2 volts is usually enough, if you take the bottom of the ripple into account.
The 317 has a maximum input voltage limit above which it might fail, even at low load current. It also has a temperature limit, though there is a feature in the chip to lower the output voltage as this is approached. Best to stay well away from both of those limits.
Sometimes it is cheaper or more reliable to dump heat from a resistor than to couple that heat to a heat sink (clamping force, thermal transfer goo, possible electrical isolation, etc.) and dump it there. But if you stay well within the data sheet specs, it is functional to dump heat through the silicon.
Both.
It's done all the time. In your case, the die must dump .3*9=2.7 watts of heat without exceeding its maximum temperature. That might take a few square inches of heat sink surface with exposure to room temperature air.
Depending on what you're building, you might consider a switching supply to save on total device power consumption. I'm thinking: EnergyStar-compliant type stuff here. LM317's are cheap, though. So if your product design doesn't benefit from EnergyStar or similar (in some cases regulatory requirement), then a linear design might prove the best option.
Lose some of that heat by choosing a filter capacitor to _maximize_ ripple but trough still staying above drop-out.
...Jim Thompson
--
| James E.Thompson, P.E. | 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 |
"The American Republic will endure until the day Congress
discovers that it can bribe the public with the public\'s money."
- Alexis de Tocqueville
On a sunny day (Wed, 22 Oct 2008 08:40:15 -0700 (PDT)) it happened snipped-for-privacy@gmail.com wrote in :
OK
Will cost you an extra resistor.
IIRC 40V is the max over a LM317. P = voltage x current, so 9 x .3 = 2.7 W or so in your case. Junction to case = 5°C / W, max junction temp = 125°C At 50°C amb that leaves for 3W: 125 - 15 - 50 = 60°C for the heatsink.... so at 3 W the heatsink should be better then 20°C/ W, try 10°C / W.
when you say transformer, do you mean 19 volts before rectification?
If so, beware that after the conversion to DC, you'll have a little less than ~ 19*1.414 DC. this puts you in the 27 DC range.
So, the W = (27-10)=17*0.300 = 5 watts.
That maybe ok with out a heat sink. It's been a while since i've checked that Reg..
Also, it maybe a good idea to place a back flow diode on the input and output in the event that the input should drop below the output.. This helps save the reg.
Those figures may not be exact but they are close enough..
That method worries me, because it depends on the stability of the capacitor under high ripple current conditions. I like adding the series resistor upstream of the rectifier, or even upstream of the transformer. Especially with cheap transformers that barely have enough iron to avoid saturation, the last option can really cool off the transformer as well as the regulator.
Ok John -- I'll take your advice...it bothers me to because I always hate excessive heat in any device.
You've enlightened me to an interesting word as well...."upstream".
Often times, this resistor being an example, I should take more time to think about the "best place" to install the component.
I would have immediately put it in front of the regulator...yet putting it in front of the rectifiers I suppose makes all the components happier.
It's really just a simple 555 switching circuit that the supply is for.
Can you clarify the filter cap stability thing for me? Are you saying that excessive voltage on the regulator input could make the cap work harder than it needs to or something?
Another thing you could do is to put the resistor between the rectifier and the filter capacitor. This will give very low ripple voltage, and low peak diode and transformer currents at the expense of really poor regulation. Pick a resistor low enough in value to meet minimum headroom requirements at maximum load. You can not calculate the resistor value from a DC analysis; it will end up at around 10 % - 20% of what it would be if it was after the filter capacitor. Tam
Makes better sense from the energy flow viewpoint.
It lowers the RMS current in the transformer winding, rectifier and capacitor back there. This helps the rectifier the least, since it has a voltage drop little changed by instantaneous current, so its heat is almost proportional to the average current, and the resistor doesn't change that.
The core losses in the transformer are not much related to the load current, but to the voltage applied to the primary. Putting the resistor upstream of the transformer has a similar effect on winding and capacitor RMS current, but also lowers the primary voltage by the amount of voltage drop across the resistor, and that lowers the core losses, too by reducing the magnetic flux swing each cycle. If you can find a fusible resistor (one that fails gracefully during an overload, rather than bursting into flame) it makes a nice fuse, too.
Still, a good excuse to think about supply principles and choices.
Jim's suggestion, to lower the regulator average power by using the smallest functional filter capacitor (lowering the average regulator input voltage by increasing the sag between rectifier peaks), works, but relies on the capacitance being stable over time. Small capacitors with large ripple currents tend to dry out and lose capacitance over time, and you risk the ripple voltage sagging low enough that, before long, the regulator output would drop out of regulation during those dips.
I like having a comfortable extra capacitance in the filter to keep operation functional even if half of the capacitance eventually disappears.
[using an input resistor to reduce three-terminal regulator heat]
It's oddly hard to make that work. When I tried, it always turned out that the smaller filter capacitors weren't rated for enough ripple current. The ripple current 'rating' is hidden in the data sheet somewhere, I found that the cheapo filter capacitors only allowed 1A per thousand uF of capacity.
--
| James E.Thompson, P.E. | 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 |
"The American Republic will endure until the day Congress
discovers that it can bribe the public with the public\'s money."
- Alexis de Tocqueville
I only did that once (1961). Kept my beasty toooob amplifier up in the attic of our apartment (third floor of an ancient house in Cambridge (corner of Tufts and Magazine Street, two blocks from Stop & Shop, for the Cambridge lurkers).
Used 0A2's as shunt regulators.
Went off to WV on vacation for two weeks, so I turned to toooob amplifier off.
Came back from vacation and switched on the power, forgetting that
0A2's, in the dark for a long time, can fail to strike... kaburp, "confetti" blown out the bottom of the big filter capacitors ;-)
Lesson learned, shine flashlight on 0A2's before applying power. Now I know there are versions with a small amount of "nuke" to aid initial striking.
...Jim Thompson
--
| James E.Thompson, P.E. | 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 |
I love to cook with wine Sometimes I even put it in the food
I remember those. I used them for screen bias and so on in some of my aforementioned 1970s tube projects.
That happens with flashtubes too--horrible timing jitter due to the stochastic character of the first few ionization events. The solution there is to run the flashtube in OA2 mode constantly--the so-called 'simmer circuit'. I didn't know that about OA2s, though. I wonder if it depended on the gas fill...something like a corona point with 10M in series with it would have solved the problem as well.
"really bad for the glass" is a line from a movie. Anyone know the name of the movie?
Actually haven't done any chips with nitride caps... mostly "poly".
...Jim Thompson
--
| James E.Thompson, P.E. | 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 |
"The American Republic will endure until the day Congress
discovers that it can bribe the public with the public\'s money."
- Alexis de Tocqueville
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.