10A H bridge design

[OM222O]

Hi everyone,
For a robotics project, we're using 12v, 8A DC motors and I can't find a suitable H bridge (most off-the-shelf modules are either out of stock or way too big for what we need), so I'm planning on building one.
We want to control it with 2 signals (PWM and direction) like most common designs. I had a look on google but only found ways of not making an H bridge as everyone was pointing out someone else's mistake. Is there a base design I can start from or what safety components should be looking to add (reverse polarity diodes, MOSFET dead time, etc.)? We're using an FPGA as a controller so adjusting frequency, dead time, etc can all be done in hardware.

[rstofer]

At startup, that 8A motor will draw a lot more current.  Measure the DC resistance of the motor and use Ohm's Law to figure the peak current.  Maybe there is a spec for 'locked rotor' current.

One big design issue is the MOSFET driver.  It needs to drive the MOSFET through the transition period as fast as possible.  It is during this switching period that the MOSFET is operating in a linear region and heating up.   Simply hanging a logic level MOSFET on a logic output seldom works very well.

I would just search the market for 50A H-Bridge and grab something off of Amazon.  Maybe something like this:

https://www.amazon.com/Controller-H-Bridge-High-Power-Single-Channel-Full-Bridge/dp/B07SH6BX97

It is an absolute certainty that if I value my time at all, I wouldn't spend it recreating something I could simply buy.  My time would be better spent working on the firmware.

[OM222O]

If you open your own link, you'll see the "currently unavailable" description. Even if I manage to buy 1 for the testing phase, I highly doubt I'll be able to buy it again in the future (we need 4, one motor per wheel). Time is essentially free as we're students so I don't mind making my own.

[rstofer]

If you open your own link, you'll see the "currently unavailable" description. Even if I manage to buy 1 for the testing phase, I highly doubt I'll be able to buy it again in the future (we need 4, one motor per wheel). Time is essentially free as we're students so I don't mind making my own.

My incantation of Amazon shows 3 in stock and delivery sucks (end of May, worst case).  There are many others...

The hard bits are guaranteeing sufficient Vgs and Ig to switch quickly.  That's the job of the MOSFET gate driver but you need to provide sufficient voltage for the gate driver.  If the datasheet shows 10V Vgs then somehow you need to have that much voltage available and with sufficient current capability.

Don't even try to use an N-channel device in the upper position.  You need a gate voltage higher than the upper rail to switch the device.

Maybe there is something in here worth knowing:

https://www.ti.com/lit/ml/slua618a/slua618a.pdf

Mouser has thousands of these n-channel devices in stock

https://www.mouser.com/datasheet/2/196/Infineon-IPD200N15N3%20G-DS-v02_07-EN-1226206.pdf

There are many other selection parameters including p-channel.  I didn't check for stock on p-channel and I just went for the TO-220 package figuring the heat sinking would be easy.

Even though the threshold voltage for the device above is as little as 3V (typical), you will notice that most of the specs call for a Vgs of 10-20V.

This may give you a step up on figuring gate current:

https://www.microsemi.com/document-portal/doc_view/14697-making-use-of-gate-charge-information-in-mosfet-and-igbt-data-sheets

You are right, there are more bogus designs on the Internet than designs that might actually work.

Don't let the rise and fall time become a significant proportion of the minimum ON time.  Again, this has the device in the linear region for a more significant percentage of the time.  That's one of the reasons that many H-bridges are limited in PWM frequency.  They switch too slow and would spend a significant percentage of the period in the switching region at higher frequencies.

What you want is switching in zero time.  This takes infinite energy but that's still the goal.  Absolute minimum switching time.

[OM222O]

Many step down converters in computer VRM (voltage regulation modules) use "power stages" where the high side fet , low side fet and the driver are combined to one IC. I think that'd be the ideal choice (as opposed to separate fets and driver etc). Most of them are also rated for much higher currents than what we need (minimum 20A I think).

[rstofer]

Yes, the fully integrated drivers are a great way to go.  But...  Are any in stock?  I couldn't buy Allegro drivers in the best of times.

This device looks interesting but it requires external n-channel devices:

https://www.allegromicro.com/en/products/motor-drivers/brush-dc-motor-drivers/a5929
https://www.allegromicro.com/-/media/files/datasheets/a5929-datasheet.pdf

There are others...

[OM222O]

mouser and rs both seem to have a decent stock for these:
https://www.mouser.co.uk/c/?type=Integrated%20Power%20Stage

I'll try to read the datasheets and put together a design a prototype using these power stages, thanks for the help.