Hi all,


I have been trying a few different approaches over the last week in an attempt to create a circuit to "safely" supply power to my NVIDIA Jetson Nano.
The issue arose when I realized that a voltage regulator I was switching with an n-channel mosfet, is actually switching through the VIN t VOUT, when
the ground is floating. Since then I have been using an n-channel to switch a p-channel mosfet that then switches the positive voltage.
Having nt yet felt comfortable enough to supply voltage this way to my (precious) Jetson Nano, I'm looking for an alternative, secondary circuit to ensure the voltage getting through never
exceeds the safe voltage range of 5.2V max.

The approaches I have taken are:
1. Built a circuit similar to the one in this image:
https://www.maximintegrated.com/content/dam/images/design/tech-docs/760/di66fig01.gif

However this does have a limitation on the power that can be switched. My Jetson Nano requires about 2.5A during normal operation.
So a circuit being capable of switching currents of up to 3.5A is needed.
For a while I have been testing a similar circuit with some of the BJTs I have available. I have tried multiple variations of the circuit using both NPN and PNP transistors and a mix thereof in conjunction with a zener diode 5.1V to realize this circuit.
(BC557, BC558, 2N2222, 2N3904, BC547, BC549)
What I have noticed is that I needed about 8V before I was able to get the diode to get conductive in the opposite direction and switch the transistor stage.

2. I have also come across the circuit below, using a triac. I have not been able to give this a try yet, as I'm waiting for parts (4 weeks and counting).
https://www.instructables.com/Over-Voltage-Protection-for-5V-Circuits/

My question to experts in this forum is, do you have knowledge or insight into a circuit, that can achieve over-voltage protection, switching off at voltages > 5.1V, while still being able to switch a load (ideally positive switching) of about 3A?
The following components are available, I just have no clue how to make a viable circuit and after spending more than a week trying on a breadboard, I finally have come to my senses.

BJT's:
24 value standard NPN/PNP transistor assortment with most common BJTs (such as BC517, 547, BC549, BC556, BC557, BC558, S-series, 2N-series, A nd C-Series)
MOSFETs:
IRF1405 n-channel
ZT2467 p-channel
Zener Diodes: 5.1V
Resistors (all sorts)

Please let me know what information is needed, or maybe you even have a schematic from a past similar circuit!?


Thank you!

 

You have a regulator, which will presumably have a current limit.
if so, use a crowbar circuit.
https://www.rfwireless-world.com/Terminology/SCR-crowbar-overvoltage-circuit.html

 

LM323 ?
https://uk.rs-online.com/web/p/voltage-regulators/6868726

 

I vote Crowbar-Circuit, but maybe not with an SCR, where the only adjustment is changing a Zener-Diode.
The LM323 might barely squeak-by, but it has ~2.5-Volts of Forward-Voltage loss.

I think a stout P-Chan FET, and a Voltage-Reference like the TL431, and a Trim-Pot is the way to go.
.
.
.

 

Below is the LTspice simulation of a crowbar circuit, which trips an SCR to clamps the output to zero if the voltage tries to go above the trip point.
The trip point is accurately determined by the TL431 programmable voltage reference, and is adjusted by pot U3.

 

Hi all and thank you very much for the many replies! This is awesome and greatly appreciated

As for

Ae2576-5t is good

I can't seem to be able to find AE Semiconductors website and the google results always point to some reseller who wants me to create an account to get to the datasheet. That's just utter BS and I refuse to purchase components like that. I appreciate the thought and I believe that you are possibly very right in it being a good device.

In regard to crow-bar circuits, I have to be very careful shorting anything, since the entire device is powered by a massive 4S LiPo.
So there needs to be a fuse that blows, could be a poly-fuse?

I have already designed a custom circuit board and figured for the (important) voltage regulator part, to use a dedicated DC/DC converter module. But that is the device that doesn't like floating ground and will just switch VIN through. Already blew a small Arduino, which doesn't matter, as long as I won't blow the Jetson Nano $$$$ (and hard to replace due to chip shortage)
Like this one from Ali-Express: https://www.aliexpress.com/item/1005001703274616.html
(DC/DC 12V 24V Step Down to 5V 5A 25W Voltage Buck Converter Regulator Transformer for Car Truck Vehicle)

Initially I thought of using a Linear Voltage Regulator similar to 7805. However I didn't want to waste my battery charge away in heat, especially coming down from a 4S (16.8V at full charge) Battery assembly.

In the above crow-bar circuit (@crutschow) I already see the use of a fuse, but it also uses a BJT. I don't think I can switch a laod of up to 3 or 3.5A this way. And that is where my difficulty lies.
I have a "voltage regulator" that ideally gives me 5V (and up to 5A), but it also could spill the Input Voltage to the output side if the ground ever gets "floaty". Now in that case I wouldn't be able to use the circuit from @LowQCab either, as that would have a significant impact on the Reference voltage. Image the voltage goes from 5V to 16.8V, then the reference would also jumps.
Yes, I guess the voltage reference should ideally be taken from before the (unreliable) voltage regulator. But since we are dealing with a 4S battery, VBat ranges from anywhere between 16.8V down to 14V on a LiPo and even further down on a Li-Ion battery pack, as those can be discharged a bit further.

So I guess a small variation of the circuit from (@crutschow) where a mosfet is switched would do the job? But how would I use that to switch a mosfet?
(Also what software are you using, and does it require a license, or is there a decent free version?)

Thanks guys!

 

Also what software are you using, and does it require a license, or is there a decent free version?

LTspice is widely supported simulation freeware from Analog Devices. No licence required.

 

I have a "voltage regulator" that ideally gives me 5V (and up to 5A), but it also could spill the Input Voltage to the output side if the ground ever gets "floaty".

How about wiring the regulator so that the input 0V and output 0V are joined separately to the regulator 0V? That way, the regulator 0V reference cannot become detached from the combined input and output 0V.

 

I already see the use of a fuse, but it also uses a BJT. I don't think I can switch a laod of up to 3 or 3.5A this way.

It does use a BJT but that's just to provide the small trigger current required by the SCR.
The current is carried by the U1 25A SCR, so no problem with 3.5A.

The fuse can be any kind you want to use.

 

@crutschow
Ah ok, I see. So correct me if I'm wrong, please:
1. I will just connect my load (Jetson Nano) in parallel to U1.
2. The potentiometer seems to show a setting of 0.498 aka 50%? So resistance about 2k5? However I will still add a potentiometer, as it makes sense, due to tolerances, etc...
3. If my voltage source (here V1) ever changes and goes above the 5.1V limit, the reference for U2 is modified and it will turn off the supply to Q1, which takes the "enable" off U1? Correct, kinda?
4. In terms of wattage for the resistors, I plan on using 0805 sized chip resistors, with 1%, 1/4W that would do, as the current flow is very marginal.

Thanks heaps, mate! Appreciate it.
Now I just need to get some of these parts to the bottom of the world here in NZ (after confirmation of above)

 

Also what software are you using

As Alec_t noted I'm using LTspice which is free from Analog Devices.
It's one of the best free analog simulators available.
It has a somewhat steep learning curve, but there are good tutorials, and many sample circuits to help with that, along with help from us on this forum.

 

3. If my voltage source (here V1) ever changes and goes above the 5.1V limit, the reference for U2 is modified and it will turn off the supply to Q1, which takes the "enable" off U1? Correct, kinda?

It's actually the opposite.
U1 and Q1 are normally off.
When the Ref voltage rises above 2.5V, U1 starts to conduct cathode current, turning on Q1 and triggering the SCR to clamp the output to ground.

4. In terms of wattage for the resistors, I plan on using 0805 sized chip resistors, with 1%, 1/4W that would do, as the current flow is very marginal.

Yes, those will be fine.

 

If you're worried about 'punch-through' of the 16v to the 5v side and/or floating grounds you need an isolated DC-DC converter with an integral output side crowbar though a properly designed isolated converter would never be able to output a higher voltage by design.