The back of the 317 and the heatsink are not infinitely smooth, so when you put them together there are small air pockets that hinder heat flow.
Thermal paste does fill the holes and has low thermal impedance so helps to transfer heat from 317 to heatsink, so it helps cooling.
Jan
Thermal paste does fill the holes and has low thermal impedance so helps to transfer heat from 317 to heatsink, so it helps cooling.
Jan
I use Sil-Pads
Mouser.com link
They are fairly good thermal conductors and excellent electrical insulators. And they are a LOT less messy than mica insulators + thermal conductive paste "goop"
Mouser.com link
They are fairly good thermal conductors and excellent electrical insulators. And they are a LOT less messy than mica insulators + thermal conductive paste "goop"
The thermal paste helps of cause, but it is really needed only for heavily-loaded parts (more than 5-10 Watts of dissipated power, I think).
As for me I use thermal electro-isolating gasket with such small-power parts (to isolate part from heatsink and to make the whole device safer).
As for me I use thermal electro-isolating gasket with such small-power parts (to isolate part from heatsink and to make the whole device safer).
+1.I use Sil-Pads
Last edited:
If the devices are barely getting warm, then it makes little difference. If they get hot, then use the TIM. By hot, meaning too hot to touch.
I dont have any paste at hand right now, but i have few silicone to220 insulators so ill try them.
Also I ordered 50pcs of these m3 screws for 0.5$ lol.
I think 5mm lenght will be ok, I measured and it should be fine.
I think 5mm lenght will be ok, I measured and it should be fine.

If the LM3*7 has an insulated tab then a SIL washer is actually detrimental.
In this instance a smear of THC is the better option.
Those heatsinks are usually isolated from the remainder of the PCB so no insulation is necessary.
In this instance a smear of THC is the better option.
Those heatsinks are usually isolated from the remainder of the PCB so no insulation is necessary.
The heatsink tab is connected to the output pin. The need of thermal paste, or even a heatsink at all depends on your application.
Did I miss the answer to the question: "Is it getting hot?"
If you can hold your fingers on it then the answer is "no - you don't even need a heatsink"
If you cannot hold your fingers on it, then the answer is "yes, you need a heatsink. If you DO need a heatsink, then you need either a SILPAD or insulator bush and heatsink goo. "
If you can hold your fingers on it then the answer is "no - you don't even need a heatsink"
If you cannot hold your fingers on it, then the answer is "yes, you need a heatsink. If you DO need a heatsink, then you need either a SILPAD or insulator bush and heatsink goo. "
Well I didnt run it yet, but it should be used for 30mA to 1A loads, so I should use heatsink anyway. Heatsinks arent connected via traces to anything on pcb so I shouldnt care about insulating. I will use silpad because I dont have any paste right now and its cleaner and easier.
What will the input/output voltage differential be? That number times the current draw will tell you the dissipated power...up to about one watt usually won't require a heat sink, assuming there's enough ventilation.
Mike
Mike
In emergencies (late Saturday night finishing a Guitar amplifie which will be used on stage 1 or 2 hours later) I have used plain, no mineral load grease I had on hand, either silicone based tape roller grease or plain red Lithium grease, what your car or water pump uses, based on it being stable at car engine temperatures.
The idea being that *anything* is better than air pockets.
After all excelent thermal insulator Styrofoam is nothing but "air pockets in a matrix".
Always with instructions to Musicians: "bring it back during the week so I can apply proper paste"
Predictably they never ever bring them back, since they "work fine".
A few need servicing, years later.
When replacing transistors, I notice silicone grease stays as new, red lithium grease has somewhat hardened but if transistor is not removed, although it does not flow easily any more, it stays in place doing its job.
In any case, when replacing transistors, it will be wiped away and fresh grease applied, the proper kind this time.
My point?
Even non "thermal" grease (meaning one with zinc oxide dust) works reasonably and better than no grease at all.
The idea being that *anything* is better than air pockets.
After all excelent thermal insulator Styrofoam is nothing but "air pockets in a matrix".
Always with instructions to Musicians: "bring it back during the week so I can apply proper paste"
Predictably they never ever bring them back, since they "work fine".
A few need servicing, years later.
When replacing transistors, I notice silicone grease stays as new, red lithium grease has somewhat hardened but if transistor is not removed, although it does not flow easily any more, it stays in place doing its job.
In any case, when replacing transistors, it will be wiped away and fresh grease applied, the proper kind this time.
My point?
Even non "thermal" grease (meaning one with zinc oxide dust) works reasonably and better than no grease at all.
This is a bit late....
I just noticed that you want to drive 12+12AC in and get +/-15VDC out.
This "should" work... mostly.
12VAC will rectify to 12*1.414V = 16.9 V (call it 17V)
Take out 0.6V for the diode drop leaves you 16.4V.
Note:
- The transformer will deliver more than 12V at low loads, falling to 12V at full load
- Your mains voltage will genrally vary by +/-10% or so, depending on where you live
The LM317 input-output drop (dropout voltage) ranges from 1.5V (at 200mA) to 2.0V (at 1A).
You may find that in some circumstances the output loses regulation, especially f the input mains is a touch low or your load high. You might want to drop the output back to 13 or 14V.
The good news for you is that heat will not be such a problem 🙁
I just noticed that you want to drive 12+12AC in and get +/-15VDC out.
This "should" work... mostly.
12VAC will rectify to 12*1.414V = 16.9 V (call it 17V)
Take out 0.6V for the diode drop leaves you 16.4V.
Note:
- The transformer will deliver more than 12V at low loads, falling to 12V at full load
- Your mains voltage will genrally vary by +/-10% or so, depending on where you live
The LM317 input-output drop (dropout voltage) ranges from 1.5V (at 200mA) to 2.0V (at 1A).
You may find that in some circumstances the output loses regulation, especially f the input mains is a touch low or your load high. You might want to drop the output back to 13 or 14V.
The good news for you is that heat will not be such a problem 🙁
This is definitely right (except if we use diode bridge you have to subtract not 0.6 V but 1.2-1.6 VDC).
Usually we can't have +- 15 VDC from 12 VAC. Even 15-0-15 VAC can be insufficient for high-drop regulators like LM317. It depends on a transformer too (its output resistance etc).
LM317 usually needs about 3.5 VDC voltage drop.
So , in general case, to have well regulated +-15 VDC we need 2×18 VAC transformer (even not 2×15 VAC).
2×15 VAC is suffucient with a light load condition (when transformer is loaded at 20-40% of its maximum).
Usually we can't have +- 15 VDC from 12 VAC. Even 15-0-15 VAC can be insufficient for high-drop regulators like LM317. It depends on a transformer too (its output resistance etc).
LM317 usually needs about 3.5 VDC voltage drop.
So , in general case, to have well regulated +-15 VDC we need 2×18 VAC transformer (even not 2×15 VAC).
2×15 VAC is suffucient with a light load condition (when transformer is loaded at 20-40% of its maximum).
Last edited:
I disagree that you can't get +/-15 volts regulated from a 15-0-15 transformer. That's 21.2 volts rectified; dropout voltage from an LM317 is at MOST 2.5 v according to the datasheet; adding in the MAX diode drop of 1.6v gives 17.1 volts. Even a heavily loaded Antec transformer at 2.9 amps out will still give 14.3 v AC, 20.2v DC so there's enough room for the diode AND regulator dropout.
Last edited:
You forget ripple voltage. Normally 2 x 15V AC will be enough for +/-15V but not with a wide margin. Normally we distract 2V for the voltage drop of 2 diodes and we calculate ripple voltage based on the size of the filter cap and mains frequency. Just to be safe one would choose a 2 x 16V or 2 x 18V transformer. I would try out 2 x 15V just to keep heat lower and then measure stuff to check if I am within limits. I would also use Schottky diodes and overdimensioned filter capacitors to keep ripple voltage low. In practice this nearly always works out OK with low loads.
2 x 12V transformers can only be used when an ideal rectifier is used and uLDO regs.
Yesterday I finished a 15V 1A uLDO PSU with a 50VA 12V transformer. The transformer gives 13.2 to 13.5 depending on mains voltage. It almost does not sag when loaded with 1.5 A. As it is of excellent quality I did not care about it being overdimensioned. It has a LT4320 rectifier with under 100 mV dropout voltage and a 5600 uF cap. The regulator is the TPS7A4700 with approx. 0.3V dropout. A few of the design goals are low heat and low loss hence the experiment with a 12V transformer.
With a today lowish 224V mains voltage and then 13.2V AC at the secondary winding it has 17V rms at the input of the TPS7A4700 (with an average load of 0.6A). Tight and I discarded it for being too tight. But as I was curious I still wanted to try it out and it is running now for hours and it still is within operating limits. I would not recommend this though. In my case it is the 50VA transformer that does not drop as the load is way under its capabilities. A 25VA would already get me in trouble I guess. The point being that the ripple voltage combined with the voltage sag can cause the regulator being unable to regulate anymore.
2 x 12V transformers can only be used when an ideal rectifier is used and uLDO regs.
Yesterday I finished a 15V 1A uLDO PSU with a 50VA 12V transformer. The transformer gives 13.2 to 13.5 depending on mains voltage. It almost does not sag when loaded with 1.5 A. As it is of excellent quality I did not care about it being overdimensioned. It has a LT4320 rectifier with under 100 mV dropout voltage and a 5600 uF cap. The regulator is the TPS7A4700 with approx. 0.3V dropout. A few of the design goals are low heat and low loss hence the experiment with a 12V transformer.
With a today lowish 224V mains voltage and then 13.2V AC at the secondary winding it has 17V rms at the input of the TPS7A4700 (with an average load of 0.6A). Tight and I discarded it for being too tight. But as I was curious I still wanted to try it out and it is running now for hours and it still is within operating limits. I would not recommend this though. In my case it is the 50VA transformer that does not drop as the load is way under its capabilities. A 25VA would already get me in trouble I guess. The point being that the ripple voltage combined with the voltage sag can cause the regulator being unable to regulate anymore.
Last edited:
To dotneck335:
Minimum datasheet differential (dropout) voltage is 3 V for LM317 (from TI, for example). And if you'll investigate it carefully you'll see that good ripple regection it have starting from about 3.5 V DC differential. Then we have to take into consideration ripple and possible mains voltage change.
So I can say from my experience that 2..2.5V differential is near enough for only low-drop type IC, like LD1086.
Minimum datasheet differential (dropout) voltage is 3 V for LM317 (from TI, for example). And if you'll investigate it carefully you'll see that good ripple regection it have starting from about 3.5 V DC differential. Then we have to take into consideration ripple and possible mains voltage change.
So I can say from my experience that 2..2.5V differential is near enough for only low-drop type IC, like LD1086.
I didn't dig into the circuit the OP used. I did have the impression that the output voltage was adjustable.
So:
-1- Option of choosing a slightly higher voltage transformer. I would think that at low loads 15+15AC would usually work.
-2- Tweaking the output voltage a bit lower - it is adjustable!
You do have some LDO options in regulators. They my drop in for those LM317's. I once started to look, but would be interested in other peoples experience.
So:
-1- Option of choosing a slightly higher voltage transformer. I would think that at low loads 15+15AC would usually work.
-2- Tweaking the output voltage a bit lower - it is adjustable!
You do have some LDO options in regulators. They my drop in for those LM317's. I once started to look, but would be interested in other peoples experience.
- Home
- Design & Build
- Construction Tips
- Does LM317,337 need thermal paste?