I'm searching for a full h-bridge driver that has the following properties:
Full bridge driver with bootstrap (or similar) to drive external MOSFETs
Ability to interface directly with a dir/pwm signal without external glue logic
"Hobbyist Friendly" that is available from common sources in small quantities and easy to solder
So far i found the NXP UBA2037t http://www.nxp.com/documents/data_sheet/UBA2037.pdf To my reckoning it is possible to use the SU input for PWM and the CLK for DIR. It has some nice features like built in bootstrap diodes and power supply. Now for the bad news:
I could not find in stock anywhere
Many suppliers had a large minimun order of several 100
Does anybody know where to get them or some alternative IC to use?
UBA2037 is not a general purpose H-bridge driver, its a buck-boost converter I think,
You won't find one that doesn't need glue logic to my knowledge, typically they just
have 4 inputs for the 4 switches. Some half-H-bridge drivers have single input + enable
though.
It is, and it is not. The datasheet states: "This circuit is designed for driving MOSFETs in a full bridge con?guration". The design is intended for HID lamps but i can see no reason not to use it otherwise.
The datasheet also states that when SU input is low both low side drivers are on and the high side drivers are off which essentially is a slow decay mode.
When SU is high and DD is high the clock input determines the bridge polarity. The high levels on CLK is typical 1,6V and on SU 1,3V and input currents are less than 1uA so both can be controlled by a pin. What has to be done is feed the VDD(CLK) and DD (maybe with divíder) pins from the VDD pin.
nilton61:
It is, and it is not. The datasheet states: "This circuit is designed for driving MOSFETs in a full bridge con?guration". The design is intended for HID lamps but i can see no reason not to use it otherwise.
Ah, I found table 3 - first glance there didn't seem to be enough inputs with relevant names.
Yes, it can be driven if DD is held high, BD low... I think. You won't get fast or slow decay
modes, only synchronous rectification, though (but that's the best mode really). Be sure never
to give SU 100% PWM or the high side drivers will lose power.
Note the drivers are deliberately slew-rate limited to about 15V/us, so it seems not to
do hard switching, yet they quote a max frequency of 200kHz, so I'm a bit confused.
The main problem with the chip is it only drives logic level MOSFETs though, the
undervoltage lock-out is 4V, so that rules it out for most devices.
Grumpy_Mike:
How about the LMD181200 or the TLE5205 or the A3986 ( for stepping motors ).
Or the FAN7382 or the IR2104
To my reckoning the first two are integrated drivers. Im actually in the process of replacing LMD18200 in my design. They cost about 22 euros here and are limited to 3A. With discretes i can go much higher than that for a lower cost. (im driving brushed dc motors)
The two lat ones are half bridge drivers. So are irf 2113.
There is not much to reverse engineer. H-bridge designs are well known. I have a design with 2 irf2184(half bridge drivers) and one 74hc00(glue logic) that works very fine. But the cost and the reliability of a circuit correspond to the number of holes/pads and the number of circuit packages.
So if i can replace the 2 irf2184 and the 74hc00 with one package there is a lot to be gained.
MarkT:
The main problem with the chip is it only drives logic level MOSFETs though, the
undervoltage lock-out is 4V, so that rules it out for most devices.
I did not quite understand this one. As i see it the high-side driver supply voltage (=bootstrap voltage) is designated Vfsl and Vfsr respectively. This should equal Vdd - Vd(bs)(bootstrap diode voltage drop, max 1,2V). That should equal about 13V with Vdd==14V. Also table 3 states that the output voltage on the high-side drivers = 1 should equal Vfsl or Vfsr.
[quote author=NXP UBA 2037 datasheet]As soon as the supply voltage on pin VDD becomes lower than VUVLO(VDD) or the supply
voltage on pin VHV becomes lower than VUVLO(HV), the IC enters the start-up state again.[/quote]
I think the undervoltage lockout is meant to prevent the MOSFETS operating at a too low gate voltage.
But i agree on that the slew rate limitation is rather conservative, probably depending on the rather high resistance of the drivers (20Ohm) but this would also eliminate the need for gate resistors. On the other hand, plan to operate the circuit at max 10kHz and the non overlap (deadband) time is 900ns which is quite long and should ensure safe operation.