through MOSFET (way inside max allowed: 33A)
• dV/dt failure Not controlling motors, so I should be right on this one.
• Excess Power dissipation Like before, using the MOSFET way inside its limit
• "Foreign Object Failure"
• Short-Circuit

Of the list above, only two of them could possibly cause a massive failure in the MOSFET:

1. Foreign Objects (leading to...)
2. Short Circuit

If a short circuit did occur (which still has a low chance when one is being extremely careful), then having an opto-coupler between the controller and the board would be nice.  I was thinking of making a sort of 'complimentary' buffer board with opto-couplers on it.  Maybe I could just buy one to save on the soldering.

I will have a poke around on Element14 AU and see what they have.

Replies to messages that came while I was typing this one (reverse order):

0. Noted.   Thanks peter.

1. I will give eagle cad free a go.

2. Thank you for all the links.  I will have a look once this is submitted.

3. More knowledge / information on  opto-couplers --> thank you for informing me on differing terminology.

This one could do the trick:

http://au.element14.com/vishay/cny74-4h/optocoupler-transistor-5000vrms/dp/1045432

But when using these with a micro-controller, you would have to provide some sort of 'boost'/buffer due to the input current being >30mA,  right?

yup, or this for the Darlington version

http://au.element14.com/vishay-semiconductor/ilq55/optocoupler-photodarlington-5300vrms/dp/1612450

I am reading the datasheet correctly:

maximum 125mA output current

20mA into the base (now an LED)

1. To use one of these do you have to buffer both the input and the output if you plan on driving >120mA and your micro-controller can only handle <15mA output?

2. Connect to controller like a normal LED?

you would have to provide some sort of 'boost'/buffer due to the input current being >30mA,  right?

Not necessarily, as this is the maximum current.

Your current transfer ratio is the 'gain/loss' in the system.

In this

In this spec they apply 5mA to the LED, and the Collector current is 2.5-30mA (50-600%) at 5v supply

At 10mA it is 6-12mA (60-120%) at 5v supply

The other one peter quoted is between 100 and 400% transfer but has a higher saturation voltage (0.9v), which may cause some problems depending what fet you use.

Since you will be driving a mosfet and its a voltage device, then high current isn't really necessary, and with the isolation, you can always use it to supply voltage to the gate, rather than drag down the voltage source.

Your investigation about failures is absolutely right.

For a production where you want nothing to go wrong having that litte extra insurance, for a few dollars is probably worth it.

It also gives you some increased flexibility, so its not wasted.

Mark

you would have to provide some sort of 'boost'/buffer due to the input current being >30mA,  right?

Not necessarily, as this is the maximum current.

Your current transfer ratio is the 'gain/loss' in the system.

In this

In this spec they apply 5mA to the LED, and the Collector current is 2.5-30mA (50-600%) at 5v supply

At 10mA it is 6-12mA (60-120%) at 5v supply

The other one peter quoted is between 100 and 400% transfer but has a higher saturation voltage (0.9v), which may cause some problems depending what fet you use.

Since you will be driving a mosfet and its a voltage device, then high current isn't really necessary, and with the isolation, you can always use it to supply voltage to the gate, rather than drag down the voltage source.

Your investigation about failures is absolutely right.

For a production where you want nothing to go wrong having that litte extra insurance, for a few dollars is probably worth it.

It also gives you some increased flexibility, so its not wasted.

Mark

Yes, for the supplying the FET there is nothing wrong with using the above Opto-coupler.

Still learning about opto-couplers ATM and getting my head around the different terminology but I understand everything that you said above.

However, I haven't been very clear on this thread and might have drifted a bit from the original question because I am working on a number of different projects at once, trying to see how I can improve each one with what is said in this discussion.  Other projects that I have done in the past have also drifted into this discussion as well.  I will just briefly outline each one for clarity.

Project One:  The Musical Set Prop Control Board (A.k,a:  "The Hot Box Sign")

Status:  Complete and FINISHED.  It is just being used as a reference for "project two" because it worked so well.

Project Two:  Christmas Lights Control Board for controlling 5050 RGB LED Strips

Status:  Planning stage.  It should be the one being discussed in this thread.  I might have mentioned it above somewhere but just for clarity I shall rehash:  I want to control seven RGB LED strips which vary in length from 0.5 meters to 2 meters.  I worked out that I need 21 channels to control each strips RGB colours individually. The power used by each 'channel' is approximately 0.6A maximum. I was going to use IRF540N MOSFETs allow me to control the LEDs from a micro-controller (possible a Raspberry Pi or Arduino).

Project Three:  Model Railway Redo

Status:  Planning Stage.  I will probably do a write up on this one later but I was going to use some of the content discussed here (such as: Opto-couplers, MOSFETs, ULN2003, etc.) on this project as well.  It involves controlling a variety of different devices (signals, point motors) with an Arduino, 98% software driven.

So in response to mcb1

Since you will be driving a mosfet and its a voltage device, then high current isn't really necessary, and with the isolation, you can always use it to supply voltage to the gate, rather than drag down the voltage source.

I understand but I have been thinking about a variety of projects as well as discussing this one and that's why I asked:

1. To use one of these do you have to buffer both the input and the output if you plan on driving >120mA and your micro-controller can only handle <15mA output?

So you don't have to buffer the input if you plan on driving < 5mA load due to the gain/loss ratio.  Thank you for answering my ambiguous questions.