--
It\'s a link to USENET and it\'s supposed to take you to
alt.binaries.schematics.electronic where I posted some PDF\'s of
incandescent lamp surge current waveforms for you.
If it doesn\'t work, then just go over there manually and check the post
with the same subject as this one.
If that still doesn\'t work then you probably don\'t have access to the
alt.* groups through your ISP.
If that\'s the case, email me your email address and I\'ll get the stuff
to you that way.
JF
Select a power resistor value such that the bulb filaments are just starting to glow. Shunt the FETs with them (one each). This not only dramatically reduces the inrush current, increasing the life of the bulb, but you get a much "snappier" turn-on (i.e., the bulb reaches incandescence quicker.)
If you can't find the right power resistors, then lurk ebay or whatever for some nichrome resistance wire, and make your own. :-)
I use news.albasani.net most of the time and access alt.* groups regularly. I also use eternal-september which I believe does binaries. The link didn't work with either (I know about abse but have never used it). I even tried GG with a browser. Anyway, I'm sending the email. Thanks.
That's exactly what I mentioned as a possibility in my reply to Jim Thompson. The drawback is that it will keep draining battery power in standby, which is where it will be more than 90% of the time.
Actually, I did mention it in my reply to Jim Thompson's post. Only I didn't use the term PWM then.
But as with paralleling the MOSFET with a resistor, it will be a continuous drain on the battery in standby mode. This is very significant because, out of 14 groups of bulbs, a maximum of only three groups will light up at any one time. And those times will be much shorter than the standby periods.
I measured the cold resistance of a 55W halogen (auto).. 0.2 ohm at
22C. Racing can happen in colder weather so there could be less R. The initial current I'm guessing will be about 60 amps. The initial power of the mosfet is about 63watts. The warm resistance is 2.6. The final power of the mosfet is 372mW. Assuming 1 second to reach max brightness. Making the false assumption of linear dynamic load, the energy put into the mosfet is 1/2*(1sec)*63Watts = 31joules (Not sure if that's right.) So... what sort of mosfet heat rise is to be expected from 31joules... I dunno.. But 31 joules of energy might raise the temperature of a Canadian penny 31.4C degrees (If I did the math right..)
Huh?? The load is wayyy more dynamic than the mosfet... iows.. For this app, assume your mosfet instantly turns into a resistor. All your heat is from the dynamic load current^2* Rdson. The material in the mosfet will rise in temperature depending on the internal energy and thermal conduction to surroundings.
One thing you might watch out for, if there is any over-current protection involved, are circuit stresses occuring if limiters act during the initial surge. This is easy to overlook.
On a sunny day (Sat, 5 Dec 2009 11:42:24 +0530) it happened "pimpom" wrote in :
Here is a differnt solution:
SRC parallel with MOSFET. To switch bulb on, trigger the SRC, a pulse a few ms wide will do. That SRC will handle the surge caused by the light bulb's cold resistance no problem. To switch the bulb off, for a short moment activate the MOSFET. The MOSFET will pull the SRC voltage close to zero, the SRC will stop conducting, the MOSFET only handles the current caused by the higher 'hot' resistance of the bulb, and that only for a few ms seconds. All you need now is an 'on' pulse to the SRC, and an 'off' pulse to the MOSFET. Much simpler. Of course, to limit voltage drop and heat in the SRC, you can also switch and keep the MOSFET on after the first few ms. Timing for that could be easily done with a PIC. hehe
I assume you meant "SCR"? Technically, it's a nice idea. Can't find any flaw in it. Perhaps a TIC126. However, I don't think I'll have time to make the substantial changes it entails - at least for this upcoming race. I'm in the process of putting the finishing touches to the pcb design. I'll certainly keep your suggestion in mind for later.
This brings up another question: Why are SCRs rated to withstand their maximum surge currents for a full mains half cycle while MOSFETs are so rated for microseconds?
On a sunny day (Sat, 5 Dec 2009 18:01:56 +0530) it happened "pimpom" wrote in :
Yes, sorry, typo.
Perhaps because MOSFETS are normally used in switchers, and SCRs in mains 50 Hz or 60Hz power circuits? Once on, in such a circuit, the SCR cannot be switched off, so 'no choice' for 20 or 16 mS.
One important thing you are overlooking... "surge" is not a _dissipation_ issue, it's a local hotspot, or wire-bond-blowing event. You're liable to end up with "cool but dead" MOS devices. ...Jim Thompson
--
| James E.Thompson, CTO | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
In wine there is wisdom,
In beer there is freedom,
In water there is bacteria
- Benjamin Franklin
I must say that I have the same problem as pimpom - click on the link but get nothing. What I have to do is use usenet replayer: (watch the line wrap)
formatting link
The effort to see your posts is always worthwhile, but having to use usenet replayer is a royal PITA. I don't think there is anything you can do at your end of things to change that. :-(
One thing I'll request: can you post a test jpeg on abse? The pdf's you post don't show up for me using news.eternal-september.org but jpegs do. I just want to make sure that the difference is jpeg vs pdf and not something else. Maybe Rich (or others who use the same thing) has better luck. Wouldn't be the first time I had a fouled up setting. :-(
I think you are right. In my example I used 50A, some people say 63A. In that case you exceed the SOA. He will ruin the chip in the end. So he don't need a heatsink, but a bigger (larger) mosfet (2 or three in parallel is also possible). Of course he can also use other methodes as suggested by others.
So continue the temporal thermal analysis. The curve is exponential like charging a capacitor. Do the rather simple integral for temperature rise. (to simplify assume no heat sink and theta ja)
Here is a really crazy idea: Can you build it such that the power transistors are thermally connected to a water mass? Like put the power tabs into a water tank? The thermal mass of even a little bit of water is truly impressive.
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.