It works for me, do you have IPv4 disabled or something?
:|
As the transistor heats up (while switched on or in desat), that current saturation will rise significantly...
Why would you let the bEMF just freewheel? I don't get that.
Tim
-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website: https://www.seventransistorlabs.com/
I like LEDs as the reference in current sources. Good tempco, and they glow!
-- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
Oh yeah, that's novel.
Now it works. Previously the session just timed out.
The datasheet shows that the saturation current does not depend much on temperature, at least not to the extent that could change anything in this application.
Table 14: for V_GS=6V I_DS=20A from +25 to +125 degrees. I'd say it's much better than necessary for a limiter.
What else could you do in the case of a pair of MOSFETs connected in series with a coil? The peak voltage is 325V, so clamping the coil with a bidirectional transil for, say, 350V would rise the MOSFET DS voltage by at least that value. The waveform is sine, so you can't introduce an "easy" back EMF discharge path, like a unidirectional transil, paying with more seconds to buy less EMF volts, because you never know which current direction is the right one at this very the moment. Well, actually you can, with a synchronous clamper, but this cure seems to be worse than the disease.
It might well be that I don't understands your concerns.
Best regards, Piotr
Well, you sure as hell can't simply let it go...
...Not for many microseconds, anyway. :-)
Well, a TVS at least saves the MOSFET, though it still doesn't buy you much energy. Still, good enough for a lot of coils, one-shot.
I put the largest coil I can find on my electronic fuse, and it didn't do much of anything. Turns on, takes milliseconds to charge up to rated current, tick, current discharges through TVS. Boooring. Short circuit testing (the waveform pictured) was a bit more interesting but just as uneventful. I tested thousands of cycles (repeat rate 5Hz) with no change in waveform.
My devices seem to work fine. :)
Tim
-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website: https://www.seventransistorlabs.com/
The main missing remark is that I'm not going to built it that way for exactly this reason. The AC fuse will be inductorless because there is no good way to deal with the back EMF. The old good serial shunt will be used instead. There is no such issue in the low voltage case, hence the 5uH inductor clamped with a unidirectional TVS or even a Schottky -- the back EMF always has the "wrong" sign.
But as far as I can see, they are low-voltage DC, at least the last one. This simplifies a lot on the voltage path (and complicates the current path considerably, no free lunch...).
Best regards, Piotr
What kind of loads will this be used on, that don't have >5uH on the AC side? Or less, for that matter.
The one isn't as easily scaled up because it depends on device SOA, but the switching one I can simply tack on more gate drivers, bigger transistors, inductor, diodes and TVSs. An industrial version (240V AC, DC bus up to
400V, precharge / soft start, etc.) would be about the size of an open hand (well, my hand anyway, which is above average).Tim
-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website: https://www.seventransistorlabs.com/
There is no "AC side", just "crippled DC". Direct, pulsating current due to the lack of filter capacitors durable enough to satisfy my requirements. If it's gonna fail within a decade, better not to have it at all from the very beginning. Call it a "universal power bus", a load must be able to work with that or perish. Mostly lamps of various kind, in quantity, but also various secondary electronic modules.
The magamp-driven rectifier is a part of that.
Durability is the name of the game, no stored charge devices, please...
Best regards, Piotr
Note that when you are using the FET to ramp up the current, you will see a different signal the first time as opposed to the following times it is turned on (remanence effect). This may lead to wrong conclusions on inductance and saturation limit values
Cheers
Klaus
Hmmm, so AC, rectified (hence pulsating), and using that "DC" bus as the supply for various loads, which hence need to be fused individually as the total bus capacity will greatly exceed what any load should see?
They make DC-capable fuses (500VDC at some amps, like a 3AG style, are pretty affordable, or else the industrial midget fuses are >$10), of course that has the problem that... you have to clear the fuse too. :)
Have you given any thought to DC-specific problems, like electrolysis (if moisture gets in) or dust collection (more a problem on kV+, not sure if it matters at ~100s V, really)?
What kind of lamps? LED I suppose? Or, anything else inverter based, including a lot of halogens actually (but not halogens on DC, because incandescent lamps on DC suffer from Edison effect).
And then, you'll still want /some/ stored charge, but more about EMI filtering than mains ripple filtering. Unless the end device does it, in which case, you would have the problem of starting a capacitive load attached to this rail.
Or it's completely different from this extrapolation...
Tim
-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website: https://www.seventransistorlabs.com/
Yes, the loads need to provide appropriate protection on their own, if possible at all. The electronic devices are usually low power ( They make DC-capable fuses (500VDC at some amps, like a 3AG style, are
Most of the available fuses are mechanic devices, and hence are insanely slow. 8ms reaction time is enough to fry the short-circuited MOSFETs a hundred times or so... :-/
The PSU must be disconnected promptly, well within the SOA of the involved devices. That wouldn't be that hard using a peak-current mode control switcher, but there is also a secondary power supply (a Pb battery) which kicks in during the mains failure. So either an old school serial breaker, or a mother of Godzilla SEPIC. The first one seems to be far simpler.
The moisture is not a problem thanks to the environment, the dust is kept away by a proper casing and the lack of ventilation slits. The best way to fight with heat dissipation is not to produce it in the first place.
Mostly halogene bulbs due to their far superior human perception. Will be replaced by LEDs when/if they reach the halogene quality.
You name it. If it can eat pulsating 12V DC (RMS) and consume less than
10A per channel, it's fine.Not a practical problem, as 3 years of continuous operation witness. I think the bulbs are even more durable compared to the ordinary AC supply, because of the implemented softstart mechanism.
I don't exclude all capacitors, just the evaporating ones, which happen to be the highest capacity beasts. I just don't want them in my system, which is designed around this requirement.
Indeed, but it is just a form of a softstart.
Best regards, Piotr
Gee, that's new one on me. The FET just slams that node to ground; the inductor's inductance does the ramping.
Cheers, James Arthur
Yeah, it's very normal for high permeability cores, and anything with high remenance obviously -- but you might not know that, if you picked up some random epoxy-coated core that you didn't know was actually stripwound Permalloy or whatever. The rapid saturation (in steady state) tells you something's funny here; and if nothing else, it tells you this part is no good for winding inductors (for energy storage) on!
Tim
-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website: https://www.seventransistorlabs.com/
Reverse the leads and do it again.
-- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
The right way to do it to apply a decaying ringing field, and then capture the first pulse applied again
Cheers
Klaus
Oh, residual magnetism in the *core*, sure. I thought he meant in the *FET*.
Cheers, James Arthur
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.