For FET-sake, I keep switching and cannot find a suitable MOSFET (or BJT)

Question

I came across this lovely term on Twitter the other day for tablet and laptop battery packs that had seen better days (usually when continuously powered/charged). It's #spicy pillow. LOL.

https://twitter.com/hashtag/spicypillow?lang=en

Anyway, this reminded me of a little project I had in mind (as shown in schema below).

As a electronics novice, this appears to be a borderline case as I know that transistors tend to work well for mA switches while MOSFETs tend to be targeted at higher voltages (and say +3A current).

So now struggling to choose a suitable component - I'm particularly interested at the 1.5A option as that tends to be the limit for many usb charger adapters:

Having done some prelim research it looks like FET's are the way to go but when working through my design criteria I'm quickly getting bogged down in the detail... In this case I'm not coming up with options as I can see that I need to minimise my resistance losses to reduce voltage drop etc... I'm also looking for a low cost <1$ SMD option.

Hence I thought to open this little design challenge to the members as I had to chuckle at aGough Lui status update I found when doing my e14 search...

Often we're lazy, choosing to use the same parts over and over. I'm guilty of this when it comes to...

So, any old favourite FETs you have which could well work here?

michaelkellett · Answer

Don't have favourites. Always use a part for a reason (usually many reasons). Having a reel of 999 of a part is a good reason if it will work OK ! In your case with 3.3V drive, 5V power supply, a load with possibly a big switch on surge, consider: Max drain source voltage rating > 10V Max gate source voltage for 5A < 3V Gate source resistance @ 5A < 0.05R (would give 312mW @ 2.5A Putting more or less these parameters into the Farnell search engine I got to the STS6NF20V STS6NF20V in SO8 package at £0.635 (10 off). It's very close but may be a little off on 5A on resistance with 3V gate drive. There are similar parts in much smaller packages but I prefer the SO8 as it can take a bit of power. MK

colporteur · Answer

Absolutely loved the title of your post! You have to start writing for British tabloids. I missed the meaning of the hashtag. I have little exposure to the sun or social media from under the bridge I only venture out for the goats that attempt to cross. I figured I write something to see if you get the reference. I read your post and will start by saying, I don't know if the module will work for your application but I have had such great success with them, I buy them by the gross. They exceed your price limit. I paid $1.27 CAN not included overhead. The design of the module matches what you napkined in your post. Other than soldering on the input pins they are good to go from the package. https://www.banggood.com/10Pcs-MOS-Trigger-Switch-Driver-Module-FET-PWM-Regulator-High-Power-Electronic-Switch-Control-B…

BigG · Answer

LOL. (British tabloids) I missed the meaning of the hashtag. Batteries look like pillows. As to where "spicy" came from, who knows, so I had to look it up (thanks Google) https://www.urbandictionary.com/define.php?term=spicy%20pillow Anyhow, thanks colporteur for the suggestion for those breakout boards. They look very useful for sure. To check the spec, I found a link to the datasheet for that MOSFET: http://www.aosmd.com/pdfs/datasheet/AOI4184.pdf Interestingly, I found a similar breakout from these guys https://protosupplies.com/product/d4184-mosfet-control-module/ They suggested a 6V as minimum voltage... their webpage makes this comment "Since the maximum gate voltage available to turn on the MOSFET is 50% of the power supply voltage, that limits the power supply to a minimum of about 6V. "

Gough Lui · Answer

It is a shame that not more datasheets have a Rds vs Vgs graph because that is often most useful for this kind of application. That's part of the reason I run my tests ... just to know that given a particular Vgs drive from a microcontroller, what kind of Rds(on) I can expect, because with that information and knowing the current of your load, you can calculate the power dissipated by the MOSFET (got to keep the junction temperature below limits given the thermal resistance) and the voltage dropped across it (to keep the load happy, primarily).

But on the whole, if you do have the graph, you really want to be on the "flat" side of the knee and not on the "steep" handle-side of the hockey stick as that would imply potentially massive changes in actually experienced Rds depending on part-to-part and temperature variations. If you don't have the graph, there's always the Id vs Vgs graph (that is more often available in the datasheet) and ultimately, you want to make sure that you're not "capped" for Id at the given drive voltage as that would suggest the MOSFET could potentially act as a current limiter and dissipate quite a bit of power (even if the load doesn't reach that level).

For high-speed switching, low Rds losses usually come with higher gate charge and capacitance which means slower switching speeds or more critical driver requirements. So choosing a higher Rds MOSFET may make sense for high PWM frequencies or where you are constrained with your driving set-up (e.g. no dedicated MOSFET driver chips, high source impedance).

- Gough

colporteur · Answer

The link explained your pillow talk perfectly. I confess, I am ignorant of the specifications for the device. I was initially building my own using irfz44n mosfets. It was not a dual design like the modules. When I discovered I could purchase cheaper than build I switched. I use the module to power 12V devices from lower power inputs like SBC and microcontroller. I have used the Raspberry Pi GPIO output to turn on a string of LED's using the module. I ways curious about what was a minimum trigger voltage. The fact it worked was good enough for me.

Gough Lui · Answer

Your choice of driving the MOSFET with 3.3V logic is definitely a big constraining factor, as it cannot be guaranteed the full 3.3V is available. The MOSFET threshold voltage (Vth) needs to be as low as possible to ensure it turns on well at the high logic output voltage. Choosing an N-channel MOSFET in this configuration is wise, as they are more efficient (generally speaking) and you would want the lowest Rds(on) that you can get for the price to reduce voltage loss and heating. As you're not rapidly switching on and off, the gate capacitance figures are perhaps less important.

Because you have a preference for SMD, these are probably the two that I thought of at first (in-stock and relatively low-cost) -

- Vishay Siliconix SI4160DY-T1-GE3 (SOIC) - https://www.newark.com/vishay/si4160dy-t1-ge3/transistor-polarity-n-channel/dp/15AC0296

- Vishay Siliconix SiRA14DP-T1-GE3 (PowerPAK SO-8) - https://www.newark.com/vishay/sira14dp-t1-ge3/mosfet-n-ch-30v-ppak-so8/dp/63W4125

Both have very low threshold voltages of ~ 1.1V and rated about 8.5mOhm at Vgs = 4.5V. This makes it more suitable for 3.3V driving, but you will suffer much higher resistances than the headline figure because you are not on the "good" side of the knee:

Ideally, 5V drive would be best with these MOSFETs but that's not to say they won't work.

I would say michaelkellett's suggestion of the STS6NF20VSTS6NF20V is definitely a safer bet with Vth = 0.6V and has Rds = 45mR at 2.7V drive, but I suspect that if you can keep the drive voltage around or above that level, the Vishay components may have slightly better effective Rds and more robustness. The ultra-low threshold does come with a bit higher price tag.

But then I came across these that might also deserve some consideration (mainly for cost)

- NXP PCMA14UNYL (DSN1010) - 11V, 14A, 13.2mR at Vgs=4.5V, Vth=0.6V - https://www.newark.com/nexperia/pmca14unyl/mosfet-n-ch-11v-14a-dsn1010-rohs/dp/84AH8275

- On Semiconductor NTLJS4114NT1G (WDFN) - 30V, 7.8A, 20.3mR at Vgs=4.5V, Vth=0.55V - https://www.newark.com/on-semiconductor/ntljs4114nt1g/n-channel-mosfet-30v-7-8a-wdfn6/dp/10N9582

The NXP PCMA14UNYL comes in a bit of an odd package I haven't seen before, but the characteristics are very nice for your needs ...

The On Semiconductor NTLJS4114NT1G seems to be another less-common (2mmx2mm) package, but it seems to be on clearance right now at US$0.167ea ...

Of course, take my suggestions with a grain of salt ... I just wanted to do some "virtual" shopping without actually buying anything .

- Gough

shabaz · Answer

Hi Frank, You're right, it's very likely solder mask will break in the blue regions here. The spacing between the rows is a bit more, so hopefully if the iron and solder are applied as shown with the red and orange arrows, then hopefully it's doable.. it's really tight though, the pitch is 0.85mm between pins within the same row. I gave the layout a shot, that is more complicated than Micro-USB since the two resistors are needed, but it is not too bad provided at least one resistor is on the underside of the board, otherwise it's a lot more awkward. These resistors are 0805 sized in the screenshot below. I might tweak this a bit more, but I think broadly I'm going with this sort of layout.

BigG · Answer

All thanks to https://oshpark.com/ , it cost me about $10 to get 3 boards made up and delivered using their express service. Very impressed. As you can see from the above schematic, the board is designed to be attached to a Nordic Semiconductor nRF52840 dongle - these are great low cost dev boards that can be slotted directly into a USB port. The MOSFET I decided to use was partly based on the fact that it was in stock and is low cost. https://www.vishay.com/docs/63302/si2342ds.pdf It works - I verified manually. Now I'm able to control the power to any device attached, using BLE to remotely switch power on and off. Well that's the theory. Now I just need to get my nRF52840 firmware sorted. Fingers crossed. {gallery:autoplay=false} USBLE Spicy Controller Board The underside of my USBLE spicy controller board My USBLE Spicy Controller attached to an nRF52840 dongle nRF52840 USB dongle inserted into laptop

shabaz · Answer

Hi Colin, Nice project! One thing you might (or might not) need (can be easily hacked in place hopefully, perhaps with surface mount resistors, is a resistance between the data pins, because the tablet/end device might not draw too much current without it - It's called Dedicated Charging Port (or other options are possible), I don't know the precise details but they should be published, there's some brief information here: SILICON LABS USB-TO-UART Bridge Controller EVM - Review (in the 'Battery Charging Negotiation' section)

Gough Lui · Answer

Indeed, it should work just fine with USB-C at least in its USB-A compatible modes by using a regular A-to-C lead. The resistance values on the CC lines using a USB-A to USB-C cable are usually configured to tell the end device to "fall-back" to other methods of detecting available charge current. Regardless, the connectors themselves shouldn't be used above 3A anyway, most USB-A devices never exceed 2.4A in the best ideal situations. However, in most cases, because of the USB-BCS rules, devices will fold-back their current if they detect the charger is being overloaded - this is done by watching the voltage. This is mainly so that if someone plugs a new phone into an old USB 0.5A charger for example, it doesn't trip the OCP. This fold-back usually works very well, resulting in slower-than-normal charging on long leads, thin leads and/or worn connectors. I'd probably be more concerned that the Nordic board doesn't have a "proper" USB connector, as the PCB-pad connections do not have the raised ridged profile of proper USB pins and the contact pressure may not be optimal. Drawing so much current this way could lead to "intermittent" resistance changes which may mean downstream devices fold-back and charge a bit slower. Of note is that USB meters also will add additional resistance due to the connectors and sensing shunts - sometimes a small wiggle of the connection and you can see the fold-back in action. - Gough

shabaz · Answer

Hi Colin, I'm in the same situation, many devices with USB-C, so I ordered some connectors - they arrived today - so that I can in future lay out boards with that rather than (in my case for USB peripheral) Micro-USB. From what I read , since the connector works in both orientations, the data pins are replicated so they need to be connected together, and then there is a pin called CC, again replicated, and each of those separately needs to be connected via a resistor, to 0V or to the supply. The computer end (host) connects it high, and the peripheral end (device) connects it low. The diagram below was from a Cypress document but I found it hard to follow so I put some colour-coded annotations on it below. Personally I went with this USB-C connector this USB-C connector which is through-hole (I preferred SMD but I didn't see one that looked easy to solder). The through-hole won't be particularly easy to solder either, since the pins are extremely close, but that's life - I've bought enough to last me a while! so I have to go with what I have.

shabaz · Answer

Hi Michael, That looks nice, with the surface mount pins not recessed in. Great for hand-soldering. Couple of mm shorter than the through-hole ones too. I'll have to make do with the through-hole ones (I bought quite a few) until I get through them all : ( Incidentally I also obtained a couple of 6-pin USB-C connectors 6-pin USB-C connectors that only have the power pins and the CC pins exposed. They are very easy to solder as a result (can even tack on wires to them easily), but I don't think I'll be using them except to perhaps quickly prototype anything charging related. Photos below of the USB4085-GF-A through-hole, and the 6-pin USB4125-GF-A :

For FET-sake, I keep switching and cannot find a suitable MOSFET (or BJT)

Top Replies