LM3915 Audio Level Meter Testing (numerous issues)

[@pev_]

Being used to PC cable conventions I would have made +12 yellow and +5 red :)

[@greaser5691]

...and blue for -12V :) Have texta (sharpie), will travel!

[@greaser5691]

@Mai Mariarti to finish your sentence for you: "...that TTL exists in." :) but we were talking about wire colours for PC PSUs...

[@simontay4851]

Yes, same here. I always follow ATX standard for 12 and 5V.

[@WolfgangMahringer]

If you look at the LM3915 datasheet carefully, they say that the LED #1 has a high leakage current of 150µA typical, while the other LEDs have 0.1µA. So yes, that effect is intentional. And the high brightness of your green LEDs helped there too **g**

[@JulianIlett]

I'll try the 10k resistor fix. But it may not be needed if I can find some lower efficiency green LEDs

[@DJignyte]

I'd say that the dim LED in dot-mode is so you can see the floor of the scale from a distance. It provides a reference for the dot, rather than a single light wavering up and down with no bearing as to how much it is going up and down.

[@urugulu1656]

actually i rather believe as the signal basically digitally created the least significant bit toggles all the time due to noise and jitter which turns the led half on.

[@Roy_Tellason]

@@urugulu1656 This has not been my experience with *real* LM3915 chips, done many years ago. Oh, the joy of chinese components!

[@urugulu1656]

@@Roy_Tellason i doubt that you understood what i was trying to explain...

[@Mark1024MAK]

The colour code that I normally use is: * Red for +5V * Yellow for +12V * Purple or violet for -5V * White or blue for -12V * Black for 0V/GND * Green for Earth/screen/chassis

[@jlucasound]

Mark! Don't forget Orange for 3.3 V! :-)

[@Mark1024MAK]

As Wolfgang Mahringer says in another comment, I thought the first LED glowing dimly was a intentional feature of the LM3915 chip when used in dot mode. And in the past, when I have used these chips, or modules based on these chips, I have had the same effect.

[@tubical71]

No, sorry to say-...that´s due to bad design...i use(d) countless of LM3914/15/16 in my stuff...i´ve never seen this... 1st: omit/eliminate offset at the input pin, as any offset would light up the first LED *everytime*. 2nd: split the rails LEDs and the LM391x should not sharing the same rail without decoupling caps. 3rd use resistors in series with the LEDs and set/use the input range/led brightness pin...don´t leave them unconected. Either set them up or put them to ground/Vcc according to the datasheet.... Anyway, it´s always a good idea to consult and read the entire datasheet before you "messAround" with a chip, as you can easily aviod pitholes whioch make you scratching your head for hours otherwise... ;)

[@ElmerFuddGun]

Colour code should follow some standard: Red = 5V Black = GND Yellow = 12V White = -12V This is pretty much standard at least for PCs except for the -12V which they normally use blue for but you only have white left. But at least 5V should always _IMHO_ be red. I think this will cause less confusion especially since you already are using red for the 5V (ish) input from the battery pack. And this is only one colour out from standard instead of three that you have at 5:47.

[@____________________________.x]

Yeh, I did suck my teeth a bit over that, especially as this PCB iteration has no connector markings, hopefully version 2 will make room for some. A crimp tool is a handy thing to have sometimes to make exactly the cable you want.

[@ElmerFuddGun]

@@____________________________.x - The biggest problem I see is using red for the 5V input and likely any other circuit including maybe for a digital supply (not sure if he will have one though) but then using that red elsewhere for 12V. That could be a real problem feeding 12V where 5V is needed (and frying something) or go unnoticed but cause problems providing 5V when 12V is needed and thus clipping the signal, etc. As for using white for -12V I think that is reasonable considering it saves a lot of time and cost of since it is already there. That's not likely to be a problem if the power wires are always bundled in the same power connectors.

[@____________________________.x]

@@ElmerFuddGun - ATX -12v is blue and -5v is white, but that would require a £10 crimp tool to fix, maybe his patreons would oblige. I'd always try to follow the ATX standard for hobby stuff, industrial control cabinets have their own DC standard.

[@frankhobsonKJ4CDM]

Julian Im glad Im not the only one turning a capacitor wrong way around occasionally. Im still shook from the last one I popped.

[@Mark1024MAK]

You mean you didn’t intentionally blow any up 😳. Blowing up removed life expired electrolytic capacitors is fun 🤣

[@ElmerFuddGun]

Your "cheerio" was quite different today. Quieter and a bit disappointed sounding. But I think most of the "issues" with this isn't so much the PCB layout but that you skipped the full breadboard test and went straight to PCB. I would try just changing out the current setting resistor (going to pin 7) first before pulling any LEDs. Temporarily wire in a variable resistor and see what range you need to be in.

[@mDeltaKilo]

I think that first LED may be designed to be on when there’s a signal so you have an indication of the bottom level when in ‘dot mode’. Otherwise, in the dark, you’d just have a moving dot and no reference where zero is.

[@tubical71]

sorry, no...it´s rather due to an neglected DC offset voltage at the input pin...mostly added by any OpAmp, in this case the dual rectifier...refer to the LM3914/15/16 datasheet...it´s all there...

[@AndyCallaway]

"Much less near"... That's Julianspeak for "Further away." ;-)

[@Thomas-rc9sc]

*im surprised it didn't blow up* - Julian Ilett 2019

[@sparkyprojects]

If you redesign the board, maybe add a spot for a resistor for the greens, a common resistor should be ok if calculated that it doesn't bun out with all greens on. As more greens turn on, it should slightly dim the display too in bar mode'

[@Hugatry]

"Negative leg to the ground" is correct way of connecting capacitors in so many cases that it often becomes a norm in our minds.

[@____________________________.x]

He did mention leaving off the silkscreen indication to force himself to check the polarity, perhaps unfortunate in a circuit using a negative rail?

[@JulianIlett]

Hmm, may I add the word 'stupid'?

[@urugulu1656]

negative leg to the more negative potential is always correct and if you think of it this way this does not happen to you

[@kyoudaiken]

The gallium arsenite green LEDs are actually more yellow than green. They look disgusting. I would just use resistors to tame the LEDs down.

[@jlucasound]

Those pins are so easy to swap out. What pev and Mark said. +12 is Yellow; +5 is Red; 0 is Black and use White for -12. (AT Motherboard is Blue for -12v; but you don't have blue, now, do you.) :-) We will go with whatever you choose Julian. In your defense, electrons are color blind. I am so glad you are "chipping away" at the Vocoder. I can't wait to hear it produce awesome sounds. Love your videos, Julian!

[@edheath4571]

Apr 11 at 10:00 AM Regarding the bottom Green LED being on when in DOT mode. I had this problem in a DOT mode LM3915 application way-back some 38-40 years (yikes!) ago. I remember adding a 10k resistor from pin 1 to VLED will shunt the 100uA auxiliary sink current away from LED #1, thus extinguishing it. I think this is shown on/about page 13 of the National Semi data sheet. Also, be sure that your supply pin is adequately bypassed right at pin 3, but you know this. Hope this helps.

[@JulianIlett]

Thanks Ed. Yes, it's in my TI datasheet as well. I'll give it a try - cheers :)

[@youpattube1]

On wild leads, like the two coming out of a battery pack, if you shorten one, it eliminates an accidental short.

[@JulianIlett]

Very sensible :)

[@Mark1024MAK]

youpattube1 - that removes the fun of watching wires arcing ☹️

[@AttitudeGames]

I would of used Black for GND, Red for +5V, Yellow for +12V and White for -12V.

[@urugulu1656]

this is the conventional way and if hes changing wires anyway it wouldnt really make a big difference...

[@XciterD]

@@urugulu1656 Well in my world -12 is blue but I agree.

[@urugulu1656]

@@XciterD yeah actually i dont know what white is used for but the rest i was quite sure. and what do you do with no blue wire at hand

[@ethanpschwartz]

I never knew how positively enthralling it would be to watch someone switch the wire order on a JST. Also, I'm so used to the BBC News alert chime that I immediately checked my own phone when yours went off.

[@____________________________.x]

3:00 If you had a laboratory power supply, you wouldn't need to wait for the magic smoke to determine if something is working, you'd have a current limit led indication... Also, a future LM723 video would be interesting.

[@____________________________.x]

@@2000jago - No, I'm saying it would avert situations like this one, a reversed tantalum causing the circuit to draw an unexpected current. Connecting up batteries capable of dumping 9 amps into a short is kinda risky, and the fault-finding process becomes one of "touching-stuff-and-seeing-if-it-is-hot". There's really no reason for every electronics hobbyist not to have a current limited PSU to work with, and ones with analogue meters are even more useful to judge current variations.

[@clintdouble-u9564]

The results of swapping the red and black was so satisfying lol, I'd better go book an appointment to talk about that :)

[@urugulu1656]

did you think of a pull resistor on the mode select (so that you have always a definite potential there and it always knows what to do)? i know though that dot mode is set by leaving that pin open (i wonder if it still works with some potential there)

[@AW-Services]

Could you not use some defraction grating to lesson the brightness and smooth out the display

[@____________________________.x]

Try removing C2, see if that first led turns off, its discharge path might be a little long? Just a guess.

[@T2D.SteveArcs]

think u can buy little vu meter kits like that on ebay for pence mate or do they need to be certain dimensions etc

[@nowoneknowsme]

The solid first green LED may have to do with oscillation. If the voltage goes down to zero for a short amount of time, this one lights up. The ones in between should flicker too, but too short for us to see. An oscilloscope should sort this out. I really like the vocoder-build, can't wait for the next video.

[@CrAzYDr1veR]

will that peak circuit work on a lm3914 "voltmeter"? The first led you are supposed to put a resistor in parallel to turn it off when it's supposed to be off.

[@pulesjet]

Swap out the IC . This one may have some leakage current going to the first LED. Simply change out the current limiting resistor or resistors. The LED's are fine.

[@DjResR]

Maybe the resistors were chosen for GaP green LED's and was supplied with InGaN ones instead?

[@bennyboi0115]

When I helped with my school play by doing the sound, we used a laptop for one of the sound affects. The teacher helping me said about how the aux basically is always throwing out sh!t which can be heard in the masters. Could that be affecting it?

[@bertoid]

I would like to have seen the outputs of that inverter module on a 'scope...

[@JulianIlett]

A small amount of high frequency ripple would be my guess :)

[@JohnnyX50]

I made a multiplexed stereo version when I was 15 (30 years ago) in my electronics course. Mine uses 2 op-amps for inputs with level VRs for input adjustment, works on 5V only and is sat right here with me bouncing to your voice lol. Thought id dig it out and power it up via usb battery supply to see if it still works. It does! 30 years on! lol My 30 year old green led's suffer the same issue on the first dot if there is audio but they are only just BARELY visible. Im not sure but my driver could be the LM3914 given its age. It only uses the one LM chip and for the life of me I can't remember how it is multiplexed to 2 channels but it does work. I'm sure it was a project from a very early Electronics magazine as mine is on hand drawn etched pcb which we had access to in the course, it looks scruffy and botched but hey, it works lol. The op-amps in mine are a pair of CA3240E .I can't tell the other 2 chips but I am sure one is a 556 timer and one is an analogue switch IC driving the 2 'bars' in a 1,2 sequence via 2 transistors and feeding the LM chip from the analogue switch IC, as my PCB is mounted upside down and bolted in, in a VHS plastic "book" case it is hard to see lol Although I am sure the PCB you have will have the proper components for its design, the input to an LM3914/5 should be a regulated audio signal (DC) as it can be fussy about AC Audio in dot mode. I would try a different green LED for your set up as the current limiting is done by the LM3915 its'self and be careful not to go above 7v on the mode input as it can warm up the IC somewhat :) As mentioned the leakage on pin 1 can be 150uA but may be as high as 450uA which would be more than enough to light your green LED at that brightness.

[@JulianIlett]

That seems like a lot of work to save an LM3915 - maybe they were very expensive back then. There is a fix for the LED1 current leakage problem (a 10k resistor across LED1). I've implemented this on my v2 PBC :)

[@stevewhitcher6719]

that link to the 5v to - and + doesn't work anymore which is a pity as i have always wanted a cheap way to get split rail voltage from batteries. Ironic that too bright leds is a problem in 2019 in the 1980s was always a problem that they were too dim

[@SidneyCritic]

My std rect green and yellow LEDs are weak, but all my water clear LEDs of all colours are super bright like those. My WC LEDs are so bright I had to use 20K on some to make them reasonable. That always on LED might need a pull-down, maybe check the spec sheet.

[@iceberg789]

please see if resistances can be put by those leds.

[@robertcoulthard7461]

I was wondering if the in and out ground on the dc2dc converter was commoned or if that could be an issue with not getting a clean 5v

[@JulianIlett]

Yes, they're both spoked onto the ground plane.

[@simontay4851]

Maybe change the 5V supply for the LEDs to 3.3V or just use a 2W resistor of suitable value (you'll need to experiment with different values) in series on the 5V. I built a stereo LED VU meter many years ago using the velleman VU meter kit. 2 boards and all the LEDs and other components are included. I put it in a transparent plastic box with RCA connectors on the back. Its powered from a single +12V AC/DC wall adaptor. It works very well.

[@littleal350]

No need for a separate 5v supply for the leds. The led outputs are current limited and the datasheet does show designs using higher than 5v for the LED supply. It does make it slightly less efficient though, and you have to be careful not to draw too much total current through the IC (increasing power dissipation) when using bar mode when all 10 LEDs could be on. If you are worried, it's possible to use a resistor in the LED supply that drops an increasing voltage as more LEDs turn on (burning power in the resistor instead of the IC), as long as the LED supply doesn't drop below around 6V with all LEDs on, then due to the constant current operation you'll not get any LED dimming from the decreasing voltage. A suitable resistor would be 56ohms 1watt for a 12v supply and maximum dissipation will be approx 0.56W with 10mA LED current. Using high efficiency LEDs with a good brightness at only 1mA solves the dissipation problem completely. This is what I did when I made one in the late 90's; high efficiency LEDs have been around for a good while now.

[@iamdarkyoshi]

Best way to buy the right flavour of LEDs is to search based on wavelength! Search for the wavelength of the older style ones and you should get the right ones. Also, have you considered an external mic on your phone using one of those 4 pole 3.5mm connectors? The internal mic on that phone just sounds so dreadful...

[@samba3403]

You need to paint the tops of the leds with a dark color sharpie maybe a blue to add some tinting

[@BajanAlan]

You could use colours from computer power supplys!

[@davedave4865]

I see your problem with the led brightness. Maybe you could use a 8PIN SIL resistor network to reduce the current on the 7 green leds. It should be small enough to fit on the same size board if u were to rework it. Great video as always.

[@Brian_Of_Melbourne]

The LED brightness problem is not caused by the LEDS, but by too much current. And that current is set in the design by the resistor connected between pin 7 and ground. Perhaps reading the datasheet would help? Changing the LED supply voltage or putting in resistors is not the way to fix this. www.electroschematics.com/wp-content/uploads/2010/02/LM3915.pdf See Tips on Reference Voltage and LED Current Programming on page 12.

[@davedave4865]

Because different types of leds are being used ,using the resistor value on pin 7 to set the LED current ' I led ' common for all LEDs on the output pins. Thus if you want to reduce the current for a section or the array. ie (7 green LEDs). Then extra resistance on the supply side would reduce the intensity so that it matches the other LEDs (brightness) in the display. I ref the LM3915 .pdf page 2. Hope this helps.

[@jensschroder8214]

see PDF: kurzelinks.de/zmph . see Figure30 on page 19. There is a 10k resistor parallel to the first LED. For the green LEDs use resistors in serie.

[@JulianIlett]

Thanks for that :)

[@metallitech]

Does it really need such high voltage?

[@littleal350]

A higher voltage is good for headroom, but he could have easily made it work from a single 12v rail supply (possibly down to 8v ish) by using a 1/2 dual opamp for half supply (bias) generation connecting that to RLo of the LM391x, and using the correct resistors to determine the RHi level... It's all in the datasheet. He'd have to use only a half wave peak detector for the audio input in that case though (other half of dual op-amp). I never liked using dual supplies for low level audio work. With the above you'd need a regulated supply. With a little more design work, it may be possible to derive a suitable bias from the LM391x voltage reference and make the design more immune to supply variations.

[@HazeAnderson]

If it's too brite ur too OLD! xD

[@SergiyGryb]

you may certainly add resistors to each green LED to make them dimmer... Would need new PCB... Or try to squeeze in SMD resistors to existing design...

[@Thomas-rc9sc]

I think the whole thing would look better with all green LEDs and a resistor

[@simonhopkins3867]

I was expecting you to get the microphone and sing to us again... ;-)

[@DAVIDGREGORYKERR]

Try 860R resistors instead of 470R.

[@martinda7446]

OK I am arriving to see a small audio level meter....Power requirements 5V, +/_12V, 62.5V, EHT and heater at 6.3 and ............OK I'm being silly, but never have I ever gone that route on a board especially a sort of non critical one. Single rail supply with no shortfall in performance would be the main design criteria. (along these days with keeping power down).

[@JulianIlett]

I take your point, but this is a faithful rebuild of a 40-year old design. Multiple supplies and a colossal power draw were not uncommon back then :)

[@martinda7446]

@@JulianIlett I knew there would be a reason....My fault. Sorry everyone.

[@Mark1024MAK]

Julian Ilett - unless the physical size of the PSU is measured in feet, it’s not considered to be a colossal unit 🤣

[@dom1310df]

RIP Headphone Users

[@Robber7]

Suprised you didnt black tape the leds as you've done before xD

[@JulianIlett]

Yes, I've got some green insulation tape :)

[@milenedejong1400]

cant you just take it out. the led... you will probably never use the device at that level.. oh no..i am just thinking i am more interested in using your vocoder where you have all the fun in designing it ;) ..

[@Chriva]

"Oooh that's very loud!" Baha

[@JulianIlett]

I must remember to start with the speaker on the floor - sorry headphone users :)

[@Chriva]

@@JulianIlett It din't really get THAT much louder. It was just hilarious to hear your distorted scream :) Certainly made my day a little brighter. Thanks :D

[@pvc988]

At first I thought that it's part of the track.

[@kryptocat4240]

Sir please help me please make a video on programming ATMEGA 16 , 32 microcontrollers

[@kryptocat4240]

@@sdffsdafdsfsdfsd no brother i searched can you give me the video link please share with me

[@kryptocat4240]

@B GG no

[@kryptocat4240]

@B GG please help me how to proggram atmega 16 and 32 and which device or programer used to proggram it and which software is used please answer thise question

[@sasodoma]

@@kryptocat4240 Do you have an Arduino? It can be used as an ISP programmer for the Atmega chips.

[@kryptocat4240]

@@sasodoma but that trick does not working with me i tried a lot could you help me

[@Max27653]

Maybe you can team up wit lookmumnocomputer... he would probably build your whole device in a couple of days^^

[@JulianIlett]

He's young, with boundless energy :)

[@Max27653]

Not only but IMHO you get lost in details a bit

[@wishicouldarduino8880]

I can't make these but I want to put it on my robot

[@deltapparo4451]

I think most viewers know how to tin wire leads. Many You-Tubers edit this sort of stuff out.

[@JulianIlett]

Whereas I leave it in :)

[@Roy_Tellason]

@@JulianIlett Which speaks to the degree of consideration fot the time spent by your viewers... :-(