HIP4080 H-bridge mosfet driver 5V logic interfaced to ESP32 3.3V logic ???

[dazz1]

Hi
I want to interface an ESP32 to a HIP4080. 
HIP4080 is 5V logic.  ESP32 is 3.3V.

I have checked the datasheets and in theory, the logic levels are compatible - just.
The ESP32 min output Hi is 2.64V.   The HIP4080 min input Hi is 2.5V.  A 0.16V margin.

I would prefer not to use a 3.3V/5V level shifter if I can avoid it.

Does anyone have experience interfacing the ESP32/HIP4080??

[redkitedesign]

It might work on the lab-table.
However, if your 3.3V is on the low side (e.g 3.23V), and the 5V on the high side (5.1V) you have just lost 0.17V of margin... Maybe a little less, because it scales.

Conclusion: its gambling, not engineering

[T3sl4co1l]

It's fine, you have the overlap.

It'll still work if they weren't overlapped -- probably -- but how well, isn't something you can know for sure from these datasheet values, and relying on that active word "probably", would be a poor idea for design purposes.  For example, the 70% threshold of 74HC CMOS (3.5V) exceeds the best-case 3.3V supply, so would not be a reliable combination.  It would likely work in a typical case (I believe HC thresholds are usually close to 40-50%?) -- but you can't guarantee that for every unit produced, under all operating conditions.

I would be shocked if the ESP's output isn't very damn close to 3.3V even with a modest (some kohms?) pull-up/down on it, it's just that they rate it at whatever current they do, and the worst case margin in that condition is quite pessimistic.  Probably, anyway; I didn't look at the datasheet.

But do be mindful of anything that will reduce that margin.  Noise, indeed; which especially means CM noise around a power converter -- use solid ground plane between them.  Variation in supply voltage; maybe you want to adjust the 3.3V regulator upward a little to increase that margin (what's the ESP rated for, 3.6V max? 3.4-3.5V would be safe enough).  Or shut down the HIP preemptively if the 3.3V supply starts to drop -- add a power monitor chip perhaps.  Which is a good idea anywhere you're doing something fairly mission-critical, like pulsing an inverter with a MCU, anyway.  Not to mention a keep-alive or watchdog timer, which might include external (analog) circuitry to lock out the inverter, in hardware, in case the MCU stops responding, or responds erroneously.  For that matter, if you're adding external logic, you might as well get the level shift for free (use 74HCT at 5V?) and avoid the problem entirely.  So there are many options to choose, as far as prudence goes.

Tim

[dazz1]

Hi
The application is a 48V bipolar driver for the clock seen here: https://youtu.be/OWvz3GIeePo
I need to provide a bipolar 48V pulse every one minute.
The plan is to interface the HIP4080 to the ESP32.  That will connect to my network via Wifi or with GPS.
I will be able to make adjustments (eg. daylight saving) from my mobile phone or desktop PC.

I like adding new tech to old tech while maintaining character.  The new tech will be invisible from the outside.

[dazz1]

It's fine, you have the overlap.

It'll still work if they weren't overlapped -- probably -- but how well, isn't something you can know for sure from these datasheet values, and relying on that active word "probably", would be a poor idea for design purposes.  For example, the 70% threshold of 74HC CMOS (3.5V) exceeds the best-case 3.3V supply, so would not be a reliable combination.  It would likely work in a typical case (I believe HC thresholds are usually close to 40-50%?) -- but you can't guarantee that for every unit produced, under all operating conditions.

I would be shocked if the ESP's output isn't very damn close to 3.3V even with a modest (some kohms?) pull-up/down on it, it's just that they rate it at whatever current they do, and the worst case margin in that condition is quite pessimistic.  Probably, anyway; I didn't look at the datasheet.

But do be mindful of anything that will reduce that margin.  Noise, indeed; which especially means CM noise around a power converter -- use solid ground plane between them.  Variation in supply voltage; maybe you want to adjust the 3.3V regulator upward a little to increase that margin (what's the ESP rated for, 3.6V max? 3.4-3.5V would be safe enough).  Or shut down the HIP preemptively if the 3.3V supply starts to drop -- add a power monitor chip perhaps.  Which is a good idea anywhere you're doing something fairly mission-critical, like pulsing an inverter with a MCU, anyway.  Not to mention a keep-alive or watchdog timer, which might include external (analog) circuitry to lock out the inverter, in hardware, in case the MCU stops responding, or responds erroneously.  For that matter, if you're adding external logic, you might as well get the level shift for free (use 74HCT at 5V?) and avoid the problem entirely.  So there are many options to choose, as far as prudence goes.

Tim

The ESP32 datasheet says min Hi output is 2.64V. 
HIP4080 datasheet says min Hi input is 2.5V. 

I plan to do some tests to measure an ESP32 loaded with a HIP4080.   If the ESP32 outputs stays above 3V, that should be enough margin for a one-off project.    I wouldn't do it for a million units.

[T3sl4co1l]

Ah, that should be very easy then. A bootstrap gate drive hardly seems worth it, a handful of BJTs would do the job -- if your coil is similar to that in the video.

This sort of thing for instance:



Add clamp diodes for the inductive load.  Heck, 1N4148s are enough.

Tim

[dazz1]

Ah, that should be very easy then. A bootstrap gate drive hardly seems worth it, a handful of BJTs would do the job -- if your coil is similar to that in the video.

This sort of thing for instance:



Add clamp diodes for the inductive load.  Heck, 1N4148s are enough.

Tim

I started with a discrete design but the HIP4080 is a cheaper, easier solution.  Specifically, it makes the software development a lot simpler because I won't have to write protection software. 

I am going use a diode voltage multiplier to step from 12V to 46V (48VDC nominal), charging a relatively large junk box capacitor.    The multiplier has an output impedance of about 20kohm so it can drive a short circuit indefinitely without releasing the magic smoke.   I won't need a fuse on the 46V supply.    The high impedance should also reduce noise on the 12V supply line.

So the use of an ESP32 and HIP4080 will simplify both the hardware and software.  If I can avoid logic level shifters, that would be great.

[T3sl4co1l]

$6.60 in singles isn't exactly cheaper...

I guess you have some on hand already though.

You can further limit short-circuit current by increasing the capacitor's ESR externally.  Also means no need for fast switching, and no danger of common mode noise, at least unless a very, very poor layout is used.

Tim

[Benta]

I've worked with the HIP4080 before (back in the Intersil days), and it's a fantastic device wtih great performance.
But exactly due to its performance, it's challenging to work with.
Don't even think about using a solderless breadboard. Even a perf board will likely give you problems.
I highly recommend sticking to the layout example shown in th data sheet and making a real PCB, even just for testing/experimenting.