A.F. spectrum bar graph display

Some of us may have seen in the past a HiFi audio frequency bar graph display, usually requiring very high "Q" filter band pass filters, Other circuit implements use a computer code to generate the display output, but in either case, cost is the issue. My idea below over comes the high "Q" band pass filter problem.

The A.F. spectrum analyser design illustrated here, is based around an audio frequency direct conversion receiver. The receiver mixer is composed of clocked gating analogue switch by the clock frequency from the audio local oscillator ( the audio local oscillator to set the spot frequency measurement ), followed by a lowpass filter to limit the zero frequency intermediate bandwidth. From this point, the LM3915 3dB increments and decrements bar graph display chip would plot the spot frequency amplitude. The circuit shown below is just one spot frequency sampling circuit, so one would need say as many as circuit required to cover each spot frequencies in your overall audio spectrum analyser home made circuit design application.

The LM3915 bar graph display chip is designed for power measurement, that is the display increments are in 3dB steps, from -27dB to 0dB. The 0dB illuminated led could represent full power on the transmitter, while the A.F. spectrum illustrates the voice signal from the microphone, an indication of the transmitted signal over the audio band plan would be illustrated, while indicating any low level or high level audio spot frequency component values. In such matters, the article relating to the LA3600 five band graphic equaliser,

url:- https://sites.google.com/site/radiohamtechnology/la3600-5-band-af-equaliser

, could then be used to level off or top any particular voice frequency component.

Should the A.F. spectrum bar graph display analyser be used for the received audio signal, then likewise the LA3600 five band equaliser could be used to again level off or top off and of the audio components, from indications given by the A.F. spectrum bar graph display.

Many other versions of an audio bar graph spectrum analyser use computer code a suitably fast enough micro-controller. The Arduino Uno is perhaps a little out of the reach to attempt this over head computer coding, for itself, but the design posted here is a purely an analogue implementation, which hopefully should provide much help for any radio ham or xbox even PS4 gamer to see their voice signal in detail, cheaply and hopefully easy to construct, and also to not to out do a shortwave listener with the project use.

Many of the readers of this article may have twigged that to just set the "555 timer" to the required spot test audio frequency is all that is needed to analysis that particular audio component. This means if perhaps say a 7 band or even a 14 band spot frequency audio analyser would mean a circuit such as above for each spot frequency, but the difference with this design, is to just tune the 555 timer to the audio frequency component you wish, as the above circuit would be the same for each spot frequency test. Each clock is used as a tuned local oscillator for the mixer staged constructed around an analogue switch using the CMOS4066 device.

Perhaps the only difference between each sample frequency should one wish to, is alter the video low pass filter, to either increase its value in bandwidth, would increase the audio frequency bandpass filtering effect of the test spot frequency, in the same way if a direct conversion radio was used for CW or SSB. The principle of operation in this respect is the same, but in this regard, the direct conversion technique is used to over come the usual AF high "Q" bandpass filters that would normally be use for such a circuit application implementation. The analogue switch is used for the direct conversion receiver mixer, an idea picked out from the "HF Rx soft rock" add on boards. The analogue switch should provide enough linearity to cater for the ranges of signal amplitudes from the audio signal source. However the analogue switch is A.C. coupled in the above diagram, but many circuit implementations of the CMOS 4066 do not use a coupling capacitor.

The circuit diagram above has these capacitors, put into the circuit for completion, however looking back, as the audio signal ranges from essentially 20Hz to 20KHz for a HiFi audio system, the coupling capacitors may give an unwanted attenuation form the capacitive reactance, to the overall audio signal as it rises across the complete audio spectrum. It may be thus best to not include the coupling capacitors, there by directly coupling the analogue switch into the overall circuit design. However experimentation would provide the answer to as not or not.to use coupling capacitors around the analogue switch used as the input signal to local oscillator mixer, particularly as we are testing for a zero intermediate frequency signal reference value.

As each spot frequency is set by the audio oscillator frequency of the 555 timer, ones construction can be adjusted for the fine adjustment of the 555 timer with a adjustable trimming variable resistor, and adjust the design specs without alter or changing component values with a soldering iron.

In this way, I am hopeful that this design is an improvement on such an equipment design of an Audio frequency spectrum analyser bar graph display, different to the norm that has gone before, costly equipment that would otherwise be the case to pay for.

Now regarding the display choice, these are many. Below are a couple of suggestions.

The two displays shown above are just two examples found Amazon, but ebay would also have the equal same. In fact, the whole project could be constructed from parts from Amazon or ebay.

The ten segment bar display specks for itself, but the 5 * 7 dot matrix display has also a use. If the 5 * 7 display was placed on its side, then the display would be 7 dot long, and 5 dots high. The display itself is as I seem to think, is stackable, thus so, if two of these dot matrix displays where place one above the other, then the complete unit display would 10 dots high in signal amplitude, as the LM3915 has ten led connection outputs, and thus the display on it side would be 7 dots along, or to put it another way, 7 spot test frequencies along the frequency domain. Now remember that the analyser circuit is identical to each one, apart if the test frequency bandwidth is different, users choice on this bit, so then each one of the seven spot frequencies would have its 555 timer set to the test spot frequency as the analyser. Now, if a second seven spot frequency unit was used, then a range spot frequencies could be easily set up, covering perhaps a very useful 14 test audio spectrum analyser frequencies along the audio frequency domain.

Such spot frequencies that could used for a 14 spot frequency analyser, each one the audio sample frequency for the direct conversion process set by the 555 timer A.F. local oscillator, could be then as follows below:-

300Hz

450Hz

600Hz

800Hz

1000Hz

1200Hz

1600Hz

2000Hz

2400Hz

2900Hz

3500Hz

4100Hz

4800Hz

5500Hz

For a 7 row display construction, covering the perhaps the spot frequencies 300Hz to 3500Hz, then the list could as follows say 457Hz apart:-

300Hz

757Hz

1214Hz

1671Hz

2128Hz

2585Hz

3042Hz

3500Hz

The band guard spacing between each frequency gives the ability to alter the AF test spot audio bandwidth, by altering the video lowpass filter cut-off frequency, so to cover to the mid point between each spot frequency. The lower spot frequencies would cover most of the lower voice audio components, while the mid range would be catered for, and then the top end of the voice components up to 6KHz, for AM broadcast stations. It may be worth thinking, that the dot matrix display I am not sure just how to interface, if each led is individual connectable, then the idea just may work. If not with each individual connections, then a degree of display led multiplexing would be required, so time to search tor a circuit diagram regarding the dot matrix display:- as below.

display no:1 - bottom level display panel display no:2 - top level display panel

( display columns 1 to 5 and rows 1 to 7 ) ( display columns 6 to 10 for rows 1 to 7 )

key:- rows for frequency domain, and columns for signal amplitude

From the diagram above of the circuit of 5 * 7 dot display matrix, the rows would be the test frequency domain, while the columns would be the signal amplitude. To achieve the display, image turning the dot matrix display on its side, rows on the bottom and columns up the side. By doing this, the correct frequency domain display would be formed, as the displays are stackable, one above the other, ten dots high for the columns thus the signal amplitude climbing upwards for a climbing signal amplitude, and then retract downwards for a declining signal amplitude, while the frequency domain of the display rows along the horizontal plane, would give the correct frequency gradient going from left to right, as the sample spot frequency climbs from 300Hz to 5500Hz.

To achieve a ten LED dot in height display, two of the 5 * 7 displays are stacked top of each other by turning the dot matrix display panel's on its side ( long side horizontal down for the frequency domain, short side vertical up for the signal amplitude ). Display no:2, is the top dot matrix display panel providing display columns 6 to 10, while display no:1 is placed below display no:2, thus the lower display panel then completes the ten LED dot height display panel, the lower place display panel providing LED dots of 1 to 5, of columns 1 to 5. The complete display panel assembly will then give a ten dot in height to match the ten LED outputs of the LM3915 signal measurement display chip, while the horizontal rows provide the frequency spotting display LED's for the step sample frequencies along the frequency domain, from left to right (300Hz to 5500Hz) in this article design spec.

To make the dot matrix display version to function correctly, all of the LM3915 outputs would need to be connected to the column led connection altogether. Although this would in effect cross connect everything, then if as the row voltage is applied in a multiplexed format, one row after the other row, the power supply to each LM3915 could then be connected to the bar graph display chip as when required, for each row after the other row, to a appointed bar graph display chip. This principle would then multiplex the complete display, one row column after the other.

While the display chip would have its D.C. supply switched on and off per the display multiplexing, the direct conversion receiver circuit would have an none interrupted D.C. voltage supply source, thus then the test spot frequency signal amplitudes to be displayed would not be interrupted, just the display chip output is multiplexed across the dot matrix display rows, from 1 to 7 and from 8 to 14, etc for a 14 spot test frequency audio analyser. The above dot matrix diagram of the dot matrix display, would be doubled with two complete display panel assemblies, providing up for the 14 individual spot test signals within the frequency domain from 300Hz to 5500Hz, relating to this article.

Another way to multiplex the dot matrix display. By using a CMOS "4017" chip to synchronise the column selection of the dot matrix display with the appointed spot frequency multiplex output analogue switch selection, by using a common connected decoded output pin selection for both the spot frequency and display column, of each individual measured spot frequency. Each of the output pins of the CMOS "4017" chip would then use a transistor switching circuit attached to the column selected of the dot matrix display, for the multiplex switching of the D.C. supply voltage to the appropriate dot matrix display column, for the appointed spot frequency measurement display. The below diagram that illustrates the spot and row selection, would be spot 1 and row 1 selected together, then spot 2 and row 2 following, up to the designed range of the number of spot frequencies and display rows ( in other words the displayed row would be thus the vertical columns of the dot matrix display ).

The CMOS 4017 chip is a chip device that is of a CMOS Decade Counter with 10 Decoded Outputs, otherwise used for a LED light chaser circuit, the circuit in this case used to multiplex the dot matrix display column selection, using the light chaser circuit principle, thus to use the 10 decode outputs to D.C. switching the individual display columns common LED pin connection. From the above circuit diagram of a dot matrix display, the common pin connection for row one is "pin 12" of each of two the dot matrix displays, to create a 10 LED high column, of column number one, and so on.

Below timing diagram represents the 10 decoded outputs "Q0 to Q9" of the CMOS 4017 chip, are used to synchronise the multiplexed display columns to the selected spot frequency amplitude sample. The below double transistor circuit illustrates the principle.

Each multiplex row is presented as and when it is required. As the anodes of the each led position is commonly connected, the power switch selection of the cathodes will determine which set of vertical rows are to be illuminated. The common connected anodes would be connected to the LM3915 LED output pins, e.g. all led 1 cathodes would be connected to D1 of the LM3915 and so for all led 2 cathodes connected to D2 of the display chip, and so on. As the CMOS 4017 clocked synch outputs are rotated from Q0 to Q9 and back around to Q0, so the correct row would then display the spot test frequency signal amplitude, derived from the intended spot frequency direct conversion process circuit in question. In other words, with this articles design spec, the output Q0 would select both row one and the analogue switch synch from Q0 to match the spot frequency 300Hz amplitude sample to the correct vertical column display, while Q1 would select both row two would be aligned with the analogue switch synch from Q1 to spot frequency 450Hz, each row thus displaying the signal amplitude of their appointed spot frequency. This is perhaps the simpler design principle overall, to use one LM3915 and together multiplex in tandem the intended spot frequency signal amplitude for the intended display row on the frequency domain.

This would use just one LM3915, and have the 7 or 14 direct conversion AF receivers, tuned by the 555 timer audio local oscillator ( the audio local oscillator to set the spot frequency ) for the direct conversion radio principle, but achieved in this case at audio frequencies rather than at R.F. frequencies, the analogue switches outputs would be multiplexed, the signal voltage from the spot frequency direct conversion circuits then presented to the input of just one LM3915 measurement display device, and then display multiplex switching used to select the appropriate display row, as only one chip would be connected to the display columns across the complete display panel assembly.

As each analogue switch is high impedance while switch off, in other words not passing through the signal amplitude voltage, a decay resistor would need to be placed across the lowpass filter capacitor, to decay the voltage over a short time period as one second or halve a second etc. This is to allow the detected signal amplitude of the spot frequency to vary in the face of the analogue switch high impedance switched off state. If the analogue switch was switched on, thus passing through the spot frequency signal voltage ( amplitude ), there would sufficient low impedance to load the video filter capacitance, and not leave the video filter output in an open circuit impedance. The spot frequency test signal amplitude voltage is essentially used as an A.G.C. voltage circuit without using the signal amplitude voltage as a feed back control circuit itself.

A point to bare in mind, the lowpass filter after the direct conversion mixer, the cut-off frequency of the lowpass filter would also govern the audio sample bandwidth, the lowpass filter referred to as the "video lowpass filter". Should the lowpass filter cut-off frequency be set to 80Hz, then the reflecting audio "A.F." sample bandwidth within the direct conversion receiver spot frequency A.F. bandwidth, would be twice that of the video lowpass filter, hence video LPF = 80Hz, then the spot frequency bandwidth = 160Hz, just as in a R.F. Spectrum Analyser. The same affect is also found with a R.F. direct conversion receiver say for the 80m band while resolving CW or SSB, the audio bandwidth is half that of the R.F. bandwidth.

The video filter cut-off frequency needs in truth to half the distance between two spot frequency channels. Example here are the spot frequencies 300Hz and 450Hz, the guard band would then be half the distance, 450Hz - 300Hz = 150Hz, thus the video filter would be a 75Hz cut off frequency. The sharpness of the video filter would relate to the sharpness of the audio sample bandwidth, in other words a video filter of 75Hz, would give an audio bandpass filter bandwidth of +/- 75Hz at either 300Hz or 450Hz. As the frequency domain spot frequencies differ in space between each spot frequency, the video filter cut-off frequency may perhaps need to be different as the frequency rises from 300Hz to 5500Hz, and the spot frequency channel separation increases. However, the sharpness of the video filters would also related to signal bleed over between each spot frequency measurement of the audio signal spectrum display, in order to provide a clean reading of the voice characteristic of the voice signal amplitude within the frequency domain of each spot frequency.

The sample and row multiplexer is essentially the same unit, in that the multiplexing timing will be used for both the spot frequency sampling and the correct column addressing of the overall display to place the signal sample amplitude on the correct column of the display in general. Should one choose to build the 7 or even the 14 spot frequency version, then second principle of multiplexing would be to use a 4 bit decade counter connected to a 4 to 16 line decoder would then could be used to multiplex time the 14 channel spot frequency analyser. The alternative to a 4 to 16 line decoder circuit from a 4 bit counter, would be to use the "CMOS 4017 decade counter". With the CMOS 4017, the ten outputs are one by one enabled.

Should one be constructing your A.F. spectrum scope in two halves, it may be an idea to use two of the CMOS 4017 chips, one device for rows 1 to 7, and a second for rows 8 to 14. If however the complete 14 rows are one display construction, then perhaps the 4 to 16 line decoder chip may be best, but constructing your audio spectrum analyser in blocks of 7 rows, would only require to place each 7 rows beside each other to make your overall display, needing to then only to just tune the audio local oscillator to the required spot frequency setting. Either way, each method would produce a logic high or low signal to use to synchronise the sample and display column selection in sequence with the projects use, to display the voice signal as a frequency domain plot using bar graph display. From the voice frequency domain plot, one will be able to determine if the signal lacks either a base or treble tone cut or boost, or even a middle tone cut or boost.

Now many of use may be aware that these bar graph display chips are able to produce either a column of illuminated led's or just a moving dot. This is useful, as on the moving dot the outlining spectrum response will be shown as a curve, but using the column option with these display chips, then the area of the display curve will filled with illuminated led's. The display mode selection is just a switch logic pin on the display chip, so a panel switch would do the job for the display mode selection.

It also may be worth noting that these chips are interchangeable, that is the linear voltage version ( LM3914 ) and the 3dB power increment / decrement version ( LM3915 ) illustrated here in the diagram, and the audio "Vu" meter version ( LM3916) are all interchangeable. One could in theory parallel connect the LM3915 and the LM3916 and switch it DC line to select either the "power reading" or a "Vu meter" version for the display reading. Although each display chip would have its own bias circuits arrangements, the display outputs of each of the chip version would common connected to the appropriate column led, but switching the DC supply line to each chip would selected the required display curve characteristic, and thus the active chip output led connection to the multiplexed display. However the multiplexed spot frequency sample voltage output would be common to each signal input of the display chips, thus the DC line select and activate the appropriate display chip.

Listening on the HF last night I heard other hams also suggest this idea in their chats on the bands.

For completeness of the 555 timer oscillator circuit, the below implementation has an equal mark / space ratio output clock signal design.

Below are a set of resistance and capacitance values for the above 555 timer using basically 5K ohm trimmer pots. If the spot frequency spread is too great, then retuning the resistor trimmer pots could be done to bring in the overall spot frequency range into an overall narrower width, thereby characterising your own design too your own specs.

Below is the BBC Basic code for the above table, placed within a text window for a copy and paste into the BBC Basic IDE.

Text Box

10 REM 555 timer tuning resistance for clock to set capacitance vaoues

20 b = 0

30

40 PRINT

50 freq = 300

60 Ct = 120E-9

70 PROC_clock

80 freq = 700

90 Ct = 120E-9

100 PROC_clock

110 freq = 900

120 Ct = 120E-9

130 PROC_clock

140 freq = 1200

150 Ct = 47E-9

160 PROC_clock

170 freq = 1500

180 Ct = 47E-9

190 PROC_clock

200 freq = 1800

210 Ct = 47E-9

220 PROC_clock

230 freq = 2200

240 Ct = 33E-9

250 PROC_clock

260 freq = 2500

270 Ct = 22E-9

280 PROC_clock

290 freq = 3000

300 Ct = 22E-9

310 PROC_clock

320 freq = 3300

330 Ct = 22E-9

340 PROC_clock

350 freq = 3600

360 Ct = 22E-9

370 PROC_clock

380 freq = 3900

390 Ct = 10E-9

400 PROC_clock

410 freq = 4200

420 Ct = 10E-9

430 PROC_clock

440 freq = 4500

450 Ct = 10E-9

460 PROC_clock

470

480 END

490

500

510 DEF PROC_clock

520 b=b+1

530 Rt = 1/(2*PI*Ct*freq)

540 Rt_display = INT(Rt *1)/1

550 PRINT TAB(5);" No.";b;TAB(15)" freq = ";freq;" Hz";TAB(35);"Rt = ";Rt_display/1E3;" K ohms";TAB(58);"Ct = ";Ct*1E9;" nF"

560 ENDPROC

Customising your audio spectrum analyser.

If you are building your audio analyser for gaming, then it was suggested on the news media to build a analyser to cover the audio spectrum up to 8KHz. Now 3 blocks of 7 rows would be 21 spot frequencies, which would then give a spot frequency of essentially around 300Hz of increments, which would mean that the lowpass video filter, will need to be tuned to a cut-off frequency of 150Hz maximum, would be the same on all of the spot frequency settings. The cut-off slope of the video filter would need to be sufficient to avoid any real amount of signal bleed over between each spot frequency reading.

The LM3915 maximum sensitivity on the lowest reading is -27dB, from the max of 0dB. Therefore, the slope cut-off of the video filter would need to be around -30dB or around such values in order to avoid the next spot frequency sample from assuming any degree of lower or higher spot frequency fall over between each frequency sample, thus from above or below the sample spot frequency in question currently on display. It may will best to outline that the LM3915 has the dynamic range of some 500:1 in ratio ( 27dB ), that is to say that the audio spectrum display between two samples would illustrate a amplitude difference within each spot frequency display of 500:1 of input signal strength, between each side by side column of LED's within the overall display. A difference between a power ratio of 500:1, would probably mean a full sound on one part of the display, a very quite sound signal on the next display column.

However the decay "R ohms" resistor would have to allow the display voltages to decay in tune to the oscillations of fire arms of say the game "call of duty". The rapid sound of fire arms would be shown on the audio spectrum display, as well as voice sounds of each soldier and any other sounds of explosive devices. Sounds of transport equipment of the sounds of the engines, be it jet or piston engines, as will the sound of opening and closing doors. All the sounds will show up on the graphic bar graph display, placed on the audio spectrum display in accordance to their sound pitch. The higher the dots on the graph rise, the loader will be the sound track of the game. Any harmonic content of the sounds would most probably show up on the audio spectrum display.

Whether if the audio spectrum analyser is made as one 21 rows display or as 3 individual 7 rows segments, would depend upon the manufacturing process, or your home construction techniques.

Now the video filter need only be a Butterworth filter design, comprising of a series resistance and a shunt capacitance. This would give a 1st order filter, so I decided to write a BBC Basic code to determine the components required, and calculated the order of attenuation. The below table illustrates a list of a 1st and 2nd and 3rd order filter attenuation, as well as the signal frequency.

Please bare in mind, that the cut-off performance of a filter is based on its octave slope, that is to say that while a 3KHz filter may have a 3dB fall at 3KHz, the next point is 6KHz, and the point after that is 12KHz. I was not sure with it was like a harmonic output such as a radio, 1st ( fundamental ), 2nd harmonic ( 2 * Fc ) or 3rd harmonic ( 3 * Fc ), or as the 3KHz example. Today watching Ham Nation ( 7th Nov. 2019 ), George Thomas W5JDX, also of "AmateurLogic.TV", demonstrated a book by "Forrest Mimms the 3rd", which illustrated the correct view point of a filter for the 3KHz octave roll-off slope, comprising of the 3KHz fundamental, the 6KHz first octave and then the 12KHz 2nd octave, performance characteristic fall-off slope of a lowpass filter. Cheers Sir, many thanks.

The series resistance is 20K ohms, because the internal input resistance of the LM3915 as well as fellow chips, has a 20K ohm input resistor, according the internal block diagram. I found that a two capacitor combination of a 47nF in parallel with a 10nF, gave the required filter characteristics. The above table illustrates that at the 300Hz point, the gap between the spot frequencies, the signal attenuation is some 30dB down from the input signal of a 3rd order Butterworth lowpass filter, shown in the below diagram.

However this is a fault with the above design, because with a frequency difference of 300Hz between the spot frequencies, the above filter really needs to be 30dB down half way between two next door spot frequencies. In other words, 30dB down at 150Hz from the spot frequency. This is to provide what is known within the radio world as "guard band". however while it is true that the cut-of point is actually around the 20Hz range, from the 3rd order filter, at the point where the video filter attenuation is required, 150Hz up from or down from the next spot frequency, the 30dB figure is achieved.

New list table below:

Thus the circuit is a follows:-

Below is a text window copy of the BBC Basic coding to use for your own 3rd video filter design.

Text Box

10 Rf = 20E3

20 Cf = 47E-9 + 10E-9

30 V_signal_start = 1

40 V_signal = V_signal_start

50 f_cutoff = 1/(2*PI*Rf*Cf)

60 PRINT TAB(5,1);"LPF cut_off = ";INT(f_cutoff *10)/10;" Hz"

70 PRINT TAB(5,2);"audio centre BPF = "; INT(f_cutoff *2 *10)/10 ;" Hz"

80 PRINT TAB(10,4);"signal freq";TAB(20);" 1st order";TAB(36);"2nd order";TAB(51);"3rd order"

100 PRINT

110 FOR freq = 1 TO 360 STEP 15

120 PROC_first

130 NEXT freq

140

150 END

160

170 DEF PROC_first

180 Xc = 1/(2*PI*freq * Cf)

190 V_first_atten = Xc/(Xc + Rf)

200 V_signal = V_signal_start * V_first_atten

210 atten_dB = 20*LOG(V_signal_start / V_signal)

220 PRINT TAB(10);"freq = "; freq; TAB(30);" 1st = "; INT(atten_dB*10)/10;TAB(45);" 2nd = "; INT(atten_dB*2*10)/10 ;TAB(60);" 3rd = "; INT(atten_dB*3*10)/10

230 ENDPROC

240

250

A 4th order Butterworth lowpass filter would give a video bandwidth of 117Hz, which is fairly close to the 150Hz guard band limit. The 4th order video filter design is shown below:-

Listening on the 20m band today 18th October 2019, one radio ham suggested a spot frequency of 150Hz spot sample, and maybe the chap said a 75Hz spot sample, I would like to say, not a bad idea at all, the low level audio signal would surely show up, if not also on an audio tap of a gaming computer or xbox or ps4. However, below are the tabled results of the 4th order Butterworth filter for a 300Hz spot frequency separation. Not at the 150Hz signal frequency line on the below list, the attenuation is some 29dB down.

Text Box

10 Rf = 20E3

20 Cf = 68E-9

30 V_signal_start = 1

40 V_signal = V_signal_start

50 f_cutoff = 1/(2*PI*Rf*Cf)

60 PRINT TAB(5,1);"LPF cut_off = ";INT(f_cutoff *10)/10;" Hz"

70 PRINT TAB(5,2);"audio centre BPF = "; INT(f_cutoff *2 *10)/10 ;" Hz"

80 PRINT TAB(10,4);"signal freq";TAB(20);" 1st order";TAB(36);"2nd order";TAB(51);"3rd order";TAB(66);"4rd order"

100 PRINT

110 FOR freq = 1 TO 360 STEP 15

120 PROC_first

130 NEXT freq

140

150 END

160

170 DEF PROC_first

180 Xc = 1/(2*PI*freq * Cf)

190 V_first_atten = Xc/(Xc + Rf)

200 V_signal = V_signal_start * V_first_atten

210 atten_dB = 20*LOG(V_signal_start / V_signal)

220 PRINT TAB(10);"freq = "; freq; TAB(30);" 1st = "; INT(atten_dB*10)/10;TAB(45);" 2nd = "; INT(atten_dB*2*10)/10 ;TAB(60);" 3rd = "; INT(atten_dB*3*10)/10;TAB(75);" 4rd = "; INT(atten_dB*4*10)/10

230 ENDPROC

240

250

The below data is from the LM3915 data sheet.

Using a 5Volts D.C. supply application.

The "R-led ohms" relates to the LED illumination intensity. I am assuming the max input signal voltage would be equal to the "REF OUT", i.e. 1·25Volts. The diagram next along illustrates the "brightness" control for the LED display. The design of the dot / bar audio spectrum analyser, could well be run from a 5Volt supply in all. This would mean that a 5Volt USB power supply would only be needed. As the maximum audio signal input to the LM3915 in the below diagram is around 1·25Volts, the isolation op-amps voltage supply of 5Volts, the audio signal range of 1·25Volts is within the power supply limits of the op-amps. The isolation op-amps are used to prevent signal cross-over between each spot frequency sampler circuit, feeding back through the power supply voltage rail. According to the CMOS 4066 datasheet, the channel separation between each individual analogue switch is some 50dB.

It maybe an idea to run the isolation op-amps from a split voltage supply, using a negative 5Volt supply invertor from the positive 5Volt supply. This may perhaps simplify the op-amp input biasing requirements, since when an AF Analyser design may perhaps contain up to as many as 7 or 14 or even 21 op-amps within the overall circuit.

Scaling the "Vref" voltage.

The LED current and hence the illumination of the LED is the lower equation of the two equations. The scaling of the "Vref" is the top equation for a scaled high signal input level than 1·25Volts.

Display multiplexer chip, CMOS 4017

The CMOS 4017 shown below is from a manufactures data sheet, showing basic operational set-up arrangement.

I have also come across an Arduino version using dsp maths with a 16dot high display, url as https://learn.sparkfun.com/tutorials/proto-pedal-example-analog-equalizer-project/the-circuit , and an other using a 8 dot high display , url : https://learn.sparkfun.com/tutorials/proto-pedal-example-analog-equalizer-project/the-circuit .

On this one if you change void setup() to as below:

Text Box

void setup()

{

// ADCSRA = 0b11100101; // set ADC to free running mode and set pre-scalar to 32 (0xe5)

ADCSRA = 0b11100101; // set ADC to free running mode and set pre-scalar to 32 (0xe5)

ADMUX = 0b00000000; // use pin A0 and external voltage reference

pinMode(buttonPin, INPUT);

mx.begin(); // initialize display

delay(50); // wait to get reference voltage stabilized

mx.control(MD_MAX72XX::INTENSITY,0);

}

the display brightness is changeable.

Also if you change a line in void loop() marked as " ADCSRA = 0b11100101; // set ADC to free running mode and set pre-scalar to 32 (0xe5) " to " ADCSRA = 0b11110111; " the sample frequency becomes 9600Hz, thus the top scan bandwidth is up to 4800Hz full scale, just right for ham radio.