AN6882 bar graph LED drivers: what instead?

[Wavebourn]

It seems to me AN6882 from Panasonic is obsolete and no longer in production.
What to use instead?

(Please don't suggest A/D and PIC, I need ONE indicator per power amp)

[Eva]

LM3914 and LM3915, but these are expensive, not very flexible, hard to find and may be close to become obsolete too due to the wide availability of advanced microcontrollers...

What do you have against microcontrollers? They are the way to go!

My choice would be a simple 8-pin PIC with built-in 10-bit multi-channel ADC, clock and everyhing, like PIC12F675, and one 74HC4094 (or CD4094) shift register per each 8 leds. These 8-pin devices are very simple to program.

With one PIC12F675 and three 74HC4094 you can do 10 level leds per channel *stereo* with peak hold and whatever dB scale you want, plus clipping (with some hold time too so that brief clipping is signaled) by sampling the two amplifier outputs plus one supply rail (used as Vref). The other three pins are required to transfer LED state to the shift registers: data, clock and strobe.

The shift register can drive the 8 leds directly if only 2 or 3 leds are intended to light at the same time (20mA/15mA), otherwise ULN2003 transistor arrays may come handy to cut part counts dramatically.

Now imagine the amount of analog electronics required to do all that.

[Irakli]

Hi Eva,

Could you give a reference where such project ot similar one is described?


Thanks

[Nordic]

Just google datasheet and that part number...

[Eva]

Originally posted by Irakli
Hi Eva,

Could you give a reference where such project ot similar one is described?


Thanks

I can't give you any reference, but I've been doing much more complex stuff with PICs (and 100% ASM) for the past 6 monts so it can't be so hard. I began from scratch too. For example, at the beginning a PIC16F690 with all its hardware (PWM, ADC, timers, serial I/O) subjectively seemed like a very complex thing to me but it no longer is. Now in one of my applications I'm sampling 7 analog magnitudes and controlling three processes concurrently (involving hardware and software PWM, state machines, and even digital low-pass filters and PID control) while communicating with another PIC too.

In a 16F675 you should set up a periodic interrupt (timer0 for example) to start A/D conversion, and an A/D end-of-conversion interrupt where you can save results to specific ram locations. Then in the main program you can analyse the results every time a new set of them has been sampled, prepare the LED lighting pattern, and transmit it to the shift registers with programmed serial IO. A HD44780 type LCD display can be driven much in the same way (just 3 I/O lines and a 74HC4094).

[Charles Hansen]

Originally posted by Eva
LM3914 and LM3915, but these are expensive, not very flexible, hard to find and may be close to become obsolete too due to the wide availability of advanced microcontrollers...

What do you have against microcontrollers? They are the way to go!

The OP specifically asked *not* to use a micro. Some disadvantages:

- The part must be programmed. This requires programming knowledge and programming hardware.

- The part will unavoidably introduce noise into the audio circuitry due to the master clock.

- If you want to adjust the brightness of the LED's additional noise will be introduced due to the necessity of using PWM.

On the other hand, the National parts (LM3914,5, and 6) are in full production, cost only $1 each, require no programming, are available with either linear or logarithmic responses, and can control the brightness of the LED's simply with a (non-RFI generating) current source. You can stack them to get up to 100 segment if so desired.

What more could a guy ask for?

[Nordic]

Hi Charles, I agree for the job and starting from nowhere the led driver chips might be the fastest solution...


However on the microprocessor side things have moved on alot
Useing a normal pic, one can learn the programming to set ports as inputs etc in an afternoon, the ADC part will take a bit longer...

I have started playing with TI's MSP chips lately, those babys are very sweet, can easily be clocked above the audio spectrum, and can put them self to sleep useing something like a microamp of current...have onboard periperals like opamps and pull down resistors etc... The learning curve is a bit steeper but I think these will leave the standard pics in their wake in the long term...

the programming is alot more compact too...
you can literaly MOV Field1 Field2
in stead of the several lines of code for PIC assembly...

[Charles Hansen]

Yes, if you want to learn programming, buy a programming machine, and install an RFI generator inside your amplifier, then you have a great solution.

But that is not what the OP asked for.

[Wavebourn]

Thanks;
LM3916 with one opamp rectifier may fit!

Why not PIC? Because I use a vactrol opto pair in series with the last very bright red LED to reduce gain before sound engineer reacted pulling faders down.

[Eva]

Originally posted by Charles Hansen
Yes, if you want to learn programming, buy a programming machine, and install an RFI generator inside your amplifier, then you have a great solution.

But that is not what the OP asked for.

This one was a very good joke. I can't stoop laughing

How much EMI can be produced by a chip that uses an internal RC oscillator (built into the die) with 5mW of total power draw?

Cell phones? WI-FI? CPUs over 1Ghz? CD players with 18Mhz clocks? Anyone?

A solution based on PIC12F675 is probably quieter than one based in LM391x. Old bipolar ICs tend to draw current spikes while modern CMOS doesn't. Remember NE555? (vs. ICM7555 that does not even require a bypass capacitor)