Signal Chain Basics #77: Enjoying the VU: Real-World Implementations

This is a continuation of the Signal Chain Basics article #73 (Signal Chain Basics #73: Audio metering, enjoying the VU) where we discussed the basics of audio metering systems. There are four main ways that manufacturers implement audio monitoring discretely, especially in systems that are low on DSP audio processing. Integration time and peak levels usually are handled by the external analog signal conditioning. However, in systems with enough digital signal processing (DSP), millions of instructions per second (MIPS), and memory, this can be handled on chip.

Low end analog — comparators

Figure 1

Using comparators for a low-end analog VU comparator matrix

Using comparators for a low-end analog VU comparator matrix

Figure 1 shows a simple system using comparators to do the level monitoring. The key advantages to using this method are cost and design simplicity.

Figure 2

TI's LM391x can be used as a precision analog comparator.

TI’s LM391x can be used as a precision analog comparator.

In Figure 2 we used the LM3914/5/6 LED metering products, which essentially are precision-trimmed comparators with additional support circuitry, such as current sources. Large range LED meters can be built by daisy chaining these devices. The datasheets for these devices have great examples of front-end circuits that can be used for this, and for the comparator and MSP430 solutions.

Many “Pro Audio” solutions on the market use the LM391x devices, as they contain 90 percent of the circuitry to implement a high quality, reliable, simple-to-implement analog VU meter.

Low-end mixed signal microprocessor

Figure 3

Using the basic ADC on a microprocessor

Using the basic ADC on a microprocessor

In my most recent design, I used the ADC10 peripheral on an MSP430G2231. The 10-bit analog-to-digital converter (ADC) allows capture of up to 60 dB of dynamic range (1024 linear steps). As I won't be listening to the output, the “distortion” really isn't critical. Absolute accuracy isn't guaranteed, but as the comparator is based in software (for example, If ADC10MEM < 0x01FF LED1 ON), it can be easily trimmed.

Figure 3 shows the basic connection. The ADC is running as quickly as possible with a basic “IF-THEN” comparator software matrix being implemented. This design can be extended by running the microcontroller with a UART interface, and allowing threshold levels to be modified via a USB-UART cable, etc. Additionally, for those feeling brave, the load on the 3V3 rail can be steadied by using a constant current source, so that a fixed 20mA (or whatever LED current you want) always will be drawn from the power rail. This also maintains a constant current through the ground plane.

High-end mixed signal

Figure 4

Using a mini-DSP codec to do the calculations

Using a mini-DSP codec to do the calculations

If your system requires any kind of audio processing, even something as simple as an audio crossover, a mini-DSP device, such as PCM3070 is an ideal drop-in solution. The Purepath Studio tools offer the ability to drop in blocks of code graphically to implement level meters. It’s even possible to use multiple blocks in parallel to monitor multiple channels, or to monitor multiple frequency bands.

Moreover, the PCM3070 can monitor both analog and digital (I2S) sources. Features usually integrated into the analog front-end can be integrated into the mini-DSP process flow. Multiple LEDs can be driven either by GPIO directly on the MSP430, or by using an external shift register, such as the SN74AHC595.

Devices with programmable DSPs are wonderful. However, consider that a microcontroller (like an MSP430) typically is required to get data in and out of the device and to drive your LED displays. That inherently adds cost and a software overhead to the solution. Yet, such a solution is capable of doing much more in real time than simply processing the audio. The PCM3070 also can do the precision rectification required in software. That immediately removes the front-end circuitry required. No more buffer op amp, rectification diodes, and so on.

It has become very apparent over the audio editions of signal chain basics that a number of audio experts and “audiophiles” also read this column. I’d be really interested in understanding how you have implemented this kind of function in your products. Are there any downsides/gremlins in these implementations that you’ve experienced? Until then, keep your signals clean and noise free.

Please join us next month when we will address one of the most common problems that engineers have with INAs and PGAs: common-mode limitations.

For more information, visit

— Dafydd Roche is an Audio Systems Engineer with TI’s High-Performance Analog team, and a graduate from the University of York (UK). When he isn’t busy pulling out his hair trying to find new ways to get an equal balance of performance, reliability, and cost, Dafydd pours his passion and knowledge of audio and music making into his work, helping designers and consumers get cleaner inputs and louder outputs. He may be reached at

12 comments on “Signal Chain Basics #77: Enjoying the VU: Real-World Implementations

  1. Michael Dunn
    May 8, 2013

    Seems a bit odd to have a VU article that doesn't explain the main points about what constitutes a VU meter – namely, the dynamic response (ballistics). 

    That's the kind of audio leadership I expect from TI – see Friday Fun: Rants & Giggles for another fun example.

  2. Michael Dunn
    May 8, 2013

    > the 3V3 rail can be steadied by using a constant current source

    I don't understand. Are you suggesting that the LED string get dimmer as more come on?

  3. Rochey
    May 8, 2013

    Hello Michael,

    regarding the use of a CCS (Constant Current Source), typically, in high end home audio systems, CCS's are used to ensure that there's no significant chance in the current flow from the main rail to ground. As LED's are 10mA each, having between 0mA and 200mA (for a 20LED array) rippling rapidly is a concern.

    There are implimentations (such as some of the ones done by Douglas Self, in his book “Small Signal Design” where folks run the LED's in a string between the +15V rail and the -15V (or 15 to GND). Further implimentations can use transistors to bypass the LED's in the string.Using the CCS means that even if you bypass 10 of the LED's in your string of 20, then all the other LED's will maintain brightness. (whereas, if you didn't use a CCS, LED brightness would vary by how many of them were on or off)

    A similar thing can be done even if you only have a 3v3 rail. The idea being that if your 3V3 rail is also looking after an ADC in your system, you can steady it by using a CCS that will maintain the current in the system while specific LED's are off.

    However, your system power efficiency just packed up it's bags and left… but frankly, I haven't seen many recording studio's worried about power consumption … (Maybe on a 96 channel mixing console, but not in individual processors etc).

    Thanks again


  4. Rochey
    May 8, 2013

    Regarding ballistics.

    fair point. I have 750 words to get a point accross, a lot is lost in the editing process. (you should see my originals!) 🙂

    For those interested – a good VU meter has very specific ballistics (speed of needle swing). It's “slow” reponse really is designed to “ignore” transients, as traditional radio broadcast equipment was incredibly forgiving about transient that would “clip” in the signal chain. Digital systems clip very nastily.

    What I've spent a lot of time working on here is a PPM system, where inputs can be monitored without the slow response of the ballistics.

    There's also a lovely article in Sound on sound from June 2000 that talks about the differences:

    Thanks again for taking the time to read the article and provide your feedback, and good luck with your blog.

    Best Regards



  5. Brad Albing
    May 8, 2013

    Oh, Michael, don't pick on our Planet Analog bloggers. I assume you read part 1 of this article:

    So alles klar, nicht?

  6. Michael Dunn
    May 8, 2013

    > I assume you read part 1 of this article

    If only the editor had linked to it…

  7. Guru of Grounding
    May 8, 2013

    Michael, I really got laugh from your story about TI and RIAA phono preamp equalization. That kind of standardized information has been almost as long as the standards for “VU” meter response. But TI seems to be flooding out information on audio, and especially pro audio, as if they invented it. And the technique of using an LM3915 with a current source and series-connected LEDs is hardly new. I did it back in 1973 when I was chief engineer at Quad-Eight, makers of ultra-high-performance analog recording consoles (users are a Who's Who of hit recordings from the 70's and 80's). Perhaps Mr. Roche can dig up some basic info I find conspicuous by its absence on a TI data sheet for the TLV320AIC3106. It claims to have a low-noise mic preamp built-in but there's nothing about input noise voltage or current. I have a customer that wants to use the preamp with a professional (that's 150-200 ohm dynamic) microphone … and wants me to tell him what turns ratio the input transformer should have. The support folks at TI can't seem to find an answer. It begs the question: Does TI intend this part only for cheap electret mics whose self-noise is so high that preamp noise doesn't matter?  Anyway, my rant about TI is that, if they're going to buy up every good analog house in the country, the least they can do is have some real experts available for customer support!!  All analog engineers don't design battery-powered portable gadgets!

  8. Brad Albing
    May 8, 2013

    Oh, like it's my fault that you can't figure out how to do a search in the search field, upper right.

  9. Guru of Grounding
    May 8, 2013

    Oh, yes, PPM (peak programme meters) have been the norm in Europe for quite a long time … and are well-specified by IEC and others. No need to re-invent the wheel. Regarding early CDs, I designed QC and other instrumentation for recording and manufacturing at Capitol Records from '81 to '88. CD's were just ramping up to replace LPs and cassettes. There was confusion, to say the least (and some persists today), about “headroom” (which analog recording has plenty of) and “reference levels”. Since a lot, if not most, master tapes were analog, the result was that digital-domain levels mastered onto early CDs was all over the map. Some were way too cautious, which made playback levels annoyingly low (and threw away lots of dynamic range) and others were way too “hot” because the mastering engineer used a VU meter, “corrected” using his own “rule of thumb” about how to infer peak levels from its reading. Metering has come a long way and, as I said, there's really no need to re-invent it.

  10. Michael Dunn
    May 8, 2013

    > it's my fault

    Then we're agreed.

  11. Michael Dunn
    May 8, 2013

    > It claims to have a low-noise mic preamp built-in 

    Tell me it's differential!!! :-}

    >there's nothing about input noise voltage or current

    Even Burr-Brown used to provide typical  specs at least 😉

  12. Guru of Grounding
    May 8, 2013

    Don't have the data sheet at my fingertips, but IIRC it IS differential.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.