Installing LED/Analog Vu Meter to Front Panel

Jun 13, 2009 at 2:36 AM Post #17 of 62
another idea is to use a cpu and an lcd display
wink.gif
wink.gif


even a 16x2 line display would be enough to have a L/R meter going horiz across the display. the problem is that you can't update the lcd 'really fast' and have it bounce if you use the slow (but cheap and affordable) 16x2 char cell displays. you can, however, average the data as you sense it at one of the a/d pins and update the lcd display every second or two with the new value.

also, while on the subject, the lm3915 style chip doesn't really do 'peak hold' stuff well. you can't get a really clean display with the 3915 chip unless you front-end it with a sample-hold kind of stage (so that you don't get really fast flicker in the led display).

by going with the cpu method, you can more easily control the 'look' of the meter, if that's one of your goals.

(I may work on the lcd VU meter idea, if I get a chance).
 
Jun 13, 2009 at 5:13 AM Post #19 of 62
Neat. Lots of nice options!

Quote:

Originally Posted by individual6891 /img/forum/go_quote.gif
where would you take the signal from without degrading the audio output?


BTW, not sure if you guys have seen this post, but Ferrari has had a couple of builds already with analog vu meters and kind enough to share his circuit:

Afbeelding645.jpg
 
Jun 13, 2009 at 5:48 AM Post #20 of 62
McIntosh's analog power meters are indeed nice. They have an extraordinary dynamic range, much more so than the typical "VU" meters that could typically manage only about -20dB to +3dB.

For a meter to be of practical use (rather than just as visual amusement), 0dB would need to be calibrated to the maximum rated output power of the amp. Let's take the β24 as an example, 0dB would be around 180W into 8 ohms. On a conventional analog "VU" meter, the meter needle would hardly move when the amp is putting out 1.8W, and that's usually already a fair amount of sound pressure level on most speakers. If you crank up the sensitivity of the meters to show movement at low/mid volumes, then the meter would simply peg on the right hand stop when the amp is called upon to deliver more of its power (not to mention that the meter then ceases to "measure" anything meaningful).

Notice that the meter on the 1200W monoblock McIntosh MC1.2KW amp, the meter has almost 60dB of range:
McIntosh | Power Amplifiers/Integrated Amplifiers
Now, THAT meter will show something from pianissimo to fortissimo!

An analog meter like that could only be achieved with a special meter drive circuit that has compressive gain. Not only that, if you have actually seen these meters in action you'll notice that they have a very fast attack time but slow decay. So they are peak-reading meters that could actually register music transients, rather than the ballistics of ordinary VU meters that could only show average levels. I've yet to see any DIY amp with an analog meter that sophisticated (I'd like to be proven wrong, though -- and am especially interested in knowing what meter mechanism is used and what circuit is used to drive it).

Btw, McIntosh is just one example. Back in the late '70s and early '80s there were a number of commercial amps with comparable analog meters (until LED or fluorescent bar graph meters became popular).

LED/fluorescent bar graph meters don't have to deal with a mechanical moving mass, so the attack/decay times can be controlled easily with electronics, and the dynamic range issue is also easy if you use the right chip/circuit to drive it (and have enough segments to give adequate resolution). I've built a few power amps with LED meters way back when, but these days I just find that flashing lights distracting.
 
Jun 13, 2009 at 10:00 AM Post #22 of 62
Quote:

Originally Posted by linuxworks /img/forum/go_quote.gif
I bet mapping a voltage range to a new non-linear voltage range and also doing peak hold attack stuff is JUST what a small cpu would be good for!


No doubt -- for an electronic display. For an analog meter, a lot of the attack performance will be in the meter mechanism itself. No electronics, cpu or not, is going to force a needle to move faster than it could.

I contend, though, using a CPU and firmware just for a meter display is overkill in a certain sense. Nevertheless, if you want the display to be accurate at 20KHz then you need to A/D the signal at >40KHz. Assuming we want 60dB of dynamic range, you can see that there is quite a bit of data to be sampled and processed. I haven't done the math, but I have a gut feeling that a little CPU like the Arduino would be insufficient.

CPUs and digital are good, but some things are easier/better done entirely in the analog domain.
 
Jun 13, 2009 at 11:58 AM Post #23 of 62
Quote:

Originally Posted by amb /img/forum/go_quote.gif
No doubt -- for an electronic display. For an analog meter, a lot of the attack performance will be in the meter mechanism itself. No electronics, cpu or not, is going to force a needle to move faster than it could.


not even if we dedicate a whole b22 board as a meter/needle driver?

then, the needle will get there even before the command to get there finishes!

wink.gif


Quote:


I contend, though, using a CPU and firmware just for a meter display is overkill in a certain sense.


I like the ability to place ANY curve you want on there, with no pain and with fast turn-around to see the results. even hot-swappable (user selectable) ranges.

you could do trick stuff like, if the signal goes below a certain range, you can go into 'expanded scale' mode. nite time listening, as one use-case.

do I want to see average, peak, rms or just a bunch of needle movement to show the channel is active? all selectable if you go the software route.

Quote:

Nevertheless, if you want the display to be accurate at 20KHz then you need to A/D the signal at >40KHz. Assuming we want 60dB of dynamic range, you can see that there is quite a bit of data to be sampled and processed. I haven't done the math, but I have a gut feeling that a little CPU like the Arduino would be insufficient.


I was *never* thinking of sampling *that* much. enough to drive a meter, which would be a fraction of the audio freq. maybe a few hundred samples per second, if even that. I'm not doing DSP, man; I'm just trying to get some needle movement trackage
wink.gif


how fast DO you have to change the needle's position in order to make a good VU meter? I'm not sure we can easily define what an ideal VU meter is, anyway; its a meter, afterall, and if you wanted a peak-hold led display you would have asked for that (the OP would have).

so front-ending a physical meter movement with an arduino class cpu sounds entirely reasonable to me. you get some sampled audio that is smoothed with a cap and read that via the a/d port as fast as you can, then for each reading you apply the mapping (linear or non-linear; probably via some lookup table to make it run faster) and then you come up with a resultant output voltage that gets sent out via another analog pin on the cpu, to the meter movement. and the cpu would do all the 'fades' that are needed (so if the signal suddenly dropped out, the user might want the meter to slowly sink down instead of dropping suddenly. or not - but it would be up to the user to specify).

the ard. can sample an IR receiver module, for example, and 'follow' its waves (38khz carrier, usually) and be able to find where the transitions are and count bits to decode the raw IR protocol. if the ard can follow an IR stream at this bit level, I bet it can keep up with enough analog sampling and also some transform mapping to a new output voltage, in enough time to keep onlooking humans amused enough
wink.gif
 
Jun 13, 2009 at 12:03 PM Post #24 of 62
Quote:

Originally Posted by zkool448 /img/forum/go_quote.gif
Neat. Lots of nice options!
BTW, not sure if you guys have seen this post, but Ferrari has had a couple of builds already with analog vu meters and kind enough to share his circuit.



that's the kind of thing I'd use as a front-end to the cpu. a rectifier stage, then a cap and then the a/d would tap in here and the meter would be at the other end of the processing/cpu chain.

want to get fancier? have a switchable C so that you can choose how much sample/hold effect you get. how much you charge the cap and smooth out the analog value, vs having a smaller cap and having a more instantaneous reading style meter.

again, bringing the cpu into it, I think you could choose a smaller charge cap and just read it more frequently (in the cpu) and have the best of both; be able to react faster to changes but also simulate the smoothing effect via software.
 
Jun 13, 2009 at 12:27 PM Post #25 of 62
Quote:

Originally Posted by linuxworks /img/forum/go_quote.gif
I like the ability to place ANY curve you want on there, with no pain and with fast turn-around to see the results. even hot-swappable (user selectable) ranges.
...
I was *never* thinking of sampling *that* much. enough to drive a meter, which would be a fraction of the audio freq. maybe a few hundred samples per second, if even that. I'm not doing DSP, man; I'm just trying to get some needle movement trackage
wink.gif



So you want all the bells and whistles, except you neglect (IMHO) the primary function of the meter -- to display the output power (either in watts or a relative dB scale) in an accurate manner? If you only sample the waveform a few hundred times a second, the frequency response of the meter would only be good for bass. Anything in the midrange or treble will be missed entirely. I don't think such a meter is very useful.

Quote:

how fast DO you have to change the needle's position in order to make a good VU meter? I'm not sure we can easily define what an ideal VU meter is, anyway; its a meter, afterall, and if you wanted a peak-hold led display you would have asked for that (the OP would have).


Actually there are industry standards governing the ballistics of a VU meter, and it's basically slow and averaging:
VU meter - Wikipedia, the free encyclopedia

A VU meter will still display high frequencies accurately, as long as the waveform is sustained long enough in time. This is unlike a meter that only samples the waveform a few hundred times a second -- on such a meter the mids and highs are simply lost.

Now, slow VU meters have their place, but a fast peak-reading one is much more useful as an indicator for amplifier output. Fast transients will not be averaged away, and potential amplifier clipping on such transients can be easily observed. How fast is fast enough? Here again there are industry standards for them. The standard terminology for this type of meter is Peak Program Meter (PPM):
Peak programme meter - Wikipedia, the free encyclopedia

Quote:

... the ard..., I bet it can keep up with enough analog sampling and also some transform mapping to a new output voltage, in enough time to keep onlooking humans amused enough
wink.gif


If amusement is the only criterion, then you win.
wink.gif
 
Jun 13, 2009 at 12:34 PM Post #26 of 62
if you use a sample/hold charge cap, why would you miss ANY frequencies?

you aren't in the freq domain here, you are simply having the cap do the peak hold for you. that gives you - I think - a lot more breathing room to scan that value for changes and update your meter display.

in terms of calibration, its just a mapping, a scale factor that you can enter or adjust.
 
Jun 13, 2009 at 12:43 PM Post #27 of 62
just read the wiki for VU:

"The rise and fall times of the meter are both 300 milliseconds, meaning that if a constant sine wave of amplitude 0 VU is applied suddenly, the meter will take 300 milliseconds to reach the 0 on the scale. It behaves as a full-wave averaging instrument, and is not optimal for measuring peak levels."

300ms is pretty slow, cpu wise. and they even say its a full wave averaging, so the idea of a bridge and a cap (just like ac to dc) works here.

and I still believe that if you use a smaller cap than one that would normally give the 300ms time figure, you could still read that stored value in the cap and set ms-level timers to start the climb-up and climb-down from any given value.

I'd have to play with it for a while
wink.gif
but my first idea would be to have an ISR routine that keeps that 300ms timer going (or some fraction of it) and looks at the current measured value and starts to fade it up or down, on each interrupt timer-tick. that would be the realtime synchronous background process. the foreground one (in a polling loop) would just read the value of the cap and note the timestamp that it took the value at, then store that in RAM and go back and do it again, endlessly. I wonder if that approach would work.
 
Jun 13, 2009 at 7:33 PM Post #28 of 62
Sorry I can't even paticipate in this conversation, I'm just way in over my head after seeing the complexity just to get one integrated to a panel. My only intent really is to mount a signal meter mostly as visual amusement and for aesthetics reasons.
smile.gif


I really appreciate you guys taking the time to give your inputs.
beerchug.gif


Quote:

Originally Posted by mattcalf /img/forum/go_quote.gif
http://i21.ebayimg.com/06/i/001/45/bb/364f_1.JPG
I bought a couple of those and also a surplus of blue LEDs so my next few projects are 'meter' covered. :P



mattcalf, I like the dial on those, do you have plans yet on how you'll implement yours? cheers.

 
Jun 14, 2009 at 1:39 AM Post #29 of 62
Quote:

Originally Posted by zkool448 /img/forum/go_quote.gif
mattcalf, I like the dial on those, do you have plans yet on how you'll implement yours? cheers.



No idea, haha and don't plan to implement them for a while sorry. Well I figure it won't be until i can understand the majority of this thread. :P
 
Jun 14, 2009 at 11:26 PM Post #30 of 62
i have a bunch of service manuals for my mcintosh gear, some if it is older (2505/2105) and uses a rotary switch to adjust the meter range (-20dB, -10dB, 0dB) but ive got the manual somewhere for my MC4000M that does have the full range meters... probably wouldent be terribly difficult to replicate if you had the right meters
 

Users who are viewing this thread

Back
Top