Maker Pro
Maker Pro

Need help diagnosing H-Bridge

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
Found it. It's an ON LM2576D2T-5. I'll be checking the outputs shortly. It's supposed to have 3.3, 5, 12, 15, and an adjustable, from 1.2 to 37V.

That's a 5V reg, proabably for the microcontroller. (the -5 on the end gives the voltage)

Ahh, OK. Well, I just got through removing the 102 and when I checked it out of circuit it is reading 1k. There's nothing wrong with the resistor, the problem is elsewhere. After I put it back in it's reading 96 ohms in-circuit. Let me see if I can chase down the rails and reg. That couldn't possibly be the FQP19N20C, could it?

OK, so the resistor is fine, but in circuit it measures substantially different to the good board?

I would be concentrating on comparing readings from nearby transistors.

Put your meter into diode test mode and measure both ways across every combination of pins on them, comparing the results with the known good board.

You're on the lookout for unusual 0L's and 0 readings as well as substantially different results.
 

eKretz

Apr 8, 2013
251
Joined
Apr 8, 2013
Messages
251
I still need to chase the 18V power source too though, right?
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
Checking all the components is probably easier :)

The board makes it very hard to trace out the circuit which is kinda what you need to do once you;ve found what generates these voltages.
 

eKretz

Apr 8, 2013
251
Joined
Apr 8, 2013
Messages
251
I think maybe I found at least part of the problem. It is not easy checking the transistors in circuit. They always seem to charge something and then go open. You really need to discharge every time before checking in-circuit, and even then sometimes I get no reading. What a PITA. I would never have found this without removing components if I didn't have the good board to compare to. Here's a shot of the one that checked differently: T1001. You can see the trace and pad repair right next door at T1005. There is just the barest sign of damage on the surface of the J3Y at T1001. I totally overlooked it, thinking it was just a blemish. Under magnification it starts to look a little more like damage. You'll have to excuse the quality of the photo, it was taken with my iphone through the scope lens.

photo.jpg~original
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
Photo quality excused :D

Another thing to look at is that solder joint at the top of the image. It doesn't look like a good one.
 

eKretz

Apr 8, 2013
251
Joined
Apr 8, 2013
Messages
251
Actually, that joint looks great on the board. It does look terrible in that picture for some reason though. Must have been the way the light caught it where there's a little probe scratch on the leg there.
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
It looks like a blob of solder, not a smooth flowed joint. Are you telling me is really is a smoothly flowed joint?
 

eKretz

Apr 8, 2013
251
Joined
Apr 8, 2013
Messages
251
It is. What looks like a blob of solder is actually the scratched leg of an FDD4141 MOSFET. The solder is under what looks like the "blob" in that photo. You can sort of see it as what looks like a shadow under the "blob." It really doesn't look like that photo at all, lol. I actually though I was looking at the wrong board when I first went to check it out after your recommendation because the solder joint was perfect when I checked it. I can post another photo if you want! :D

On another note, I just removed the J3Y transistor that was damaged and a spare from my scrap board and soldered the spare into place on the new board. SUCCESS!! It's working normally again. Just did a 10 minute race to try it out and no problems so far. Thanks a million for the assistance!! You are the man!
 

eKretz

Apr 8, 2013
251
Joined
Apr 8, 2013
Messages
251
Now, as far as making H-bridge changes...lol. I have done a lot of temperature probing on the different areas of the board and really, the weak link seems to be the motors themselves. They are sealed cans, and it's tough to get the heat out. I think I was crossing over 130-140 degrees Fahrenheit at the can surface with 15-20 minutes use. So the internals must be baking. The H-bridge and everywhere else remained under 115 Fahrenheit. This is running at 24VDC. When I up the voltage of course things get a little warmer, but it still looks like the weakest point is the motors. I have a guy working on custom motors to try to increase torque while reducing amperage. I would like to be able to upgrade the motor drive components to beef things up a little and cut temps to as low as possible. IIRC, the RDS on of the 4141's is something like .0025 ohms. That new ST I linked is slightly less, but in reviewing, I've discovered that it's also a different package, and they don't make it in DPAK.
 
Last edited:

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
It is. What looks like a blob of solder is actually the scratched leg of an FDD4141 MOSFET.

Oh yeah, I see that now.

On another note, I just removed the J3Y transistor that was damaged and a spare from my scrap board and soldered the spare into place on the new board. SUCCESS!! It's working normally again.

Excellent!

Thanks a million for the assistance!! You are the man!

In all fairness, you did most of the hard work. All I did was point vaguely at stuff.

You have successfully played the game of "spot the difference", an amazing technique which can solve a lot of problems like this. :D
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
Now, as far as making H-bridge changes...lol.

Well, it's a good time to look at that now.

This is running at 24VDC. When I up the voltage of course things get a little warmer, but it still looks like the weakest point is the motors.

Well, we do know that the original problem with the board happened at a higher voltage.

1) The first thing to do is to look at the ratings of these transistors to see if increasing the voltage is going to cause problems.

2) Without the schematic, that's not easy, but we're pretty sure that the logic is operating at 5V, so the signal is level-shifted. This will *probably* work at higher voltages.

To check this out, we might want to probe the gate voltages in the H Bridge at a *reduced* input voltage first. We see 6V and 18V at 24V. Do these remain at 0+6 and Vcc-6, or are they 1/4Vcc and 3/4Vcc? This will become important as we raise Vcc because we don't want to risk exceeding Vgs(max). Measuring Vgs at a motor voltage of 12V as well as 24V will indicate what is likely to happen at (say) 36 volts.

3) Current capacity of the mosfets. Did you blow some of these already? It is important to know if the extra current will be safe for them.

4) other factors -- Rds at the Vgs the driver provides -- this will determine the amount of power lost in the H bridge. Gate charge will determine how long the mosfets take to switch. If this is too long, shoot-through may become an issue. Voltage ratings of capacitors etc on the board. dissipation in resistors on the board.

I have a guy working on custom motors to try to increase torque while reducing amperage.

Well... He could reduce the RPM, or increase the voltage.

IIRC, the RDS on of the 4141's is something like .0025 ohms. That new ST I linked is slightly less, but in reviewing, I've discovered that it's also a different package, and they don't make it in DPAK.

The best thing to do note the specs of the 4141 that are important, then to use a parametric search tool (like the one Digikey has) to search for parts which match the higher specs you desire.
 

eKretz

Apr 8, 2013
251
Joined
Apr 8, 2013
Messages
251
On the motors, we're trying a whole different approach. Increasing the size of the can and armature to get more leverage with the same magnetic repulsion/attraction, as well as switching from alnico magnets to neodymiums. I'm hoping to increase the brute strength of the motors enough so that I can current limit them to keep heat down. It's very important to reduce the heat in the motors because as they heat up they weaken a lot. This makes for a bad driving experience since you end up with differing strength in the torque of the wheel, which makes precise control of the racecar difficult. Basically, as the race progresses, it gets easier and easier to turn the wheel, so you can't use muscle memory to train yourself lap by lap. Another approach I'm trying is to cut holes in the sealed cans and run air cooling ducts right through the side of the can and ducting that air outside the wheel housing so carbon dust from the brushes doesn't foul everything. This way I'd cool down the armatures, brushes, coils and magnets directly and carry the heat away immediately instead of having to wait for the heat to conduct through to the can.

I have already looked at the specs for the FETs and they are max rated at 40V and 15A. I have even ran the wheel at voltages up to 40. I started small, went from 24 to 26, then when that was fine, to 28, etc. all the way up to 40. I will do the proper checks now that I have a little better understanding of what is happening in circuit. None of the MOSFETs have blown yet. I originally thought I had blown one when this problem started, but it turned out to be the T1001.
 
Last edited:

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
My experience with some mosfets (when testing them) was that they suffered avalanche at only a whisker above the rated voltage.

I would recommend using mosfets rated about 50% higher than your highest expected voltage. So 60V if you want to use a 40 volt rail.
 
Top