I went through the same search about 2-3 years ago to find anybody that had the Velleman K4300 kit or the discontinued IC (U2067b). I contacted the company who sent me the same schematics and the U2067b datasheet from Telefunken Electronics, the company who made the IC. No luck ... then I went through the same quest to find a display circuit to go with the active filter from the info I had received ... I made the mistake of waiting too long to buy the kit... $79 was just too much at the time on my budget. Then when I was ready to order ... it was discontinued. I built a 3 channel color organ many years ago from a 1969 Popular Electronics article and still have it. I reworked the box I had it in about 14 years ago. This spectrum audio display was going to be my next attempt at audio to light displays for my stereo. Sea-Shadow and Animeguru are first two I've found who were doing the same... Did either of you get your projects completed?? If so, could you post your results? I know I'm new to the group ... I would have posted the same request that Ainmeguru did 3 years ago if I would have know about this forum. Thanks
Well hello there, welcome to the forums. I have gotten rather close to completing my schematic, but as I got down to the finer points of the layout I was making it more and more complex and more confusing than it needed to be. Also I have been planning the projects in which it will go into, then disaster struck. My 200 GB HDD flatlined so I spent the better part of a month recovering the data, and then I have had to deal with a few RMA's seperate from my current projects and my 4x40 LCD for my HTPC died due to a moment of stupidity. But now that I have the time again I think I will resume the work on this project. Thank you for bringing this back up As a point of interest it was observed that if I were to complete and use this schematic I would have to use the LM3915 in place of the LM3914 as the 3914 is on a linear scale and the 3915 is on an audio logarythmic scale.
Hmmm... Velleman.. Just remembered that I had a Velleman catalogue from 2000-2001 with the K4300 kit circled out (I thought about purchasing it). But then I discovered on a page before it a mono VU meter (model K4304) which can be used for the purpose. The VU meter uses the LM3916N IC.
new product with 2 100 LED-spectrum analyzer on speakers IMHO the coolest product with your spectrum analyzer can be found at www.rocket-technology.com.
aha! Now i know what new project i'm going to start on. Multiplexing the output of an LM3916 and 10 filters shouldn't be to hard and expensive. Anyone know if switching the input on an LM3916 is a bad idea? (i'm thinking noise) edit: the LM3916 has become pretty cheap, so there is no need for multiplexing of inputs. Maybe i should go for a solution to switch the LEDs, though. (one bar at a time). I guess 100LEDs@20mA could heat up things a bit. edit 2: what's the difference between lm 3916 and lm3915?
An old ETI circuit used a single multiplexed LM391x to drive 8 x 10 bars, here's the block diagram. The post#19 Velleman circuit is also multiplexed.
the difference between them is that one is log and the other is not. cant rember off the top of my head, quick google will tell you
I guess you want to use so many of the LM3914 in order to let every single LM3914 show another frequenzy band? If so, you won't have luck with your circuit. Cascading the LM3914 will create an analyzer that only displays ONE frequenzy band over a broader range. In order to achieve what you want you'll have to build different band filters for every single LM3914. A block diagram would then look similar to this: PC Line-Out -> impedance converter (OP-Amp) -> band filters (Butterworth) => LM3914's
LM3914 has linear comparator stages LM3915 has logarithmic comparator stages I'm not sure about the LM3916, but it seems to have double logarithmic comparator stages. There are no other differences than that.
The datasheet doesn't spell it out AFAICS but according to Maplin, 3916 is a vu-meter scale, loggish but expanded around 0dB for more sensitivity setting recording levels at 0dB. 3914 : linear 3915 : -3dB steps 3916 : -10, -7, -5, -3, -1, 0, 1, 2, 3dB
Yeah, but won't the sensitivity get divided by 10? What will be best to use for a spectrum analyzer? It doesn't have to be all correct as long as it looks good (3916 is about half the price of a 3915)
Why? It's only like taking a few different multimeter readings, just faster. Voltages in parallel are the same. Currents would be reduced. You may need to turn the LED current up a bit, multiplexing may make it a bit dimmer as they're only on 12.5% of the time.
i was thinking of a 10 band analyzer. anyway. i thought the 3916 ICs had some kind of smoothing on the input. I have no idea where i have that from... anyway. i wonder if the advantage of multiplexing is worth it. Cost vise its not that big of a difference. Those analog switches is pretty expensive (of those i have found) 10x lm3916 100x leds 10x opamps and a few caps and resistors should get me pretty far. If i go trough with this i'll post schematics and PCB layouts. edit: i found a usable switch. a 4067. This would require some more counters and stuff, but it's doable. I'm not sure how much it will save space-vise, though. Is there anything i should think about when making the oscillator. what frequencies is good to use? is 3-4KhZ ok? I see that i'm a bit rusty on the design side nowdays.
The mutiplexing side of the ETI circuit is a 555 clock (750Hz), with a transistor inverter giving the second chip anti-phase clock pulses, a 4051 8-way analogue switch (36p at Rapid), 4024 binary divider (35p) and a 4022 1-in-8 decoder. Can't find the last chip anywhere. A handful of switching transistors are used to boost the logic current for the LEDs. Each filter uses 2 opamps from a cheap quad LM324 and also rectifies the AC music signal with zero loss (diode in feedback loop), followed by a unity-gain buffer, so the most expensive item is the 3915 Parts - logic, 3915, LEDs, 1 single opamp, 4 dual, 4 quad, 10 PNP, 9 NPN transistors, a few diodes. I have some scans of the pages if you need more info.
I have figured that i'll probably use a redisegned verson of the velleman kit that animeguru posted. it just sound a tad bit less complicated (and it have 2 more channels). I don't think there will be a big problem to replace that obsolete IC (IC13) with a 3916 (or 3915 still haven't figured out what's best to use) one small question though: What's the point of that switch at the very bottom uf the MUX board for? (the one controlled directly by the oscillator)
Yes, this is right, if you want more then 10 LEDs you would use a lm3915 for all the lower steps and the 3916 for the last 10 leds so you have the expanded range.
Now that you say it. 3916 can't be cascaded, while 3915 can. Still a bit unsure what to use, though. For a normal VU meter 3916 is best, i guess, but for an spectrum analyzer it might end up a bit odd?
Hi, I appologize for once again resurrecting a dead topic, but I have just recently decided that this is something that I would like to build for my car, since I am unable to find something that is commercially built. Unfortunately I'm not much of a circuit guru, and I was just wondering if someone had worked out a completed design that works. Thanks.
Wow, I can't believe this thread is still getting posts, I remember reading this when it first was created. Anyway, a while ago, I built an 8-channel spectrum analyzer for a school project, that worked reasonably well, save for one thing. The device is to be plugged in parallel with the speakers, however, I am getting distrotion from the circuitry. My best guess at this point is the cheap op-amp I used for the input buffer/pre-amp stage, but I ended up moving before I could test it, and had to pack it away until now. Right, well anyway, here is a schematic of a single band (no multiplexing for this project, although it wouldnt be that hard to implement) The Filter I used (MFB, documentation for this filter, and general filter design and construstion guidelines can be found here: http://sound.westhost.com/project63.htm, here http://www.technick.net/public/code/circuits.php#filt2bp, and here http://ourworld.compuserve.com/homepages/Bill_Bowden/opamp.htm#bandpass.gif) Overall Layout As you can see, Im just using a simple 741 in an inverting amp configuration for the pre-amp/buffer. Anyone have any ideas to decrease noise produced by this circuit, or what could be causing the awful feedback?
Keeping the thread alive I've recently needed information on spectrum analyzer design for a 3d cube project I'm doing. As this subject comes up from time to time, I figured I post my proposed idea in case it is helpful to anyone else, of if anyone should have any thoughts on it... It seems to me that there are two primary methods of deriving the spectrum data; either through parallel fixed analog filter networks as has been discussed in this thread, or through coded fft algorithms using some type of processor (either dedicated dsp or generic cpu/ucontroller). I will be using a ucontroller for my project, but the fft routines are very processor intensive, and I'd prefer to use the horsepower for other tasks. My current theory is that I could use a single pin programmable bandpass filter like the MAX263 or MAX260, and cycle through the desired frequency centers. This leaves one analog output to sample for amplitude, rather than requiring analog multiplexing of individual filters. One could then multiplex the display as previously posted. I've only scanned through the datasheets, but it looks like if you desired a free running circuit (no processor control), you could hardwire the filter Q to your need and use a bit of 74xx logic glue clocked by your display multiplex to provide the needed frequency selection bits. This potential solution would minimize parts count (eliminating the multiple opamp/res/cap filters), though would probably not be cost efficient for a small number of bands (MAX263 ~ $7). But for a larger display of 16, 32 or even 64 channels, it's probably cheaper and much less complex. I am thinking of trying it on my project to save boardspace and wiring complexity, and also potentially allow more flexibility in analysis by selecting the frequencies I want to look at on the fly. That way I could have dedicated spectrum displays for music, voice, ambient noise, etc. To be continued...
