I am not looking forward to the timing belt change on my Mitsubishi. I guess I'll let a shop take care of it.
HC14. Way to go. That's how I did a lot of switchers. Usually after having to design out some vaporware chip. If you want to pinch a few pennies you could use a TL431 instead of the TLV but it needs more current.
Thanks. Always good to hear from someone who has actually used a company. Seems they still have comms difficulties after the earthquake damage. Another service that is less often to come by is a place that can reliably assemble fine pitch SMT in prototype quantities, 0402, 0201 and stuff like that. A place that has the usual jellybean parts on hand.
"Terry Given" schrieb im Newsbeitrag news:1169346205.482736@ftpsrv1... | >
=PN/5332938>
| > | or | >
| > Much shorter here: | >
formatting link
| >
| > - Henry | | interesting, but not what JL was talking about. This circuit wont work | very well at all if you intend the MOSFET to stay on for any length of | time, as the shunt inductor will bugger up the gatedrive. Useless for | the majority of FET drivers. | | I'm also pretty sure I have seen a very similar technique before, so the | patent is probably worthless. | | | John, why stop there - why not a bridged-T network....
Yes. Seems technical a little difficult.
I copied the patent to a pdf only. It is the same!
news:45b29698$0$9772$ snipped-for-privacy@news.>> | > It has just ocurred to me, after all these years duh, that if you
Tha patent is goofy. If the magical current source has enough peak current capability and enough voltage compliance to whack the fet gate adequately, you're home free already.
Bridged-tees are great, except that I don't understand them. Can you suggest a circuit?
A simple series L will change the gate step from first-order (exponential) to 2nd order (critically damped or underdamped), which means it will slew faster when it does get going and get to +5 quicker, maybe even overshoot a tad. This is classic "series peaking". It could be useful when gate drive voltage is limited, as when using a cmos gate as the driver.
Without the L, you might have...
_____________ / / / / / / __________/
at the gate, and with it,
___ / \\_____________ / / / / / / __________/
or something.
The Miller plateau changes things some, but the L still might help.
Oh, ok, I'd forgotten that it was a shunt - thought I remembered a series inductor. The general idea of using L to boost the initial gate drive was what I recalled.
You could handle the DC component by providing an R/C parallel pair (virtual ground) in series with the L...? The whole thing is kinda like steering a bicycle - initial left to lean for a RH turn.
Luckily my Mazda Bongo Friendee (what a name!) is a doddle. I was likewise surprised when I removed the radiator in 5 minutes flat. c.f. an old mini.....
I've simulated the controller, and hacked up a dead-bug version to prove it pretty much works (large cap instead of switch), but bunging one inside a 10kW psu might be a slightly different story, hence the UC38xx "oops" plan.
My first lot of PCBs was ordered Dec22, arrived Dec29. Luckily CourierPost neglected to pop a card in my letterbox (I was away at the time) so I didnt know they were sitting at the post office until 7 Jan, when I went a-huntin'. I'm very happy with the PCBs, they look good, and were cheap.
I know of a place in Auckland, but I bet that wont help you ;)
where L-r-c is the LRC load to be peaked, and L1 & L2 are coupled (but not the same). Addis states "Bob Ross wrote the equations for the asymmetrical bridged T coil which could peak a series RLC load. The derivation is quite complex, and the equations were held as a trade secret (by Tektronix) until quite recently". AIUI Zin = R.
its kinda analagous to a technique I have read about recently, which uses the -ve mutual inductance of a pair of coupled coils to cancel out the inductance of a capacitor.
I havent gone thru the math yet; I looked at using it a couple of years ago to speed up a laser driver, but went for brute force instead (it was cheaper), and so never did the maths.
add a pair of diodes to the npn/pnp emitter follower, fiddle with the bases separately, and voila - resonant (lossless) gate driver, that also has the advantage of doubling Vcc into Vg, so FETs turn on harder. Shame about the added base drive complexity but.
-- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona Voice:(480)460-2350 | | | E-mail Address at Website Fax:(480)460-2142 | Brass Rat | |
formatting link
| 1962 | I love to cook with wine. Sometimes I even put it in the food.
Why not? I found they work as well or better than some PWM chips. Of course, with a 10kW supply the cost savings will vanish in the noise.
Who knows, maybe some day. Right now I need one closer, preferably near Los Angeles. Or some place where customs won't hold up the shipment of the non-jelly-bean parts to the shop. Which kind of rules out at least Australia from what I heard.
I've pulled several cute tricks, therefore Murphy is gunning for me. I'm pretty confident it will work, and extremely well (I can go a LOT faster than a UC38xx). But the effort required to slap in a UC38xx is negligible, as is the additional cost of the little sub-board I have hacked up (which actually sits in an unused corner of the main PCB, and will get sliced out if need be)
IME providing such a backup plan guarantees the original will work.
indeed, but the "damn, no stock of that magic part" solution remains ;)
and NZ too ;)
Customs can be relied upon to provide such an inconvenience!
How can any kind of impedance in series with the gate (or in parallel, for that matter) improve risetime? Unless there's more to turn-on speed than just pumping as much charge into the gate as quickly as possible.
Hmmm... or does it work like this: At first the gate drive current is indeed hampered a bit by the series L, but as the gate voltage approaches the driver voltage the current is kept going strong by L's freewheel action, which more than compensates the sluggish initial turn-on?
McEwan patented a lot of "borderline" stuff, like driving a common circuit from another common circuit and calling that an invention. He did some fairly original work in impulse radar while at LLNL, and left to start his own company to commercialize it, but I haven't heard much from them. I recall something about IP disputes on the radar stuff.
Yeah, here they are...
formatting link
See "about us" for info on the "world renowned" inventor.
I note that the mandatory "global leader" appears in its proper place.
Right. The step response at the gate changes from first-order exponential (Rgen times Cgate) which is sloppy as hell, to 2nd order, much crisper and steeper. If
L = K * r^2 * c
and K something like 0.5, the step response becomes critically damped, which is much nicer and sharper past about mid-voltage. If K=1, it's underdamped and will actually overshoot the drive voltage, great for turning on a fet from a logic gate that's marginal on voltage swing. Too big an L will of course slow it down again.
This is the first baby step towards a transmission line:
Consider a 1 nf capacitor being driven from an infinitely-fast, 50 ohm step generator. Tau is 50 ns, so risetime is about 2.2 times that, 110 ns. Now try the same thing with a 100 foot piece of coax that has the same capacitance; the output step is, ignoring losses, as fast as the input. The inductance of the trans line "cancels" its capacitance.
Another way to look at it is that the inductor stores energy to allow a steeper charge of the cap, essentially trading time delay for risetime.
You can simulate a step generator + resistor + inductor in series, charging a grounded cap, and play with K. It's sort of cool... you can see the tradeoff between risetime and delay. If K=0, it starts off steep but peters out. As K increases, it starts slower, gathers its strength, then shoots up.
Or look at it from a series resonance point of view. That's essentially what makes wireless power transfer work, except that there it is the leakage inductance that you'd be compensating with a series cap.
I like your trading more delay for less risetime. This only works in a coax if the time delay of the coax is greater than 4*Tr of the source. Give me an infinitely long coax an I can make DC look like a step function. One of my old PhD bosses wanted me to believe that coax's characteristic Z was constant down to low frequencies. Good theory but we still don't have a higher speed replacement for the
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.