H-Bridge with level shift - MOSFET gate voltage issues?

You need to redesign it, no good with oveestressing Vgs

Use level shift with capacitive method, if you have guaranteed minimum duty cycle

Otherwise, add gate to upper part for reshaping of the drive signal

Cheers

Klaus

maximum load current?

You could just buy a driver chip, IXDN602 or LTC4442 or one of those.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Oh - about 1A tops

--
Email does not work

Wow - I can buy that for little over a quid (first one) - a lot less than the parts to do it the "hard way")

2A, 35V - that's more than enough.

Thanks very much :)

--
Email does not work

It's fun to design stuff, but then some jerk usually comes along and designs a cheap IC. So we have to move up the abstraction stack.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

better read those specs carefully, the output resistance of the first one will dissipate several watt at 1A, that's a lot for such a small package.

for something simple, old, and cheap how about L298

Right, calculate the dissipation. Those are intended to be mosfet gate drivers, but often work fine as load drivers. One can parallel sections for more current.

The garden lights may be LEDs that don't need a lot of current.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

You need a high-side driver and/or level shifter. Don't try to put 30V on the gates.

HIP4081 is not a bad choice - just add the external MOSFETs.

Clifford Heath.

In real life it would be OK. Most fets blow out around 75 volts.

But the resistors will dissipate about 5 watts each, more if the input ever sticks high or low.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

that would major overkill for a few amps at a few kHz

He's only gonna build two, not 2 million. The price doesn't matter, so why not overkill? Anyhow they're only a couple of bucks at Aliexpress.

TR1 was supposed to be the level shifter - or at least that's what I was led to believe in the various "H-Bridge for dummies" guides I found...

I did see some warnings about gate voltages and some vague mention of zeners but I did not see a practical example anywhere and I can't quite see how to protect the gates but leave them functional.

I'd like to know how to solve this, even if I use one of the many excellent driver chips that have been suggested.

We might still do it the "hard way" as this is a project for education (my son and clearly me too ) rather than solving a real world problem.

========

After that, we're be on to "NeoPixel" strings to complement these lights and have some real fun :)

I envisage a number of PiZeros in small waterproof boxes local to each string set, maybe even as much as one high quality 50V DC SELV supply back in the shed to keep the current and volt drops down (but still safe) and local DC-DC converter modules in the boxes for both the 30V lights and the NeoPixel strings. I'll knock up a simple client-server protocol over WiFi. We'll deploy everything at Haloween next time for fun and keep over Xmas. The NeoPixels at least won't need drivers (well maybe a trivial 3.3-5V shifter - I forgot the data bus specs).

We're not planning a lightshow, just some tasteful fun!

--
Email does not work

The full 30V goes into the stages of the complementary pairs from TR1, so all the transistors except TR1 will have Vgs max violation

Cheers

Klaus

That was ocurring to me too when I saw one of the transistors had a max gate V of 20V relative to drain(IIRC).

Do you happen to know how this is solved by discrete design - have I just found a bad example of a level shift circuit (perhaps one intended to level shift to a more modest load voltage)

--
Email does not work

so all the transistors except TR1 will have Vgs max violation

The gate drives gets more complicated. Essentially, you can't tie the gates together and have to provide the voltage swings appropriate to each gate,

The first stage is straight-forward enough.

You could put in a +15V rail, and drive the gate of Tr3 directly from the 1

00R returned to that rail, but it would be cuter to use a bipolar NPN with its base tied to the +15V rail, return R1 to the emitter of that device (gi ving you a 0V to 14.4V swing on the gate of Tr3) while a second 100R resist or from the collector to +30V would give you a +30V to +15V swing on the ga te of Tr2. 91R might be a better value for the second resistor - it would keep the bip olar cascode out of saturation. A 13V swing would be quite enough to turn t he part on hard.

The output stage transistors needs higher currents, and you don't really wa nt resistive pull-ups there, so it gets messier.

I once used a capacitor to move enough charge through a cascode into (or ou t of) a gate to guarantee rapid turn-on and turn-off - gate charge versus gate-source voltage is documented on MOSFET data sheets - while relying a r esistor to keep the gate on or off. There are quite a few other options.

--
Bill Sloman, Sydney

Thanks Bill.

It does sound that the correct solution is non trivial and a driver chip to a pair of power MOSFETs is warranted here.

It's an interesting case, trying to bridge drive at 30V from a signal

1/10 of that.

--
Email does not work

Why an h-bridge? You should be able to switch/dim a string of LEDs with a single logic-level-gate mosfet.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

antiparallel strings: different polarity, different color