Some people at DIY I take with a pinch of salt and the same them of me. I can be friends with them regardless. Good drinking buddies. I tend to say a lot of things I only half believe to get a discusssion going. Andrew was a rare person. He could understand that idea whilst always sticking with the facts. I always felt Andrew loved the more specialised type of design like this one for exactly my reasons. He had ways of saying the same as we would from a highly researched study.
Andrew liked mathamatical coincidences. Not facts, just very close to reality. For example, pi^3=31. Or 1.2phi^2 = Pi ( can be said 1.2( phi+1)=Pi ). Also 355/113=Pi. Phi is ( [root5+1] /2 ). Most of these little ideas are as good as engineering needs it to be. It always seems logical the irrational numbers should relate. A good amplifier is close to reality.
I think Tesla said maths killed experiment. I never felt Andrew put maths first although he spoke that way. Andrews racing cars seemed to smoke. I suspect it was a way of keeping the engine healthy. It happened on two video's. Andrew was too fussy to let it be an engineering weakness. That Coventry Climax engine was very remakable. A fork-lift maker who made a racing engine for familly cars.
Andrew liked mathamatical coincidences. Not facts, just very close to reality. For example, pi^3=31. Or 1.2phi^2 = Pi ( can be said 1.2( phi+1)=Pi ). Also 355/113=Pi. Phi is ( [root5+1] /2 ). Most of these little ideas are as good as engineering needs it to be. It always seems logical the irrational numbers should relate. A good amplifier is close to reality.
I think Tesla said maths killed experiment. I never felt Andrew put maths first although he spoke that way. Andrews racing cars seemed to smoke. I suspect it was a way of keeping the engine healthy. It happened on two video's. Andrew was too fussy to let it be an engineering weakness. That Coventry Climax engine was very remakable. A fork-lift maker who made a racing engine for familly cars.
Here is a very cheap thing I made today which might be useful. It has quite a high output so would need a series resistor on pot control for most users. I was a little surprised I got these frequencies from these parts. As very little advice is giving the interplay of the 10K and 100K is the likely cause. All the same a pleasing result. The 68pF NPO measured 64pF. Sorry the graphs didn't copy well.
I might do a 38Hz, 1200Hz , 9600Hz and 19 kHz filter to have sine waves. If so TL084/74 I think would be ideal. Hope somone finds this useful. Be very very careful with tweeters if testing with it. The voltage will change the frequency. A fresh cheap zinc PP3 is ideal. In some ways the 74HC4060 is better, on the other hand it needs a 5V PSU.
That looks great. So it's essentially a multiple 555 timer all in one chip, on the 40-series platform? DIY sig gen, here we come! In terms of making a PSU, I think most here could knock up a 5V regulated PSU in an evening. I know I did aged 18, still have the thing. It was only a simple Zener circuit. Hook up a PP3 and you have a nice steady 5V for as long as you want.
I haven't got to grips with the timing of the 4060 yet when an RC circuit. I have a hunch the 74HC4060 will give a higher frequency with the 10K 68pF although the buffer looks much like the 4060 ( pin 9 ). In the notes I think I read 10K to 1M and adjust cap to suit. Some say R2 is 10 times R1 as I have it.
I would guess a 555 with a 50/50 mark space ratio could make an excellent master oscillator for the 4060. It would feed to clock in below MR. The 4040 and 4020 could do the same. They have more useful outputs. Alas they are becoming rare, they have no oscillator onboard. Using reasonably accurate parts 555 is reasonably predicatable. The 4060 needs playing with to get what you want. One way to do it is to make one output 100Hz ( or 120Hz ). Uses a turntable strobe to compare. The other frequecies would be multiples. 1.6384 MHz/2^14 = 100Hz. Pi computer and others seem to be keeping 4000 series alive.
I have a 6MHz crystal. I have worked out and idea using 2 x 4060 to give very useful frequencies. Being a crystal there is no guessing the frequency. 4060 drives crystals very well. No need to use the fancy circuits. One crystal and one 1M resistor will work. Pin 9 Q1 isn't used ( some makers state Q1, 1/Q1 as it is an inverting buffer I think ).
The advantage of the 4060 is it will work from many voltages up to about 15V. The 74HC4060 is best at 5V. An interesting fact is the 4000 series are very tollerent of abuse. Whilst the 4060 is a low current design it probably will give plenty of current as long as full output voltage isn't required. Pin 9 if carefull is a buffered Q1 output as said. That can be very useful. From memory if a diode is connected from QX to MR ( master reset ) that will work like a NE555 as a one shot timer, it must need a pull down resistor I guess to start the clock. Very long periods can be arrived at. I had one giving days, in fact 1 week if all LED's showing.
If driving lets say a Chebishev filter I think the 74HC4060 best. From my best understanding the the 74HC4060 has an Ron of 50 ohms ( 50 ohms current limiting and more if abused ). Thus if a red LED was used the 74HC4060 would try to give 66mA. Both devices well above where they should work. I have a hunch it could work as LED's often can work on pulsed output at a higher current, almoist certainly Ron > 50R if so. If a 51R was added that might be an ideal turntable strobe using a 3000mcd 10mm high bright LED ( Kingbright ). The parts are so cheap that it's worth the risk. 120R would be totally safe. Nowhere in the specs will it say this. That's because it isn't full rail to rail output.
The good old 7805 is a great 5V source.
I would guess a 555 with a 50/50 mark space ratio could make an excellent master oscillator for the 4060. It would feed to clock in below MR. The 4040 and 4020 could do the same. They have more useful outputs. Alas they are becoming rare, they have no oscillator onboard. Using reasonably accurate parts 555 is reasonably predicatable. The 4060 needs playing with to get what you want. One way to do it is to make one output 100Hz ( or 120Hz ). Uses a turntable strobe to compare. The other frequecies would be multiples. 1.6384 MHz/2^14 = 100Hz. Pi computer and others seem to be keeping 4000 series alive.
I have a 6MHz crystal. I have worked out and idea using 2 x 4060 to give very useful frequencies. Being a crystal there is no guessing the frequency. 4060 drives crystals very well. No need to use the fancy circuits. One crystal and one 1M resistor will work. Pin 9 Q1 isn't used ( some makers state Q1, 1/Q1 as it is an inverting buffer I think ).
The advantage of the 4060 is it will work from many voltages up to about 15V. The 74HC4060 is best at 5V. An interesting fact is the 4000 series are very tollerent of abuse. Whilst the 4060 is a low current design it probably will give plenty of current as long as full output voltage isn't required. Pin 9 if carefull is a buffered Q1 output as said. That can be very useful. From memory if a diode is connected from QX to MR ( master reset ) that will work like a NE555 as a one shot timer, it must need a pull down resistor I guess to start the clock. Very long periods can be arrived at. I had one giving days, in fact 1 week if all LED's showing.
If driving lets say a Chebishev filter I think the 74HC4060 best. From my best understanding the the 74HC4060 has an Ron of 50 ohms ( 50 ohms current limiting and more if abused ). Thus if a red LED was used the 74HC4060 would try to give 66mA. Both devices well above where they should work. I have a hunch it could work as LED's often can work on pulsed output at a higher current, almoist certainly Ron > 50R if so. If a 51R was added that might be an ideal turntable strobe using a 3000mcd 10mm high bright LED ( Kingbright ). The parts are so cheap that it's worth the risk. 120R would be totally safe. Nowhere in the specs will it say this. That's because it isn't full rail to rail output.
The good old 7805 is a great 5V source.
I have some of ceramic resonators. Alas most places do not stock them now. 465kHz I seem to remember was AM radio. Even crystals are becoming rarer. If this 6MHz idea works I will show it. This came to mind as my bench squarewave oscillatror isn't 50/50. This seems to be because it can work with a modulator. If the modulator input is carefully set it is just about 50/50, no mention in the book of words. It wasn't cheap so a surprise.
An old car radio is a good place to find a 450-465kHz ceramic resonator but you still need a frequency divider and some way to ensure a decent 50:50 mark/space ratio.
Many years ago, there were a couple of analog chips designed specifically for low cost, general purpose function generators. There was Exar XR2206 and ICL 8038 and surprisingly, they're still around though probably copied versions. They work very well in a carefully designed application but there are a number of minimalist kits around that for the price, have to be worth the effort of simply stuffing the pcb, wiring up and using as it is, for fun.
I built two fully featured kits based on XR2206 as I figured this chip was the better. These kits included 8 digit counters, tri-wave options and voltage controlled auto sweep - quite expensive for kits in the 1980s but without accurate frequency measurement and a very steady amplitude, an oscillator (or tone generator if you like) tends to be of limited use.
The point of using the chips rather than discrete circuitry, is that frequency and amplitude settings could be made quite stable and you didn't need external temperature compensation to keep them that way, at least in the short term. Here's a really cheap kit to start the fun. Just note the cheapo touches though, like the trimpots supplied for adjustments instead of quality pots. They would have to be upgraded pronto since trimpots don't take kindly to repeated adjustment nor are they consistent:ICL8038 Function Signal Generator Kit Multi-channel Waveform Generated Electronic Training DIY Spare Part Sale - Banggood.com
Many years ago, there were a couple of analog chips designed specifically for low cost, general purpose function generators. There was Exar XR2206 and ICL 8038 and surprisingly, they're still around though probably copied versions. They work very well in a carefully designed application but there are a number of minimalist kits around that for the price, have to be worth the effort of simply stuffing the pcb, wiring up and using as it is, for fun.
I built two fully featured kits based on XR2206 as I figured this chip was the better. These kits included 8 digit counters, tri-wave options and voltage controlled auto sweep - quite expensive for kits in the 1980s but without accurate frequency measurement and a very steady amplitude, an oscillator (or tone generator if you like) tends to be of limited use.
The point of using the chips rather than discrete circuitry, is that frequency and amplitude settings could be made quite stable and you didn't need external temperature compensation to keep them that way, at least in the short term. Here's a really cheap kit to start the fun. Just note the cheapo touches though, like the trimpots supplied for adjustments instead of quality pots. They would have to be upgraded pronto since trimpots don't take kindly to repeated adjustment nor are they consistent:ICL8038 Function Signal Generator Kit Multi-channel Waveform Generated Electronic Training DIY Spare Part Sale - Banggood.com
I was using that in 1980 in FSK modems.
I think its cousin chip was the XR2211 in the modem.
I just had a little play with the 4060 circuit. It seems imune to 17 to 37C ( holding the crystal ). The squarewave distortion is mostly the analyser. Not bad for a cheap idea. The 47kHz square wave is almost perfect. If I get a low cost filter to give good results I will post it. I would predict a JLH would look better showing the squarewaves as it would be a filter in it's own right. I saw 160kHz - 3db on my birdsnest JLH. To be clear, The JLH could look better than the input wave using FFT type scope. That suggests 10 kHz squarewave on a JLH should still look reasonably square with a small rounded edge. 5 kHz should be near perfect. A CD player will struggle after 1 kHz. If you take my results using a 3.2768 MHz chip the frequencies are reduced by a 3.2768/6 ratio. 2/6 for 2 MHz. The 3.2768 gives 50Hz if ratios correct.
Now Analogue guys here is a big truth. I have just switched on my analogue scope. What a shame my Bluetooth dongle has given up, a photo would surprise a few people. The little 4060 gives almost pefect squarewaves on an analogue scope. Pin sharp and ruler flat with 50/50 ratio. The 47kHz as good as the LF ones. I can clearly see the lower gain scope setting is more linear. A 1K load no worse except slight voltage drop and that at 47kHz.
Now Analogue guys here is a big truth. I have just switched on my analogue scope. What a shame my Bluetooth dongle has given up, a photo would surprise a few people. The little 4060 gives almost pefect squarewaves on an analogue scope. Pin sharp and ruler flat with 50/50 ratio. The 47kHz as good as the LF ones. I can clearly see the lower gain scope setting is more linear. A 1K load no worse except slight voltage drop and that at 47kHz.
BTW. Being that the JLH is class A I suspect squarewaves are of no concern? Not your speakers I hasten to say. 5 watts 1kHz should kill the tweeters quickly. You might say the crossover is 3 kHz so no problem. A 6 Vrms 1 kHz square wave gives 2 Vrms at 3 kHz more or less and a volt at 5 kHz, that's plenty to kill a tweeter.
I just saw a few scopes on eBay ( 20 MHz dual trace £50 ). None were showing good squarewaves, some even were not 50/50 ( 47/53? ).
HAMEG HM 203-7 20mhz Oscilloscope | eBay
I just saw a few scopes on eBay ( 20 MHz dual trace £50 ). None were showing good squarewaves, some even were not 50/50 ( 47/53? ).
HAMEG HM 203-7 20mhz Oscilloscope | eBay
If you're yearning for a little more complex signal generator, here's a 4060/4046 based PLL circuit that might be adaptable for audio range output.
$12 AM Band Frequency Synthesizer For Part 15 transmitters and AM Broadcast Band Frequency Generator
$12 AM Band Frequency Synthesizer For Part 15 transmitters and AM Broadcast Band Frequency Generator
I wouldn't pay much attention to displays of the probe adjust oscillator, particularly with the base 20MHz models. Same with all popular 20MHz analog models, as I recall. It's a different matter when you pay thousands for a 'scope and you can reasonably expect every feature and function to be just so.
I did something that I speculated about years ago. That is the current output of the original CD4060 could be much higher than suggested. The reason being the On ( Ron ) resistance of the CMOS output becomes high and limits voltage and current. The speculation was that Ron would be enough to prevent the CD4060 and an LED ( 10mm 3000mcd Kingbright red ) from damage as a direct connection. That is infact true. I tested the LED direct to the Q14 3Hz output. The output when into a 1K load is Ron of 240R. It is hard to say what the LED is running. It's going to be ( 10-1.7 / 240 ) = < 35mA. By eye I would guess it to be 10mA. If I put 1R in series with the LED I could work it out using a scope. 1R is low enough not to change Ron much. This has run for 24 hours now.
If the CD4060 outputs were bipolar ( standard transistors ) this would not be possible without some precautions which might distort the waveform. Of greater interest is that the waveform is still very square here! I will test at lets say 23.5 kHz to see if it remains happy, that's a tough test from many points of view. Most is the output heat wise is like DC.
This test was to confirm if it could drive a filter. Driving many LED's like this from various Qout might be possible. As the devices are cheap I won't say the obvious. My one LED could be 300mW chip dissipation which is as far as I would go. It's using the whole chip area to get rid of the heat. From these test I will recomend CD4060 over 74HC4060. Buy some now, tomorrow might be too late. 6 MHz is about the external input limit ( I dare say higher, 74 series will ). This one worked down to 3V at 6MHz !!! This test circuit cost about £1.
One very odd quality of the CD4000 series is the unbuffered Hex Inverters can be converted into inverting amplifiers. CD4069A, CD4069UB or 74HCU04 ( Note U for unbuffered ). Although the output is zero bias class B the distortion is better than it should be. I suspect this is either because the CMOS never completely switches off at the mid point or hiss is working as a dither. CD4007 has one inverter output and some spare devices. Two gain of 1000 amplifiers can be made from a Hex Inverter ( less in reality if above 1kHz ). I did build a filter from one just to see what it would do. I say about this as it resembles the JLH MOS FET idea. CMOS are very like the modern FET's like the IRF series. The input is biased at the mid point like a single rail op amp with input blocking capacitor. Feedback resistors are usually in the 1M range with 100K input for gain of 10.
http://www.ti.com/lit/ds/symlink/cd4069ub.pdf
If the CD4060 outputs were bipolar ( standard transistors ) this would not be possible without some precautions which might distort the waveform. Of greater interest is that the waveform is still very square here! I will test at lets say 23.5 kHz to see if it remains happy, that's a tough test from many points of view. Most is the output heat wise is like DC.
This test was to confirm if it could drive a filter. Driving many LED's like this from various Qout might be possible. As the devices are cheap I won't say the obvious. My one LED could be 300mW chip dissipation which is as far as I would go. It's using the whole chip area to get rid of the heat. From these test I will recomend CD4060 over 74HC4060. Buy some now, tomorrow might be too late. 6 MHz is about the external input limit ( I dare say higher, 74 series will ). This one worked down to 3V at 6MHz !!! This test circuit cost about £1.
One very odd quality of the CD4000 series is the unbuffered Hex Inverters can be converted into inverting amplifiers. CD4069A, CD4069UB or 74HCU04 ( Note U for unbuffered ). Although the output is zero bias class B the distortion is better than it should be. I suspect this is either because the CMOS never completely switches off at the mid point or hiss is working as a dither. CD4007 has one inverter output and some spare devices. Two gain of 1000 amplifiers can be made from a Hex Inverter ( less in reality if above 1kHz ). I did build a filter from one just to see what it would do. I say about this as it resembles the JLH MOS FET idea. CMOS are very like the modern FET's like the IRF series. The input is biased at the mid point like a single rail op amp with input blocking capacitor. Feedback resistors are usually in the 1M range with 100K input for gain of 10.
http://www.ti.com/lit/ds/symlink/cd4069ub.pdf
I read some very interesting ideas from the music industry using hex-inverters.
Half way down this PDF is an idea I tried. I got the same results. I re-engineered it with the diabolical LM324N and had better results. 324 is OK if lightly loaded or forced into SE class. All the same this is a reasonable result considering.
If a crystal with bleed resistor is arranged from out to in of a hex-inverter you have one of it's function, an oscillator. If you look at the 4060 that's it's pins 10/11 ( triangle with circle at the point ). Pin 9 is a further inverter for driving RC oscillators. The next is a special inverter called a Schmidt. This is something like a mousetrap that has a very quick action. It has a special non linear switching that removes noise. As far as I can see it's the only one used. From the output waves from the following flip flops it does a good job. Each flip flop halves the frequency. Although a dinosaur it has a ton of circuitry for it's low price. Notice diodes used to protect the inputs. Even so take static precautions when touching them, Notice also pin 12 is a double inverter, This gives the correct logic to stop oscillation if pulled up.
I was asked if 4060 could do very low frequency square-waves. Yes they can and they can be days long.
I am trying to work out a good circuit to translate the 4060 square-waves into sine-waves. A 5 pole Chebyshev filter looks OK using TL074. The only problem is the distortion of the op amps could degraded the theoretical - 66dB ( -75 dB 1/3 harmonic including 1/3 F3 of -9.5dB ). The JLH which some describe as acceptable in distortion terms can beat that. What planet are people from who think 0.03% THD is high distortion. Not least when it have no switching elements. I guess they can hear 100 kHz without any effort.
https://m.eet.com/media/1131048/13232-60905di.pdf
https://www.intersil.com/content/dam/Intersil/documents/cd40/cd4060bms.pdf
Half way down this PDF is an idea I tried. I got the same results. I re-engineered it with the diabolical LM324N and had better results. 324 is OK if lightly loaded or forced into SE class. All the same this is a reasonable result considering.
If a crystal with bleed resistor is arranged from out to in of a hex-inverter you have one of it's function, an oscillator. If you look at the 4060 that's it's pins 10/11 ( triangle with circle at the point ). Pin 9 is a further inverter for driving RC oscillators. The next is a special inverter called a Schmidt. This is something like a mousetrap that has a very quick action. It has a special non linear switching that removes noise. As far as I can see it's the only one used. From the output waves from the following flip flops it does a good job. Each flip flop halves the frequency. Although a dinosaur it has a ton of circuitry for it's low price. Notice diodes used to protect the inputs. Even so take static precautions when touching them, Notice also pin 12 is a double inverter, This gives the correct logic to stop oscillation if pulled up.
I was asked if 4060 could do very low frequency square-waves. Yes they can and they can be days long.
I am trying to work out a good circuit to translate the 4060 square-waves into sine-waves. A 5 pole Chebyshev filter looks OK using TL074. The only problem is the distortion of the op amps could degraded the theoretical - 66dB ( -75 dB 1/3 harmonic including 1/3 F3 of -9.5dB ). The JLH which some describe as acceptable in distortion terms can beat that. What planet are people from who think 0.03% THD is high distortion. Not least when it have no switching elements. I guess they can hear 100 kHz without any effort.
https://m.eet.com/media/1131048/13232-60905di.pdf
https://www.intersil.com/content/dam/Intersil/documents/cd40/cd4060bms.pdf