Design review: H-bridge

Other Hardware Motors, Mechanics, Power and CNC
1

I'm just going to post my whole schematic cause its far to difficult to do anything else.

This is fairly complex and you probably need to look up the spec sheets to fully understand parts of it so I don't expect most if any to bother.

I'm basically looking for any mistakes like missed connections, bad logic, and so on. I've reviewed it myself several time and I think I have them all.

The tps2111 is a power multiplexer, its being used in a slightly unconventional way to how it was intended, there is a little power blip should you have the USB and external power (POW) at the same time and you pull the POW line, which is why I have the large caps to bridge that.

board_motor_drive V2.sch (552 KB)

2

Can you post it as a PDF ? Most people don' t have anything to open a .SCH file.

3

Eagle PCB is what it was made with.

I can export as a dxf file (autocad)

I exported to clip board and got a image file, probably not the best quality.

If you install eagle PCB you can view it as it was built.

motor drive board.png
motor drive board.png4041×1851 241 KB

4

You need to run ERC and DRC, you've got a lot of missing joints.

5

Although it tells you you are missing them, there not actually needed. Electrically they are all connected and it knows this, there more of a visual aid to be able to see the electrical connections at T's and area's where they graphically cross a lot of other parts.

I'm not even sure why it gives me a error on the +5V line, they are clearly all connected, and is confirmed on my .brd file.

I find the DRC to be quite unnecessary and stupid at times.

6

Without a blob there is no guarantee the nets are the same net. ERC will list all
the cases where one net ends on top of a wire from another net.

The fact that lots of blobs are missing suggests something drastic is wrong, like
failing to use the net tool and using the wire tool instead.

It certainly suggests an ERC check hasn't been done.

7

I'm just curious, why is this design so complex ? (the post title refers to H-bridge but clearly there is much more than that)
What is the purpose of this design ?

8

Your right I used the wire command, its how I first started and it kind of just stuck. It makes the same electrical connections which is why I haven't bothered to change.

The only way you can make a graphical connection without a electrical connection is if you say no to the merge prompt that comes up when you connect 2 different nets.

I'm just curious, why is this design so complex ? (the post title refers to H-bridge but clearly there is much more than that)
What is the purpose of this design ?

There is a H-Bridge design there. :wink:

Taking the time to separate the h-bridge from the whole schematic was not something I wanted to waste time on.

Here is a updated one, more or less the same except I changed a few small mistakes on netnames, and added the not so relevant junctions. Any other warnings I didn't think mattered.

board_motor_drive V2.sch (555 KB)

9

As .png please - that .sch doesn't load in my version of Eagle, many people don't have eagle...

10

I'll try later.

Its a free download and install. :slight_smile:

11

I have it, I don't have a compatible version to yours and I have 300+ board designs that
currently are known to work with my version and I don't want to break them just
to look at your picture!

Post non-proprietry opne formats on a public forum if you can, after all these threads will
be archived for decades to come.

12

Nothing about this is really proprietary, its mostly derived parts of other circuits and recommended configurations from spec sheets all thrown together on one sheet. Aside from the h-bridge which in the end was also a modification on a different design I found.

If anything was proprietary, it would be in the routing which I'm not sharing.

h-bridge.png
h-bridge.png4041×1851 237 KB

13

harddrive123:
Nothing about this is really proprietary, its mostly derived parts of other circuits and recommended configurations from spec sheets all thrown together on one sheet. Aside from the h-bridge which in the end was also a modification on a different design I found.

If anything was proprietary, it would be in the routing which I'm not sharing.

Eagle file formats are binary and proprietry, not open-source. At anypoint
CadsoftUSA could decide to start charging for all versions of Eagle, they own the format.
They could (realistically) be bought out and closed down by a competitor...

Courtesy wikipedia:

Portable Network Graphics (PNG, pronounced "ping"), is a raster graphics file format that supports lossless data compression. PNG was created as an improved, non-patented replacement for Graphics Interchange Format (GIF), and is the most used lossless image compression format on the Internet.

PNG is also an international standard. It will be supported decades/centuries into the future.

....anyway.... back to the plot.

I see n-channel / p-channel bridge, presumably the PWM input is to the low-side
switches via the FAN3122 drivers (which are overkill, 100mA drivers would suffice
for those small Si4564's - still give 100ns switching).

The top-side p-channel circuit isn't engineered for PWM, those 10k pullups are woefully
slow, but if just used for direction and braking it doesn't really matter (I personally
would make them 1k anyway).

The layout is important too, low inductance paths to decoupling. Adding more decoupling
on the power rails would be good, 10uF ceramic is a start, but more electrolytic
bulk decoupling can reduce the noise on the rail - large voltage spikes on the rail could
overload the Vgs rating on the MOSFETs, note. Some people add TVS diodes to the
rail to help here.

How are you sensing current? Or is the supply or load current limited enough to prevent
blowing the MOSFETs? Overcurrent events when using powerful battery supply can
explode MOSFETs - for instance a big motor pulling double-stall current when reversed
at full power would be a worst-case situation to consider (try to prevent this ever happening)

You bridge is

14

The p-channels are for switching only, the n-channels are for driving.

FAN3122 drivers (which are overkill, 100mA drivers would suffice
for those small Si4564's - still give 100ns switching).

The Si4564 is a 9A mosfet, at 12V gate driving the math suggests up to 9A pulse with no external resistance.

FAN3122 spec sheet recommends using 1uF or larger, which makes sense as we are using this cap to charge a smaller cap on the mosfet gate (in the nF). Whats most important is a low series resistance of the cap. The FAN3122 also rated to supply about equal the current of the mosfet gate without the need of a resister in series.

15

9A to the gate? Why? 0.1A will do fine.

A MOSFET rated at 9A source-drain will never be used at that sort of current
in practice, certainly not continuously, since its really a marketing gimmick rating. If
you want to water-cool the thing then yes you'll be able to pass 9A just....

Choose a MOSFET by the Rds(on), and the amount of heat dissipation you can
live with.

The gate current just has to charge the gate fast enough for the switching speed
you'd like, and 100ns is fast already.

16

Hi, why U8 and U10 drains connected together by a cap, then they don't go anywhere.
Same U9 and U11.

Tom..... :slight_smile:

Have you built this, does it work, i would protoboard this section by section, looks like a circuit proposal built from a datasheets.

17

Hi, why U8 and U10 drains connected together by a cap, then they don't go anywhere.
Same U9 and U11.

Because those are the motor leads. You'll notice that they are labeled M1-1, M1-2, M2-1,M2-2, and they make a connection to the pin outs up above.

There SMT you can't prototype them. But lucky me I read the spec sheets.

9A to the gate? Why? 0.1A will do fine.

Ya but based on the spec sheets gate charge, gate series resistance and the driving voltage the calculated value is ~9A pulse current. Driving it slower would only increase the power losses and make the chip get hotter, plus I need to use a series resister on the gate.

I picked that mosfet chip because it has nearly double the rated current rating that I require (5A motor stall, 9A rating mosfet). Naturally this also equals a low Rds(on), which means it will run cooler, will have less heat runaway which equals low de-rating when in operation and generally will last longer in all hopes.