The LM317 is quite the hot head ...

[EasyGoing1]

Re: The LM317 is quite the hot head ...

Those were marked as discontinued/obsolete around 2011 I think

This one should work, right? I can get 10 for like $8

[mariushm]

Re: The LM317 is quite the hot head ...

Those were marked as discontinued/obsolete around 2011 I think - some guy probably has a few thousand of them old stock (from some project he didn't finish or something like that) and he put them as Amazon listing.

The modern replacement seems to be FQP30N06L which is 11$ for 10pcs : https://www.digikey.com/product-deta...06L-ND/1055122

Add 6-8$ for shipping and you're at the same price but at least you know you get genuine and relatively new parts (these don't stay on shelves for a long time at a distributor like Digikey)

If you can live with only 30v Vds , this one is 0.6$ in 10pcs and will work just fine : https://www.digikey.com/product-deta...893-ND/2606361

I gave you Digikey and the link to a huge selection of mosfets, just use the filtering mechanism to pick what you want. If you're afraid of surface mounted components just filter them out (select through-hole only, there's an option there to the right)
Here's all n-channel mosfets that are through hole , play around with filters and check datasheets like I explained and you'll find plenty that would fit your needs.

[EasyGoing1]

Re: The LM317 is quite the hot head ...

The pmv16xnr is very tiny but the pins are spaced far apart that you can get some prototyping board, put three blobs of solder on some through holes on the prototyping board and then just tin the pins and then heat up the blobs of solder and make contact between the pins and the blobs of solder. It's easy.

These are being marketed as “Arduino Power MOSFET Switches” etc.

They're $16 for 10 on Amazon.

[budm]

Re: The LM317 is quite the hot head ...

1. By looking at the curve for V(GS) = 4V. The vertical axis is Drain current; the horizontal axis is V(DS). With a V(GS) of 4V, the curve doesn't go higher that ~1.5A.

2. Those curves are for junction temperature, not ambient temperature. Elsewhere in the datasheet, it specifies a junction to case thermal resistance. That means that if the device is dissipating power, the junction temperature - the temperature of the silicon die - will always be higher than the temperature of the case. On top of that thermal resistance there is the thermal resistance between the case and the heatsink, and the thermal resistances from the heatsink to ambient air. All those thermal resistances together mean that if the silicon die is dissipating 30W, that silicon die has to be hotter than the room ambient temperature (unless the device is being cooled by a refrigerated liquid).

MOSFET makers publish that junction-case thermal resistance. Insulator manufacturers specify the thermal resistance of their insulators. Heatsink manufacturers publish the thermal resistance vs. airflow curves for their dissipators. Fan manufacturers publish the airflow vs. back pressure curves for their fans. Actual airflow can be measured with anemometers. All in all, design engineers can calculate the junction temperature rise for their power devices to choose heatsinks and fans that can keep the junction temperature within a desired range (if the ambient is 40C instead of 25C, the junction temperature will be ~15C higher).

3. You would do better to go to device manufacturers' websites, pick several possibilities, and see what is available. Also, I'd go with Digi-Key or Newark over Amazon.

Right now, EG1, I think your focus needs to be on what different types of devices do, how they can be used, and how they operate. I've noticed that even university engineering programs (e.g. San Jose State) give those fundamentals short shrift. Without that foundation, you'll have trouble recognizing and understanding the importance and meaning of information such as the thermal resistance or datasheet curves. I'm not being critical. You're probably wa-a-a-a-aaaaayyyyy over my head when it comes to network and computer stuff; OTOH, power electronics has been my career field for over 35 years, and much of what I've learned was by doing, hands on, over that time.

+10, especially 'I think your focus needs to be on what different types of devices do, how they can be used, and how they operate."
You cannot skip the basic and there is no stop in learning, and hands on is the great learning tool.

[PeteS in CA]

Re: The LM317 is quite the hot head ...

1. How did you determine that at 4 Vgs, there will never be a Drain current larger than 1.5 amps?

2. 25C is 77F ... and if it can sustain 10 amps and remain at that temperature, then I can’t see why it would be improbable for it to sustain 3 amps at room temperature, unless the amount of current it can handle significantly decreases with each degree of temperature increase, but in most cases, the temperature that the device will be in (in this case anyways) wont be much higher than 77F.



3. Do you have any in mind that I can get at Amazon?

1. By looking at the curve for V(GS) = 4V. The vertical axis is Drain current; the horizontal axis is V(DS). With a V(GS) of 4V, the curve doesn't go higher that ~1.5A.

2. Those curves are for junction temperature, not ambient temperature. Elsewhere in the datasheet, it specifies a junction to case thermal resistance. That means that if the device is dissipating power, the junction temperature - the temperature of the silicon die - will always be higher than the temperature of the case. On top of that thermal resistance there is the thermal resistance between the case and the heatsink, and the thermal resistances from the heatsink to ambient air. All those thermal resistances together mean that if the silicon die is dissipating 30W, that silicon die has to be hotter than the room ambient temperature (unless the device is being cooled by a refrigerated liquid).

MOSFET makers publish that junction-case thermal resistance. Insulator manufacturers specify the thermal resistance of their insulators. Heatsink manufacturers publish the thermal resistance vs. airflow curves for their dissipators. Fan manufacturers publish the airflow vs. back pressure curves for their fans. Actual airflow can be measured with anemometers. All in all, design engineers can calculate the junction temperature rise for their power devices to choose heatsinks and fans that can keep the junction temperature within a desired range (if the ambient is 40C instead of 25C, the junction temperature will be ~15C higher).

3. You would do better to go to device manufacturers' websites, pick several possibilities, and see what is available. Also, I'd go with Digi-Key or Newark over Amazon.

Right now, EG1, I think your focus needs to be on what different types of devices do, how they can be used, and how they operate. I've noticed that even university engineering programs (e.g. San Jose State) give those fundamentals short shrift. Without that foundation, you'll have trouble recognizing and understanding the importance and meaning of information such as the thermal resistance or datasheet curves. I'm not being critical. You're probably wa-a-a-a-aaaaayyyyy over my head when it comes to network and computer stuff; OTOH, power electronics has been my career field for over 35 years, and much of what I've learned was by doing, hands on, over that time.

[mariushm]

Re: The LM317 is quite the hot head ...

The pmv16xnr is very tiny but the pins are spaced far apart that you can get some prototyping board, put three blobs of solder on some through holes on the prototyping board and then just tin the pins and then heat up the blobs of solder and make contact between the pins and the blobs of solder. It's easy.

The AON7406 is scarier but again it's easy to solder. You only have 3 wires going to it, gate (1 pad) , source (3pads) and drain (the big metal stuff) ... so you can put the three wires flat over the pads and solder the wires easily to the pads.


As for lm317 alternatives... There's literally tens of thousands of chips, linear or of switching type

60.000+ linear regulators : https://www.digikey.com/products/en/...ors-linear/699

26.000+ switching regulators : https://www.digikey.com/products/en/...regulators/739

For linear regulators,
pick maximum current,
pick dropout voltage (in order to provide smooth output voltage, input voltage must be higher or equal to dropout voltage + output voltage),
pick minimum input voltage the regulator must be able to accept (for example you filter out regulators that can only work with up to 6v, if you want to adjust voltage up to 10-11v)
pick package to account for how easy it would be to keep them cool (regulators dissipate the difference between input voltage and output voltage as heat - some regulators say they can do so much current, but in reality that's only if the input voltage is very close to output voltage, for example max 6v in , 5v out at 3a for a total of 3 watts of heat dissipated)

[EasyGoing1]

Re: The LM317 is quite the hot head ...

https://www.digikey.com/short/3d24j0[/url]

For example, see PMV16XNR: https://www.digikey.com/product-deta...0-1-ND/5395602

here's another good example: AON7406 : https://www.digikey.com/product-deta...7-1-ND/1856080

... Package looks scary but it would be easy to solder.

Yeah it does ... surface mount stuff ... haven't messed with components like that yet...

Do you know if there is an alternative to the LM317 that is just as cheap and easy to use? Just curious about what else might be out there in terms of programmable / changeable voltage regulators.

[redwire]

Re: The LM317 is quite the hot head ...

This guy made an animated javascript circuit simulator, seems like some fun: http://www.falstad.com/circuit/circuitjs.html
There is a mosfet to play with, but not a logic-level one.

[mariushm]

Re: The LM317 is quite the hot head ...

Digikey has a huge selection of mosfets, but maybe shipping is more expensive.

Digikey can also filter by Vgs and Rdson and average current all those specs that you can play around with

Here's ALL n-channel mosfets filtered by lowest price if you order 10 of them : https://www.digikey.com/short/3d24j0

The price for 1 may be a bit higher (you can see it on the mosfet details page), I selected 10 pcs minimum for you so that you wouldn't get in the list of results mosfets that can only be bought in full rolls of 1000-2500 pcs.

Play with Vgs(th) (Max) @ Id (ex select all <= 3.5v for example) and then select the maximum continuous current you want (so that you'd filter out parts that can only do less current than what your project will do) and then pay attention to drive to source voltage max drain / source voltage and then look in datasheets to see the actual performance for your Vgs of 3.3v (and try to get mosfets with lower gate capacitance but don't stress yourself too much)

For reasonable currents, i see something as cheap as 30-40 cents and easy to use.

For example, see PMV16XNR: https://www.digikey.com/product-deta...0-1-ND/5395602

It can do at least 4A continuously with very low Vgs (1v and up) , up to almost 7A with higher Vgs (let's say 2.5v and up) as long as there's some surface around the chip to act as heatsink

here's another good example: AON7406 : https://www.digikey.com/product-deta...7-1-ND/1856080

max 2.4v Vgs (at 3v it can do 10A or more with Vds above 1v) and it's also cheap, less than 0.5$ if you buy 1. Package looks scary but it would be easy to solder.

[EasyGoing1]

Re: The LM317 is quite the hot head ...

Corrected the url linking, you had a weird " character there.

Yeah, that mosfet should work fine with 3.3v though the gate capacitance is a bit on the high side (it should work fine i think).

I picked that one because pf the four logic level power mosfets that Mouser had, it was the least expensive at $1.20 or so. The most expensive one was like $11

[EasyGoing1]

Re: The LM317 is quite the hot head ...

Open your datasheet:
LTC1154 datasheet , especially at the bottom of page 6 where you see the block diagram, which shows you how the IC has a gate charge and discharge control logic and fast/slow gate charge mechanisms.

Interesting, thank you for taking the time to explain that to me.

Mike

[mariushm]

Re: The LM317 is quite the hot head ...

What about something like this? FDP7030 N-Channel Logic Level Power MOSFET?

Corrected the url linking, you had a weird " character there.

Yeah, that mosfet should work fine with 3.3v though the gate capacitance is a bit on the high side (it should work fine i think).

[budm]

Re: The LM317 is quite the hot head ...

BTW, if you look at that graph, at 10A of Id and Vds of 3V, that is 30W of power dispensation, so you will have temperature rise above the room temperature, that rise is based on thermal resistance of the device rating in xxC per Watt. You also do not want the junction temp to be above the limit as shown in the spec.

[mariushm]

Re: The LM317 is quite the hot head ...

Open your datasheet:



Determine your Vgs value.

Page 2, in the Electrical Characteristics (note above the table it says Tj = 25C unless otherwise specified) :

Vgs (to) GATE THRESHOLD VOLTAGE : min 2.0v , typ 3.0v , max 4.0v

This tells you that typically , the mosfets will turn on fully at 3.0v but they don't guarantee ALL will turn on at 3.0v or that the ones that turn on at 3.0v will turn on at 3.0v under all circumstances.
Basically, if you buy 100 of them, some won't turn on fully with 3.0v, but all should turn on fully with a voltage above 4.0v (the maximum value in that table).

Also, note again that all values are for Tj = 25C. The mosfet has some internal resistance so as soon as there's some current flow through it, it will slowly warm up.

The mosfet has some mass to it so it won't heat up instantly, it will slowly warm up and some heat will be radiated to the air around it, and some heat will also be radiated through the leads in the circuit board ... but basically you know that if you started the circuit with the circuit board and the mosfet at a temperature of 25 C, after a while that mosfet won't be at 25 C

Now go down in the datasheet and look at Fig.5. Typical output characteristics. on page 3 :



Here's how you read this chart :
On vertical (left side) you have the maximum current that will flow through the mosfet (in amps)
On horizontal (bottom) you have the Vds (voltage between drain and source pins)
Each of those lines (curves, whatever) tells you how the mosfet behaves for a specific Vgs value. Note that they don't even bother to use the typical Vgs of 3.0v, they use the 4v value as minimum because that's the one guaranteed to work for any transistor at the default 25c temperature (see chart corner)

So with a Vgs of only 4v, you're looking at the most bottom line: regardless of how much the Vds will be, the mosfet will only allow less than 2A of current to flow through it

With a Vgs of 5v, you're looking at the top line from the set of 6 lines. So you can see that with a Vds of only 1v, the mosfet may only allow up to about 5A to flow through it, but if you design the circuit properly to allow at least 3v between Drain and Source pins, then the mosfet will allow up to 10A. For anything above Vds of 3v, you can see that mosfet will basically "stabilize" itself to somewhere below 10.5A of current flow.
If you want more than 10.5A you need to have a Vgs higher than 5v - for the maximum of 10v (top line) you can see the mosfet scales almost linearly with the Vds voltage : it can do over 12A with a Vds of 3v and almost 20A with a Vds of 5v

Now I have to note again that values are valid only for 25 C : think of the chart being made with the mosfet at 25c , then power the circuit for 1-2 seconds (so that the mosfet won't have time to heat) and literally take a snapshot of the values and power off the circuit. As the mosfet and the circuit board will heat up, those lines will drift a bit (but overall the numbers should still be "in the ballpark"

You can scroll down in the datasheet some more and see how the mosfet will behave with heat (on page 4, Figure 9 and Figure 10 in the picture below) :



So you can see that the warmer the mosfet will be, the easier it will turn on fully.
It's important to know these, because for example you may say "oh, i'll just use 4v and then I'll know for sure the mosfet will turn on fully" but what if your project/design will be outside in a winter when the ambient temperature will be -20c .. 0c ? You can see that the 4v max is only guaranteed at 25 C - at 0C you already need about 4.2v for the mosfet to turn on fully.

The other way is also important .. for example you may say "I'll use Vgs of 4v because then according to the chart above I'll know for sure the current between drain and source will be limited to less than 2A so basically my mosfet will limit the current through the LEDs and I won't need a led driver or a resistor to limit the current"

BUT as the mosfet heats up, you can imagine it behaving like having 5v on the gate, it will basically let more than 2A flow between drain and source pins so in the theoretical example above you'd damage your leds.
Also you can see in figure 9 how with Vgs of 4v, it barely lets 2A at Tj of 25c but at 150c it almost crosses the 5A threshold.

Something else you can see from the datasheet, scroll up to page 3 and look at Figure 6 Typical on-state resistance :



You can see that with a Vgs of 4v even with as little as 1A of current between drain and source, the internal resistance will be somewhere around 0.4 ohms
So you have P = IxIxR = 1x1x0.4 = 0.4 watts of heat dissipated in the mosfet.
But if you use a Vgs of 5v then even with as much as 10A of current between drain and source, the Rds (on) resistance will be below 0.3 ohm and at 1A it will be below 0.25 ohm ... basically with higher Vgs, the mosfet will heat less.

So basically, this mosfet is really designed for a Vgs of around 5v, even though it says it typically can turn on at 3v... if you want the advertised currents and rds(on), you drive it with at least 5v.
To drive a mosfet directly with 3.3v, you would normally look for one mosfet that has less than 3.3v on the MAX column for the Vgs threshold.

The input capacitance is important as well.
Microcontrollers can only output a limited amount of energy per pin, and then a limited amount of energy over all the pins - for example a PIC chip may output up to 15-25mA per pin but only 100-150mA through all of its pins.

If the gate capacitance is too high, in those microseconds when you want to turn on the mosfet the gate capacitance will suck a huge amount of current from the microcontroller and in some rare cases that could cause glitches in the microcontroller or even resets. With such mosfets you either accept a slower switch time by limiting the charge with a resistor, or you use a mosfet driver between your micro and the mosfet itself.

Also high gate capacitance means the mosfet will discharge slower or not discharge at all when you stop sending voltage to the gate pin .. that's why you typically put a resistor between the gate and source pin, so that the capacitance between gate and source will discharge into the resistor.
The resistor will quickly discharge the gate and bring the voltage down to below the threshold required to keep drain-source link "active".

But the resistor will then be also between your output pin and ground, which means while you keep the mosfet on you'll also waste some power in the resistor that's on the gate-source ... ex with a 1kOhm resistor you have V=ixR .. 5v = i x 1000 = 5mA of energy being wasted on a resistor between mosfet gate and ground.
Mosfet drivers would have circuitry inside to discharge the gate and then disconnect that discharge mechanism while the mosfet is kept ON.

For example, see LTC1154 datasheet , especially at the bottom of page 6 where you see the block diagram, which shows you how the IC has a gate charge and discharge control logic and fast/slow gate charge mechanisms.

[EasyGoing1]

Re: The LM317 is quite the hot head ...

Now, if your available G-S voltage is just 3.3V, you're going to need a logic level MOSFET, which may turn with 3.3V

What about something like this? FDP7030 N-Channel Logic Level Power MOSFET?

[EasyGoing1]

Re: The LM317 is quite the hot head ...

You need to understand the device a bit and think about what that graph tells you. And remember that ohm's law still applies. Each curve on that graph tells you what a typical device will do with a particular Gate-source voltage applied with a junction temperature of 25C. If you apply 4V G-S, the Drain current will never be greater than about 1.5A, no matter how high your Drain voltage is. The 10A, 3V D-S point you picked is with 5V G-S, if the silicon inside that nice TO-247 package is 25C. Well with 3V across the D-S and 10A Drain current, that silicon is going to be much warmer than that, even with a massive heatsink and a lot of air.

How did you determine that at 4 Vgs, there will never be a Drain current larger than 1.5 amps?

25C is 77F ... and if it can sustain 10 amps and remain at that temperature, then I can't see why it would be improbable for it to sustain 3 amps at room temperature, unless the amount of current it can handle significantly decreases with each degree of temperature increase, but in most cases, the temperature that the device will be in (in this case anyways) wont be much higher than 77F.

Now, if your available G-S voltage is just 3.3V, you're going to need a logic level MOSFET, which may turn with 3.3V

Do you have any in mind that I can get at Amazon?

[EasyGoing1]

Re: The LM317 is quite the hot head ...

Well it is never too old to learn, you are so lucky that you have internet to get all kind of information at your finger tips right away I did not have that over 40 years ago.

I sill have my original Electronics textbook by the way ... it's in storage, but I took a look at it a few years ago and was quite impressed with how much I apparently knew at one time... lol

[EasyGoing1]

Re: The LM317 is quite the hot head ...

Well it is never too old to learn, you are so lucky that you have internet to get all kind of information at your finger tips right away I did not have that over 40 years ago.

My first exposure to the Internet was my first job as a paid computer tech. I worked at Phillips Laboratories which could be considered the “North Campus” of Edwards AFB. Its the area where they did all of the Apollo rocket tests and a lot of astronaut training etc. Very VERY interesting facility complete with underground command centers and cable tunnels that went for DAYS AND DAYS ... data acquisition systems that were interconnected using these tunnels. In some cases, a tunnel would be far too long to see the other side, and barely wide enough for one person because on each side of the tunnel, there were sometimes 5 cable trays running the length of the tunnel, each COMPLETELY PACKED with cable of all kinds. I can't imagine how much weight those trays had to sustain ... not that it mattered since the trays were welded from some very rigid metal. They weren't going anywhere anytime soon.

But I digress...

I used the Internet for the first time on that job back when Mozilla / Yahoo was the only GUI browser in town. Google was non-existent, and our LANs were interconnected using RG-59 coax or “Thinnet” ... we had special cables that connected the computers using something called a Vampire Tap which would leave a port in the wall that these special cables would use. The coax would be connected through the tap in the wall and when you plugged one of these jumpers into the tap, it was wired in such a way that it would break the connection in the wall, but then feed it through the jumper, leaving the entire run still connected in series. Not sure how much you know about that stuff, but at each end of one of those runs (which could include I believe up to 255 computers), the coax needed 50 ohm terminators and each computer was slaved off the coax in series (like the way they wire telephone in a house). They ran Ethernet when I was there because they had just upgraded from Token Ring. They also used TCP/IP ... but thats right when CAT-5 started making a name for itself as the better option - the government seemed to always be years behind everyone else when it came to adopting new technologies in areas that didn't require clearance, and especially when they had to provide infrastructure to contractors fulfilling contracts.



There were 7 of us techs supporting the whole area, and we had to rotate help desk duties (cause it was boring) and I would often be at the helpdesk and in between calls, researching UFO's, or looking at porn (no filtering back then nor any kind of fancy firewall stuff like they have now)...

Amazing how much the Internet has exploded in the last few years.

[PeteS in CA]

Re: The LM317 is quite the hot head ...

You need to understand the device a bit and think about what that graph tells you. And remember that ohm's law still applies. Each curve on that graph tells you what a typical device will do with a particular Gate-source voltage applied with a junction temperature of 25C. If you apply 4V G-S, the Drain current will never be greater than about 1.5A, no matter how high your Drain voltage is. The 10A, 3V D-S point you picked is with 5V G-S, if the silicon inside that nice TO-247 package is 25C. Well with 3V across the D-S and 10A Drain current, that silicon is going to be much warmer than that, even with a massive heatsink and a lot of air.

Now, if your available G-S voltage is just 3.3V, you're going to need a logic level MOSFET, which may turn with 3.3V

[budm]

Re: The LM317 is quite the hot head ...

Well it is never too old to learn, you are so lucky that you have internet to get all kind of information at your finger tips right away I did not have that over 40 years ago.