H bridge design

Discussion in 'The Projects Forum' started by cl10Greg, May 18, 2012.

1. cl10Greg Thread Starter Active Member

Jan 28, 2010
49
0
Hi everyone,

I am designing my own H-bridge to make my own riding scooter. I am using 3x12V 12Ah batteries and the motor is going to be a 1000w 36V motor. I am using an additional 12V battery for controls. All the grounds will be tied together. I have the basic design setup but I have a few questions.

I attached my schematic for the LSD and the HSD. Ignore the names, I just wanted to present a schematic. Those are not my actual part numbers or values.

My HSD schematic basically takes a 5V signal from the microcontroller to drive a NMOS. The NMOS is hooked up to a 12V supply that is used to drive a PMOS. The PMOS is hooked up to the 36V battery load to turn on the motor.

The LSD schematic takes a 12V signal and uses a boost circuit to convert it to 60V (used Ladyada boost calculator). The 60V is used to drive the 36V load of the motor.

My question is basically a review of basics of transistors.

HSD schematic:
Will the 5V and 12V signal combine for a 17V signal or will it be a negative -7V signal?
Can the negative voltage on the gate turn on the NMOS?
Will the 12V to 36V difference result in a -24V on the gate?
Will this turn on the PMOS?
Any concerns?

LSD schematic:
Are there any problems with using a boost circuit to activate a NMOS?
My understanding is the gate voltage must be higher then the source voltage (hints the 60V) is this correct in this application?

Other notes/questions:
I realize that there will be some voltage drops because on the on resistance of the transistors and the motor.

Power dissipation of a transistor:
I read that the power dissipated by the transistor can be calculated with the on resistance and the current, is this true?
My application should draw a peak of 27A so that gives me a maximum power rating of (22mOhm)*27^2, is that correct?

Sorry this is very scattered brained but I am just analyzing thing over and over again and have some uncertainty. Anyone that can give me some insight would be much appreciated.

File size:
38.3 KB
Views:
45
File size:
17.1 KB
Views:
34
2. wayneh Expert

Sep 9, 2010
12,361
3,220
Lots of problems. The IRF540 you drew needs a voltage of at least about 10v to be fully on, to get the Rds nice and low. Otherwise it will get hot with even a little load current. Your calculation of power dissipation in a transistor is correct, I^2*Rds. But that low Rds must be achieved. If your control voltages are low, choose a logic-level fet which will be fully on at a much lower gate voltage.

Why is that diode on the gate? It will drop another 0.7v or so and adds to your low voltage problem.

3. wayneh Expert

Sep 9, 2010
12,361
3,220
If you tie the grounds together, then you won't get 36v. You need them in series. Maybe you mean the controller 5v power supply and the 36v supply?
You don't need a power MOSFET to drive the gate of the p-MOSFET. I think you could use any small NPN transistor, by small I mean just enough current handling to drop the voltage across R3, to turn on the IRF9Z24S.
Including the entire plan on a schematic would be VERY helpful to the pros here. You'll get much better help if you post that.

4. cl10Greg Thread Starter Active Member

Jan 28, 2010
49
0
I was under the impression of at least +-10V for the Rds. The maximum is around +-25V. So ideally anywhere between 10 and 25 should be ok?

The diode is hooked up to a 5V output of a microcontroller. I didn't want any voltage to damage the microcontroller.

I meant tie the 12V battery for control to the 36V (3*12V) battery so they have the same reference.

I am designing this right now so hopefully I have the attachment ready within 10 minutes.

5. wayneh Expert

Sep 9, 2010
12,361
3,220
My datasheet says ±20v for absolute max. Otherwise, yes, it'll be on from 10-20V.
That could only happen upon a MOSFET failure, but fair enough.
Got it.
Great. You'll get more eyeballs helping.

6. cl10Greg Thread Starter Active Member

Jan 28, 2010
49
0
Hello Again,

I have attached two documents. The first one is the overview of the batteries and what the different voltage potentials are. It is a simple block diagram.

The second one is the h-bridge configuration and signals the way I see it at least.

Notes:
R1 and R2 are pull up resistors to keep the PMOS off when the driver gate is not active. When the drive gate is active, the gate is driven with 12V which is a -24V potential and should turn the gate on.

R3 and R4 are used to hold VGS to 0 when the micro is off and have a VGS of -7 when it is activated.

R5 and R6 are pull down resistors that keeps the LSD off when the driver gates are off. When the driver gates are turned on, the 60V is applied to the NMOS gate and results in VGS being 24 volts and turning on the NMOS.

The diodes are used for the EMF feedback on the collapsing magnetic field. Each driver is driven with a PWM to control the amount of power the motor is able to draw.

There are some things in here that I am not sure about though. Some of the voltages or the flows. For example, will the HSD PMOS really just observe the 12V on the gate because it is the path of least resistance or will the resistor hooked up to the 36V effect it?

I hope that helps clear my idea up and I hope for some feedback on the application and if it will work.

File size:
52.5 KB
Views:
38
File size:
52.8 KB
Views:
28
7. wayneh Expert

Sep 9, 2010
12,361
3,220
The PWM signal and the power supplies share the same ground, right? So Vgs is -12V when the pwm signal is low, and -7V when high. It'll never turn on.

8. wayneh Expert

Sep 9, 2010
12,361
3,220
If the driver gates are turned on, won't this apply +60v to the gates of the power MOSFETs (ruining them)?

Let's back up a bit. I'm sure there are good H-bridge designs available here. I'm pretty sure SgtWookie has provided one that is often recommended here. Perhaps it would be wise to start with a known good design and modify it it necessary.

9. shortbus AAC Fanatic!

Sep 30, 2009
4,082
1,662
I've got a dumb question for the O/P. If it is a scooter why do you need an H-bridge? A H-bridge is only needed for running the motor in reverse. A scooter like a motorcycle is usually pushed for the short distance needed to back up. Eliminating reverse also allows you to use a single low side Nmos to drive the motor.

Your also better off using one of the mosfet driver chips between the micro and the mosfet.

10. JJAngleton New Member

Aug 21, 2010
14
1
On the upper left and lower right Mosfets, you have the internal diode and the external with opposite polarities. I am not sure if you simply drew the internal ones improperly, but if the schematic holds true and you build this, you will get a lot of smoke fast.

11. cl10Greg Thread Starter Active Member

Jan 28, 2010
49
0
This is a little blurry to me as well. It is my understanding though that for a NMOS to turn on, the gate voltage minus the threshold voltage needs to be greater then the source voltage to be in saturation. Can anyone enlighten me on this? I believe if you also look at MOSFET drivers, they have a higher output voltage then a source. I also believe I read multiple items on "pump" or bootstrap circuits to drive the MOSFET. I think this is the right nomenclature but can anyone help clear this up more? It also makes sense what you say because it does have a limitation on the gate voltage based on the potentials.

12. cl10Greg Thread Starter Active Member

Jan 28, 2010
49
0
Well that would depend. Could I put a big enough resistance to get a voltage drop that is under the threshold voltage? Or maybe I am thinking of the other way. If I have a very small voltage drop then the gate would see 12V and it would be compared to the 12V on the souce so the potential would be 0 ish when the PWM is off (or maybe it needs to be grounded)?

13. cl10Greg Thread Starter Active Member

Jan 28, 2010
49
0
This is also true, I am starting to think also that reverse maybe overkill. Your saying if with forward only I would just have to hook motor to the voltage and then have the NMOS turn on or off to connect the motor to ground? That would be very simple but also can be an issue for failure modes. So there are pros and cons.

14. cl10Greg Thread Starter Active Member

Jan 28, 2010
49
0
You are correct. I just grabbed parts from Altium for place holders so they are not representative to what it really is. I am using the same NMOS and PMOS for both sides and I believe they are both rated at 60V and 35A.

15. cl10Greg Thread Starter Active Member

Jan 28, 2010
49
0
Thanks for all the feedback so far. These are a lot of questions that I have had for awhile and its hard to find a reliable source for. I have read many different applications that seem to have contradictions.

I do have one general question for making these boards now. So I have been into etching my own board for the last couple months and I feel ready to go for this application. I will have a bunch of breakout type boards that will interface to each other (Arduino, Boost circuit (if needed), H-bridge Driver, H-bridge, battery monitor, etc...). The question I have is current from the board to the MOSFETs. So I know how wide I need to make the traces and the distance between them for the voltage and current. The current can peak up to 28A based on 1000W/36V=Current. The cable equivalent is 10AWG that I will be using to go from the batteries to the PCB. So how in the world do the small legs of a MOSFET allow that current to go through it? Is that even possible?

The obvious answer is that it can't and the current rating is based on the junction capability internally. But I don't see how this can be the case when you can find plently of off the shelf H-bridges that are designed with similar chips and similar packages. Can someone tell me how they make that possible? I think I have read that some MOSFETs have the heatsink pad on the back tied to the source or drain that that can be used for conduction.

Any insight would be great. Thanks guys/gals!

16. shortbus AAC Fanatic!

Sep 30, 2009
4,082
1,662
Unless your using a mosfet in an audio amplifier circuit, just ignore the "threshold voltage". That is the very minimum voltage that will turn the mosfet on, it only lets a very small voltage conduct. To use a mosfet as a switch you need to supply 10 or more volts to the gate to turn it on.

The ten volts is referenced to the source terminal. In a high side switch, that is the reason to use a driver IC, though it can be done other ways. The 10V reference to the gate/source does not mean 10V more than source voltage though.

The low side switch method is probably the easiest and safest way to go IMO. As most times a mosfet fails in a "open" mode, it will just shut the entire scooter down. The low side mosfet can be PWMed to control the speed of the scooter.

17. cl10Greg Thread Starter Active Member

Jan 28, 2010
49
0
I understand the threshold so that's good.
If this is true, then why do gate driver IC's have a boost circuit that puts out a higher voltage then the source that is on the high side switch? I was looking at some IR IC gate drivers on digikey and they all put out higher voltages on the high side drive then the source. http://www.datasheetarchive.com/IR2106-datasheet.html

Maybe you can help me clear this up?

I understand this method and control and it is pretty simple. On this same note, how does a leg from these MOSFET able to support the equivalent of a 10AWG wire based on cross sectional area of their legs? I know the rating is based on the junction but I still look at numerous designs for H-bridges and they use similar MOSFETs and claim to have power up to 50A. Are there certain packages that you can use the body pad instead to conduct?

18. wayneh Expert

Sep 9, 2010
12,361
3,220
To allow current to flow thru the high side, you need to have a gate voltage 10V above the high side voltage. Drivers use a bootstrap capacitor to supply this voltage briefly while the high side pulse goes thru. The capacitor gets recharged when the low side wave goes thru.
I've wondered the same thing and speculate it comes down to distance. The legs are not so hefty, but they are also very short compared to a PCB trace or a hookup wire. The loss and attendant heat dissipation are minimal.
Yes, the metal pads are usually connected to the drain or source pins. But the current path still includes one leg.

19. shortbus AAC Fanatic!

Sep 30, 2009
4,082
1,662
I was under the same idea about gate voltages in the beginning. With a high side driver circuit there is a "boot strap capacitor" and a diode. The diode is wired to ~10V and then to one leg of the capacitor, the other leg of the capacitor is wired to the source,with me so far?

When the source is at ground/common level the capacitor charges to ~10V. Then when mosfet is turned on and the source voltage rises to the drain level voltage, the ~10V in the capacitor is then transfered into the gate. The diode keeps the ~10V from transferring back into the supply voltage. When the gate is turned off, the source voltage again drops to ground, and the 'boot strap' capacitor is then recharged back to ~10V for the next pulse. The boot strap capacitor gives the "reference barrier" to the gate.

When I say gate voltage is "~10V" it should be at least that voltage. Anything 10V to close to the limiting voltage from the data sheet is fine. A lot of circuits use 12V and that is a good voltage to use.

In a low side switch you can use a "wire lug/terminal" (like the crimp on type) to connect the mosfet "tab" to it. The 'tab' is electrically the same as the drain terminal. BUT- when using a heat sink, which you need, The heat sink will be at the 'drain' voltage potential. You must either make allowances for this in your design OR insulate the tab and its screw from the heat sink. They make bushings and insulators to do this.

Here is a link to explain mosfets better than I can - http://reibot.org/2011/09/06/a-beginners-guide-to-the-mosfet/

20. Bernard AAC Fanatic!

Aug 7, 2008
4,230
412
It would seem advisable to provide a soft start ckt. At turn on motor could draw 100 or more amps.