Using FETs in parallel

Have you considered using a do it yourself bucker instead of the clamp?

By time your got the parts for doing it as a linear regulator, you've most likely got the cost of a bucker tied up in heat sinking etc.

[...]

Vicor makes 46V nominal ones that are good from 21-56V. You could have 2 DC-DC converters and shut off the power to the low voltage one when the voltage goes high enough that the high voltage one can take over.

[...]

Its not completely like that but kind of. You will have to have a huge over design for the low voltage case because of the need to handle the power in the high voltage case. Both the bipolar and Mosfet will have a low drop because of this. The MOSFETs drop will be greater than the bipolars, partly due to the need for more bipolar transistor to meet the SOA requirements. Look up "second breakdown".

--
--
kensmith@rahul.net   forging knowledge

Why? Ed

Tee hee hee. You are a funny guy.

Where is your "reference design", geek boy?

[Referring to lateral power MOSFETs]

Yes, I used 16-off Hitachi lateral MOSFETs in a parallel/push-pull 1250VA inductor test set, generating sinewaves up to 50KHz, in the late 80's. Nasty inductive loads. The 2 test sets were still in use certainly up to last year, still running on their original MOSFETs.

I do not think there is entirely 100% self- -protection, but the negative dId/dVgs at all Id's above 100mA means no thermal runaway and full confidence that paralleling devices can be done without any one device hogging current.

There was a thread here a few years ago when I realised that vertical MOSFETs did not reach a negative dId/dVgs until Id had reached several Amps..... requiring temperature-compensated Vgs biassing, and difficulties when paralleling.

In fact I had a seriously good rant in that thread about the deceptions in (say) the old IR HEXFET manual, which used phrases like "intrinsic freedom from thermal runaway" in linear applications. Claims that were just not true. AFAIR Siliconix also made similar claims for their VMOS devices.

If you need MOSFETs for paralleled linear operation then lateral MOSFETs are well worth looking at.

--
Tony Williams.

"John Larkin" "Phil Allison"

** Sarcasm will get you nowhere.

Snipping and ignoring what was posted just for you is a good way to start a flame war.

** How arrogantly asinine.

Only two lateral MOSFETS are needed to meet that spec.

The BUZ900 /905 complements.

** Lateral MOSFETS range up to 250 volts breakdown and 250 watts dissipation - using dual chips in the one pack.

Notice that they are all made in P and N channel complements - unlike switching types.

Bias setting and bias stability are a doddle.

There is no need for source ballast resistors in most cases.

There is NO need for thermal protection either - lateral MOSFETS inherently self protect, allowing fewer devices to be used than otherwise.

Your MOSFET knowledge is sadly lacking.

........ Phil

Why not seek suitable BIG mosfet that handle all the current requirement? I tend to avoid parallel Mosfet due to difference in VGS threshold. Perhap bipolar transistor in parallel might do better job than Mosfet, you can control base current for optimal sharing...perhap.

Regards

Riscy

Phil Allis> >

Not available in the UK at that time. It was 2SK135 and 2SJ50 afair.

--
Tony Williams.

"John Larkin" "Phil Allison"

** Lack of reply noted.

** You are simply full of shit - John.

You arrogantly said 20 devices were needed.

I say it can be done with just 2.

A conservative designer would use 4.

Long bloody way from 20 - right ?

** You still fail to recognise the unique thermal qualities of the lateral MOSFET.

Negative tempco prevents bias runaway.

Inherent thermal shutdown prevents damage from overtemp.

BTW

Until you have some experience with the devices - better put a plug in it.

** Irrelevant s**te.

** You really are full of it.

Right up to the eyeballs.

The "audio market" being alluded to is in fact the professional one.

** Until you have some experience with the devices - better put a plug in it. ** The real life scenario is when an overload causes the devices to go overtemp or say the fan cooling fails.

Once the chip temp reaches about 170C, its transconductance drops dramatically and Id reduces to a low level, just enough in the circumstances to keep the chip at that temp until the overload condition is removed.

A lateral MOSFET can remain in this "self protection" mode almost indefinitely with no damage.

Seen it happen in the field.

I have contrived to make it happen on the test bench.

Amazing to watch.

** Obscurus jokus.

How appropriate.

........ Phil

On the subject of robustness, whilst the lateral fets are certainly very good, they are not totally unbustable. Audio amplifiers made by C-Audio are a case in point. The heatsinking is so poor that I have measured case temps in the 140C region ! Fortunately they fail open so don't affect the function of the amplifier other than to restrict its output.

Graham

"Winfield Hill" = the snipping TROLL

** More trolling.

** Try learning to read with regard for the context - Win.

Try not snipping so much.

** No idea what the word "irrelevant " means - Win ?

Thought so.

.......... Phil

"Anthony Fremont"

** With nothing but mindless abuse and really dumb lies as usual.

No surprise for a f****it radio ham.

....... Phil

"Tim Williams"

** Huh - most people have been ignoring you for ages , Tim.

Got you figured for a dumb turd in seconds.

......... Phil

"John Larkin"

** Pretty much exclusively used in high powered audio amps ans some industrial apps.

Driving SONAR transducers is another app.

The Hitachi 2SJ56 and 2SK176 types were the most popular TO3s - now deleted.

** Try asking the cat.

** The sales outlet for "Semelab" of the UK ( Scotland actually)

The other major maker of lateral MOSFETS.

Parts from the same UK factory are sold as SML (Semelab) or else Exicon.

........ Phil

Sigh. Let's re-phrase that.

dId/dT at constant Vgs is negative at all Id's above about 100mA.

--
Tony Williams.

"Tony Williams"

** Bet they were 2SJ56 and 2SK176 TO3 MOSFETS. ** Precisely.

** Bad enough in manufacture - a complete nightmare for later, non factory servicing. ** Naughty, naughty.

** N ch switching MOSFETS with very high power and voltage ratings are cheap and plentiful so there is great pressure to use them. P ch ones are very scarce by comparison and no true complements exist.

The situation with lateral MOSFETS is quite different and that is why they remain in production for specific applications where their unique characteristics make them very much the device of choice.

........ Phil

"Tony Williams" <

** Ok - they are close cousins.

You would only need 12 ( 6 of each) if the up rated ones were used.

In a typical audio power amp, 10 x 56/176 were used to make an near unbustable 1000 watt @ 4 ohms stage.

The Semelab dual chip types are even more robust.

......... Phil

Not Crest by any chance ? Those 2r2s IIRC ?

They're actually there to stop VHF oscillation ( a la mosfet ).

Graham

< snip .

Hmmmm. Yup, you're right.

In the low MHz region IIRC.

Don't recall now. Would have made sense. Most of the Rs around my own bipolar outputs ended up as fusibles.

I experienced the oscillation problem myself once btw when I suspect On Semi changed the process on the MJ15023/4s.

Thanks for the good info btw.

Graham