LM7815 power supply question

[nickb]

Since you're choosing more powerful components, why not go with one of these just to shoot the shit?

[ATTACH=CONFIG]49417[/ATTACH]

I'm sure you'll agree it would dissipate the same amount of power so we should expect the same result.

Still don't believe me huh? OK, here's a LM7805, 15V in well over 1A out. Also quite happy.

[jmaf]

Still don't believe me huh? OK, here's a LM7805, 15V in well over 1A out. Also quite happy.

[ATTACH=CONFIG]49419[/ATTACH]

I've tested hundreds of LM regulators in real production settings and they're not reliable at 1A.

Not much else I can say except that it's getting really pedantic to try and share some experience here and be confronted by random experimenters all the time.

[g1]

jmaf, I have great respect for your posts, but nickb is hardly someone I would call a 'random experimenter'. Never mind the implication that the forum is full of them. That seems somewhat of a low blow to be honest.

[jmaf]

jmaf, I have great respect for your posts, but nickb is hardly someone I would call a 'random experimenter'. Never mind the implication that the forum is full of them. That seems somewhat of a low blow to be honest.

I didn't imply the forum is full of them. I might have implied that I've been rebuffed by this kind of reply more than once where one thing is cherry picked out of a general post where you're trying to help.

nickb's entire point here is that a component survived some time running at max rated current on his rig. That's all. But if he ever builds anything like that professionally, it'll be back for repair in less than a month. My advice to the original poster is to not do that professionally, because I've seen these components break, for years, since they don't really stand their max rated current in continuous operation. I hope it helps the member who asked originally, that was the only point of me posting here, trying to help.

[g1]

Thanks for the clarification. I appreciate your sharing your practical experience.
I agree with not using a 1A regulator in a circuit that needs 1A most of the time. (In this case we are off on a bit of a tangent as the circuit requirement was stated as 500mA, but tangents are par for the course around here anyway. )
For a circuit that needed constant 1A, would you be comfortable with a 1.5A reg. or go with 2A ?

[jmaf]

Thanks for the clarification. I appreciate your sharing your practical experience.
I agree with not using a 1A regulator in a circuit that needs 1A most of the time. (In this case we are off on a bit of a tangent as the circuit requirement was stated as 500mA, but tangents are par for the course around here anyway. )
For a circuit that needed constant 1A, would you be comfortable with a 1.5A reg. or go with 2A ?

It depends. Would it be mass produced and low cost device? If so, then I'd go the extra mile to make it work with a 1.5 amp regulator to save costs. If it were a boutique/custom or personal hobby project with just one unit, then I'd use 2A or more, or whatever I found in the drawer in the latter case. How will it be used? Is it critical? Will it run 24x7? Does someone die if it fails?

For guitar amps I used to use the LM350 (3 amp) @ 1 Amp draw for preamp tube heaters.

[J M Fahey]

I've tested hundreds of LM regulators in real production settings and they're not reliable at 1A.

If you chop your post down to the useful part, which is what I left above, Karma will be better for everybody

FWIW I implied something similar, based on roughly similar reasons, but didn´t turn it into a pissing contest, just restricted it to the OP apparent needs

self quoting:

The board manual only says "10 - 15V DC power supply, 500 mA min."

Then it can be fed by a 7815.

You need at least 3V more than +15V feeding the 7815, including ripple, so aim at some 20/22V DC RAW to account both for ripple and drop under load.

If you can get a 15/16VAC transformer, rated 1 to 1.5A, it will be perfect, so you don´t need more than, say, 22/25VA .

All of this is very conservative, the board may not even take 500mA by a large margin, if a "500mA supply" was suggested.

Didn´t even mention 1A but was happy that it would take 500mA or less

And on the other side suggested reducing Raw Voltage input to decrease dissipation .

On the other side, understand JMAF´s position very well, very often walk away in disgust from other Forums (won´t mention names) when TV couch "designers" steer some poor OP straight into the quicksand pit and boo boo my efforts to save him ... oh well.

Just not here

[Mick Bailey]

It's interesting to see how components in mass-produced, commercial designs are run at or close to their maximum rating - including voltage regulators. So we see 15v supply rails with 16v caps, power amps with transistors run right to the edge of their SOA, 300v tubes run at over 400v, 3w resistors run at 2.7w. In many cases, this is all down to cost and the component manufacturer's device specification. Why install components in (say) a set-top box to ensure 25-year reliability when it only needs to last maybe 3 years? There also factors into this the predicted failure rate and warranty costs. Fitting higher spec components may give increased reliability but the cost may outweigh fitting cheaper components and taking a hit on the increased failures.

Think of all the Fender amps with cooked boards where the 16v zeners and dropper resistors are fitted. Fender have made them with that failure area from day one and they started to fail after about three years. They have to know after a short while that almost every Deville/Hotrod Deluxe would fail in that area. They just kept building them like that, year-in, year-out. They never relocated the resistors and stuck a couple of aluminium-clad 15W devices on a piece of angle bolted to the chassis. They never uprated the zeners and mounted them off the board surface, they never installed 'proper' regulators. They saved money instead. Perhaps they only need to last the warranty period - maybe 5 years design life but no more.

When it comes to 78xx regulators, I find these to be pretty robust. They're short-circuit and thermally protected. So what's the worst that can happen if it's over-current or thermally compromised? It just shuts down. I see them running so hot they'll sizzle spit and they run for years like that on tiny (or no) heatsinks. They must be right on the edge. Often, if the load is too high, they won't conduct in the first place. A ready-made 500mA board does not need over-thinking.

[jmaf]

It's interesting to see how components in mass-produced, commercial designs are run at or close to their maximum rating - including voltage regulators. So we see 15v supply rails with 16v caps, power amps with transistors run right to the edge of their SOA, 300v tubes run at over 400v, 3w resistors run at 2.7w. In many cases, this is all down to cost and the component manufacturer's device specification. Why install components in (say) a set-top box to ensure 25-year reliability when it only needs to last maybe 3 years? There also factors into this the predicted failure rate and warranty costs. Fitting higher spec components may give increased reliability but the cost may outweigh fitting cheaper components and taking a hit on the increased failures.

Think of all the Fender amps with cooked boards where the 16v zeners and dropper resistors are fitted. Fender have made them with that failure area from day one and they started to fail after about three years. They have to know after a short while that almost every Deville/Hotrod Deluxe would fail in that area. They just kept building them like that, year-in, year-out. They never relocated the resistors and stuck a couple of aluminium-clad 15W devices on a piece of angle bolted to the chassis. They never uprated the zeners and mounted them off the board surface, they never installed 'proper' regulators. They saved money instead. Perhaps they only need to last the warranty period - maybe 5 years design life but no more.

When it comes to 78xx regulators, I find these to be pretty robust. They're short-circuit and thermally protected. So what's the worst that can happen if it's over-current or thermally compromised? It just shuts down. I see them running so hot they'll sizzle spit and they run for years like that on tiny (or no) heatsinks. They must be right on the edge. Often, if the load is too high, they won't conduct in the first place. A ready-made 500mA board does not need over-thinking.

You're right about everything working on the limit. It's all about planned obsolescence and it moves our markets ever since Edison and friends came up with the idea. But that's a market thing, not technical good practice, and it creates mountains of electronic junk in Africa.

Quoting myself:

Would it be mass produced and low cost device? If so, then I'd go the extra mile to make it work with a 1.5 amp regulator to save costs. If it were a boutique/custom or personal hobby project with just one unit, then I'd use 2A or more, or whatever I found in the drawer in the latter case. How will it be used? Is it critical? Will it run 24x7? Does someone die if it fails?

If a fellow member is building a circuit from zero, why give advice to work on the limit when they can buy regulators for 3X, 5X the rated current cheaply? What's the point of running a toaster in a circuit?

Was it really bad advice to tell OP to not use 78xx's @ 1A?

Fender tube abuse is part of Leo's magic. But tubes are different beasts. Silicon fails sharply, vacuum doesn't. Abused tubes can last for 5000 hours which is a lot for a musician's instrument. 5000 hours is not a lot for a set top box or a PC USB port. They're different applications. Leo didn't have the cheap readily available components we have today, he had to adapt. It's bad electronic design, but it's what he had back then. He ran tubes like that because he had to adapt but then it became part of his tone recipe. Today you have 100's of different regulator IC's you cab buy by clicking a mouse, why use anything at 100% rated current if you're building one piece for yourself?

Didn´t even mention 1A but was happy that it would take 500mA or less

If anyone is curious, test a LM7812 to regulate 3 x 12AX7 150mA heaters. 450mA total, right? Safely under 50% Imax? Then take a thermometer gun and measure temperature after 30 minutes. You'd need a truck sized heatsink. Can't even fit a heatsink that big into most projects which demand 500mA. (PS. I also picked the useful part of your reply .)

[Mick Bailey]

If a fellow member is building a circuit from zero, why give advice to work on the limit when they can buy regulators for 3X, 5X the rated current cheaply? What's the point of running a toaster in a circuit?

1. He isn't building a circuit from zero.
2. I have never, anywhere, advised him to work on the limit - those are your words, not mine. In fact, I advised him on sizing for a transformer that will reduce thermal dissipation.
3. The regulators are already fitted.
4. He was asking for advice on transformer sizing for the boards he already has

[jmaf]

1. He isn't building a circuit from zero.

This is unnecessary pedantry. My point was clear, we're electronic hobbyists, you can employ components any way you want to.

2. I have never, anywhere, advised him to work on the limit - those are your words, not mine. In fact, I advised him on sizing for a transformer that will reduce thermal dissipation.

You wrote a long post about how everything uses components on the limit. Pardon me if I understood your implication wrong. What was your point then?

3. The regulators are already fitted.

I assume he has a soldering iron.

4. He was asking for advice on transformer sizing for the boards he already has

Which he's going to use somewhere on his own projects. Right?

[Mick Bailey]

You wrote a long post about how everything uses components on the limit. Pardon me if I understood your implication wrong. What was your point then?

Just illustrating the commercial approach and wishing I'd never bothered to post anything on the subject.

[nickb]

Thanks Nick. After some toil, I found these: https://www.mouser.com/ProductDetail/546-187E16 16v 56va, 3.5a, a little bigger than needed, I think, for a 1a PS. Allows for some voltage drop and still stay above 17v min input.

For some reason Hammond haven't given the regulation for those. At 56VA I'll guess 15%.

With a 3300uF and taking the the worst case line voltage of 115V less 7.5% at the input to the regulator would be about 17.1V, only just enough. Given your wall voltage is higher and the current is 500mA you should be fine. The power dissipation at max line volts of 115 + 7.5% and 1A is 7.4W so you'd need a 7.8C/W heatsink or better. If you want to limit to 500mA then 20C/W should do. The cooler you keep it the more reliable it should be.

[J M Fahey]

For some reason Hammond haven't given the regulation for those. At 56VA I'll guess 15%.

Take manufacturer´s regulation with a pinch of salt.

The "number" they quote is for AC regulation into resistive loads (and maybe it holds for choke input supplies) , but capacitor input ones get charged by narrow pulses with quite high peak current and narrow duty cycle.

Since loss is proportional to I *squared*, 1A 50% of the time requires 2A 25% of the time just to keep *average* current the same, but RMS current will be twice as high, with matching doubled loss.
And even worse with narrower pulses.

My point being that rectified voltage drop is higher than expected.
I have used overbuilt 5% regulation transformers where DC under full load dropped some 15%

[mikepukmel]

Take manufacturer´s regulation with a pinch of salt.

The "number" they quote is for AC regulation into resistive loads (and maybe it holds for choke input supplies) , but capacitor input ones get charged by narrow pulses with quite high peak current and narrow duty cycle.

Since loss is proportional to I *squared*, 1A 50% of the time requires 2A 25% of the time just to keep *average* current the same, but RMS current will be twice as high, with matching doubled loss.
And even worse with narrower pulses.

My point being that rectified voltage drop is higher than expected.
I have used overbuilt 5% regulation transformers where DC under full load dropped some 15%

Wow, didn't think it would drop 15%. Is that a smooth drop off, or a hockey stick thing that goes down quickly near the limit?
EDIT: Oops, sorry, you wrote clearly above "loss is proportional to I squared".