LM3915 Buffering and Powering

Redtns · Sep 19, 2019

Hi all,

I'm wanting to add some LED output level metering to some mic pre's (1073s and alike).

The LM3915 seems a good shout for LED metering - pretty much a plug and play design, and it works off a single rail power supply (like the 24VDC one I'm using to power the pre's).

I read here (https://groupdiy.com/index.php?topic=48577.msg613543#msg613543) that I need to rectify the audio signal before it reaches the LM3915. Assuming this is the case, my first instinct would be to buffer the signal with a 5534 style op-amp...but this needs +/- V to power, which isn't a supply I have available.

So, my questions are:

- Do I really need to buffer the input to the LM3915? Why is this?

- If I do need to buffer, is running an op-amp on a single rail supply a clever thing to do? I've never done this, but I've read there are a few ways to do it.. I'm worried this might mess up the supply though - obviously the priority is quiet mic pre's.

- Is using the 24V mic-pre power a good idea for the LM3915 in the first place, or again is it just going to mess up the supply?

- Is there another (better) way to do all this that I'm missing? Space is limited so 5 or 6 LEDs (just for a rough indication of level) was my plan.

Thanks folks,

R

abbey road d enfer · Sep 19, 2019

Redtns said:
- Do I really need to buffer the input to the LM3915? Why is this?

You need to rectify the signal because the 3915 reacts only to positive voltage and, more importantly you need to hold the signal; if you don't the display will flicker like crasy and you won't have time to capture the signal. Rectifying without buffering is inadequate because it produces a non-linear load to the signal, which results in distortion.

- If I do need to buffer, is running an op-amp on a single rail supply a clever thing to do?

It is definitely feasible - in fact there are many "single-rail" opamps, but the signal will be offset by a significant amount, which is difficult to handle.

I've never done this, but I've read there are a few ways to do it.. I'm worried this might mess up the supply

Not a concern.

- Is using the 24V mic-pre power a good idea for the LM3915 in the first place, or again is it just going to mess up the supply?

depends how robust is you PS. The LM3915 introduces current glitches that can be very tricky to get rid of. I would suggest a separate PS, and while you're at it, make it bipolar.

- Is there another (better) way to do all this that I'm missing? Space is limited so 5 or 6 LEDs (just for a rough indication of level) was my plan.

You could use a one-transistor buffer-rectifier and whatevet LED driver you choose; the 3915 is as good as anything.

PRR · Sep 19, 2019

> I need to rectify the audio

Did you read the datasheet??
http://www.mouser.com/ds/2/405/lm3915-443929.pdf

They knew they had to make it "easy" (so they could sell some) so they listed many suggestions.

Top of page 12, fig 17, shows a single-supply rectifier which, while not Ideal is plenty good for your purposes.

OTOH, the Asian toy-sites are FULL of pre-assembled "LED VU" boards at very low prices. I built LM391x stuff back in the day and it was tedious.

abbey road d enfer · Sep 19, 2019

PRR said:
> I need to rectify the audio

Did you read the datasheet??
http://www.mouser.com/ds/2/405/lm3915-443929.pdf

They knew they had to make it "easy" (so they could sell some) so they listed many suggestions.

The suggestion to connect audio directly to the input is just plain inadequate, utterly useless, except for just showing some signal is present or entertaining the user with light animation, but gives no valuable indication.

PRR · Sep 20, 2019

> inadequate, utterly useless

Maybe you see something I don't. I see rectification, filtering, and buffering. Also nominal thermal compensation.

Winston OBoogie · Sep 20, 2019

abbey road d enfer said:
The suggestion to connect audio directly to the input is just plain inadequate, utterly useless

I think that suggestion in the datasheet is just to get the punters in the door. As PRR said, they wanted to sell chips so, after that suggestion, they go on to numerous better ones.

As you also said, these things will take a dump on your ground so keeping it off your clean audio reference is important.

abbey road d enfer · Sep 20, 2019

PRR said:
> inadequate, utterly useless

Maybe you see something I don't. I see rectification, filtering, and buffering. Also nominal thermal compensation.

From the datasheet:
"TIPS ON RECTIFIER CIRCUITS: The simplest way to display an AC signal using the LM3915 is to apply it right to pin5 unrectified. Since the LED illuminated represents the instantaneous value of the AC waveform, one can readily discern both peak and average values of audio signals in this manner. The LM3915 will respond to positive half-cycles only... ...True average or peak detection requires rectification."

In reply to: > I need to rectify the audio
You answered: Did you read the datasheet??...
...They knew they had to make it "easy" (so they could sell some) so they listed many suggestions.
Which, to me, implied that you advocated a solution without rectification, which is "possible", but only according to an overenthousiastic marketing intern.
Sorry for the misunderstanding, please chalk it up to the language barrier.

moamps · Sep 20, 2019

Redtns said:
I'm wanting to add some LED output level metering .....
.....but this needs +/- V to power, which isn't a supply I have available.
....Is there another (better) way to do all this that I'm missing? Space is limited so 5 or 6 LEDs (just for a rough indication of level) ...

I would use here BA6144 (2Euro on ebay) or similar 5 LED VU driver, it doesn't need additional buffer nor rectifier.

ruffrecords · Sep 20, 2019

If you only have a 24V supply and do not want to add another just for the VU display but want to avoid adding current spikes to the audio supply I would recommend something based on the Soundcraft constant current design. If yu scroll down to the fifth page of this schematic:

file:///home/chronos/u-817c9231946e3e73bcd42ae09847997204442289/MyFiles/Downloads/MIXER%20-%20Spirit_FX8_Schems.pdf

you will see the circuit on the right hand side. It basically uses a string of LEZs fed from a constant current source and just switches them on and off depending on the signal level. Because the current is constant there is very little noise generated. I think it uses LM339 quad comparators. A pair of these and a TL072 will give you an 8 LED display.

Cheers

Ian

Redtns · Sep 20, 2019

Thanks folks - loads of great information! Best forum on the internet this - no attitude, no snobbery, just positive suggestions and advice!

I'll return with my findings...

R

abbey road d enfer · Sep 20, 2019

moamps said:
I would use here BA6144 (2Euro on ebay) or similar 5 LED VU driver, it doesn't need additional buffer nor rectifier.

yes, the implementation is quite simple, but I would still be cautious about the current spikes. Although the schemo shows constant-current sources, the overall current draw varies according to the number of LED's actually lit. Nothing to be really worried about, but careful ground circulation is de rigueur.

L´Andratté · Sep 20, 2019

ruffrecords said:
If you only have a 24V supply and do not want to add another just for the VU display but want to avoid adding current spikes to the audio supply I would recommend something based on the Soundcraft constant current design. If yu scroll down to the fifth page of this schematic:

file:///home/chronos/u-817c9231946e3e73bcd42ae09847997204442289/MyFiles/Downloads/MIXER%20-%20Spirit_FX8_Schems.pdf

you will see the circuit on the right hand side. It basically uses a string of LEZs fed from a constant current source and just switches them on and off depending on the signal level. Because the current is constant there is very little noise generated. I think it uses LM32 quad comparators. A pair of these and a TL072 will give you an 8 LED display.

Cheers

Ian

Ian, your link goes 404 on me?

moamps · Sep 20, 2019

abbey road d enfer said:
... but careful ground circulation is de rigueur.

As always.

moamps · Sep 20, 2019

L´Andratté said:
Ian, your link goes 404 on me?

It is link to a Ian's local disk.

Try here
https://groupdiy.com/index.php?topic=44786.msg561388#msg561388

abbey road d enfer · Sep 20, 2019

The Soundcraft meter can be made even simpler and more compact, with only one LM393 (IIRC), dispense with the CCS and use the single-transistor rectifier I posted earlier. It will have only 4 LED's, powering them via just one resistor, the current variation won't be enormous so the luminosity of the LED's will be almost constant.
One thing to remember is that LED's of different colour do not provide the same luminosity for a given current; however, since reds are more sensitive than greens, the change of luminosity is welcome.

JohnRoberts · Sep 20, 2019

This is a well plowed field, with many discussions on this forum.

The LM391x is long in the tooth and several variants are no longer available.

As others have offered there are many ways to skin this cat, and simple designs based on inexpensive comparators (like LM339) or even cheap op amps, can be fairly easily designed to deliver exactly what you want/need.

JR

ruffrecords · Sep 20, 2019

L´Andratté said:
Ian, your link goes 404 on me?

Oops, that was on my hard drive. Here it is attached.

And I should have said LM339 not LM324. (double brain fart)

Cheers

Ian

ruffrecords · Sep 22, 2019

I have been thinking some more about this because I am interested in designing an LED VU powered from a 12V tube heater supply. The last thing you want on a heater supply is fast spikes because they do get coupled all tot well to the cathodes of tubes as I know from personal experience.

The series string of LEDs could be hard to do with only 12V to play with. For an 8 LED display the bottom four LEDs would all be green. With a typical 2.1V forward voltage drop at 10mA, four in a string would drop 8.4V leaving 3.6V for the CCS which is probably OK. But the maximum forward volts drop of a green LED at 10mA can be as high as 2.7V which would leave a mere 1.2V for the CCS.

So an alternative is just to feed each LED via its own current limiting resistor -say 1K for about 10mA from a 12V supply. You now draw 40mA from the supply rather than 10mA but this is insignificant if you have a few amps of heater current anyway. What might be significant is the change in current from all LEDs off to all on which I reckon as about 8mA if we use the Soundcraft short out the LED method). Double this for an 8 LED display.

So then I got to thinking about isolating these current spikes from the 12V supply. First thing I though of was to use a regulator. We don't need 12V to run a single LED, 5V is plenty so why not tun the display off a sub-regulated 5V supply? And the answer is because this just isolates voltage variations but not current spikes. This now begins to look like a real application for a shunt regulator which aims to maintain a constant current draw from the supply.

Is this over complicating things - do we just need a local chunky reservoir cap?

Cheers

Ian

abbey road d enfer · Sep 22, 2019

ruffrecords said:
I have been thinking some more about this because I am interested in designing an LED VU powered from a 12V tube heater supply. The last thing you want on a heater supply is fast spikes because they do get coupled all tot well to the cathodes of tubes as I know from personal experience.

The series string of LEDs could be hard to do with only 12V to play with. For an 8 LED display the bottom four LEDs would all be green. With a typical 2.1V forward voltage drop at 10mA, four in a string would drop 8.4V leaving 3.6V for the CCS which is probably OK. But the maximum forward volts drop of a green LED at 10mA can be as high as 2.7V which would leave a mere 1.2V for the CCS.

So an alternative is just to feed each LED via its own current limiting resistor -say 1K for about 10mA from a 12V supply. You now draw 40mA from the supply rather than 10mA but this is insignificant if you have a few amps of heater current anyway. What might be significant is the change in current from all LEDs off to all on which I reckon as about 8mA if we use the Soundcraft short out the LED method). Double this for an 8 LED display.

So then I got to thinking about isolating these current spikes from the 12V supply. First thing I though of was to use a regulator. We don't need 12V to run a single LED, 5V is plenty so why not tun the display off a sub-regulated 5V supply? And the answer is because this just isolates voltage variations but not current spikes. This now begins to look like a real application for a shunt regulator which aims to maintain a constant current draw from the supply.

Is this over complicating things - do we just need a local chunky reservoir cap?

Cheers

Ian

The real problem is current spikes, if they are allowed to circulate in more or less sensitive circuits. That's often the case in mixers, when ground currents are not properly isolated from the audio reference.
I would think the heater supply currents are using a different "ground" than the audio ground, so there is no risk of cross-pollution. In that case, yes, a well dimensioned capacitor and asmall value isolation resistor should be enough.
However, you'll have to connect the meter's voltage reference and Rlo to the audio ground and the V- pin of the 3915 to the heaters ground.

LM3915 Buffering and Powering

Help Support GroupDIY Audio Forum:

Member

Well-known member

Attachments

Well-known member

Attachments

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Well-known member

Attachments

Well-known member

Well-known member

Similar threads

Join the conversation!

Join today and get all the highlights of this community direct to your inbox. It's FREE!