low cost - rather low power mosfets

Anything not related to STM32
ag123
Posts: 1322
Joined: Thu Jul 21, 2016 4:24 pm

low cost - rather low power mosfets

Post by ag123 » Wed Mar 13, 2019 7:29 pm

recently i've been searching for low cost rather 'low' power logic level mosfet transistors, on various sites digikey / aliexpress etc
i stumbled into http://www.aosmd.com/products/mosfets/single
the particular part numbers i took an interest are
AO3400 30V N-Channel MOSFET 5.8A on resistance 52 mohm at 2.5v
http://www.aosmd.com/res/data_sheets/AO3400.pdf
AO3401 30V P-Channel MOSFET -4A 85mohm at 2.5v
http://www.aosmd.com/res/data_sheets/AO3401.pdf
AO3402 30V N-Channel MOSFET 4A 85mohm at 2.5v
http://www.aosmd.com/res/data_sheets/AO3402.pdf

the nice things about them are the turn on voltages are around 1.5v and by around 3v the characteristics curves are reaching 10s of amps though we'd need to limit them to the spec currents in real use. AO3402 is apparently a matching n channel mostfet for ao3401. however, i decided to go with that ao3400 for the high amps even though they seemed 'unbalanced', i'm not sure if there could be any implications

they easily beat bipolar npn and pnp transistors such as 2n2222 npn or 2n2907 pnp that max out around 600ma or so
the amperes sized current handling capacity are really attractive

i'm finding prices as low as about $1.5 per 100 pcs for the AO3400 shipped and around $1 per 100 pcs for AO3401 shipped
https://www.aliexpress.com/w/wholesale-ao3400.html
https://www.aliexpress.com/w/wholesale-ao3401.html

unfortunately they are sot23 smd transistors and i'm really noob with smd soldering, in the end i ordered some smd to 0.1" adapter boards that are easily priced a couple of times more than a single piece of transistor each, what an irony. hopefully i can try them out to see how these mosfets work and to see if they are really as performant as the specs claim to be.

ag123
Posts: 1322
Joined: Thu Jul 21, 2016 4:24 pm

Re: low cost - rather low power mosfets

Post by ag123 » Wed Mar 13, 2019 7:40 pm

i'm wondering if these 4amp mosfets work if i'd be able to make buck converters out of them and use that 'overpowered' stm32f103 timers to pwm the mosfets for the buck converters. i'd then have a 'digitally controlled power supply' rather than using buck conversion ICs
:lol:

misfet
Posts: 44
Joined: Sun Jan 14, 2018 8:06 am

Re: low cost - rather low power mosfets

Post by misfet » Thu Mar 14, 2019 11:01 am

5 amps through a sot 23 is bending my brain. I'll have to get some. Easy to solder, heat the pad on your board (with a soldering iron), add a bit of solder, not so much it is a high bubble. Put the sot on with some tweezers and then put the tip of the iron on each sot leg. With a bit of pressure you can feel it when the solder melts.

ag123
Posts: 1322
Joined: Thu Jul 21, 2016 4:24 pm

Re: low cost - rather low power mosfets

Post by ag123 » Thu Mar 14, 2019 1:46 pm

thanks for the smd soldering tips. i've a couple of these pieces of cheap buck converters
https://www.aliexpress.com/wholesale?ca ... ter+LM2596
my original purpose is to make some power supplies with them and use variable resistors to tune the voltage

after i stumbled into the sot 23 mosfets it is giving me a twitch that i should after all use that 32 bits 72 mhz arm m3 stm32f103 + that 5 built in timers to pwm the mosfets and perhaps make buck converters out of them.

i can in addition run an ili9341 lcd to show the voltages and currents and to run regulation of the buck converter and perhaps to shutdown if the output is shorted out (over current protection).

and i realise the true benefit of this approach using the stm32f103 to drive the buck converters, whatever voltages you need, it becomes just a matter of sending commands to that stm32f103 buck converter and you get the voltages you need - no potentiometers or variable resistors, a digital power supply
:D

User avatar
BennehBoy
Posts: 886
Joined: Thu Jan 05, 2017 8:21 pm
Location: Yorkshire
Contact:

Re: low cost - rather low power mosfets

Post by BennehBoy » Thu Mar 14, 2019 3:07 pm

Sounds cool.
-------------------------------------
https://github.com/BennehBoy

ag123
Posts: 1322
Joined: Thu Jul 21, 2016 4:24 pm

Re: low cost - rather low power mosfets

Post by ag123 » Thu Mar 14, 2019 5:28 pm

btw i had to agree with @misfet that 5 amps on sot23 seemed too large to be true, hence 'bending the brain'
but at 1c - 1.5c each, it is probably worth experimenting, either way through hole bjt npn or pnp i doubt it is possible to even go anything higher than one amp. a tip120 takes a TO-220 package 5 A max
https://www.mouser.com/ds/2/149/TIP120-890130.pdf
that i think is due to Vce saturating around 1v which implies a TIP120 burns a pretty high 5 watts of power at the peak amps
that 5 amps sot23 mosfet is possibly only possible due to low Rds on resistance, but with sot23 it makes one wonder if the traces may be so thin that it is the copper itself that produce all that heat (due to higher resistance), i.e. the mosfet runs cool and the copper burns
:lol:

misfet
Posts: 44
Joined: Sun Jan 14, 2018 8:06 am

Re: low cost - rather low power mosfets

Post by misfet » Fri Mar 15, 2019 1:43 am

that 5 built in timers to pwm the mosfets and perhaps make buck converters out of them.
The switching regulators do more than switch the outputs. There are different modes of operation. http://www.ti.com/lit/an/slva057/slva057.pdf This info from ti is a good intro.

The regulators also regulate, which means your micro will not need to keep tracking the voltage and adjusting the pwm every time the load changes. I would expect if you attempt this, you will be faced with a very slow reacting result, or over and undershoot, and most likely both, as the load changes with different amounts and speeds. I may be wrong, but I would be thinking the supplied regulator will probably do a much better job.

Having a programmable and interactable control over the voltage certainly has appeal. But I think you may have a more satisfactory success if you look at ways to interact with the regulator rather than the power output. - Keep the buck converters. It will have a sense pin that the adjustable pot is attached to, you could use pwm and a rc filter to adjust the regulator, and let that keep control over the output fet.

mf

ag123
Posts: 1322
Joined: Thu Jul 21, 2016 4:24 pm

Re: low cost - rather low power mosfets

Post by ag123 » Fri Mar 15, 2019 5:08 am

thanks, but a guess is that as buck converters basically use the inductor to store and release energy, a larger inductor bigger L may help to smooth large current swings. but i think you are right about the responsiveness of regulation to changes. nevertheless, i'd think it is fun to try an experiment as a digitally controlled buck converter is pretty useful. apparently most PC motherboards use dedicated specialised digital buck converters so that you can change the Vcore to the cpu - for overclocks, low power etc.

apparently someone has been there done that
https://github.com/ksarkies/SMPS/tree/m ... -STM32F103


even st provide a 'reference design' with a 'massive' stm32f334
https://www.stmicroelectronics.com.cn/r ... 108524.pdf

:)

misfet
Posts: 44
Joined: Sun Jan 14, 2018 8:06 am

Re: low cost - rather low power mosfets

Post by misfet » Sat Mar 16, 2019 12:21 pm

Mmmmm. I could not really glean anything from the you tube. I am curious as to how you get on. Can you have it finished with a performance spec before the forum's demise? :lol:

I have been toying with a related idea - holding photo voltaic cells in their efficient voltage range and using the switch/ inductor to convert the resultant output to current for a battery stabilized power supply.

I would think of it as a task more naturally suited to an analogue based design, it is just that I don't know how to do it.

ag123
Posts: 1322
Joined: Thu Jul 21, 2016 4:24 pm

Re: low cost - rather low power mosfets

Post by ag123 » Sat Mar 16, 2019 3:25 pm

the formulas for buck converters is 'surprisingly simple',
D = Vo / Vi
where D is the duty cycle
i think this is based on 'steady state', basically non-fluctuating power draw
https://en.wikipedia.org/wiki/Buck_converter

if you have power draws that goes up and down like a bungee jump, stm32 may not be able to catch up with regulation
but if otherwise for 'slow' regulation, it is quite possible stm32 handles it.
i've been looking around for mosfets. however, i tend to find 'big' high current ones e.g. IRLZ44N
http://www.irf.com/product-info/datashe ... rlz44n.pdf
they are good but that the currents provided seemed much higher than would a 'normal' case of say a lipo charger say 1 A and that the package is TO-220 which is 'a little big'

searching for 'smaller' mosfets is fraught with all the catches as it seemed the brand name companies would rather simply sell power regulators, motor controllers that are all integrated with the necessary regulation and control logic. i'd think it is simply that in terms of costs and simplicity, the integrated power regulators / motor controllers are far simplier and cheaper than do discrete logic and that there is more margins selling the integrated 'plug and play' solutions. discrete mosfets turned out to be 'expensive' or hard to find parts that hit the few amps specs and yet are prevalent and low costs.

those from alpha and omega semiconductors seemed to be a decent find and they seem to be willing to mass produce the small mosfets handling a few amps, but that they mostly only do smd.

i happen to come across this article from microchip
http://ww1.microchip.com/downloads/en/a ... 00792a.pdf
one of the rather interesting data is in table 3 which shows that sot23-3 has a thermal resistance of 336 degrees C per watt
assuming that the ao3400 can carry 5 amps of current and that its Rds on is 50 milli ohms that gives ~ i^2 * r = 25 * 0.05 = 1.25 watts !
using the microchip app note, this would imply that that the AO3400 sot23 mosfet could literally burn at > 336 degrees C (most plastics would melt and i'd think could burn) the moment it does 5 amp continuous currents

it is somewhat suspect how far the AOnnnn could go in terms of mosfets but i'd think they are sufficiently low costs to experiment.
the low costs makes it feasible to do 'low medium power' power management applications. e.g. if i need only 1 amps, and to charge a 4.2v lipo battery
i'd think it is pretty feasible to use that AO3400 and have a stm32 to pwm that AO3400 and use that as a lipo battery charger. i.e. instead of using a dedicated lipo charger IC, it becomes quite feasible to just have that AO3400 that modulate the 5v (usb) - 4.2v lipo battery charging
then stm32 is in turn supplied from that same 4.2v lipo battery, put a diode so that you get 3.6v VDD from the lipo battery.

that seem like an interesting use case as all you spent is one of the 4 timers (maybe + an adc channel to monitor the charging) on the stm32 and an ao3400. i.e. stm32 handles its own lipo charging. there is a catch though, it may be necessary to use ao3401 p channel instead as ao3400 may not turn on if it is done as a high side mosfet

Post Reply