Logic level HV fet...

At voltages such as this, Qg (and Ciss, Coss) are more important than Rds(on) although obviously it has to be reasonably low. The reason is the large voltage swing of the source when the Switch turns on. It will go from Vout to Vin, and capacitively couple that delta to other parts of the SMPS with possibly harrowing results. At the load level you suggest, a device in the hundreds of milliohm range would probably be suitable (but check the SOA curve).

Most FETs at this breakdown are normally specified at 10V Vgs. If 9V is not sufficient, you can always make a simple capacitive charge pump to run the boost driver (Linear tech has excellent app notes on this sort of thing).

For a typical device list, see this page

which lists the excellent range of IR HexFets(tm). Similar devices and capabilities are also available from Vishay-Siliconix and Fairchild (although the selection from fairchild is more limited).

As for footprints, many high voltage devices come in D-PAK or D2-PAK surface mount footprints, to name the more common ones.

Cheers

PeteS

Mike Deblis wrote:

maybe 3A

in an

to give

I know

3A

give

know

Use a voltage boosted boost converter. You tap the inductor so that the FET sees a fraction of the output voltage.

----nnnnnnnnnnnn----->|----- | | |-- |

3A

give

know

Hi there. Unfortunately the pickings for 200V or more logic level MOSFETs is exceedingly slim. For the most part your only potentially feasible choices I'm aware of are the IRL620, IRL630, IRL640 (TO-220 devices) and the D2Pak versions of those same parts: IRL620S, IRL630S, IRL640S. Here is the datasheet for the IRL630S which is a 200V 0.4 ohm Rds(on) D2Pak part.

Digikey sells all of these device types. They aren't really very good, noticeably inferior to their standard gate drive newer cousins.

The LTC1872 is a nice part, but the switching frequency is really unreasonable for this application. Efficiency would be hideous at 550kHz operation, no matter your MOSFET selection. Without doing any major calculations before spouting off conclusions, I would bet efficiency would be so low as to preclude the use of an SMD MOSFET for thermal reasons alone, despite your low 3.6W output power. In addition to abysmal efficiency, you could also expect tremendous EMI production, likely way beyond FCC limitations. If you listen to the radio at all, you quite possibly might find that all but the most powerful AM and FM band stations will have become too static filled to listen to.

The problem is high dV/dt radiates. A little 3.3V to 5V step up converter can operate at high switching frequency with very fast transitions without producing very much EMI. The drain of the MOSFET would only swing from

0-5V. With 180V output the swing suddenly becomes 0-180V. Unless the switching speed is reduced drastically, the dV/dt will be greatly higher and hence much stronger EMI radiated. At first it may seem strange that a switcher operating at 550kHz could interfere with the FM band which operates around 100MHz. It is important to remember that a 550kHz squarewave with fast edges contains large amounts of higher frequency content.

I would strongly recommend using something much closer to 20kHz as your switching frequency. Make sure to use discontinuous conduction mode since the control loop gain gets really large (IE: impossible to make stable) for large step up ratios (which 9V to 180V certainly qualifies) in continuous conduction mode. With a 20kHz switching frequency and 9V input an inductor value of vaguely around 150uH would seem to be reasonable.

the

the

...

for

inductor

Hi Fritz,

Thanks for the reply. Normally, I use a MAX1771 which runs at 50kHz and gives me about 87% efficiency in this configuration - really a very nice chip.

However, getting the MAX chips in Europe and elsewhere is often a pain, and I was looking at emulating its performance using LTspice when an LT FAE mentioned their devices, which are emulated for you in LTspice, and he suggested I look at those as well, hence this investigation.

I agree with everything you say about the LT chip - it is a very nice device in a very small package, but the maximum efficiency I've been able to get is in the order of 65%.... Also, unlike the MAX1771, the LT device can only handle up to 9.8V in, which really means logic level FETs - if I was going at add a charge pump or whatever for the gate drive, I might as well use the MAX device which is more forgiving anyway...

Ho, hum... just a thought...

Thanks

Mike

Although diode sees a multiple of reverse output voltage, the duty D can be significantly reduced from the 95% required of single inductor 20:1 multiplication of 9V->180V. Good idea.

maybe

an

FET

Yes, that's the caveat.

DNA

the duty D can

Hi Fritz, This is a great application for running a simple boost in Zero Voltage Switching mode. Add more capacity in parallel with the FET to control (resonate) rise and fall times to about 200nS. Operate the PWM in constant "off time" and current controlled "on time". Most control IC can be configured for this. I use this mode of operation so I can lisiten to the radio as I check it out. Piece of cake. Harry

>

9 volts is plenty of gate drive for most regular mosfets. Drain current will be well below 1 amp, no big deal.

John

3A

give

know

The 2SK2350 is 200V, 0.26ohms, with 4V gate drive.

Siliconix Si4490DY Nchannel 200V 0R09 in SO8, is characterized with

6Vds.

RL

Once you've decided to tap the inductor, is there any practical difference between that and a transformer. Probably still want different size wire in the two sections anyway. mike

--
Return address is VALID.
Wanted, PCMCIA SCSI Card for HP m820 CDRW.

I agree. Once you have decided to spring for a custom magnetic part, you have all sorts of choices that trade cost and complexity in one place for cost, efficiency and simplicity in others. I would consider a 3 winding transformer, with the primary rectified and both secondaries separately rectified, and all 3 stacked in series, so that the transistor would see only about 60 volts and schottky rectifiers could be used all 3 places. 100 volt fets are more common than 200 volt units in SMT low gate voltage versions.

--
John Popelish

3A

an

give

know

Hey now that is an interesting part. I wonder how much it costs...

I agree, ZVS would be very nice for this application. This is a good idea.

The last time I was fooling with 150V parts, top of the line SO8 packages were about US$ 1.70, when they hit their intended volume. Small quantities and early parts could be three times that.

Never inquired about Si4490. The 150V parts were 0R05 or less. RL

Genome wrote: snipped all the bigoted stuff

Why'd you have to go ruin a perfectly good suggestion with all that crap.

Of course, you have to do the math for all the options. But you can at least be objective about it.

mike

--
Return address is VALID.
Wanted, PCMCIA SCSI Card for HP m820 CDRW.

It's a boring part. Yawn.

--
 Thanks,
    - Win

It appears the 0.26ohm rating is with 10V drive. That's a boring part. Nothing special. Yawn.

--
 Thanks,
    - Win

Yes but only on a casual reading. Look at the graphs. This fet has very nearly the same Ron at a 5V gate. This is in sharp contrast to many Fets advertised as being 'logic-level' when in actual fact they are barely starting to move, hence incapable of doing any real work at this drive level.

Go on then, suggest a logic level fet, easily available at a reasonable price, that should be preferred for this switching job. An insulated case type would be nice. :-). regards john