aiken cmos switch

[R.G.]

if im understanding correctly the data sheet
1) coil resistance = 50 ohm
2) relay draw current / power consumption= 500mw
3) rated current = 100 Ma

http://www.futurlec.com.au/transistors/2N7000.jsp
1) im not sure but i think "Continuous Drain Current" is = 200Madc
so if i interpreted the data sheet correctly does that mean 1 x 2n7000 can only continuously run 2 of the relays im using?

That's right. Actually, that's going to be a marginal case, and doing that on a long term basis will be unreliable because of the edge-of-the-cliff heating, as you correctly found out.

You were also correct in your suspicions in the previous post: the 2N7000 is marginal in this application, running at 100% of its continuous current rating. Get a bigger horse - or use more horses.

Since the input of a 2N7000 is a tiny capacitance, and it's not fussy about the exact voltage once it's turned on, you could use one 2N7000 per relay, and parallel all the gates. That gives you one device running at 100ma per relay, and expandability all the way out to you can't afford the relays.

The single "bigger horse" approach is to get one MOSFET in a much more thermally and current-capable TO-220 package and drive multiples from it.

How to decide what to do? Both work electronically, you need mechanical and monetary data to decide the issue in practice. Here's a trick of mine that I use a lot. It seems obvious when I say it, but I've seen very few other people actually do it. When evaluating alternatives and you have a database-style sorter available, define the selector terms, then sort by price. See if you can use the cheapest thing that comes up. If no, go to the next more expensive alternative. Mouser's search makes this VERY easy, as you can set performance and packaging alternatives first, then sort by price.

First I looked up the price of one 2N7000. It's about $0.40. Then I looked for MOSFETs with a TO-220 package, and sorted by price. Cheapest TO-220 MOSFET at Mouser is the 511-STP24NF10, which is a 100V/24A device for $0.54. So for $0.14 you could move to a device that would theoretically run a hundred or more relays simultaneously. But will it work?

That depends on the necessary gate-source voltage to turn it on. You only have 5V, and that's low for power MOSFETs. Look at the datasheet: The guaranteed maximum Vt is 4V. Hmmm... marginal. Looking at the typical curves shows that typically a device will do 10A at 5V on the gate-source. That's a maybe. Some devices would not work.

Next least expensive... 511-STP24NF10 is $0.59, but has Vt = 4V.
Next... 844-IRF610PBF is $0.67, but Vt is again 4V
Next... 771-PSMN02230PL127 is $0.72, but Vt is 2.45V and the curves say 4.5V will saturate 10A to 0.2V. We have a winner.

The TO-220 package will dissipate 2W without a heat sink, so for an additional $0.32, you get a device that will thermally drive 7.5A down to 0.2V, dissipating 1.5W in the process, and driving up to 75 relays.

I would stop there. There may be even better devices for just a bit more money, but you're approaching where you'd have to add a heat sink. And .... taa-daaH! it's cheaper than two 2N7000's!!

[merlinb]

do i need an Irf Power mosfet equiv?
ie Irf510?  or am I still wandering down the wrong path?

It sounds like you're not saturating the transistor properly. A saturated transistor shouldn't get hot unless it's passing an amp or more. I'm not sure why he chose to use MOSFETs... It's a lot easier to ensure a BJT is saturated. Maybe he just wanted to save a few pennies on base resistors. I use BC337 for anything under 800mA.

[R.G.]

It sounds like you're not saturating the transistor properly. A saturated transistor shouldn't get hot unless it's passing an amp or more. I'm not sure why he chose to use MOSFETs... It's a lot easier to ensure a BJT is saturated. Maybe he just wanted to save a few pennies on base resistors. I use BC337 for anything under 800mA.

I think he's OK as far as it goes.  The curves for the 2N7000 show it will pull 200ma down to 0.8V with 5V on gate/source. That's 160mW. The package thermal spec shows up to 350mW with Tc=25C and 2.8mW/C, which, after some modestly repetitive calculations show that 160mW would be warm-to-hot at a fingertip, but not disastrous with only two relays.

A BC337 is better from the thermal standpoint in that it won't get as hot. It would work OK, and even with a base resistor would be marginally cheaper.

The real winner might be a uln2004 darlington array. $0.53 gets you seven 500ma drivers in one DIP, complete with catch diodes, and integrated base resistors.

We're well into the viable alternatives here. 

[njkmonty]

Firstly  can I say, Thank you for everyones suggestions.
Im still learning, and trying to do as much as I can myself before asking for help.

after a little thinking ive come to the conclusion that if possible go the TO-220 package as appose to Darlington Array suggestion.

My only problem is ....
I usually order through futurlec and Tayda, as Mousers shipping etc quite high and i have to spend over $200 to get free shipping, So my preference is through Futurlec.
That being said , they dont have an easy search /comparison parameter  tool , which makes it a little tougher.

Now the suggested PSMN022-30PL   http://www.mouser.com/ds/2/302/PSMN022-30PL-353491.pdf
specs i can mostly understand, however when looking through futurlecs offerings, im having difficulty distinguishing a substitution.
Ive been looking at IRF Series
http://www.futurlec.com.au/test13.jsp?category=TRANSMOSFET-IRF&category_title=IRF%20Mosfet%20Transistors&main_menu=TRANSISTOR&sub_menu=TRANSMOSFET-IRF
and i find  acouple similar but their Rds On - Drain-Source Resistance  is in Ohms not MOhms like the PSMN02230!!!!
PSMN022-30P
Transistor Polarity:                                   N-Channel   
Vds - Drain-Source Breakdown Voltage:   30 V   
Vgs - Gate-Source Breakdown Voltage:   20 V   
Id - Continuous Drain Current:                   30 A   
Rds On - Drain-Source Resistance:           22 mOhms   
Configuration:                                           Single   
Maximum Operating Temperature:          + 175 C   
Pd - Power Dissipation:                          41 W   
Package / Case:                                    TO-220-

Am I trying to get a part that very few make or Im not doing it properly!?
Without spoon feeding me , is it possible to find an alternative at Futurlec, & if so am I on the right path with the Irf Section?
or BU Section???

i just found this one that Futurlec has
BUZ72 10A 100V N-Channel Power MOSFET Transistor
Manufacturer:                                         Fairchild Semiconductor   
Product Category:                                 MOSFET   
RoHS:                                                 No   
Transistor Polarity:                                 N-Channel   
Vds - Drain-Source Breakdown Voltage:   100 V   
Vgs - Gate-Source Breakdown Voltage:   20 V   
Id - Continuous Drain Current:                9 A   
Rds On - Drain-Source Resistance:        250 mOhms   
Configuration:                                        Single   
Maximum Operating Temperature:      + 150 C   
Pd - Power Dissipation:                         40 W   
Mounting Style:   Through Hole   
Package / Case:   TO-220-3

[R.G.]

I usually order through futurlec and Tayda, as Mousers shipping etc quite high and i have to spend over $200 to get free shipping, So my preference is through Futurlec.
That being said , they dont have an easy search /comparison parameter  tool , which makes it a little tougher.

Mouser's search/comparison tool is so good, I cheat with it. I just took some suggestions from futurelec and popped them into Mouser's tool for a look.

If Futurelec didn't make a pricing mistake, there's a goodie hidden in there. They list the STP40NF10L for $0.79 (AuD, I think). Mouser lists that as a 100V 40A device in a TO220 that has a Vthreshold (turn on voltage) of 2.5V. If the part number and price are right, that's a steal for your purposes. Mouser wants US$2.38 for it.

Now the suggested PSMN022-30PL   http://www.mouser.com/ds/2/302/PSMN022-30PL-353491.pdf
specs i can mostly understand, however when looking through futurlecs offerings, im having difficulty distinguishing a substitution.

You're looking for (1) over twice your supply voltage; no problem there, they're all going to be more than 10V; (2) current rating; at 100ma/relay, you're getting four relays per ampere; (3) low saturation voltage at the desired current; since MOSFETs "saturate" to a resistance, you have to first determine how many relays you're going to run from one MOSFET, convert that to a total current by adding up 100ma per relay; a TO-220 can dissipate around 2W in free air, so you divide 2W by the square of the current, and the result is the maximum Rds you can stand in your MOSFET with the drive voltage you're going to give it. With the rds in hand, you look at the curves in the datasheet for Rds versus Vgs, and see if with 5V gate--source you can get low enough rds. Another way is to look at Id verus Vds curves for your gate voltage and pick out the point where the Vgs cujavascript:void(0);rve crosses your drain current, and multiply that drain current times the voltage at the point. If that's under 2W, you have a good candidate. Finally (4) look at the datasheet for the threshold Vgs where the device just begins to turn on. With a 5V maximum gate drive, this had better be less than 3V, ideally a lot less, for the guaranteed maximum.

When you've identified a candidate, look at price, cry some, then start over. 

Am I trying to get a part that very few make or Im not doing it properly!?
Without spoon feeding me , is it possible to find an alternative at Futurlec, & if so am I on the right path with the Irf Section?
or BU Section???

It's a complex selection process. You have to get at least enough and not too much of several variables: voltage rating, current rating, power *at the working current*, package, threshold voltage, and price. It is hard to do when you start.

I went over and contended with the futurelec web site. It's remarkably difficult to find info to make good decisions on. It's almost enough to make me pay high shipping charges. The simpler way here is to change your criteria. Look up the MOSFETs in all the categories that are in TO-220, have enough voltage and current (I'd start at about 10A on this), and then pick out the lowest price candidates. With your list in hand, search the net for datasheets on each one, and look up the Vgs threshold, and eliminate any over 3V in the "max" column. With the field narrowed down, look at the remaining candidates for the power considerations I mentioned. If they can't get the I²*Rds or Volts*amps under 2W with 5V Vgs and your desired current, they won't work and you'll have to go to the next higher prices. I'm guessing from what I've seen you can come in under AuD$1.00

[PRR]

> ie Irf510?

Sounds good.

[karbomusic]

It's sad but true that a minimal level of programming experience is becoming as necessary as learning to solder

As you probably remember RG, when you and I were growing up there was this thing called the three Rs

http://en.wikipedia.org/wiki/The_three_Rs

Since I spend most of my professional time in the programming, debugging and computer technology industry, I tell someone most every day that there is now a fourth R and that not having any clue about that side of things or at least simple scripting is quickly becoming yesterday's near equivalent of not being able to read and write.

[njkmonty]

Thank you again for such detailed info, 
I was a little afraid that i may have annoyed those here  with my questions.

i was getting confused when comparing as I saw mohms, thinking it was mega ohms and then seeing < 0.033 Ω    Diiirrrr!!!  milli ohms!
well thats clarified that bit up!

the prob with my design is that each mosfet may change the amount of relays being used. But im hoping maximum of 15 at anyone time

so..   15 x 100 miliamps (relay draw) = 1.5amps

a TO-220 can dissipate around 2W in free air, so you divide 2W by the square of the current

2w  divide by  1.5 x 1.5   or do you mean square root of 1.5


a) 2w divide by 2.25   or       = 0.88888888   max rds
b) 2w divide by 1.22             = 1.63934426   max rds

am i lost?
apart from learning this , (which I still want to grasp)  STP40NF10L   should do the job?  for 15 relays etc

With the rds in hand, you look at the curves in the datasheet for Rds versus Vgs

which one is that?



4 R's  sounds like a Pirate convention.

6 munths ago i could even spell ingwish teecher  and now I are won

[PRR]

> 15 x 100 miliamps (relay draw) = 1.5amps

a TO-220 can dissipate around 2W in free air, so you divide 2W by the square of the current

> 2w  divide by  1.5 x 1.5   or do you mean square root of 1.5 Huh
> a) 2w divide by 2.25   or       = 0.88888888   max rds
> b) 2w divide by 1.22             = 1.63934426   max rds
> am i lost?

What he said. Your "a)". 0.9 Ohm max Rdss.

>> look at the curves in the datasheet for Rds versus Vgs

I do not see that curve in the datasheet you posted.

But look on the front page. Rdss is usually printed BIG. You can't quite trust the front-page number for all conditions. However the boast for STP40NF10L is 0.033 Ohms. This is "way better than" the 0.9 Ohms you need. Even if you must derate for heat and low gate drive, it is on-its-face plenty good.

If it is way better than you need, the next question is "can you afford a too-good part?" In many cases, the cost is trivial. This is a 2-buck part. I would not spend an hour to find a 1-buck part for a one-off. (When you build millions, you will hunt and negotiate for days for a part-penny difference.)

Your Fig 4 shows current ability starting at 30 Amps and going over 100A (tho Fig 2 shows >40A pretty restricted). A 40A part carrying 1.5 Amps is, to use my analogy of the week, a porch rated for 40 sleeping dogs but only carrying a dog and a cat. It's plenty ample.

Fig 7 shows Rdss. It is only for one, high, Vgs. But the number is (for <10A) 22 MILLIohms, 0.022 Ohms. This is less than the 0.9 Ohm max computed above. It will for-sure work if Vgs is 10V.

However you have that runty 5V supply. Go back to Fig 4. 5V Vgs with 1.5 Amps flowing is.... hmmmmm.... a hard squint between zero and 30A. It is also VERY low on the Vds scale. Say the 5Vgs curve passes 1.5A at 0.3Vds. That says your power dissipation: 0.3V*1.5A= 0.45 Watts. 2W in a TO220 is pretty hot, I like 1W or less, 0.45W is plenty cool.

[njkmonty]

[R.G.]

You're on the right track, njkmonty. You're also getting a very valuable lesson in reading datasheets and making educated guesses about what they mean, which is in most cases necessary.

EE's use the old saw about statistics to refer to datasheets: lies, d@mned lies, and then datasheets. Manufacturers view datasheets as advertising, and so does the government, in that courts have held for plaintiffs suing for damages because of incorrect statements on them. So the manufacturers are in the position of having to not tell straighforward untruths in datasheets, but making the datasheet so that engineers choose their part instead of a competitor's part. So the exact nature of and method of presenting the absolute truth gets a lot of variation. It was almost a coming-of-age test for an EE back when EEs designed circuits with bought parts to make the EE pick transistors or ICs from datasheets. That's one reason that picking the right data out of a datasheet is so very hard for people who have not both had formal EE training and also had some mentoring in reading the datasheet and trying to pick out the information you need from the information that the manufacturer put there to make the part look good. For experienced EEs, it becomes a game on a par with doing the New York Times crossword. For untrained and inexperienced people, it can be like the first lesson in speaking Mandarin for a native speaker of English or a first lesson English for a native speaker of Mandarin.

The thing about power comes from the two ways to compute power P = I²*R and P = I *V . If you the power - in this case the quasi-limit of a TO-220 - and want to know the resistance, then you juggle the first equation to be R = P/(I²) . If you have only a power and a voltage, you have to juggle those two, by guessing between the lines on plots.

As for which curve is the one with Rds versus Vgs - on this device, there isn't one. Sigh. Some MOSFETs have them and this one don't. On this one, you look at the Rds verus current curve, and note that its remarkably flat, so you're reduced to trying to get the thing turned on. At least you can tell it's good enough for you.

You go look at the curves in figure 4 and 5. Figure 4 is what you want, but blown up so you can see what happens in the first, smallest square. Sigh. Looking at figure 5 says that Vgs = 5V will enhance it to about 30A, so you make the guess that it's probably good enough. Well, I made that guess.

As an aside on thermal capability.  Practically every TO-220 MOSFET you look at is good for 30-150W if properly attached to a really good heat sink, but only has its own metal tab if not attached. In between no heat sink and a finned monster, there is a world of marginal heat sinks that are much better than nothing. You can double the 2W in free air easily by attaching a flat piece of aluminum with 3-4 times the surface are of the exposed tab with a bit of heat sink goo and a bolt.

[njkmonty]

if I understand correctly, Data Sheets are a bit like nutritional advertising on a bag of jelly Beans / Snakes, with a big logo saying...
"100% Fat free" ! ?  You have to learn to see through the bull?

I guess  i was hoping  that every small package component had the exact equivilent in a larger package.
a bit like J201
Smd Size
TO-92 size

and a "xyz201"TO-220  which was like a higher rated bigger brother

[PRR]

Possible typed-too-fast-s.....

> If you the power - in this case the quasi-limit of a TO-220
"If you know the power..."

> Figure 4 ... blown up so you can see what happens in the first, smallest square.
"so you can't see..."
_____________________________________

Step-back point:

> TO-220 .... good for 30-150W if properly attached to a really good heat sink
> You can double the 2W in free air easily by attaching a flat piece...

Yes, but he's talking a 5V 1.5 Amp load. The "useful power" is 7.5 Watts. (Really just hot relay coils, but the side-effect of clacked connections is worth that much.)

We would probably like the "extra" losses to be much less. Maybe they can't be, but maybe they can. Much-less than 7.5 Watts is say 0.75 Watts.

Which is also nicely much-less than an un-aided TO-220 can sweat.

> ways to compute power

Good reminder on how to scramble Volts Amps Ohms and Power.

If we want say 0.75 Watts, at 1.5 Amps, we probably want 0.5 Volts max. Check: 0.5V*1.5A= 0.75 Watts.

Given 0.5 Volts at 1.5 Amps, we want a part with 0.5V/1.5A= 0.333 ohms resistance. Or rather, somewhat less, in case rounding-error and data-sheet fudges work against us as Murphy demands.

So while many-many Watts is perfectly possible in a TO220, in this case we prefer a part-Watt solution. Not just to avoid heatsinking, but to avoid excess waste.

And since many 5A-50A MOSFETs boast under 0.1 Ohm Rds, this is possible.