Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Protection diodes for H-Bridge MOSFETS

Status
Not open for further replies.

Speakerguy

Active Member
I'm building an all N-channel H-bridge. I figure I should strap some diodes around the FET's so the body diodes don't blow, yes?.

I have 1N4002, 1N4148, and 1N5819.

Should I use the Schottky's because they kick in faster and at a lower voltage or will either of the others be fine?
 
Last edited:

dknguyen

Well-Known Member
Most Helpful Member
I'm building an all N-channel H-bridge. I figure I should strap some diodes around the FET's so the body diodes don't blow, yes?.

I have 1N4002, 1N4148, and 1N5819.

Should I use the Schottky's because they kick in faster and at a lower voltage or will either of the others be fine?

1N4148 may suffice if it can handle the current but schotky would still be better (Schottkys have soft recovery which is another aspect along with recovery time). 1N4002 is likely too slow to do much and you might as well just use your body diode in that case.

You could also just rely on the body diode if the circuit is not demanding and the body diode is sufficiently fast.

You don't really care about the body diodes blowing (well you do), but the reason is that if they blow the current spike fries the rest of the MOSFET. And if the body diodes do blow you just lost the mechanism that is keeping the rest of your MOSFET silicon from misbehaving (that produces the body diode as a side effect).

What are your current levels?
 
Last edited:

Speakerguy

Active Member
Mkay, I'll use the Schottky's. These are some heavy duty TO-220 FET's - the kind that if you want something bigger you get a it in a TO-247 case :) They'll be on a solderable breadboard but the traces won't handle the current, so I'll be point to point wiring the power connections straight from the power supply.
 
Last edited:

dknguyen

Well-Known Member
Most Helpful Member
Yeah, I don't think a 1n4148 would survive that. Make yer connections short!
 
Last edited:

dknguyen

Well-Known Member
Most Helpful Member
You can rely on the body diodes if while you are using a resistive load unless you got crappy traces and wiring introducing lots of inductance and really fast fall times

But yeah...for current levels you might use those FETs at, a 1n4148 wouldn't cut it. The body diode characteristics look quite good- you might be able to rely on them if you can't find large schottky diodes (depending on your inductances, current levels, and your fall times). An RC snubber to take the initial pulse in combination with the body diodes to do the rest might work if you can't find the right diodes.
 
Last edited:

mneary

New Member
My favored approach is to use an RC snubber to give the body diodes enough time to work. It's an extremely short time, and usually not much energy.

The hard part is understanding the pulse capability of a resistor. ;)
 

Speakerguy

Active Member
And tuning the darn thing. When I was at Cirrus part of my job was getting PWM amp boards to be EMI compliant. Tuning PWM amp chip output's RC snubbers was fun, I used an oscilloscope and an FM radio as my tools :)
 
Last edited:

mneary

New Member
Ever encounter an action request where 'your design fails EMI' and I reply that's not my design, my design had snubbers. They say 'it worked when I removed the snubbers so they must not matter' and I say then why does it fail EMI? My design had snubbers and it passed.
 
Last edited:

mneary

New Member
Snubbers were deemed an unnecessary 0.045 expense. :p

Until we failed EMI. :mad:
 
Last edited:

Sceadwian

Banned
I picked up a high current FET from NTE that's designed for inductive loads no freewheel diode required as long as the kickback voltage isn't above breakdown, a modest bypass capacitor should be enough to soften the kickback to avoid breakdown if your switching rate isn't too high.
 
Last edited:

Speakerguy

Active Member
I saw so much retarded @#$% from customers it was unbelievable. I mean, the prototype HAD to have clip-on test points in between the chip and the snubbers, how else could we test the outputs? And why does it now fail EMI?

What was worse was the idiot who got layed off before me (he deserved it based on competency). I was getting back layouts he had approved that had such tremendous **** wrong with them it was mind blowing. Ground plane split right beneath the output traces? No problem! Those return currents will just magically find a way back. How do you tell a customer "Yeah, we told you that, and you spun boards and went through testing, but that guy was an idiot and we fired him"?

Meh. I quit that job after 3 months. I was promised at least 50% PCB layout/design. Instead I coddled customers and wrote data sheets, too little real engineering.
 
Last edited:

Hero999

Banned
Many MOSFETs are avalanche rated and will be fine even if they breakdown, providing the stored energy in the inductor doesn't exceed the maximum rating.
 

Ubergeek63

Well-Known Member
I saw so much retarded @#$% from customers it was unbelievable. I mean, the prototype HAD to have clip-on test points in between the chip and the snubbers, how else could we test the outputs? And why does it now fail EMI?
Sounds like my associate, while he over designs snubbers, he sees no problem connecting 5-20 little "ground planes" together through 0.020 vias and 0.025" traces and calling it a "ground plane".

I keep trying to tell him what he has is not a ground plane but a multiphaser antenna array but it takes the owner of the company getting frustrated trying to explain the theory and just tossing up his arms and saying " I don't care I want it this way" to get him to redo it right!
 

Speakerguy

Active Member
LOL. First thing I ever learned about ground planes was ground plane meant ground PLANE. One layer, Vias/Through Hole components OK, but signal routing better just be a jumper trace and very, very, very short if you had to put it there at all. And done while being conscious of where return currents were flowing based on top layer traces.
 

Ubergeek63

Well-Known Member
My first engineering supervisor was the same way... a 1.9" diameter "shield" with feed through caps, fine so far.

However, the "shield" is a PCB and there are wires soldered to the inside down to the PCB and then through hole EMI beads soldered to the board and turret pins to solder wires to leaving very little ground plane for a shield.

And if that was not bad enough, the main ground connection to the case (a hermetically sealed steel aircraft instrument case) was through a 3" wire and a brass clip. Can you say antenna? I knew ya could!

While I could not rip up my boss's (the company's EMI "expert") shield, I did get away with specifying a multi fingured berilium copper EMI gasket be soldered to the perimeter of the board so it at least had a solid connection to the case. I probably would not have gotten permission to do that had the cheesy brass grounding clip not kept breaking.
 
Last edited:
Status
Not open for further replies.

Latest threads

Top