DIY RTA - Do it Yourself Real Time Analyser
22 May 2011 Project has been contributed and published in the ESP site. You can see it here: Project 136
Please bear in mind the amount of work needed to complete this project. There is already a lot of information in here and on the Project 136 Page in ESP. I won't be able to offer much help since I also have other things to do. I will try to help but, Please! Read everything here before asking questions.
With all the available software based RTA available to buy (or for free) there's not much reason to make your own RTA from electronic parts. But where's the fun in that?
This page will document the 31 Band RTA that I will be building from scratch. I based my circuit from various sources (mostly ESP)
The basic RTA can be broken down into this circuit (click for full size):
It contains a buffer (not all channels will have a buffer), A MFB filter, simple full wave rectifier and the bargraph driver
Here's a link to the spreadsheet I used for this project (~74kB). It calculates both ideal and actual values based on available resistor values and also includes suggested values for 10 band, 20 band and 31 band filters.
There are two LEDs per output because one is a 3mm size and the other is a 0805 SMD type. I'm not sure which I'll use but probably the 3mm kind.
The PCB can be seen here (click for larger view):
The RTA will be composed of 31 of these PCBs. Only a few will have the buffer populated as one buffer can drive several filters.
Estimated supply draw will be +15V @ 600mA, -15V @ 300mA for the analog and LED drivers and +8V @ 3.1A max (bar mode) 310mA max (dot mode) for the LEDs
There is also a small signal FET used as a switch to select fast or slow response time for the display.
A solder jumper is used to select bar or dot mode.
I used mostly SMD to keep size to a minimum. Board size is a little over 2 inches square.
I'm going to have the boards professionally made. Soldering 31 boards is one thing, etching 31 identical boards is a little too much for me.
I also have made a spreadsheet that helped me calculate all the resistor and capacitor values for the MFB filters.
This page will continually build up as I move forward with this project.
15 Mar 2011: Board layout has been sent to boardhouse. Lead time is around 3 weeks.
Here's the schematic for the planned PSU. It's a basic triple output SMPS using TL494 in push-pull topology.
Schematic (click for full size view).
I plan to power it from a 12V battery pack to make it portable that's why the input has to be 12V. It will be computed to regulate down to 9V.
PWM Feedback is taken from +15V output which is the highest draw that is most constant. The LED supply output (+8V) does not need to be tightly regulated. The LED drivers will work as long as it is above about 4V but not too high to limit dissipation.
18 Mar 2011: Board is scheduled to ship on 29 Mar 2011.
Here is the plan on how the PCBs will be wired....
This is a block diagram of one PCB.
It contains a buffer with two paralleled op amps (for high current capability) and a MFB filter + rectifier + led driver section.
There will be 31 boards. All will have the Filters, Rectifiers and LED drivers used but only 11 buffers will be populated and used. Here's why:
Here's a block diagram/signal flow of the complete RTA.
The first preamp has high gain. It is simply used to provide enough gain and buffer the mic output to be able to drive ten further buffer inputs.
Its output fans out to 10 more buffers (each with 100K input impedance, 10K overall) then further drives the 31 filters. If a line level source is connected, the first preamp/buffer is bypassed and the source drives all ten buffers. The ten buffers simply provide enough gain so that most line level inputs can drive the display to full scale.
Each buffer will be designated to drive three filters to limit the max loading on the buffer output but still not have to populate all 31 buffers to save time and parts. The first buffer will drive four bands as 31 divided by 3 is ten plus one remainder. I chose to group the first four to be driven by one buffer because those bands have the highest input impedances so the buffer would not have a hard time driving them.
The reason for using such a high input voltage for the LED drivers is because the simple rectifier circuit does not overcome the diode voltage drop. If the input voltage is low, The diode drop will cause error in the first few LEDs. Using a higher input voltage lessens this effect.
21 Mar 2011: I decided to change the block diagram a bit....
Here's the revised block diagram.
Instead of making a mic from electret capsules, I decided to go with a more accurate low cost Behringer measurement mic. Since it is balanced and needs Phantom power, I was able to acquire a copy of the Behringer DCX2496 schematic diagram and copied the mic preamp section (used for auto align feature of the DCX) which also has +15V phantom power.
Here's the schematic of the planned preamp board.
Here's the board layout.
Line input will be a BNC connector and the mic input a female XLR connector.
It is not final though as I feel the low frequency rolloff might be too high. Increasing the input caps or ESP's P122 appears to be a good candidate. With just a little bit of modification, I could make it work with the Behringer mic.
Hardware is still a mess. I don't know yet how I will keep all those boards together.
28 Mar 2011: PCB's have been shipped. Hopefully arriving before the weekend....
I have built the mic preamp module. I used 100uF instead of 47uF in the input caps to lower the LF cutoff.
Piccy of the module.
I have also completed the power supply board.
Piccy of the PSU board.
I expected the unregulated outputs to not differ from the regulated output but voltage fluctuation was too much resulting in very low output (with the +15V unloaded) to over voltage (with slight loading on the +15V output) so I scrapped the idea and made a new one from a couple of DC-DC bricks I found in the parts bin. The new PSU will provide +15V and -15V at 1000mA each and an unregulated 12V output (connected to the 12V input but after the main fuse).
Piccy of new PSU module.
Powering the LEDs at 12V is fine but would result to having too much dissipation in the LM3915s in BAR mode so I would be limited to DOT mode operation until I find (or make) a suitable DC-DC that takes in 12V in and output anywhere from 4-8V at 3amps.
03 Apr 2011: Very busy weekend....
The boards arrived the evening of 30 Mar 2011.
I set up and assembled one complete band (1kHz) and tested the timing (sort of attack and release) of the rectifier circuit so that the movement of the LEDs would be smooth and the speed just about right. It ended up that the initial design was too slow. such that it will hold on the peaks and stay there for ages. I changed it to 3.3uF and the resistor 1M to 270K, the 100K to 56K. I also found a mistake in the schematic. (published schematic is now correct, the one with the mistake was lost in the old website) One of the powersupply caps started to smoke then popped while testing. It was drawn in the schematic in reverse polarity and was not discovered up until this time. My decision to make one test board to iron out the bugs before mass production paid off. Imagine 31 caps all popping at the same time!!!
Here's a snapshot of the 1st completed board.
Once it's good, Time to set up for mass production... Here's my bench with the parts within reach.
After a while I started to get dizzy due to the solder smoke so something has to be done. I needed a fume extractor....
And after 10 hours of soldering, all SMD parts for all 30 boards have been installed.
The next day, All through hole parts are installed. Started with the LEDs but I forgot to take a pic of the jig I used when mounting the LEDs so that they are all in the proper position. It's simply a piece of acrylic with holes drilled to hold the LED's in place while I solder them. Here's one while installing the LM3915 LED driver chips. The piece of blank FR4 material is used to guide the pins into the holes while I press down on the chip. I was supposed to use the IC inserter thingamajig but the IC is to long such that 2 pins aren't guided into the holes properly.
Here's the bench jig used to hold the PCB in place while I solder the chips.
Cleaning the board was done using lacquer thinner, toothbrush and compressed air. My process is dipping the brush in a can of thinner then brushing the flux off the board and immediately blowing with compressed air afterwards, repeat if necessary. If the thinner was left to dry on the PCB, it will leave sticky flux residue so it has to be removed quickly.
31 boards all done and the chassis will be done next weekend.
09 Apr 2011 Here are a few pics of the partial assembly of the chassis so that I could test the RTA with the whole 31 Bands wired up...
Drilling 62 Holes into an acrylic base plate to hold all the boards together, a jig makes quick work of the tedious job.
The baseplate finished.
Mounting the boards one by one using right angle brackets.
Stringing them together for the common electrical connections.
Connections to my triple output bench supply for testing.
Calibration pots and the labeled frequencies of each band.
AND IT WORKS!!!!
Current consumption is much lower than I was expecting. Supply current for the +15V is 340mA, -15V is 160mA, and +5V is a little over 300mA with 1 LED of all frequencies lit.
10 Apr 2011 Here are a few pics while continuing with the chassis work...
You can see the buffer circuits under the assembly.
The chassis is 50% done. Just need to make a box out of sheet metal to enclose the top, bottom, sides and back.
bottom ready for wiring up...
All wired up.
Powersupply: I stuck heatsinks because the DC-DC bricks get pretty warm. They are rated 1000mA each. One is driving 160mA, the other 340mA. I have no idea why they get hot.
I also added a 5V buck converter module for the LED supply to lessen current draw.
The bricks had trouble starting under load. Turned out the undervoltage protection is triggered by the inrush current so I had to add a 2200uF capacitor at the DC jack.
Fired it up.
Side view of the open chassis. The whole assembly would be inserted into a metal box through the front. I plan to have the front acrylic laser etched for the labels and then paint some portions sometime in the future.
I tested the RTA in the car playing pink noise and using the Behringer measurement mic and powered by a Li-Ion battery pack.
Here's a couple of animated gifs of the RTA in the car while I was playing music...
23 Apr 2011 Chassis is finished! RTA is almost done. I just need to get some car window tint to add contrast to the display.
Labels are done using rub on transfer lettering.
Drilled holes on the sides for ventilation. A fan was added inside for circulation. Works very well.
The back is closed. The four screws hold the subchassis inside the box.
Added window tint to the LED display for better contrast (camera flash was on).
05 Mar 2013:
A reader, Scott, Built his version of my Diy RTA and sent me some pictures a while ago and almost forgot about it due to busy work schedule, sorry about that. He used my original Eagle files for the PCB and had them made. He also used blue LEDs for the display which are quite nice.
27 Jan 2014:
I decided to play with the RTA with my arbitrary waveform gen. One pic is a collage of the RTA displaying a fixed sine wave of the center frequency of each band and the other an animated GIF of a logarithmic sine sweep from 20Hz to 20kHz.
Page updated and copyright R.Quan © 22 May 2011.