Question about minimum input voltage using Voltage Regulator L7809?

[Jasonmatthew911]

Another thing to consider is that on my PCB design I only have space for a small 21mm x 15mm x 10mm Aluminum heatsink (See pic) which claims a Power dissipation of 2W @ 30 degrees C rise...I saw the formula to calculate if a heat sink is required is by subtracting the output voltage from the input voltage and then multiplying the voltage difference by the mA consumption, and if the result is less than 2W, then no heat sink is needed...See my pic of the heatsink I had in mind for my space...I'm wondering if it will be enough now, considering I input 18V into a 7806 for 6V, this will probably produce a lot of heat...Otherwise maybe using a 7809 would be a better deal to control heat with 18V input supply...I just wonder if anyone here knows, if a 9V input to a 7809 just drops the voltage to around 7V but still works properly or if it can get sketchy being 2V below the minimum input voltage specs?...Also, how exactly can I calculate my mA consumption?

[Jasonmatthew911]

Yes 2V dropout is about right.

You can let the 7809 dropout.    Using a 7805 is probably wasting too much output from the LM386.

Slighty better is to increase the output of the 7805 with diodes,
https://www.deeptronic.com/electronic-circuit-design/converting-5v-7805-linear-voltage-regulator-to-produce-6-4v-output/

I'd suspect three diodes would give you about 7V.

It depends if your 9V rail is from battery (which could drop to 7V), an unregulated supply, or a regulated supply.

Now I'm starting to consider using a 7809 and just let it drop out with a 9V supply, the difference between 9V and 18V is less than going down to 6V with a 7806, so this should produce less heat, considering I only have room on my PCB design for a small 21mm x 15mm x 10mm aluminum heat sink (See pic)...This heatsink has power dissipation 2W at 30 degrees C rise...I saw that in order to calculate if heatsink is needed, you subtract the output voltage from the input supply and multiply the voltage difference by mA consumption, and if result is more than 2W, then heat sinking is needed...I'm not sure how to calculate my mA consumption?...I also still wonder if the 7809 will function properly if I feed it 2V below minimum input voltage?...BTW, all my 9V and 18V supplies would be regulated.

[Rob Strand]

The amount of dissipation depends what headphone impedance you are going to use, and to a small degree if you are using one amp to drive both left and right channels or two LM386's.

If you go down to 8 ohms, that will be a 4 ohm load.   The problem with headphone amps is the 8 ohm head phones are really loud and 32 ohm (or more) phones will be softer.    The way this is dealt with is to add a resistor in series with the output of the amplifier.  It affects the 8 ohm more, dropping the level, and the 32 ohm less,  over all giving a better balance of levels with different phones.

The point I'm making here is if you add a series resistor to balances the level then the LM386 is driving a higher impedance.   That will reduce the minimum load impedance and reduce the dissipation of the LM386 and the regulator.

Given low supply voltage side is going to be the most limiting in terms of output swing reducing the dropout is probably going to the be the best approach and that means a 9V reg which is allow to drop out.

[anotherjim]

If you use diode lift, the diode anode goes to the middle 0v pin of the regulator and cathode to 0v. This means that the tab of a TO-220 package cannot connect to 0v and must be insulated from 0v. The diode can be pretty much any silicon since it's only the regulators bias currents passing through it. Personally I would fit a small power diode like 1N4001 - I just think small signal diodes like 1N4148 look wrong in power supply areas!

Consider what Rob is saying about reducing headphone power. If you don't want hearing damage, you only need milliwatt power levels in headphones. For this project, I would use a TO-220 regulator without heatsink. You can create an isolated pour on the pcb for it to lie down on, but you also stand it upright to save board area. The regulators have thermal shut down that can actually be useful in a fault situation and it will work sooner without a heatsink.
Also, 386 have a lot of voltage gain - signal will need attenuating with its own volume control, so the signal headroom requirement is less than you might think - I would not worry about maximising the supply voltage so much.

[Jasonmatthew911]

The amount of dissipation depends what headphone impedance you are going to use, and to a small degree if you are using one amp to drive both left and right channels or two LM386's.

If you go down to 8 ohms, that will be a 4 ohm load.   The problem with headphone amps is the 8 ohm head phones are really loud and 32 ohm (or more) phones will be softer.    The way this is dealt with is to add a resistor in series with the output of the amplifier.  It affects the 8 ohm more, dropping the level, and the 32 ohm less,  over all giving a better balance of levels with different phones.

The point I'm making here is if you add a series resistor to balances the level then the LM386 is driving a higher impedance.   That will reduce the minimum load impedance and reduce the dissipation of the LM386 and the regulator.

Given low supply voltage side is going to be the most limiting in terms of output swing reducing the dropout is probably going to the be the best approach and that means a 9V reg which is allow to drop out.

Ok thanks...The signal is actually Mono so I'm just creating a fake stereo for the Headphones...I'm just gonna share the schematic I made for the headphone amp so you guys can help me improve it or make recommendations, or tell me if it's fine, considering there will be to separate TL072 Pre-Amps being fed into the headphone amp, either guitar, bass or a synth, but this is kind of a mixer that mixes 2 instruments to the same outputs, splitting between an unbalanced out for an amp and a balanced XLR out with CabSim On/Off for Live or direct recording, so it works like a splitter to record a mic'd sound and direct sound simultaneously...Let me know if I have to do something else to this?...I can put the small aluminum heat sink on the 7809 just in case, I have the space...Also, I'm still not sure if the Max Voltage for the LM386N-4 is 18V or 22V, because on the data sheet I think I saw 22V for the 386N-4...Is that right or wrong?

[anotherjim]

22v is only for the -4 as a maximum. Sustained shouldn't be over 18v. Reducing the supply voltage is a good idea though as a way of limiting output power.
Are you going SMD with this build? For future proofing, note that the through hole versions are obsolete. NJM386 is still available thru hole but they only allow 15v absolute max and 12v operating.

On your circuit...
Good to see proper provision of supply bypass capacitors
Stock 386 has x20 gain, won't that be enough? The components on pins 1 & 8 may not be needed. With this chip, too much gain makes them noisy.

A pro addition to those regulators is a back feed diode. If the supply input got a short circuit and there is charge stored in the circuits capacitors, it can force current back into the regulator and damage it. A diode fitted between the regulator output (anode) back to its input (cathode) will stop damage from back feed.

[Jasonmatthew911]

Ok cool, thanks...Yea I have the Gain set at 50 with the resistor and Cap between pins 1 & 8, so I'll eliminate that, as you said, I guess it's too much if I want a cleaner sound that doesn't get too loud...For the Diode on the Regulator, how should I orient the Diode, should it be the negative side of a 1N4001 Diode from input (leg 1) of regulator to output (leg 3), or should the Diode be oriented the other way around with negative side of Diode coming from output (leg 3)?

[anotherjim]

See the cathode & anode I mentioned. They are the diode ends. It's so the charge stored in the circuit can feed back to the shorted input. Having the diode the other way would bypass the regulator in normal conditions.

[Jasonmatthew911]

Sorry, I missed that, my bad...Thanks again...Everything else seems good to you right?

[Jasonmatthew911]

22v is only for the -4 as a maximum. Sustained shouldn't be over 18v. Reducing the supply voltage is a good idea though as a way of limiting output power.
Are you going SMD with this build? For future proofing, note that the through hole versions are obsolete. NJM386 is still available thru hole but they only allow 15v absolute max and 12v operating.

On your circuit...
Good to see proper provision of supply bypass capacitors
Stock 386 has x20 gain, won't that be enough? The components on pins 1 & 8 may not be needed. With this chip, too much gain makes them noisy.

A pro addition to those regulators is a back feed diode. If the supply input got a short circuit and there is charge stored in the circuits capacitors, it can force current back into the regulator and damage it. A diode fitted between the regulator output (anode) back to its input (cathode) will stop damage from back feed.

This is kind of a different question, for my PCB's I'm using wide Traces thinking it's better, but I'm not sure how wide is better?...I'm using 60mil traces for my signals and 80mil for my +V Power traces, I still have a good amount of spacing between all my traces, so I figured I'd make wider traces, and all my GND's are attaching directly to GND Planes on both sides of PCB...Let me know if I'm good using these Trace widths when you get a chance...Thanks again!

[anotherjim]

PCB design isn't my thing. Maybe worth starting a new thread for that? I will add that if you have any doubts about the electronics, then breadboard it first. If you do SMD, it should be worthwhile mounting some common IC's mounted on adapter boards or get the thu-hole versions to fit the breadboard.

[Jasonmatthew911]

The amount of dissipation depends what headphone impedance you are going to use, and to a small degree if you are using one amp to drive both left and right channels or two LM386's.

If you go down to 8 ohms, that will be a 4 ohm load.   The problem with headphone amps is the 8 ohm head phones are really loud and 32 ohm (or more) phones will be softer.    The way this is dealt with is to add a resistor in series with the output of the amplifier.  It affects the 8 ohm more, dropping the level, and the 32 ohm less,  over all giving a better balance of levels with different phones.

The point I'm making here is if you add a series resistor to balances the level then the LM386 is driving a higher impedance.   That will reduce the minimum load impedance and reduce the dissipation of the LM386 and the regulator.

Given low supply voltage side is going to be the most limiting in terms of output swing reducing the dropout is probably going to the be the best approach and that means a 9V reg which is allow to drop out.

I keep reading that if you don't input at least 2V over the voltage regulator output that it simply doesn't work...When they say it doesn't work, are they simply referring to the fact that it just won't regulate or it won't let the voltage pass at all?...I'm assuming based on what you've said, is that it just doesn't regulate if I feed a 9V supply into a 7809, and that the 7809 will take 2V from that 9V and will just end up outputing 7V, but that my circuit can work just fine with those 7V...As long as the 7809 voltage regulator always outputs either 7V - 9V with 9V - 18V input supply, then that's what I want...However, if a 7809 with 9V supply is possibly gonna give operation issues/noise, then I'd rather just go down to a 7806, and just always feed 6V into the 386 headphone amp...If I had the 2 regulators right now I'd breadboard and try it, but I'm trying to decide which to order...Help me decide if I should stick with the 7809 for those extra Volts to the 386 or just compromise with the 7806?

[Rob Strand]

I keep reading that if you don't input at least 2V over the voltage regulator output that it simply doesn't work...When they say it doesn't work, are they simply referring to the fact that it just won't regulate or it won't let the voltage pass at all?.

Typically the regulators work in that they produce output.   The output voltage is the drop-out voltage below input voltage.

If you want some evidence, see figure 8 on page 10, (LM340 is virtually identical to the L7805).
https://www.ti.com/lit/ds/symlink/lm340.pdf

In this case we a lucky that the manufacture provides such a graph.   Many device datasheets don't give this info but all regulators do possess this behaviour.  If you look at the graph you can see also at 3V input the regulator no longer works, as you would expect as the low voltage cannot keep the internal circuit working.

What is not given is the same graph for a 12V regulator.     That might be available in other datasheets.

As far as what better, a 7809 or an 7807, it depends on what you want.   The 9V regulator gives you more headroom when operating at 18V.   The 7V regulator will give you the same performance operating at 9V and 18V.    The thing is 7V is fairly low and you might find the output level is little compromised.   You might be willing to accept that at 9V.   The other way to spin the requirements is "we *need* full output at 9V and 18V"  in which case the 78xx regulator is out and the search would be a for a low dropout regulator, and possibly not use an LM386 but use a more modern amplifier which has output swing closer to the supply rails.   What seems simple has a whole lot of trade-offs and consequences.  For DIY projects you might be happy to aim low but for professional products the requirements force you to aim higher.

[PRR]

> I keep reading that if you don't input at least 2V over the voltage regulator output that it simply doesn't work...

Don't believe all that you read. Do NOT burn a PCB until you breadboard!! You are maybe in uncharted swamp here.

It will "work". Maybe like a loose connector. The regulator will not hold any sort of solid voltage, not even "2V less". Probably bop around a half volt or so.

I do not understand why this is getting more complicated when higher-voltage audio chips are readily available. LM383, bam, done.

[Rob Strand]

I do not understand why this is getting more complicated when higher-voltage audio chips are readily available. LM383, bam, done.

It was sort of mentioned in replies #22 and #23.   When you run at a higher voltage you need a larger series resistor to limit the output power.  However when the supply drops that causes a large change in the available headroom.


You can see what happens in this table,




The base-line level is the recommended power levels for headphones from standard IEC 61938, which is a 5V rms source driving a 120 ohm series resistance.    To meet that standard directly you need a +/- 9V supply rail.  For lower supply rails the output power must increase with low impedance phones and decrease with higher impedance phones.   Typically using 32 ohm as a nominal design center.

Anyway without a regulator you use a higher resistor value to limit the power at 18V.  However when the supply voltage drops the loss in headroom is somewhat larger.

Obviously if the PSU stays regulated the headroom stays the same with 18V/9V input

However, that imposes a lower supply voltage and you get more output power variation with different phone *impedances* (because it forces you to use a smaller series resistor.)

[PRR]

I've been blowing-up headphones since LBJ. You make good points but IMHO not compelling.

[Rob Strand]

IMHO not compelling.

100% agree.

As a user I find limiting the output too much is really annoying.   
Especially when listing to raw bass guitar, if the peaks clip it sounds bad.

[Jasonmatthew911]

I do not understand why this is getting more complicated when higher-voltage audio chips are readily available. LM383, bam, done.

It was sort of mentioned in replies #22 and #23.   When you run at a higher voltage you need a larger series resistor to limit the output power.  However when the supply drops that causes a large change in the available headroom.


You can see what happens in this table,




The base-line level is the recommended power levels for headphones from standard IEC 61938, which is a 5V rms source driving a 120 ohm series resistance.    To meet that standard directly you need a +/- 9V supply rail.  For lower supply rails the output power must increase with low impedance phones and decrease with higher impedance phones.   Typically using 32 ohm as a nominal design center.

Anyway without a regulator you use a higher resistor value to limit the power at 18V.  However when the supply voltage drops the loss in headroom is somewhat larger.

Obviously if the PSU stays regulated the headroom stays the same with 18V/9V input

However, that imposes a lower supply voltage and you get more output power variation with different phone *impedances* (because it forces you to use a smaller series resistor.)

Hey I actually had a L7809 laying around so I just breadboarded the 386 Headphone amp...I had to make a few changes with the input components to get rid of extra noise, I also raised my 2 output resistors from 10 Ohms to 100 Ohms, though when I measured with my multimeter my 100 Ohms resistors were more like 300 Ohms, but raising those values of those resistors takes out a lot of extra noise, but they also attenuate more sound...Oddly, with a 9V Boss adapter through the 7809 it sounded very clean after I atenuated with the resistor going to GND at the input (5K - 10K)...10K would be better as it atenuated less, but was noisier, 5K was my in between point, cuz 1K got rid of all background hum, but I felt it atenuated too much...So with 9V the 386 headphone amp works pretty clean, and I could put volume all the way up and not distort, hoping the TL072 Pre-Amps will push it a little louder so headphone volume doesn't have to be almost maxed out to get a loud enough sound...Then I tried my 18V Dunlop adapter through the 7809, and it just added more background hum that I ha gotten rid of with the 9V Boss supply...I could do more attenuating to kill noise with the 18V supply, but then it wouldn't be loud enough, especially when going back to the 9V supply...Is the extra noise from the 18V Dunlop supply because it's not really regulated, or does that even matter with a 7809?...Is there anything else I can do to get rid of background noise with the 18V supply without attenuating more sound...I tried the extra gain through pins 1 & 8, but that adds a little bit more background noise...In the end I got the 9V supply to sound very clean, barely any background noise at all, but the 18V changes all that and adds a bit of background noise...See my new schematic below after breadboarding, this is how I left it for now...BTW, I tried it using my Shure SRH440 headphones which have a 44 Ohm impedance, so I guess that's why I had to raise my 2 output resistors to get rid of extra noise...I tried all I could think of when it comes to 386 chips...Any further recomendations?

[Jasonmatthew911]

> I keep reading that if you don't input at least 2V over the voltage regulator output that it simply doesn't work...

Don't believe all that you read. Do NOT burn a PCB until you breadboard!! You are maybe in uncharted swamp here.

It will "work". Maybe like a loose connector. The regulator will not hold any sort of solid voltage, not even "2V less". Probably bop around a half volt or so.

I do not understand why this is getting more complicated when higher-voltage audio chips are readily available. LM383, bam, done.

LM383 for the headphone amp?...Isn't 7W a bit much for what I'm doing?...Also eliminates chance of using any battery supply if I used an LM383...Anyhow, as I just mentioned in my previous post, I just breadboarded the 386 headphone amp with a 7809 I had lying around...I got it to sound nice and clean with the 9V Boss supply 500mA, but with my 18V Dunlop supply 1A, volume didn't really get louder, it just added extra background hum, which I had mostly gotten rid of using the 9V supply...Any idea what else I can do so there's less background hum with the 18V through the 7809, without attenuating more sound?...If I atenuate more when using 18V, then the volume will drop too much going back to the 9V supply which actually sounded a lot cleaner...The thing is that the rest of my circuit will have more headroom and sound better with the 18V, so I'd like to get 18V to sound just as quiet as my 9V through the 386 headphone amp and 7809 regulator...If you know of a fairly simple headphone amp schematic that would work better for what I want just let me know, but I just don't have much space for a lot of components, that's why I opted for the simple classic 386...Let me know, thanks again!

[Rob Strand]

Any further recomendations?

The 100 ohms series resistor for a 9V rail is too high.   

If you need to increase the series resistor to reduce hum then that's kind of pointing to a problem elsewhere.  Either hum is getting into the circuit via the power, which it shouldn't with a regulator on the power rail ( you also have a bypass cap on pin 7 which helps), or it is getting into the input.

Are you getting the noise with the volume pot set to minimum?

The problem could simply be the gain is too high, normally you would put a resistor in series with pin 3.

The 5k-10k to ground is going to load the input a lot when the volume pot is set to full.   Like, even the 100k pot is quite a low load for a guitar.