With the full potential of the transformer, the LM317 yields 2.69W of heat when the pot turned all the way (3V output).
It yeilds 1.645W of heat under normal conditions (6.3V output / pot at 13%).
Now, I know nothing about thermodynamics but; my heatsink is aluminum and is an L shape that is found here. Also, it is 900mils tall.
I am no expert, but 2W of heat doesn't seem too bad with a heatsink.
I also read in the data sheet that an increase in heat produces less dropout voltage. That is good news!
It yeilds 1.645W of heat under normal conditions (6.3V output / pot at 13%).
Now, I know nothing about thermodynamics but; my heatsink is aluminum and is an L shape that is found here. Also, it is 900mils tall.
I am no expert, but 2W of heat doesn't seem too bad with a heatsink.
I also read in the data sheet that an increase in heat produces less dropout voltage. That is good news!
Ok, I have done extensive design work today. Here is what I have. I managed to fit my EQ circuit on the same board and cut it after. I wouldn't call it cheating.. I'd call it using my resources 
Unfortunately, I need to purchase this board soon. That means that I need to stop dilly-dallying and pick a design.
I'm pretty sure that I'll be using the full 15V CT design. It seems much safer rather than hoping that the mains doesn't drop too low... and 2W of heat should be fine for my little LM317 strapped to it's cute little heatsink.
A question I'd like to ask is; do heatsinks need to be grounded? If so, I will need to use mica insulators for some of my TO-220 parts. Also, do I need thermal compound?
Unfortunately, I need to purchase this board soon. That means that I need to stop dilly-dallying and pick a design.
I'm pretty sure that I'll be using the full 15V CT design. It seems much safer rather than hoping that the mains doesn't drop too low... and 2W of heat should be fine for my little LM317 strapped to it's cute little heatsink.
A question I'd like to ask is; do heatsinks need to be grounded? If so, I will need to use mica insulators for some of my TO-220 parts. Also, do I need thermal compound?
An externally hosted image should be here but it was not working when we last tested it.
Attaboy. Engineering. As my academic advisor once said, "you can hope it won't happen, or you can just design the circuit so it doesn't happen".
I think the engineering on the 15 V CT transformer solution is solid.
That's the only outstanding issue. So... 2 W from the LM317. How much power will be dissipated in the LM7815, LM7915? That's the total power dissipated in the heat sink.
Somewhere I did stumble across a graph or equation that showed the thermal resistance of an aluminum sheet vs area. You might try asking The Google for that... Or you could perform an experiment. Set up a transistor in a TO-220 package to dissipate, say, 1 W in the heat sink and measure the rise in temperature. That should give you the thermal resistance.
2~3 W is probably OK to dissipate in a strip of aluminum like the one you've drawn, but it would be nice to do the math.
A PCB heatsink could be floating. I personally prefer them to be grounded, but I don't think there are any safety requirements that demand this. However, if the heatsink touches the metal chassis - or is mounted to the chassis - it will need to be grounded.
You will need insulating washers and shoulder washers for the bolts. Recall that the middle pin on the TO-220 is connected to the tab of the package. On the LM78xx it's ground. On the LM79xx I think it's the input voltage. And if I recall correctly, on the LM317 it's the output voltage. So unless you want those voltages shorted together via the heat sink, you'll have to use insulating washers and shoulder washers.
I hate the thermal goop. I much prefer the silicone washers. It could be argued that at the relatively low power levels you're dealing with, you could forgo the goop. But if you're using mica washers, I'd probably do the goop as well.
~Tom
I made a little oopsie. I ordered a 12VAC transformer before I really thought about using a 15V regulator. The good news is that I can use an LM7818 and LM7918 by using the 24VAC CT on the transformer. It's current capabilities are 250mA. 24V/6VA.
I am powering 10 LM741 OpAmps for an EQ circuit, so +/-18V will give me more voltage swing!
The LM741 data sheet states that the MAX power consumption is 85mW. Therefore the total is 850mW MAX.
850mW = (18V*I); I < 50mA ???
I'm a little confused how to use a CT transformer for a bipolar supply with a positive and negative regulator..
I am powering 10 LM741 OpAmps for an EQ circuit, so +/-18V will give me more voltage swing!
The LM741 data sheet states that the MAX power consumption is 85mW. Therefore the total is 850mW MAX.
850mW = (18V*I); I < 50mA ???
I'm a little confused how to use a CT transformer for a bipolar supply with a positive and negative regulator..
Last edited:
Figures...
First off, LM741 was fine when it was first invented (it was the first IC op-amp ever) but more modern op-amps such as LME49710, OPA134, et. al. are vastly improved by comparison. There's a vast improvement both in specs and sound quality between those and an LM741.
That sounds way off. Are you sure that's not the max power dissipation allowed for the package? The National LM741 spec sheet lists the supply current as 2.8 mA max for the LM741C. So I wouldn't think 10 of them would draw more than 28 mA. In fact... Depending on how much current you intend to drive with the op-amps, you could easily use the 78Lxx and 79Lxx series of regulators. The 100 mA, TO-92 versions.
Well... If you can get a positive voltage by pointing the diodes one way, wouldn't you think you could get a negative voltage by reversing the diodes?
You won't be able to hit 18 V, though -- even with the center tapped transformer. But ya know..... Modern op-amps and 10 V regulators wouldn't be so bad either.
~Tom
Slow slew rate -- I think it's asymmetrical as well. Low bandwidth. Low full-power bandwidth. High THD. Yeah... Lots of improvements have happened since 1967 when Fairchild designed the uA741 (later LM741 when National bought Fairchild).
Yep.
Except you don't have 24 V available. You have 2x12 V or 24 V CT. So you can make 30-ish V and run the op-amps on a single supply. You'd have to cap couple inputs and outputs. I would not recommend that route.
Or you can make +/-14-ish V by hooking a diode bridge from end-to-end on the trafo and grounding the center tap. Accounting for some ripple, you just don't have enough headroom for an LM7812/7912 +/-12 V regulator, but you do have enough for an LM7810, LM7910 +/-10 V regulator. That'd be my solution with the parts you have.
Or... You could run the supply to the op-amps unregulated and let the op-amp's PSRR take care of it. Of course, then you'd have to guarantee that under worst case conditions (mains AC at 126 V, no load, etc) the supply voltage won't exceed the ABS MAX for the op-amp.
The 78xx series needs 2.0 V drop-out. So 323 mV won't cut it.
The power dissipated in the regulator is (Vcap - Vout)*Iout. Figure a couple of volts at 28 mA --> 100-ish mW. No heat sink needed. However, that does not take load current into account. The op-amps will supply current to whatever they're driving. If they're driving a high-impedance input, say 10 kOhm, you can neglect the current (a few mA her and there), but if they're driving, say, 600 ohm it's another story.
Welcome to power supply design. It's not complicated, but there are a lot of i's to dot and t's to cross. Do it right. Do it once.
~Tom
Haha, yeah. Didn't anticipate these many factors. I expected to use a transformer>rectification>cap>regulator and move on. Not the case!
The board needs to be ordered by Friday or so, therefore I need to plan to finish making choices.
I was interested in getting as much voltage swing as I can since I want the EQ to actually be audible and effective. Maybe I will still have a decent amount of boost/cut even with +/- 10V.
I am absolutely wanting to use VRs, it's just which ones.
The board needs to be ordered by Friday or so, therefore I need to plan to finish making choices.
I was interested in getting as much voltage swing as I can since I want the EQ to actually be audible and effective. Maybe I will still have a decent amount of boost/cut even with +/- 10V.
I am absolutely wanting to use VRs, it's just which ones.
No need to tape out garbage, though. Engineer it. Make sure it will work. Then tape out.
One simplistic way of looking at it: The difference between 18 V and 10 V is 20*log(18/10) = 5.1 dB. So you'll have 5 dB less boost available. Cut is no problem.
You won't need more than 1~2 Vrms to drive a power amp into saturation. So the output voltage isn't an issue. Whether you'll saturate some of the intermediate stages in the EQ circuit could be found out by doing the math or running a sim.
With the tranny you have, you can get +/-10 V regulated (with LM7x10's). If you find a different set of regulators, you may be able to push that up some. You know how to do the math on that one and I've pointed you to sites with LDOs. I suspect you'll find that positive regulators are easy to find, but negative LDOs are not as plentiful. But even then, you won't get much past +/-12 V. If you absolutely want +/-18 V, you'll have to use a different trafo. You know how to calculate the voltage for it. Just do the math...
Anyhow. Best of luck with your project. I hope it turns out well. During the past couple of days you've been through the hoops of power supply design. You have one supply left to design (the one for the op-amps). You know how to do it. You just need to get off the computer, onto a piece of paper, grab a calculator and do the work.
Have fun.
~Tom
As a general rule of thumb, I create a 1 ma drain, so I take the voltage and multiply by 1,000 to get the resister value, so for a 300 volt power supply use a 300,000 Kohm resistor. I place one at the end, the last capacitor, to get a quicker discharge I will use a second one one capacitor upstream. So when I model a power supply I add the one or two 1 ma currents to the model.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.