isolated gate drives / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

isolated gate drives

1 2 3 4 next last

Move Thread

LAN_403

ragnar

Tue May 09 2006, 03:45PM

Registered Member #63 Joined: Thu Feb 09 2006, 06:18AM
Location:
Posts: 1425

Just what the world needs: ANOTHER permutation of an isolated gate-driver. ^^

This is a circuit board I've put together to facilitate close-and-snugly drive of a FET with complementary UCCs... BUT, instead of isolating the FETdrives from each other by means of a gate-driver transformer, I thought it might be clever to give each FET a dedicated set of drivers, then isolate the drivers at the power end so each has its own lil 15V supply.

My board looks like this:
at the extreme right are four ultrafast diodes, a smoothing cap, then an LM7815 voltage regulator. This lil bridge is for rectifying a winding off a HF supply I haven't designed yet; this is where the board draws its isolated 15V supply from.

Underneath the regulator on the right you'll see a 100-ohm resistor, a schottky, and the apparent "signal in" connections. An isolated signal (e.g. from an oscillator) from another multifilar isolating transformer is connected between ground and the schottky. The schottky ensures the input pin of the UCCs don't cop a negative voltage, and the 100-ohm resistor allows the input capacitance to discharge so the things can turn off again.

Next you'll notice a complimentary pair of gatedrivers in an unusual configuration. The bottom one (UCC**) has had all its pins tortured around upside down such that when I socket the chip, I have effectively mirrored the left/right pinouts. This allows me to have a very neat layout for the inputs/outputs of the chip. Yes, I mean the legs are bent 180-degrees so you can see the underside of the chip when inserted.

You can see provisions for power bypass capacitors to the right of each chip; electrolytic, monolithic, and ceramic, each.

The top chip's output goes to an optional bank of DC blocking capacitors (which will probably be shorted out by a jumper), then to the source of the FET.

The bottom chip's output goes through a 10-ohm resistor to the gate of the FET. There's a TO220 freewheeling diode in there for the fun of it, too.

Out the bottom left of the board is a handy connection to the FET's drain. Out the top left, similarly, is a big pad for its source.

I'm hoping to push a bridge fairly hard with this setup, hence individual drivers for each transistor. Yes, I have to make four of these boards.

I'm looking for feedback on the board layout, and this isolated drive concept in general... =)

What do youse think?

Matt

...

Wed May 10 2006, 04:27AM

Registered Member #56 Joined: Thu Feb 09 2006, 05:02AM
Location: Southern Califorina, USA
Posts: 2445

No comments?

I will admit that it is considerably more complicated than a strait gdt, but whatever floats your boat...

For the minimal cost of making these, even if they each save one fet they are worth building...

Hmm, maybe you should add over current shutoff... Seems pretty easy, just put a very small sense resistor in series with low side of the fet and hook it up to the enable of the ucc

Add a mov and you would have created an invincible switch

ragnar

Thu May 11 2006, 08:24AM

Registered Member #63 Joined: Thu Feb 09 2006, 06:18AM
Location:
Posts: 1425

... wrote ...

...but whatever floats your boat...

as opposed to floating my mosfet's gate... =D

First problem I can see with my own board is that the connections on the edge for power and mosfet drain/source are actually outside the line projected by the edge of the MOSFET package, which means I can't stack four drives in a row with the FETs real close on the heatsink.

MUR860s might also not be the ideal diode packages - should I make provision for double-diode packages or diodes in series with the FET (to isolate the body diode)?? hrmmm

Desmogod

Thu May 11 2006, 08:26AM

Registered Member #139 Joined: Sat Feb 11 2006, 11:01AM
Location: Perth, Western Australia
Posts: 358

Considering that you run a PLL, is there any real need for the MUR's at all?

ragnar

Thu May 11 2006, 10:31AM

Registered Member #63 Joined: Thu Feb 09 2006, 06:18AM
Location:
Posts: 1425

First, this isn't copping a PLL VCO, it's copping a MAX038 VCO (which does have an internal phase detector, but meh)... AND there's deadtime, AND there's signal modulation... so those diodes definitely have to freewheel =P

So, how do you like the layout and the idea of isolating the FETdrives at the power stage instead of output, to get that "hard" push/pull feel? =P

Steve Conner

Thu May 11 2006, 10:41AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

As far as I know, the MURs are pointless unless you also have a schottky in series with the MOSFET, to make sure the body-drain diode gets isolated. I once did some experiments where I put a fast diode in parallel with a MOSFET's body-drain diode to see how the current shared between them. The fast diode had a higher forward voltage on account of being optimised for speed, so the body-drain diode ended up still passing all the current anyway.

Admittedly I used a 1200V fast diode (RURG30120) and 600V MOSFET (IRFP460) so the comparison may have been unfair. Some high voltage fast diodes can be two diodes in series internally.

ragnar

Thu May 11 2006, 10:56AM

Registered Member #63 Joined: Thu Feb 09 2006, 06:18AM
Location:
Posts: 1425

wrote ...
As far as I know, the MURs are pointless unless you also have a schottky in series with the MOSFET, to make sure the body-drain diode gets isolated.

Eek, my MURs have a forward-voltage drop of ~0.4V, which will rise in duty, so I'll make that schottky a must... I scored some absolutely awesome 15A and 30A 60V schottkies off a laser printer board, but if this is running rectified mains, high voltage schottkies are expensive...

By the way, I just re-read the UCC datasheet, and on the ENBL pin they're happy down to -0.3V, and on the IN pin they're happy from -5V to VDD+0.3V.. so I can actually use a signal transformer without a 100-ohm resistor and schottky etc, so long as I've got an appropriate reduction down to +/-5V.. that makes things a lot neater and much more elegant, and just significantly changed my layout

I should also make provision for connecting that enable pin to something in the event that I actually decide to use it =)

So far I've fixed:

#1. input/output pads too close to board,
#2. allowing TO-247 double-diode packages to fit,
#3. making space for body-diode-isolating schottkies,
#4. removing the 100R and schottky on IN since it can take negative voltages,

Please, critique the crap out of my board. I didn't just put it here for flattery or to confuse y'all

Steve Conner

Thu May 11 2006, 12:22PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

The schottkies don't need to have a voltage rating equal to the supply voltage. 30v ones will do just fine, in fact they don't make high voltage schottkies (except for the mythical silicon carbide ones)

HV Enthusiast

Thu May 11 2006, 01:42PM

Registered Member #15 Joined: Thu Feb 02 2006, 01:11PM
Location:
Posts: 3068

wrote ...

in fact they don't make high voltage schottkies (except for the mythical silicon carbide ones

Actually, they do. I happen to have some 1500V Schottky diodes right here on my desk in front of me. And they
aren't SiC either . . .

ragnar

Thu May 11 2006, 01:52PM

Registered Member #63 Joined: Thu Feb 09 2006, 06:18AM
Location:
Posts: 1425

Hey, EVR, good to see someone looking at my lonely lil thread here.

Perhaps you could utter a few words of guidance regarding my [strange] ideas... ^^

1 2 3 4 next last

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob