Dual VU meter using color LCD or TFT displays

Hello,

I am looking for people, who are interested in building a "nice" colorful VU-meter with peak indicator and at least 80 dB dynamic range using some analog circuitry plus a uPC and a color LCD or even a compact TFt display. I want to build 32 of them for an old but valuable analog mixing desk.

We could co-develop or if you know about a prject already done, please point me in that direction.

Cheers Hermann

Something similar to this?
http://sch-remote.com/OLED-display-based-VU-meter-Oscilloscope-RTA-Envelope-PID-EVOR02x.php

Maybe better suited for the "Truth Table" subforum?

Jakob E.

madherm said:

Hello,

I am looking for people, who are interested in building a "nice" colorful VU-meter with peak indicator and at least 80 dB dynamic range using some analog circuitry plus a uPC and a color LCD or even a compact TFt display. I want to build 32 of them for an old but valuable analog mixing desk.

We could co-develop or if you know about a prject already done, please point me in that direction.

Cheers Hermann

Click to expand...

I've been designing meters for a few decades while not in that exact display technology (I think Andy Peters has messed around with LCD so may share his results.).

I did a nice 100 segment vacuum fluorescent Peak & VU meter back in the early late '70s early '80s but that was a one-off prototype using discrete digital logic and I wouldn't swear to the linearity of the 100 dB range.

More recently I did a LED output microprocessor front end Peak & VU meter currently in use in a production console.

In that design I did 4 meters completely using 1 cheap microprocessor. The dynamic range was clearly limited by the processors 12 bit internal A/D  convertor to way less than your target 80 dB.

For your design I am inclined to suggest using a microprocessor with a 16 bit A/D built in which will easily deliver your 80dB dynamic range with room to spare. The one chip family I am aware has a dual 16B A/D built in so perhaps organize the meters as stereo pairs.

I am a big proponent of displaying both peak and VU simultaneously. Using a dB linear display scale, you can impute crest factor or how dynamic the source material is from the distance between peak and top of VU display.

I see little need for much external analog circuitry. Any processor with 16b a/d will be powerful enough to crunch a basic peak and ave (VU) meter.  Note: I even coded up a true RMS calculation for the VU portion of my meter but couldn't discern any difference visually between RMS and simple average on program material with identical time constants, so I didn't use it (It required a custom square root routine that consumed processor overhead with no discernible benefit. Square root is not a common micro instruction).     

JR

shot said:

Something similar to this?
http://sch-remote.com/OLED-display-based-VU-meter-Oscilloscope-RTA-Envelope-PID-EVOR02x.php

Click to expand...

very nice 8)

madherm said:

Hello,

I am looking for people, who are interested in building a "nice" colorful VU-meter with peak indicator and at least 80 dB dynamic range using some analog circuitry plus a uPC and a color LCD or even a compact TFt display. I want to build 32 of them for an old but valuable analog mixing desk.

We could co-develop or if you know about a prject already done, please point me in that direction.

Click to expand...

I rigged up a multi-channel meter to learn about how to use a Silicon Labs ARM processor and monochrome graphic LCD (128 x 64). It could be extended to more channels but it was limited to the number of analog inputs on the development board. That processor's ADCs are 12 bits so dynamic range is 72 dB.

Extending it to more channels is the easy part. Driving a higher-resolution display can be tricky.

-a

Use a pcm2900 USB and a raspberry pi. Job done

Andy Peters said:

I rigged up a multi-channel meter to learn about how to use a Silicon Labs ARM processor and monochrome graphic LCD (128 x 64). It could be extended to more channels but it was limited to the number of analog inputs on the development board. That processor's ADCs are 12 bits so dynamic range is 72 dB.

Extending it to more channels is the easy part. Driving a higher-resolution display can be tricky.

-a

Click to expand...

A minor point about bits and display resolution, after you full wave rectify the AC signal you end up with one less bit.. At least for the peak data. The average data can actually increase resolution from the multiple samples being averaged over time. In practice when using input multiplexing to a single A/D there are settling time and sample and hold constraints that affect display resolution floor.

For a premium design using independent 16b conversions would be more than adequate and nowadays pretty reasonable. External lo-fi stereo codecs are like $1 so A/Ds should be cheap but I suspect there is not much between 12b free inside micros and say 16b(?)  cellphone codecs.

For my next generation tuner platform I'll probably use a micro with 16b built in...

To load up a console meter bridge I wouldn't try to make this a single processor design even if not that hard (for the processor), X1 or X2 meters would be more practical IMO.

JR 

JohnRoberts said:

A minor point about bits and display resolution, after you full wave rectify the AC signal you end up with one less bit.. At least for the peak data. The average data can actually increase resolution from the multiple samples being averaged over time. In practice when using input multiplexing to a single A/D there are settling time and sample and hold constraints that affect display resolution floor.

Click to expand...

.. but!

If the goal is to display multiple meters on an LCD, one needs to take the display resolution into account.  Say it's 128 pixels high. You need some pixels for decorations, so now you're down to 100 vertical pixels for display. Mapping the 4096 values of a 12-bit ADC into that area is the trick, as you need to decide how much granularity you want at either end. That, and peak hold time, are the sort of preferences about which we'll argue endlessly!

-a

(Of course none of the above is really any different from scaling an LED bar-graph meter.)

Andy Peters said:

JohnRoberts said:

A minor point about bits and display resolution, after you full wave rectify the AC signal you end up with one less bit.. At least for the peak data. The average data can actually increase resolution from the multiple samples being averaged over time. In practice when using input multiplexing to a single A/D there are settling time and sample and hold constraints that affect display resolution floor.

Click to expand...

.. but!

If the goal is to display multiple meters on an LCD, one needs to take the display resolution into account.  Say it's 128 pixels high. You need some pixels for decorations, so now you're down to 100 vertical pixels for display. Mapping the 4096 values of a 12-bit ADC into that area is the trick, as you need to decide how much granularity you want at either end. That, and peak hold time, are the sort of preferences about which we'll argue endlessly!

-a

(Of course none of the above is really any different from scaling an LED bar-graph meter.)

Click to expand...

But you do not want to map the 4096 linear quanta directly to a display or you would end up the same display issues as a mechanical VU meter. Nice resolution for top several dB but a full scale of only + 3VU  and nothing much visible  going on below  -15 or -20 VU

A Log display to cover 80 dB gives more than 1 pixel per dB for modest 128 step display. ( I am not convinced you need 80dB while more is always better. and I would go for as much size as practical to be able to read from a few feet away.)  If you use a multi color display medium, color is also useful to represent regions of level on a meter

While many meters do not use the fixed dB spacing I advocate for.

Besides hold time there is attack and release.  I used PPM standard for attack and much slower for release and ave/VU (something like 200 mSec).. Another trick for combination peak/VU display is to make release time exactly the same for both so you can see the ratio of peak to average stay constant after the signal goes away and display decays. Of course if the dB spacing isn't constant there is not much benefit from making the two release times the same as apparent crest factor will vary with absolute level.

JR

Hi all,

thank You for the messages so far.

A few more explanations:

When I wrote " some analog circuitry" I was assuming to build the rms of the analog signals with a 4301 rectifier from THAT corporation and then do the VU and peak detection in analog...just because I am more a hardware man than somebody enjoying to write software...sure in these days a software solution is more elegant. Even further I was tempted to create the log of the signals by using a log-amp...nowadays the conformity to the expected log curve is better than 2 db over 80 db....again in software more precise and easier....just needs a somewhat bigger uPC.

So I see two things to discuss:

The display: I came along "intelligent TFT displays" for example
http://www.alibaba.com/showroom/7%2522-tft-lcd-module.html
which exist from 100s of suppliers at incredible low prices up to real professional industrial products including professional prices. Interfacing to the uPC is in RS232, I2C or other serial protocols. Several companies deliver libraries with basic graphics commands.
Depending on the size of the display 2, 4 or even 8 VU/peak meters will fit.
Has somebody from you practical experience with such hardware?

The uPC: This is the problem for me: The last time I programmed a uPC must be in the 80s (an 8 bit 6502 programmed in assembler). I had 1 kbyte of RAM!!
I am lost in the present choice of hardware/software and I do not clearly see how I would go from my prototype (which would almost certainly be made with a commercial evaluation board) to my small series. Somehow all the PCBs I do end up having most elements in classical form factors, SMDs only when I have to....
So Ideally the uPC should exist as evaluation board, but also in small physical size as ready target system.

All the other arguments about 16 or more bits resolution are correct, but not a critical issue.

So let us see were we go?  Cheers Hermann

madherm said:

Hi all,

thank You for the messages so far.

A few more explanations:

When I wrote " some analog circuitry" I was assuming to build the rms of the analog signals with a 4301 rectifier from THAT corporation and then do the VU and peak detection in analog...just because I am more a hardware man than somebody enjoying to write software...

Click to expand...

I'm an old analog dog and don't enjoy writing software (that's what I should be doing instead of writing this post), but I also don't like using external circuitry for something i can do with relatively simple math, for free inside a micro.

On the plus side using a THAT rectifier/log conversion front end will dramatically reduce your dynamic range A/D resolution bit depth requirement so you're back into a cheap jelly bean processor. You might need a calibration (0VU) trim but that's no big deal. 

sure in these days a software solution is more elegant. Even further I was tempted to create the log of the signals by using a log-amp...nowadays the conformity to the expected log curve is better than 2 db over 80 db....again in software more precise and easier....just needs a somewhat bigger uPC.

Click to expand...

You don't really need a full precision log conversion even without the rectifier front end. In the digital domain you can do iterative approximations based on the dB output resolution you need, while for an 80 dB FS meter with 100 or 200 steps the finest is still a large fraction of dB resolution (say first parse to 10 dB steps, then 1 dB steps, then ,1 dB steps.. For a finite number of output steps just calculate a full output dB map once, and look up where actual levels map out to for real time output.  Calculating a 200 step output map could be done in a faction of a second during program start up.

So I see two things to discuss:

The display: I came along "intelligent TFT displays" for example
http://www.alibaba.com/showroom/7%2522-tft-lcd-module.html
which exist from 100s of suppliers at incredible low prices up to real professional industrial products including professional prices. Interfacing to the uPC is in RS232, I2C or other serial protocols. Several companies deliver libraries with basic graphics commands.
Depending on the size of the display 2, 4 or even 8 VU/peak meters will fit.
Has somebody from you practical experience with such hardware?

Click to expand...

No not yet.. but i will eventually.  A 7" 800x480 is a TV screen not a meter display. I'd sure aim lower (smaller/cheaper/etc).

The uPC: This is the problem for me: The last time I programmed a uPC must be in the 80s (an 8 bit 6502 programmed in assembler). I had 1 kbyte of RAM!!
I am lost in the present choice of hardware/software and I do not clearly see how I would go from my prototype (which would almost certainly be made with a commercial evaluation board) to my small series. Somehow all the PCBs I do end up having most elements in classical form factors, SMDs only when I have to....
So Ideally the uPC should exist as evaluation board, but also in small physical size as ready target system.

Click to expand...

You might even be able to find an evaluation board with the exact display you want.

For modest sized PCB there are free ware PCB layout programs and modest cost prototype raw board makers, but assembling SMD DIY is a challenge when starting out and using modern fine pitch parts. All the hip new parts are being made even smaller to fit inside an I-watch. The micro I'm currently using is smaller than my pinky fingernail.

All the other arguments about 16 or more bits resolution are correct, but not a critical issue.

So let us see were we go?  Cheers Hermann

Click to expand...

Good luck, if you were coding micros back in the '80s it has only gotten easier with more choices today.

I would strongly advise against 8 bit, it't hard to represent the world in only 256 levels. There was a 16b version of the 6502 but i don't think it was widely embraced. I won't push you toward my favorite, there are several with similar features and low cost.

I prefer working in assembler, but it seems the rest of the world is writing in C or some higher level language.

JR

madherm said:

When I wrote " some analog circuitry" I was assuming to build the rms of the analog signals with a 4301 rectifier from THAT corporation and then do the VU and peak detection in analog...just because I am more a hardware man than somebody enjoying to write software...sure in these days a software solution is more elegant. Even further I was tempted to create the log of the signals by using a log-amp...nowadays the conformity to the expected log curve is better than 2 db over 80 db....again in software more precise and easier....just needs a somewhat bigger uPC.

Click to expand...

You can do the ADC and the rectification/averaging/peak-detection/EVERYHING in a $3.00 microcontroller. Really. The only analog you need is something to scale the incoming signal voltage to something that's within the range of the ADC.

So I see two things to discuss:

The display: I came along "intelligent TFT displays" for example
http://www.alibaba.com/showroom/7%2522-tft-lcd-module.html
which exist from 100s of suppliers at incredible low prices up to real professional industrial products including professional prices. Interfacing to the uPC is in RS232, I2C or other serial protocols. Several companies deliver libraries with basic graphics commands.
Depending on the size of the display 2, 4 or even 8 VU/peak meters will fit.
Has somebody from you practical experience with such hardware?

Click to expand...

Electronic Assembly has TFT displays which you can talk to over SPI. Others do too. Others use horizontal and vertical sync controls and a pixel clock which loads bytes for Red, Green and Blue simultaneously. None of these are particularly hard to drive from a micro.

The uPC: This is the problem for me: The last time I programmed a uPC must be in the 80s (an 8 bit 6502 programmed in assembler). I had 1 kbyte of RAM!!
I am lost in the present choice of hardware/software and I do not clearly see how I would go from my prototype (which would almost certainly be made with a commercial evaluation board) to my small series. Somehow all the PCBs I do end up having most elements in classical form factors, SMDs only when I have to....
So Ideally the uPC should exist as evaluation board, but also in small physical size as ready target system.

Click to expand...

Silicon Labs and NXP have ARM eval kits which include a graphic LCD and lots of example code. Yes, it's all in C, and yes, if you're not a firmware guy there is a LOT to learn. But the kits give you everything you need. Development software is free, too.

You could do the gLCD interface in an 8051 with sufficient RAM (you need at least enough for a frame store). I'll shill for SiLabs again. They have this "unified development platform" which includes the gLCD and you plug in the processor board of your choice.

Then the thing is to take the stuff you need and spin your own board. It's not that hard ... although I say that because I do that for a living.

And yes, it'll all be SMT. That's the way of the world.

-a