dual coil relay coil polarity

[VEGETA]

I however checked its table and showed Vgs th to be max of 3v so I said it could work

Max Vgs th is defined by output current (typically very small, here 1mA).
Practically you need max Vgs th about 1V below voltage you will be using to drive it.

so you mean i need 2.3v if i use 3.3v to drive the gate? my understanding is the datasheet puts min and max Vgs_th, more like best and worst case, meaning the worst case it will need 3v to be fully activated which means my 3.3v should be totally fine.

is this correct? what do you suggest instead?

[Psi]

You maybe able to turn the fet on, but the Vgs also effects Rds which affects current.
So if you are too close to the Vgs your fet will be on but the Rds will be too high to flow much current.
Does that make more sense?

[VEGETA]

You maybe able to turn the fet on, but the Vgs also effects Rds which affects current.
So if you are too close to the Vgs your fet will be on but the Rds will be too high to flow much current.
Does that make more sense?

well, that is known but as you mentioned it does not clearly show the behavior in the graphs. 7000 shows like 4.5v Vgs having something like 2.8 ohms of Rdson, thus for 3.3v maybe i can say 4 ohms it will be the resistance.

the coils are 100 ohms of resistance for the relay mentioned, which means a maximum of say 5 ohms from the mosfet won't do much effect. what do you think of this on this application?

still, it would be nice to have a better performance, so maybe reverting back to npn with some resistor at the gate is better. 2n2222 is cheap and with 220 ohms at its gate it will be fine. esp32 can deliver such very low current as well.

do you recommend going with this npn solution or mosfet?

the only concern with npn is 0.7v drop which makes coil voltage 4.3 instead of 5v. PRL6-5V-DC-1A datasheet says minimum voltage is 4v, which i think is good margin.

[Peabody]

I am just a hobbyist, but it seems to me that what matters here is the current the mosfet can pass at the voltages involved.  The datasheet will have a diagram of Vg/s versus drain current.  However, that curve is probably "typical", not worst case.

It seems that EEs often use the lowest voltage at which behavior is characterized as the lowest voltage for which the part can be used (4.5V in this case).  It's not clear that manufacturers intend for that to be the case.  But I guess experience suggests that's a good rule of thumb since behavior at lower voltages is not specifically defined.  Maybe one of the experts here can clarify that.

But for a hobby project, if I found a convenient (TO-92) mosfet like the 2N7000 or BS170 that convincingly switches the relay, I would use it.

[PGPG]

my understanding is the datasheet puts min and max Vgs_th, more like best and worst case, meaning the worst case it will need 3v to be fully activated

Where from have you taken 'fully activated'? This term is even hardly possible to define. I think fully activated = have minimum Rds is when Vgs is the highest that is allowed (so 20V in this case).

See bottom of page 2 in:
https://www.onsemi.com/pdf/datasheet/nds7002a-d.pdf
It is hardly to believe that you see 3V but don't see Id=1mA in the same line.

In worst case it will need 3V to drive 1mA what is certainly not enough for your relay.

[VEGETA]

my understanding is the datasheet puts min and max Vgs_th, more like best and worst case, meaning the worst case it will need 3v to be fully activated

Where from have you taken 'fully activated'? This term is even hardly possible to define. I think fully activated = have minimum Rds is when Vgs is the highest that is allowed (so 20V in this case).

See bottom of page 2 in:
https://www.onsemi.com/pdf/datasheet/nds7002a-d.pdf
It is hardly to believe that you see 3V but don't see Id=1mA in the same line.

In worst case it will need 3V to drive 1mA what is certainly not enough for your relay.

I mean that it has low Rds_on when provided with the VGs I send. Yes, maybe you can get lower resistance with higher Vgs but if I send 3.3v and the mosfet gets reasonably low Rdson then it is good for me. Yes, "fully activated" is not an objective term, I use it casually. Sorry for that.

I did not understand the 1 mA and I did not see anything in the graphs to indicate the behavior at 3.3v gate voltage.

figure 2 of the datasheet you linked has Vgs of 4v having Rdson of 2.5 ohms at ~ 250 mA, so estimating it i get 3.3v will have resistance of 2.5 ohms at only 200mA of drain current.

so i see that it does not allow current to be high at low gate voltage like 3.3v.

is my understanding here correct? what do you suggest?

[PGPG]

figure 2 of the datasheet you linked has Vgs of 4v having Rdson of 2.5 ohms at ~ 250 mA, so estimating it i get 3.3v will have resistance of 2.5 ohms at only 200mA of drain current.

I don't understand how you are trying to estimate anything.
If you see (at figure 2) the distance between line for 4.5V and for 4V you can expect 3.3V line being totally out of this chart.

Than this chart is TYPICAL.
As typical Vgs th is 2.1V and max is 3V to get characteristics for device having max Vgs th you should read this chart as each Vgs value would be 0.9V higher than written at figure.

When you design anything you should always think of worst case. Of course if you want to do only one device you can measure 10 transistors to select two with lowest Vgs th and they will be working for you.

[VEGETA]

figure 2 of the datasheet you linked has Vgs of 4v having Rdson of 2.5 ohms at ~ 250 mA, so estimating it i get 3.3v will have resistance of 2.5 ohms at only 200mA of drain current.

I don't understand how you are trying to estimate anything.
If you see (at figure 2) the distance between line for 4.5V and for 4V you can expect 3.3V line being totally out of this chart.

Than this chart is TYPICAL.
As typical Vgs th is 2.1V and max is 3V to get characteristics for device having max Vgs th you should read this chart as each Vgs value would be 0.9V higher than written at figure.

you are correct, i did not estimate 3.3v correctly. somehow i saw 4v and did not focus on the next line which is 4.5v. still what confuses me is that there is no 3.3v curve or any reliable info besides the VGs table row showing 3v max.

but even if taking figure 2 with 3.3v it will be allowing very little current, mostly not reliable at all.

When you design anything you should always think of worst case. Of course if you want to do only one device you can measure 10 transistors to select two with lowest Vgs th and they will be working for you.

the issue here is i want small through hole package, generic cheap part, and available from different manufacturers. this kinda limits the choices.

i designed a device which has different types of mosfets for different power functions, they performed fantastically because selection wasn't limited by this, but rather smt and soldered by pcb assembler.

___

thus, 7000 and 7002 won't do the job. my previous suggestions seemed not very suitable too.

do you have other n-mosfet suggestions which comply with the requirements? or reverting back to 2n2222 with 1k at base and 100k pulldown is the choice?

[ledtester]

Will the jellybean AO3400A work?

EEVblog 1736 - TOP 5 Jellybean MOSFET's -- EEVBlog
https://youtu.be/eI5epLNfXf0?t=14m47s

#2548 AO3400 Tiny Powerful FET -- IMSAI Guy
https://youtu.be/yJj8vcUpsI8

[PGPG]

do you have other n-mosfet suggestions which comply with the requirements?

What about DMG1012 ?

or reverting back to 2n2222 with 1k at base and 100k pulldown is the choice?

'Since always' I am using BC847 (BC547 in the last century).
You can consider BC847S (= 2x BC847).

[VEGETA]

AO3400A
DMG1012

would be good if i decided on smt, right now i am still trying to have through hole. looks like true logic level ones are made for smt, very few are through hole and expensive.

BC847

that is an npn, and yes it would be a good choice. my npn was 2n2222 and 3904, but i am trying to check for a mosfet. if i could not, then npn it is.

[PGPG]

but i am trying to check for a mosfet.

What mosfet gives you beyond the uncertainty of Vgs th ?

With mosfet it is rationale to use 2 resistors while with npn one is practically enough.
Searching for low Vgs th and low Rds you also get a small Vds max in the set (DMG1012 has 20V).
While Vce=50V seems being a standard among small signal npn transistors.

If you need for example 20mA Ic than loosing additionally say 0.5mA for Ib seems not being a great loss.

[Peabody]

I don't think you will find a TO-92 mosfet that's logic level for 3.3V, at least using the worst case specs.  But the 2N7000 and BS170 are widely used in hobby 3.3V circuits because they typically work ok in low current.  It turns out that the "typical" datasheet specs are indeed typical.

What is the incoming power supply that is regulated down to 3.3V in your circuit.  Would it be possible to feed that higher voltage to the relay coils instead of 3.3V?  It doesn't matter if it isn't regulated, and the higher voltage should remove any doubt.  Edit:  No, I guess not.  Never mind.

My memory is that the 2N7000 and BS170 do not have the same pinout.

I wonder if it would be possible to gang two of these mosfets in parallel - in the same through-holes.  That would in theory give you more current, and possibly let you eliminate the series resistors, plus doubling the odds that at least one of the mosfets switches at the typical voltage.

But I suspect a single 2N7000 or BS170 would in practice work fine.  You might have difficulty finding one that doesn't work.

Edit:  The P-channel TP2104 mosfet comes in TO-92, and it has a -2V maximum threshold voltage.  You could  put it above the coil to make a high-side switch.  However, its Rds is still characterized only down to -4.5V, and it's 10 ohms at that voltage.  Maybe the experts could opine on that option.

[PGPG]

You could  put it above the coil to make a high-side switch.

Have you read the first post in this thread?

[Peabody]

Yes.  Sorry.  I should have left well enough alone.  Besides, it's a bad idea.

[VEGETA]

but i am trying to check for a mosfet.

What mosfet gives you beyond the uncertainty of Vgs th ?

With mosfet it is rationale to use 2 resistors while with npn one is practically enough.
Searching for low Vgs th and low Rds you also get a small Vds max in the set (DMG1012 has 20V).
While Vce=50V seems being a standard among small signal npn transistors.

If you need for example 20mA Ic than loosing additionally say 0.5mA for Ib seems not being a great loss.

the coil is 100 ohms, and at 5v it will be 50mA current. npn like 2n2222 will be more than enough for this load and with 220 ohms base resistor it will be 3.3v-0.7 = 2.6v /220 = 11.8 mA base current which is safe for esp32 output pins especially at pulse not continuous.

npn will reduce the voltage across the coil from 5v to 4.3v but looks like it is ok. for mosfets usage i was aiming to never have that big drop.

I don't think you will find a TO-92 mosfet that's logic level for 3.3V, at least using the worst case specs.

looks like it is the case.

But the 2N7000 and BS170 are widely used in hobby 3.3V circuits because they typically work ok in low current.  It turns out that the "typical" datasheet specs are indeed typical.

yes, but looks like as people suggested here, it should not be used when reliability is needed.

Edit:  The P-channel TP2104 mosfet comes in TO-92, and it has a -2V maximum threshold voltage.  You could  put it above the coil to make a high-side switch.  However, its Rds is still characterized only down to -4.5V, and it's 10 ohms at that voltage.  Maybe the experts could opine on that option.

the pulse must be positive so pmos won't do by itself.

[Peabody]

npn will reduce the voltage across the coil from 5v to 4.3v but looks like it is ok. for mosfets usage i was aiming to never have that big drop.

The base/emitter voltage will be 0.7V,  but the collector/emitter saturation voltage will be about 100mV or possibly less.  The collector voltage will be lower than the base voltage.  I think close to the full 5V will appear across the coil.  So since the base current will be very short-lived, there's really not much benefit to using a mosfet here unless you're switching the relay very often.  That's assuming it is actually a latching relay.  I doubt the difference in battery life, if any, will be noticeable.

Edit: I'm assuming you would place the transistor below the coil, as shown in schematics posted earlier in the thread.

[VEGETA]

npn will reduce the voltage across the coil from 5v to 4.3v but looks like it is ok. for mosfets usage i was aiming to never have that big drop.

The base/emitter voltage will be 0.7V,  but the collector/emitter saturation voltage will be about 100mV or possibly less.  The collector voltage will be lower than the base voltage.  I think close to the full 5V will appear across the coil.  So since the base current will be very short-lived, there's really not much benefit to using a mosfet here unless you're switching the relay very often.  That's assuming it is actually a latching relay.  I doubt the difference in battery life, if any, will be noticeable.

Edit: I'm assuming you would place the transistor below the coil, as shown in schematics posted earlier in the thread.

ok, i will use npn 2222 version (i think other generic ones will work). will build a small prototype board.

yes, it is as schematic, last element and used as a switch.

is 220 ohms ok or should it be increased to 470? i mean for esp32 to operate it without problem or the transistor heating up.

[PGPG]

is 220 ohms ok or should it be increased to 470? i mean for esp32 to operate it without problem or the transistor heating up.

Parameters for transistor in saturation are typically specified for Ic/Ib = 10 or 20.

There are small transistors having higher Ic max current. They should have smaller Vce sat.
In SMD it is BC807 versus BC847.
I think 2N2222 is not the best choice if you want small Vce sat for Ic=50mA but since 30 years I'm not using THT transistors so just don't know which one to chose. Spend some time looking through parameters.