Options for creating a configurable high- & low-side load driver?

[HwAoRrDk]

I am trying to think up some options for creating a general-purpose load driver that is configurable (via microcontroller) for either high- or low-side operation. It will need to work with supply voltages of 12 to 24 volts, and drive small inductive and resistive loads (< 1A) - e.g. relays, solenoids, small motors, bulbs, LEDs. I want to it to also be suitable for PWM operation and to have some kind of short-circuit or over-current protection.

My first thought is to make it up out of discrete components. An N- and P-channel MOSFET pair, an NPN BJT to drive the P-channel, resistors, some diodes for flyback protection. Then for over-current protection, use a bi-directional current-sense amplifier on the output coupled to an analog comparator built-in to the micro. Maybe a polyfuse too as a fail-safe to the current-sensing. Possibly use an inverter logic gate (74x04) on the control inputs to assure one side is always acting the opposite to the other. A complementary pair MOSFET package in SOIC-8 would probably be good, if perhaps a little overkill. I have nearly all of these parts on hand already to prototype with.

But then I thought: are there more integrated, more convenient solutions?

Use a dedicated half-bridge driver IC? Or maybe an H-bridge motor driver IC? I would guess some of the simpler ones would work as a dual half-bridge if I wanted more than one channel.

There's also a bunch of 'smart' high- and low-side load driver ICs, which look nice because of all the built-in protection (over-current, short-circuit, over-temp, etc.) and fault reporting (some just a 'fault' output, some with serial control giving more detailed status). Could I use one high-side and one low-side IC and pair them together by common-ing their outputs?

Can anyone suggest which direction I should go in, or any parts that may be ideal?

[David Hess]

Are you expecting the output voltage to be adjustable?  For instance you would generally not want to drive a MOSFET gate with more than 15 volts on a 24 volt supply.

Do you want the output to also support a high impedance state like a three-state driver?

What kind of transition times were you expecting to support PWM operation?  The transition time question is important because linear operation becomes difficult at high speeds and more complex output structures must be used.  1 microsecond is feasible in a linear design and 100 nanoseconds in a lower current linear design but faster may require an open loop output which is more complex and less intuitive.

ATE (automatic test equipment) pin drivers generally do what you want and there are some ASICs which are suitable for them like the Linear Technology (Analog Devices) LT1970 which almost does what you want but does not quite reach 1 amp and may not be quite fast enough without help at 10 microseconds.

[max_torque]

Off-the-shelf drivers tend to have an "enable" pin and an "in" pin.  By controlling how you switch those pins by the logic controller you can switch between High/Low drive, and High/Hi-Z or Low/Hi-Z. 

If you use half bridge or full bridge drivers you can then control in software what actually is driven and how it is driven.   Drivers like the classic TLE5206 and TLE5205  (if you want a fun task, work out the difference from the datasheets between those parts!  It's subtle but important.....) include vital safety's like over current, over temperature and open load detection etc, and ime are really robust.  In one device i use them for, the output can be running a motor at fully current (around 3A continuous depending on heat sinking) and you can jam a screwdriver directly across the output (hard dead short!) and the device just switches off without damage.  To get to that level of robustness in your DIY version is going to be very time consuming and expensive!

[HwAoRrDk]

Are you expecting the output voltage to be adjustable?  For instance you would generally not want to drive a MOSFET gate with more than 15 volts on a 24 volt supply.

No, just to work with whatever the supply voltage may be. Good point about gate voltage, that had slipped my mind. Given that it seems common for many MOSFETs to have a V_gs(max) of 20V, if I want this to work up to 24V, I would need to factor in some gate voltage clamping on a discrete solution.

Do you want the output to also support a high impedance state like a three-state driver?

At this moment, I'm not sure. How will it affect things if I do?

What kind of transition times were you expecting to support PWM operation?  The transition time question is important because linear operation becomes difficult at high speeds and more complex output structures must be used.  1 microsecond is feasible in a linear design and 100 nanoseconds in a lower current linear design but faster may require an open loop output which is more complex and less intuitive.

Are you talking about edge rates? Or PWM frequency? For the latter, I'm not envisaging anything too strenuous - from 10's of Hz up to only 10's of KHz. For the former, I'm not sure whether a slow transition time will have any bearing on my intended application. What I'm aiming to do with PWM output is twofold: a) do things like variable duty-cycle speed control for a motor or reduced hold current for a relay/solenoid; and, b) simulate a square-wave frequency-output sensor, to be used as an input to some other system. With the latter, I would imagine so long as the transition time is not so slow as to glitch/flap the receiving input, anything's okay, and for the former I assume a faster transition time is better for MOSFET power dissipation, but has little other benefit.

Drivers like the classic TLE5206 and TLE5205  (if you want a fun task, work out the difference from the datasheets between those parts!  It's subtle but important.....)

Umm... the TLE5205 can put the outputs in to a Hi-Z state, and the TLE5206 can't?

Speaking of Infineon, I have been looking at the TLE4207G, which looks great but appears to be discontinued at all the major distributors, even though Infineon on their website say it's an active part. Anything similar out there?

Edit: Also TLE6208-3G, which is NRND, although I have also found the ON Semi NCV7703, which seems to be a part of identical function (possibly a second-source for some OEM, I suspect), and thankfully is fully 'active' (at least, the 'C' revision).

[David Hess]

So it seems what you really want is a high voltage tri-state output driver which has current limiting although this ignores the issue with driving a MOSFET gate with a restricted voltage but that can be handled externally.

Are you expecting the current limiting to be adjustable or just fixed at 1 amp?

Usually the current limiting in this sort of open loop driver acts by disabling the output for a short time if the current limit is activated.

I think what you are looking for is an integrated fault protected half-bridge.  They can have two digital inputs for enable and level, output current limit protection, and usually digital outputs to indicate faults.

[HwAoRrDk]

Are you expecting the current limiting to be adjustable or just fixed at 1 amp?

Fixed. Basically, at whatever level will be suitable to prevent overload and damage to the circuitry.

I think what you are looking for is an integrated fault protected half-bridge.  They can have two digital inputs for enable and level, output current limit protection, and usually digital outputs to indicate faults.

Yes, I am coming to the conclusion that that would be the best route. But such ICs seem to be thin on the ground. Most I have so far found are either for low voltage (< 10V) or high-current (> 10A) applications. Or they have a very hard to work with package (tiny pitch and leadless, e.g. WSON, QFN, etc).

I really like the look of the Infineon TLE4207G, but still puzzled as to why some places show it as obsolete/discontinued, I looked into it some more. Turns out Infineon have revised the part number, and it is the TLE4207GXUMA1 that is discontinued, replaced with the TLE4207GXUMA2 revision. Some vendors do not have the newer one listed for some reason, only the old one.

Think I might get hold of some of these and try them out. But it's dissatisfying that this seems to be the only chip out there that ticks all my boxes. There must be something else similar. Anyone know of anything?

By the way, I couldn't help thinking that the TLE6208-3G seemed familiar somehow... until I realised it's because I've already used this chip! I was playing around with it a couple of years ago for some X-Y motor control. And I still have some in the parts box! But, stupid me has just realised you can't do PWM control with their SPI interface, so this chip is sadly off the menu.