Bought Some BS170 Mosfets for 3.3V GPIO Switching... That Was A Mistake lol

[TomKatt]

I usually get away with generic 2N222's for things like relay control with my microcontroller projects but needed something fast for a 3.3V project...  Did some Googling and saw a few references to the BS170 mosfets, so I grabbed a few.  No workie lol.  Despite low Vgs specs, they don't fully saturate at 3.3V.  I did have luck with the IRLB8721 mosfets - although being much bigger 30V / 62A TO-220 components, they seem to work fine for driving low power opto leds.  I know there are better options, but they all seem to be smd and I haven't quite escaped through hole world yet.

I know there's a million threads on logic level mosfets, and here's one more telling you what not to do lol.

[nonius_]

saw a few references to the BS170 mosfets, so I grabbed a few.  No workie lol.

Where did you order them from? On a local forum there recently was a discussion about those BS170's being fake, with new legs being welded on... Basically the entire batch of 20 pcs was useless. Unless bought from a reputable supplier, I'd question whether they actually were BS170s....

https://www-circuitsonline-net.translate.goog/forum/view/172349/2?_x_tr_sl=auto&_x_tr_tl=en&_x_tr_hl=en-US&_x_tr_pto=wapp

Gate threshold voltage for genuine BS170's is 3V max, so I expect it should've worked....

[TomKatt]

Where did you order them from? On a local forum there recently was a discussion about those BS170's being fake, with new legs being welded on... Basically the entire batch of 20 pcs was useless. Unless bought from a reputable supplier, I'd question whether they actually were BS170s....

Gate threshold voltage for genuine BS170's is 3V max, so I expect it should've worked....

I get all my parts from reputable sources - these BS170's came from DigiKey, so I'm sure they are genuine.  I've been burned too many times on cheap parts and I never buy from eBay or Amazon.

And I agree - by the datasheet they should have been fine.  But my circuit that pulls 100ma direct was only getting about 60ma with the BS170.  Swapped in the IRLB8721 and it was fine, pulling 96ma (I'm guessing accounting for slight fet resistance).  Went back with a different BS170 and same results.  Oh well.

[thm_w]

microchip has a few, eg TN0604N3-G but they are not cheap.
I'm sure if you look around on LCSC you'll find something better and cheaper in TO92.

Is the 1-5 ohm rds on going to cause enough of a drop in your circuit to reduce the current?

[TomKatt]

Is the 1-5 ohm rds on going to cause enough of a drop in your circuit to reduce the current?

I'm not sure - have to test or do some calculations...  The IRLB8721 Rds is max 16 milli-ohms at 4.5V, so that's an order of magnitude greater and might influences things given my 3.3V supply.  I'm driving a fancy opto-isolator that has an led with built in current limiting resistor rated for 5V max and suggested 3V operating and I'm running 3.3V through it.  The datasheet says 100ma at 3V and that's pretty much exactly what I observe when I run it directly off the bench power supply for testing.

[PCB.Wiz]

Did some Googling and saw a few references to the BS170 mosfets, so I grabbed a few.  No workie lol. 

The BS170 is ancient, and there is another thing to watch these days, not 'fakes' per se, but Asian die upgrades, as they simply do not bother making any softer/older parts. Customers will order BS170, so they label their cheapest/smallest die with that number.

Here we compare two parts labeled BS170, the original OnSemi one, and another Digikey one, from EVVO

These die-upgrade variations do mean it can be hard to be sure even a part-code that is 'not fake', will work the same.

Another issue often seen, is spec-laziness, where typical curves simply are copied, or fail to agree.
If we look at the last BS170FTA_LCSC_HXY, the left graph shows 500mA Id at 3V, yet the right graph shows 500mA at ~ 2.6V 

[TomKatt]

The BS170 is ancient, and there is another thing to watch these days, not 'fakes' per se, but Asian die upgrades, as they simply do not bother making any softer/older parts. Customers will order BS170, so they label their cheapest/smallest die with that number.

Here we compare two parts labeled BS170, the original OnSemi one, and another Digikey one, from EVVO

These die-upgrade variations do mean it can be hard to be sure a part-code that is not fake, will work the same.

Geez - as if things weren't bad enough and I'm thinking that buying from DigiKey is the safe bet, now I need to vet all the options for a part number... bah lol.

Those are quite different curves there !!!   This may be what happened to me, as I probably sorted by price.  But for cheap stuff, I usually look for a name brand like OnSemi or something.  I'll have tp check my order to see which one I actually ordered...

EDIT - So I checked and these were purchased from Mouser.  The Mouser part # identifies these as oem OnSemi components.  For what that's worth I guess lol.

[thm_w]

Not for OP, but general warning, never buy EVVO parts: https://www.eevblog.com/forum/chat/warning-lm317k-duds-from-digikey/

Other than the Rds on the BS170s look fine, you could verify with a multimeter what the drop across them is and if it matters, and the resulting resistance.

Most interesting TO92 on LCSC I see is this one UTC UK2962L-T9N-B: https://www.lcsc.com/product-detail/C2978930.html
2V Vgs th, 1A Id, <1ohm Rdson. But yeah still nothing compared to modern SOT23 SMD parts.

[Geoff-AU]

If we look at the last BS170FTA_LCSC_HXY, the left graph shows 500mA Id at 3V, yet the right graph shows 500mA at ~ 2.6V 

Yes and those are also *typical* graphs, even onsemi's original datasheet specifies the 250uA gate threshold is *typically* 2.1V but can be as low as 0.8V or as high as 3V. 

So if an in-spec onsemi part can flow as little as 250uA at 3V it's really not suitable for being driven by 3V3 GPIO any way you slice it.

BSS123 is my go-to for a low voltage fet switch.  Max Vgs is 2V. 

[TomKatt]

BSS123 is my go-to for a low voltage fet switch.  Max Vgs is 2V.

Yes - I believe this may be what I saw as the better option, but I was looking for through hole components.   Though I now have an easier way to make pcb boards at home with my engraving cnc, so it may be time to get with the 20th century lol.  Or at least get some kind of SOT smd to DIP adapter module if I can.

Thanks for the suggestion!

[Peabody]

You could use two BS170s in parallel, and maybe get enough current.  Just double up the leads in each through-hole, and see what you get.

Also, what is the effective resistance of the mosfet when you're getting 60mA?  The voltage drop across it should let you calculate that.

[TomKatt]

You could use two BS170s in parallel, and maybe get enough current.  Just double up the leads in each through-hole, and see what you get.

Also, what is the effective resistance of the mosfet when you're getting 60mA?  The voltage drop across it should let you calculate that.

I am entirely too lazy to figure all that out lol.  The IRLB8721 mostfets I swapped in place work fine even though they are entirely overkill for the application.  And their Rds is an order of magnitude less than the BS170 anyway.  For low voltage applications, the IRLB8721 is probably a better choice than the BS170 I started with, other than TO-220 vs TO-92 packaging.

EDIT - I just noticed that Adafruit sells the IRLB8721 under their own name brand as a '355' mosfet.  Curiously, Mouser also sells them as 'Adafruit 355' mosfets.  If you look up the datasheet on either web page it shows you the Infineon IRLB8721 part I have.  Interesting.

https://www.adafruit.com/product/355

https://www.mouser.com/en/ProductDetail/Adafruit/355?qs=GURawfaeGuB9%2FKqnwClJ%252Bw%3D%3D

[Geoff-AU]

You could use two BS170s in parallel, and maybe get enough current.  Just double up the leads in each through-hole, and see what you get.

Uh, a mess?   

If you were desperate, sure..  but tomorrow you can build it with a single mosfet and the day after you'll need 5.

I was looking for through hole components.

Ah sorry I forgot that detail..   Digikey's parametric search is really excellent for this.

Product Index > Discrete Semiconductor Products > Transistors > FETs, MOSFETs > Single FETs, MOSFETs

Filter by "In Stock", Exclude Marketplace, Mounting Type "Through Hole", then you can start whittling down. 

As others have said, with Digikey you now have to watch out for manufacturers you've never heard of, stick with the big names if your time is important to you.

The best-practice way of speccing a switching mosfet is to look at the specified drive voltage, not the threshold voltage.  If you're running at 3.3V then filter by "Max Rds(on)" below your gate drive.  Immediately it cuts the field down to 7 TO-92 parts, all by Microchip (which is a bit of a red flag in itself for future availability).  One example is TN0104N3-G.

You're occupying a vanishing niche by trying to use through-hole parts below 5V though..  Going SMD has a bit of a learning curve but a lot of advantages.  You can use adapter boards but that eventually gets quite cumbersome.  With a SOT23-3 you can be quite generous with the pad size and your CNC will handle it easily.

[TomKatt]

You're occupying a vanishing niche by trying to use through-hole parts below 5V though..  Going SMD has a bit of a learning curve but a lot of advantages.  You can use adapter boards but that eventually gets quite cumbersome.  With a SOT23-3 you can be quite generous with the pad size and your CNC will handle it easily.

Yeah, I see that ;-)  My cnc is working out better than I expected and milling sot pad layouts won't be an issue.  Now whether my hands can solder it on the board is another story lol.  But there is no question that this is the future.

[iMo]

..also mind the Vgs=x.xV means the mosfet "at Vgs=x.xV just starts to conduct" - imagine nA/uA, afaik.
It does not say at x.xV it fully opens to the saturation.

[TomKatt]

..also mind the Vgs=x.xV means the mosfet "at Vgs=x.xV just starts to conduct" - imagine nA/uA, afaik.
It does not say at x.xV it fully opens to the saturation.

Yes - I think that is the most confusing parameter.  It would be helpful if they specified both saturation start and also the voltage where the device is fully enabled.

[Kleinstein]

The is no sharp thereshold for fully on. R_on continues to go down with higher gate voltage, though less and less when the gate voltage is high.
One can use the R_on vs U_GS graph to estimate how much above the threshold one should go to get something like 2 x the nominal R_on. For the 2N7000 (essentially the same as BS170) this are some 2.5-3 V extra. With lower threshold / lower voltage fets this may be a bit less.

[gamalot]

saw a few references to the BS170 mosfets, so I grabbed a few.  No workie lol.

Gate threshold voltage for genuine BS170's is 3V max, so I expect it should've worked....

If you read it carefully, you'll notice that that was measured at only 1 mA of current.

[gamalot]

Yes - I think that is the most confusing parameter.  It would be helpful if they specified both saturation start and also the voltage where the device is fully enabled.

In fact, it doesn't really have a sharp threshold, which is why you see those curve plots in datasheets.

[RoGeorge]

You should look for "small Vgs", not for "small signal", or better search for "logic level MOSFET".

Logic level MOSFET is a dedicated term for low Vgs switching MOSFETs, they are meant to be controlled by an on/off digital pin at 5V or 3.3V, or sometimes even lower than that.  The datasheet of BS170 says "small signal MOSFET", which means they are for small analog signal, audio, radio and such, but small.  That doesn't mean Vgs is small.  From the datasheet, it looks like Vgs on is about 10V or so:  https://www.mouser.com/datasheet/2/308/BS170_D-1803008.pdf

[Terry Bites]

AO3400 anyone? Cheap as.
www.lcsc.com/search?q=AO3400&s_z=n_q_AO3400
 

[David Hess]

I usually get away with generic 2N222's for things like relay control with my microcontroller projects but needed something fast for a 3.3V project...  Did some Googling and saw a few references to the BS170 mosfets, so I grabbed a few.  No workie lol.  Despite low Vgs specs, they don't fully saturate at 3.3V.  I did have luck with the IRLB8721 mosfets - although being much bigger 30V / 62A TO-220 components, they seem to work fine for driving low power opto leds.  I know there are better options, but they all seem to be smd and I haven't quite escaped through hole world yet.

The 2N2222 was not fast enough?  As a saturated switch and without any charge control, they turn off faster than 250 nanoseconds.  Add a speedup capacitor and they could reach 50 nanoseconds.

The BS170 was one of the first commonly available TO-92 MOSFETs, so was not a low threshold voltage device.

The least expensive low theshold voltage part I found with similar on-resistance is the Microchip TN0604N3-G at $1.39.

[iMo]

This is a picture you may perhaps find in the datasheets..
Here in the below simulation the Vgs is aprox 1V.
In order to get Ids = 1mA you would need 2.3V..

[Jay_Diddy_B]

Hi,

Look for the gate charge waveform on the datasheet, this is from the onsemi BS170 datasheet:




The plateau shows the Vgs required, in this case, for Id = 500mA. It is around 4.2V.

This section of the datasheet shows the Vgs required for Id = 1mA, a very small drain current.





Regards,
Jay_Diddy_B

[Jay_Diddy_B]

AO3400 anyone? Cheap as.
www.lcsc.com/search?q=AO3400&s_z=n_q_AO3400

Terry Bites,

I bought some aliexpress AO3400 MOSFETs a while ago. I shared the test results on the forum here:

https://www.eevblog.com/forum/projects/smd-test-fixture-for-the-tektronix-576-curve-tracer/msg4553651/#msg4553651

I did get MOSFETs, but RDSon was high and BVds was low.

If you use low cost parts, you need to determine what you get.

Regards,

Jay_Diddy_B