DIY DC-DC converter to supply high-side N-channel MOSFET gate drive power.

[SoQ]

Hey, so I'm a bit of a noob at all of this although I know which end of a compiler to hold on-to when programming.

As mentioned in another post I'm building a small spot-welder for hobby purposes.  My research suggests my best bet for the basic topology is an N-Channel MOSFET driven on the high-side, or perhaps an IGBT.  I want to do this with what I have on-hand if possible.  My parts box is not big, but I have a variety of parts -- some salvaged from consumer equipment -- such as a H30R1103, which might be a workable choice if I find I'll likely get large voltage spikes on the output side.  (I understand this device is somewhat specialized and so might be better saved for something else.)

In any event, and because I am learning I would like to avoid a high-side MOSFET gate driver with integrated charge pump or whatever.  So what I might do is use a GPIO clock signal to drive a voltage boost circuit of some kind.  There's probably more, but I'm aware of three basic circuits to do this:  charge pump, diode voltage doubler, or a boost-converter feeding an op-amp slash power transistor (I have MJE3055 if that will work) to set the gate-drive rail.  Perhaps a charge pump alone can be designed to provide a stable voltage under varying load, but I digress.

My lack of expertise (and lack of care for parts count, within reason) makes it difficult to know which topology would work best.   A tiny boost converter is obviously the most complex, but if can of course be set to any voltage by design and a COTS unit from Amazon or whatever might help with debugging my eventual circuit.  Charge pump circuits seem a little anemic for a (upper-limit for my purposes ATM) 200kHz MOSFET gate driver and I don't understand if they scale to ~1-5A (a guesstimate).  A half-wave voltage doubler looks like it might be pretty simple, but I only saw the circuit for the first time today and I'd probably have to drive it with a square-wave to keep it simple.

I'm just looking for a trivial comparison of the different schemes in this application and not so much with their part-count complexity as this is a learning exercise.

[Andy Chee]

Page 29 of this app note has a summary of gate drive techniques.  The rest of the document is good reading as well.

What kind of consumer equipment are you salvaging?  I assume you're talking about switchmode power supplies?  SMPS have plenty of components which you could potentially use.

[SoQ]

Page 29 of this app note has a summary of gate drive techniques.  The rest of the document is good reading as well.

Looks good; will read thoroughly.

What kind of consumer equipment are you salvaging?  I assume you're talking about switchmode power supplies?  SMPS have plenty of components which you could potentially use.

Usually whatever opportunistically presents.  People leave all sorts of crap out in alleys and apartment buildings all the time.  I'm not a fan of large TVs.  stereo components, computers; inkjet printers are ok, but I don't need any more tiny 24V motors.  I'm hoping for a 2-stroke motor since battery-powered hedge trimmers and whatnot are now more common.

After a handful of years I have more salvage than new, excepting resistors and some other trivial components.  But the nice parts are the rare ones: Nespresso: heater, pump, etc.  MFP: quadrature encoders and associated media.  Heat sinks, etc.  I save a lot of money (against time) and get to see what actual engineers are selecting for their commercial products.  Case in point:  8 AR251 diodes from an oldold Canadian Tire 10A automotive battery charger and used as the bridge rectifier:  280A/4000A surge if I were to use them all in parallel.  Not sure why the BR was designed to handle 70A except if Vf is low at a fraction of the rated current (~1.1V on the data sheet).  Perhaps the transformer output voltage was marginal for the application or something.

[Andy Chee]

Make sure you salvage the humble walwart and power brick.  They will contain components usable for your high-side gate driver.

[David Hess]

I have a couple of thoughts.

Why not use the n-channel MOSFET to switch the negative side?  Or are you producing the output with a source-follower?

The gate only draws DC leakage current, so a small stack of coin cells can be used to level shift the drive to the gate.  Bypass the battery with a capacitor for good AC performance.  Note that the battery does *not* power the drive circuit; it is just in series with the gate.

Do the same thing with a photovoltaic optocoupler.

[Buriedcode]

When you say spot welder - is this simply a high current supply (battery or bulk caps) switched briefly with a MOSFET/IGBT?  If thats the case then your switching frequency is clearly very low, and as such, the average power for driving the gate will be low, albeit with large current for switching. 

I don't see why a charge pump with a large output cap wouldn't be sufficient for driving a highside MOSFET.  A charge pump can be as simple as some inverters, or even a 555, with a couple of diodes and caps.
It's not regulated, its not high current, but should be more than enough to keep a largish value cap charged that can provide enough current to quickly switch the MOSFET.