High speed mosfet for High Voltage Power Supply

[Max DN]

Hello,

I'm working on a solar powered nixie watch (similar to the Kopriso Nixie watch).

It's crucial that I dramatically reduce stand-by power as much as possible and within 2.5uA, which is where the solar panels will balance off the stand-by current.

Regardless of which power supply I use, even using the shutdown pin, the power supply still takes 2.5uA. This is perfectly in line with the electrical specifications of the datasheets.

So, my question is, can I 'brute force' a shutdown, effectively removing the power from the VIN pin via a PNP high speed mosfet such as MMBTA42 or it will introduce too much of a delay in starting the power supply?

I'm going to try it on the fly but if anyone has any suggestions on how to bring to 0 or to 0.01uA the stand-by current that would be great. 

As an example, I'm using the LT1308B DC/DC converter as on page 17 of the datasheet https://www.analog.com/media/en/technical-documentation/data-sheets/1308abfb.pdf

Any suggestions will be much appreciated.

Thank you,

Max

[gregebert]

I used the LT3561 to provide the 3.3V logic supply from the battery (3.7 to 4.1V Li-ion), and it's standby leakage is practically zero. All I know is the leakage current of this regulator, plus the operating current of the DS3232 RTC and leakage of the HV driver, makes a very slight movement of my 50uA meter. The RTC adjustment every 100 seconds pegs the meter.

The only MOSFET I have is for the HV inverter (DMN6040SVT) , and it's leakage is rated at 100nA.

The standby current of my nixie watch board is so low that the battery lasted 6 years on a single charge. I should also mention this battery was used in my cellphone for several years, so it's capacity is significantly less than the original 1100ma-hr. This board is in my garage, and every so often I would push the display button to energize the display. Alas, I finally had to charge it last month because it could no longer light all digits.

[Max DN]

Thanks for your reply. That's impressive. 

As a 3.3V regulator I'm using a simple MCP1700, but that takes about 6.5uA, that's "too much" vs 1.6uA declared in a the datasheet, I tried a few parts, maybe not a genuine batch...

Then 2.4uA is taken by the power supply DC/DC converter. This can be improved, but if I improve the efficiency of the power supply when on, then I find that the chip uses more current in stand-by. I mean, 2.4uA is nothing really but as I'm trying to break that even with solar panels, I need to reduce it further and as close to 0 as possible when in shutdown or I can try through a 'true shutdown'.

The uC itself should take just about 1uA or less when in sleep mode.

That is a total of 10uA stand-by current vs my goal of 2.1uA (I know it can be done, just not sure how yet).

Perhaps  think I need something like this, I believe the FQU11P06 is used as a high side load switch, although it may not be the best in terms of efficiency...

https://www.facebook.com/Zx81NixieClock/photos/a.361754507268684/361754510602017/

Or a better solution could be to implement a high-side load switch as suggested by TI here (as a true shutdown):

https://www.ti.com/lit/gpn/tps27081a

I guess I'll have to run some experiments on the bench. If anyone has already done it, feel free to share more suggestions. Thanks again gregebert!

[David Forbes]

Max,
The LT1308B has a .01 uA typical, 1 uA maximum standby current when
shut down, according to the data sheet, shown on page 2 as Iq (the
third line with Vshdn = 0V).

I don't think you have a problem here.

> --
> You received this message because you are subscribed to the Google Groups
> "neonixie-l" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to neonixie-l+...@googlegroups.com.
> To view this discussion on the web, visit
> https://groups.google.com/d/msgid/neonixie-l/d45edcd8-0176-4205-95f1-b04eb34a32f1n%40googlegroups.com.
>

[Max DN]

Thanks David, indeed you are absolutely right. The power supply with LT1308B is very reliable and not as sensitive to layout design as other ones that I have tried.

LT1308B has in Iq of 0.01uA in stand-by, as you say. I've built your nixie watch (picture attached) and in my design it seems to use 2.3uA. If I build the HVPS on a breadboard, I can measure 0.1uA or less, which is great (I guess the resolution of my multimeter Keysight U1241B cannot measure less than 0.1uA accurately but that's great at 0.1uA or less). Of course I have tried to troubleshoot my LT1308B HVPS with no success. In my quest for a more efficient HVPS,  I have designed a power supply using the MAX668 chip (as per datasheet and similar to many commercial HVPS), this was tricky to design with no noise, it works well now (after 6 different designs) however the stand-by power is 2.4uA, as per datasheet. So I can have higher efficiency than with LT1308B but also higher stand-by current (I believe because MAX668 has an internal LDO).

Which is why I was thinking whether I can use a high load switch to truly bring shutdown current to zero using a power supply built around MAX668. Of course, this is a bit of a (pointless) extreme challenge to reduce stand-by current as much as possible to break even with solar panels, but that's where the fun is for me. And of course I could keep on troubleshooting the LT1308B design, it actually works perfectly other than it gives me higher stand-by current than on the datasheet, not sure if that depends on the design layout. But then again, I'm only a hobbyist and not an engineer, so I may be missing something.

So any suggestions on why my LT1308B draws more than 1uA with SHTN=0V welcome (I have tried 4 different chips, same thing on the PCB, fine on the breadboard). Or any other suggestions to use any other power supply with perhaps a high load switch to bring standby current to less than 1uA, which I worry is unnecessary additional components and introduces a potential start-up delay.

[Christian Riise Wagner]

I've successfully used a P-channel MOSFET as a high side load switch for a LM3478 based tapped inductor booster. IIRC the off current was immeasurable, so sub 100nA.

[Max DN]

That's great to hear as a confirmation. As soon as I get a chance to spend some time on the bench, I'll give it a try. It should be quick enough. I'm going to use a standard MMBTA42, please let me know what you have used, if you can share that piece of information.

Thanks Christian.

[Christian Riise Wagner]

The MMBTA42 is a bipolar transistor and not a FET. Secondly it's NPN, making it unsuitable as a high side switch, because the base needs to be at least at 0.7V higher than the voltage at the emitter for the transistor to turn on. I used a Winsok WST2339 because that was the lowest Rds(on) P-channel FET I could find in a SOT-23 package. However you will no doubt be able to find something similar from Mouser, Digikey, etc. It will be active low, so the gate should be connected to the input voltage rail through a pull-up resistor. Your microcontroller will then have to pull the gate low to enable the power supply. If the supply voltage of the power supply is within the voltage rating of your microcontroller's GPIO pins, you can connect it directly. If not, you should use a N-channel MOSFET or an NPN transistor to pull the gate low.

[Christian Riise Wagner]

Sorry, it's early morning here and I'm apparently not fully awakened yet. Meant to say that it was the lowest Rds(on) P-channel FET I could find in a SOT-23 package from LCSC, which is where I shop most of my components. You'd be able to get away with a slightly higher Rds(on) FET. If it's too high though, it will start to affect the efficiency, leading to a higher current consumption when on, due to the boost converter compensating for the voltage drop.

[Max DN]

Thanks Christian, that's very helpful and hopefully useful to others here too. 

Yes, of course MMBTA42 is a NPN transistor and not a FET, what was I thinking??? I have a few mosfets around (N-channel and P-channel) I'll check their datasheet, so great to have yours as a reference point.

I've also ordered a high side switch such as TPS27081A, it seems to control inrush currents and doesn't take too much real estate on the pcb.

I'll report back as soon as I have received the parts and tested them.

Thanks everyone.