Your original complaint was that 12V was always being applied to the load.

Why are you measuring voltages between the MOSFET terminals? Measuring with respect to round is sufficient

We only need to know what the voltages on the gate and drain are. If you have things connected properly, the source will always be ground.

The drain-source voltage is an order of magnitude higher than it should be. 8A * 7.3mΩ = 58.4mV, you measured 500mV.

I measured all the voltages, unsure of which ones would be useful to anyone. I didn't realize the extra data would be a distraction. I guess now that you say it that way, you maybe just need to know gate-ground (which is gate-source) and drain-ground (which is drain-source).

500mV drain-source was on the full scale; if I need to get that number more precise, my voltmeter has a <300mV option. Does that unusual reading indicate a particular failure that I should investigate differently?

 

It rather acts like the source and drain are interchanged, even though the board looks correct.
Is it possible that the MOSFET you have has a different pin configuration than the one on the breadboard?

Before building another, check for any shorts between all the traces that should be separated.

 

It rather acts like the source and drain are interchanged, even though the board looks correct.
Is it possible that the MOSFET you have has a different pin configuration than the one on the breadboard?

Before building another, check for any shorts between all the traces that should be separated.

Just finished building another. Behaves the same. I let the board get to 150F before I pulled the plug. And this one is reading ~600mV drain-source.

It's certainly **possible** that the MOSFET I'm using is a different pin configuration. But I read the label on the bag and looked up the datasheet specifically to make sure I had that right. Then I double-checked by reading the label on the device itself. It's definitely possible I've gotten it somehow backwards. Just thinking in writing: it wouldn't be upside-down, like on the wrong side of the board -- that would switch 1 with 3, and it's behaving like 2 and 3 are switched.

I can pull this MOSFET out and run leads to the board, switching pins 2 and 3. That would certainly test that.

 

Are the ones you used on the breadboard exactly the same as on the PCB?

 

EDIT: i was wrong here, ignore this post. Thanks crutschow for pointing that out.

You have the drain and source backwards. Look at the board, the negative terminal of the supply goes to the drain and the load is connected to the source. The kickback diode inside is conducting (which is why you have 600mV across there). You gotta somehow switch the drain and source around.

According to this datasheet anyhow
https://www.infineon.com/dgdl/Infin...N.pdf?fileId=5546d462533600a401535619ea0e1e68

 

You have the drain and source backwards.

No.
]If you read all the posts, you will see that this has been gone over.
The top connector is the supply with the left contact ground, and the bottom connector is the load, so the source is connected to the supply ground, as required.

 

I think that the MOSFET is dead.
If I were making this I would have added another diode across the load, just in case the load or its cables are inductive.

 

The D-S measurement with the switch open makes no sense.
There is no way that voltage should be 0 if the load is connected.
It should be at least 11V as you suggested.

Are those voltages all positive in the direction indicated.?

 

Ok, so voltage testing semi-successful. I slipped with a probe and made sparks, after which the behavior of the board changed, so I think I'll have to build out another one to do any more testing.

But here's what I have so far:
Input line seems to be doing the right thing -- it's 11V anywhere I expect it to be.
The switch seems to be switching -- it's 0 when there's no magnet near it, and 11V when I put the magnet on it.

For the transistor:

	Switch Open (no magnet)	Switch Closed (magnet)
G-D	0	10.5
D-S	0 (should be 11?)	0.5
G-S	0	11.1

View attachment 278874

The load is a heated vest, and it definitely starts heating up as soon as I connect it, with no magnet on the switch. Also, small correction to above -- the board gets hot (~95F) when the load is connected. With no load connected, it doesn't appear to be passing any current, so it does not heat up.

It's not clear to me where the reed switch is, so I redrew the circuit.
Here's what I think it should be.
1. S1 (PWR) is supposed to remove power from the whole circuit. It needs to be 12A switch to support an 8A load.
An alternate location for the S1 (PWR) switch is between the +12 and the load, with one side of the reed switch connected directly to +12.
2. The N.O. Reed switch only controls the mosfet, so it passes only a small amount of current when closed.
3. R1/C1 provides some switch debounce at the gate input.

EDIT: labeled high current traces

 

When you measured 500mV between drain and source, was the drain positive, or was the source positive?
i.e where was the red lead? Was there a negative sign on the display?

 

I think part of the confusion arises because traditionally red denotes positive. Red ain't necessarily so with this board design.
Where did you source your FETs? There's a lot of fakes around, apparently.

 

A lot to reply to, here -- thanks for all the inputs. Let me just roll through the list, here:

- Bad FET is a possibility. I got them from Mouser, but they've been sitting in a cardboard box in my office for a year or two probably. But the fact that two of them in different boards behaved the same makes me think there's something else going on here. The one installed in the breadboard is identical to the ones that I've been putting on the PCBs.
- Crutschow and Ian0: I'll double-check those voltages, and will make sure to orient the + and - leads appropriately.
- Alec_t -- on the board picture, red and blue are the Eagle-selected colors for top and bottom traces, respectively. Understand that doesn't help keep the problem clear.
- EETech00 -- thanks for the redesign. I see you've change around the resistor network a little. Is that significant? Or is the biggest change the addition of the master switch?

 

I see you've change around the resistor network a little. Is that significant? Or is the biggest change the addition of the master switch?

That depends on the intended functionality of your design.
As I understand it, the reed switch is supposed to control the mosfet, and the mosfet controls current thru the load.
The reed switch isn't shown on your circuit schematic, so I made an attempt to place where I though it should be. S1 appears to be a small mechanical power switch of some kind, so I also placed it where I thought is should be. There's no need for R1 to be 1k and changed it to a size for a safe current limit, and the pull down R2 is usually placed at the mosfet gate to ensure the mosfet turns off. I added C1 to smooth the reed switch contact bounce since the mosfet is carrying a significant load. C2 is supposed to represent an existing supply filter.

 

Bad FET is a possibility.

From your measurements in post #17, your initial complaint of 12V being applied to the load regardless of whether the reed switch is closed has resolved itself.

The current problem seems to be what you're calling overheating when the MOSFET is on and you're measuring a drain-source voltage drop that's an order of magnitude larger than it should be. With a Vgs of about 10V, on resistance should be 7.3milliohms or less so the drain-source voltage drop should be more like 60mV, not 500mV.

sitting in a cardboard box in my office for a year or two

Have you been handling them properly?

 

From your measurements in post #17, your initial complaint of 12V being applied to the load regardless of whether the reed switch is closed has resolved itself.

Just re-tested, and I'm definitely still getting power to load when the reed switch is open. The board heats up, and so does the load vest.

The current problem seems to be what you're calling overheating when the MOSFET is on and you're measuring a drain-source voltage drop that's an order of magnitude larger than it should be. With a Vgs of about 10V, on resistance should be 7.3milliohms or less so the drain-source voltage drop should be more like 60mV, not 500mV.

I can confirm that voltage drop is there, and it's consistent. It's possible that the vendor is providing an inaccurate value for their load current, but it would uncharacteristic for that to be wrong by an order of magnitude.

Weird thing, and maybe this is relevant. The load voltage appears to be negative. With the black probe on ground and the red probe on the positive side of the terminal, I was showing -10.6V. (No magnet applied, and switch showing high resistance.) Is it possible I've just swapped battery terminals or something?

Have you been handling them properly?

They've been sitting quietly in a cardboard box, inside the little foil packs they come in from Mouser. The house doesn't generally get above 80F, or below 60F, with humidity in the 40%RH to 65%RH. It doesn't seem like that should make them fail, but it's a possibility.

 

Weird thing, and maybe this is relevant. The load voltage appears to be negative.

Perhaps your circuit is oscillating? That might also cause unexpected board heating. I note that your circuit doesn't include supply decoupling capacitors.

 

Holy ****. Yup, that appears to be it. I've swapped the battery terminals. The one labeled + was actually the -. If I just rotate the head around the "wrong" way, it works as expected. Magnet triggers the reed switch which gates the MOSFET, and the board doesn't even overheat. I knew it was simple.

... A little further detail is the wiring between the battery pack and the terminal; I appear to have gotten confused in routing those wires and ran them to the wrong terminals on the board.

I believe Ian0, dl324, and Alec_t, and crutschow deserve credit for solution, in pointing out that I needed to pay attention to the sign.

Thank you, everyone!

 

I was going to say it sounds like a reversed battery connection, but you already found it.
When measuring DC voltages, polarity is as important as the magnitude.
Glad you got it working.

 

I was going to say it sounds like a reversed battery connection, but you already found it.
When measuring DC voltages, polarity is as important as the magnitude.
Glad you got it working.

We were all kind of latched on to "D and S just seem swapped!" which was such a compelling case. Really grateful for all the help; would not have caught that without a bunch of people slowly coming to the same realization.