Audio Signal Generator with Specific Requirements

[Frankentronics]

Greetings,

My desire is to build an audio signal generator for testing amplifier circuits. I did find quite a few circuits online, but I am unable to find a circuit that meets some of the requirements that I have in mind. So, I would like to know if I should keep looking or if my requirements are not technically possible or practical.

1 - I don't quite like the idea of having a switch that selects one range at a time. I would much more prefer to have a multi turn pot that could be dialed in from about 20Hz to about 20KHz.

2 - I don't like the idea of having a ganged pot so I am hoping that there's a way have a circuit that has a single gang pot.

3 - I would prefer if the sine wave was not a converted square wave or converted triangular wave. And I really like to idea of using a light bulb in the circuit.

So far I have not been able to find a schematic that meets these requirements.

Is what I am looking for technically possible, or practical.

Thanks...

[gbaddeley]

You probably can’t find a design that meets all your requirements because a practical and good performance design like this does not exist. Can you think why most (non synthesized / analogue) designs do use a ganged pot (or variable cap) and 4 switched ranges to cover 20-20K with reasonably low distortion?

[Manul]

For what kind of testing? One method for noise and distortion measurement is to subtract input from output. I guess in that case it does not matter so much, how good is the quality of input signal, but I'm not an expert about that.

[fcb]

ICL8038/MAX038 IC's were always the staple of homebrew 20-20K signal generators.  You can set these up to sweep the full band with one pot.  The sine-converter isn't half bad either.

[TimFox]

I am aware of two single-range/one knob R-C tuned analog audio generators from the good old days.
1.  HP 207A (vacuum tubes) 20 - 20kHz in one range.  See http://hparchive.com/Manuals/HP-207A-Manual.pdf
2.  General Radio 1313A (transistor), 10 - 50 kHz in one range.  See  https://www.ietlabs.com/pdf/Manuals/GR/1313%20A.pdf
(General Radio seemed to prefer beat-frequency generators for wide-single-range devices.)
You may find the manuals instructive, but good luck finding one of these very rare items for sale.
The wide-range R-C networks are extremely interesting in themselves.
If you literally want a single control, you might look at the G-R beat-frequency generators for ideas.

[Frankentronics]

Thank you all for your replies.

I actually breadboarded a few circuits that I found online but the results were never good. I did manage to produce sine waves, but there was always something not working right with each circuit.

For example, last one I built was this one:


http://www.next.gr/oscillators/sine-wave/Audio-sine-wave-generator-l12331.html

I did get a sine wave but when I adjusted the output level pot, R16, the sine wave would disappear towards both ends of the pot rotation. I got similar problems with other circuits and got frustrated.

At one point I would like to build a function generator and for that project I would like to use one of the usual choices of IC's but for the audio generator project I kind of like the idea of building a circuit that doesn't use a function generator IC. Using a light bulb kind of appeal to me, but I don't know why. I guess it feels more organic.

I recently built a dual adjustable power supply. It started as a circuit that would use a ganged pot and then I discovered that you can use a single pot that have an OP Amp track the pot setting for the other rail. I was wondering if that same trick can be used in a sine generator to eliminate the ganged pot.

I really like the vintage HP 207A and General Radio 1313A models that TimFox posted. I am a sucker for vintage equipment and I am trying to resist always buying vintage stuff. Now I can't get those out of my mind and I already checked eBay. But like TimFox said, I can see they are rare so they are not showing up in eBay history.

I guess I'll keep breadboarding circuits that I find online. Eventually something's going to work out.

Thank you all...

[gbaddeley]

Breadboarding until something works out? You could waste a lot of time and get nowhere. There are plenty of working wien bridge designs out there that use a bulb for amplitude stabilisation. Read the Wikipedia article on wien bridge.

[Frankentronics]

Breadboarding until something works out? You could waste a lot of time and get nowhere.

If I am interpreting your comment correctly, I think you are saying that some circuits should not be built on breadboards. Is that so?

There are plenty of working wien bridge designs out there that use a bulb for amplitude stabilisation. Read the Wikipedia article on wien bridge.

Along the same lines, I am interpreting your suggestion that I should just pick a circuit and build it straight on a PCB.

One thing that I find difficult with building any designs in general, that I find on the internet, is the fact that I often need to order specific IC's or transistors, before I test a circuit, and then if it doesn't work out I need to order more components. To avoid paying for multiple shipping I try to compile a few candidate circuits and order all at once, then wen the time comes to build I still seem to be missing something. And often for a specific circuit that I find there's one component that none of the sellers stock. But many times I need to order some components from Mouser, some from DigiKey and there's still one or two components left that I need to look for elsewhere. There must be a better way.

I've read a lot of online articles about these oscillators and I will definitely read the Wikipedia pages also.

I'm still wondering if it's possible to replace the ganged pot with a single pot and OP Apm tracking circuit.

[ArthurDent]

...If you literally want a single control, you might look at the G-R beat-frequency generators for ideas.

I believe this is the most practical idea. The General Radio units are ancient but the idea is sound and has been used in many applications. This link below describes how it works. A crystal oscillator could be used for the fixed oscillator if you're interested in greater accuracy. A high stability variable oscillator could be made if it is made without switching and to cover a wide range. For instance, a 2Mhz crystal oscillator and a variable 2Mhz to 2.020000Mhz oscillator could be manually or voltage controlled to sweep any portion of the range.
 
https://www.eeeguide.com/beat-frequency-oscillator/

[Frankentronics]

I was actually working on a circuit all morning and it's a good example of what I've been struggling with.

I thought I'd give it a shot at building the simple figure 4 circuit from this page...

https://www.nutsvolts.com/magazine/article/op-amp-cookbook-part-3



It looks simple enough and I have all the components. I tested the light bulb and it was within specs and I know the OP Amp works because I recently used it on a breadboard for another project.

At first I was getting a square wave output. The output amplitude adjustment pot was working in a way that the amplitude would be increased 100x (or so) from minimum to max. and I could not adjust the 1K RV2 to get a sine wave. The ganged frequency pot worked but it was also moving too extremely.

Then, after a while the square wave disappeared and I got a distorted wave on the scope. No matter how much I tried adjusting I could no longer get anything that resembled a square or sine wave.



I checked and triple checked all the connections and even swapped some parts. No luck.

That has been my experience trying to build a simple audio generator for the past 2 years. So frustrating I had to walk away from it all several times.

Am I just having some really bad luck?

[magic]

 

Use a soundcard if you really need high quality sine waves.

DIY if you want to DIY for DIY's sake or if you need 1ppm linearity (you probably don't).

[Cubdriver]

That output looks like a differentiated square wave, though why it's doing that I've no idea at a glance.

https://learnabout-electronics.org/ac_theory/filters84.php

-Pat

[Cubdriver]

What happens if you DC couple the scope input?  What frequency is that?  (Can't see what the time base is set to.)

-Pat

[Frankentronics]

Sorry, I already dismantled the circuit and started building figure 8 from that same page...

https://www.nutsvolts.com/magazine/article/op-amp-cookbook-part-3



I used the same 741 OP Amp and was able to get a sine but it's slightly distorted. Also, when I turn the output pot near either extreme the sine disappears.

I'm not really interested in that circuit because it doesn't have an adjustable range.

I guess what would be helpful if someone could post a link to an audio generator circuit that should not give me trouble, so that I can wrap up this project and start building amps and learning how to use the audio generator to work on amps.

I did look closely at the schematic of the General Radio 1313A unit https://www.ietlabs.com/pdf/Manuals/GR/1313%20A.pdf and I would be up for building it, but I think that it uses a FET that is no longer available. The parts list calls it a U-147 and I can't find it anywhere online. I can't even find a equivalent through the https://alltransistors.com/ web site.

If there is a way to build that circuit with some equivalent components that are still available today I'd be up for it. But I don't know how to go from here.

Thanks...

[bob91343]

That FET doesn't appear to be critical.  If you put in one rated for 40 V or more it probably will work.  The U-147 is only rated for 20V.  But try what you can find cheaply and it should be okay.

[Frankentronics]

That FET doesn't appear to be critical.  If you put in one rated for 40 V or more it probably will work.  The U-147 is only rated for 20V.  But try what you can find cheaply and it should be okay.

Looking through the parts list and trying to compile an order, there are quite a few other items that would have to be substituted.

Transistor 2N2188 (equivalents 2N2189, 2N2190. 2N2191) none I can find available.

Mica caps 267p, 453p, 634p, 649p, 787p.... I can't find.

Diode 1N4009 I also can't find and I am wondering why they used that diode and if 1N4007 might work just fine.

At this time I'm also wondering why they list two types of resistors, composition and film, and how important it is to use the exact ones.

Thanks...

[ArthurDent]

Diode 1N4009 I also can't find and I am wondering why they used that diode and if 1N4007 might work just fine.

1N4000-4007 are common rectifier diodes, 1N4009 is a fast switching diode. 1N4148/4448 might be a good choice.

At this time I'm also wondering why they list two types of resistors, composition and film, and how important it is to use the exact ones.

these old composition resistors had terrible aging specs. some would jump several hundred % over time. The film ones were stable and were required in the frequency generating circuitry. The reason ancient composition resistors are mentioned is I'm guessing the book you're using might be from about 1970.

[Frankentronics]

1N4000-4007 are common rectifier diodes, 1N4009 is a fast switching diode. 1N4148/4448 might be a good choice.

Thank you for this info...

...these old composition resistors had terrible aging specs. some would jump several hundred % over time. The film ones were stable and were required in the frequency generating circuitry. The reason ancient composition resistors are mentioned is I'm guessing the book you're using might be from about 1970.

I am using the PDF https://www.ietlabs.com/pdf/Manuals/GR/1313%20A.pdf, so it is outdated.

So, would I use any modern resistors or do I still need 2 different types?

Also, what do I do about the mica caps? The values they mention sound very specific and precise.

And what about that 2N2188 transistor? What would be a suitable replacement?

[TimFox]

At the time when the 1313A was designed, silver mica was the normal type of high-precision stable capacitor for critical applications, with Q > 1000.  Now, NP0/C0G ceramic (avoid X7R and Z5U for these applications) is typically used instead.  For larger values, polypropylene film capacitors are good, but physically larger, with higher Q than polyester film ("Mylar").   Also, molded carbon composition resistors were the common resistor for non-critical applications, but they have the problems discussed above.  You can certainly use metal-film to replace carbon resistors in audio-frequency applications, being careful about power rating.  In my own audio work, I use metal film for lower-power applications, and wirewound for higher powers.  The worry about self-inductance of wirewounds in audio applications is often overstated. 

[Vovk_Z]

Wide range oscillators often are made as voltage-controlled oscillators (VCO). Look at that way.

[Frankentronics]

At the time when the 1313A was designed, silver mica was the normal type of high-precision stable capacitor for critical applications, with Q > 1000.  Now, NP0/C0G ceramic (avoid X7R and Z5U for these applications) is typically used instead.  For larger values, polypropylene film capacitors are good, but physically larger, with higher Q than polyester film ("Mylar").   Also, molded carbon composition resistors were the common resistor for non-critical applications, but they have the problems discussed above.  You can certainly use metal-film to replace carbon resistors in audio-frequency applications, being careful about power rating.  In my own audio work, I use metal film for lower-power applications, and wirewound for higher powers.  The worry about self-inductance of wirewounds in audio applications is often overstated.

I am still having trouble finding capacitors of any kind in values 267p, 453p, 634p, 649p, 787p. Since these are not rounded off number I assume that it's important to use a cap of 267p instead of 265p, and so on. Or do these numbers represent old standard values and it's OK to use rounded off values? I know that if I have to recap an old radio it doesn't make much of a difference if I use a bigger cap. But I never touched mica or ceramic caps in old radios, so I don't know if those have to mach the factory values closely.

I would hate to spend a couple of hundred dollars on part and have my circuit behave oddly. For some reason I'm not having much luck building oscillator circuits.

[TimFox]

General Radio could order mica capacitors with close tolerances made to their specific values.
If you want to proceed along these lines, you need to obtain a good LCR meter (such as the DE5000, much discussed on this forum).  Purchase standard-value C0G/NP0 capacitors lower than the required value and measure them.  Also purchase lower-value units and carefully measure them.  To make 267 pF, for example, start with 220 or 240 pF, and add maybe 47 or 27 pF, depending on what you measure, in parallel.  For many such circuits, the match between two capacitors is more important than the value itself (e.g., shoot for 0.2% match between 1% values).  The cost of the extra capacitors will be much less than the DE5000, which is roughly $100.  To be authentic, you could stick with silver mica, but they are now expensive and rare compared with the C0G ceramics which are cheap and plentiful.
Old corollary to Murphy's Law:  self-starting oscillators won't.

[MarkF]

At one point I would like to build a function generator and for that project I would like to use one of the usual choices of IC's but for the audio generator project I kind of like the idea of building a circuit that doesn't use a function generator IC. Using a light bulb kind of appeal to me, but I don't know why. I guess it feels more organic.

When you get to building a Function Generator with a DDS chip, I can offer the circuit I did with a AD9834 DDS and PIC18F2550.  It has Sine, Triangle, Square, Sine Sweep Up & Down, and PWM.  Frequency range is 1Hz to 15MHz.

I overlaid a picture of the OLED display.

[Frankentronics]

Thank you again for the info.

Let's say that caps can be resolved in that way. That still leaves the JFET that 2N2188 transistor.


I can't find any 40V JFETs but I did find a 30V SMD JFET on DIgiKey, lsited as PMBFJ177,215. Mouser's filters don't filter by voltage, so I don't know how to search there. But I guess the PMBFJ177,215 should be good, right?

However, I still don't know what replacement to use for that 2N2188 transistor.

When you get to building a Function Generator with a DDS chip, I can offer the circuit I did with a AD9834 DDS and PIC18F2550.  It has Sine, Triangle, Square, Sine Sweep Up & Down, and PWM.  Frequency range is 1Hz to 15MHz.

That function generator looks good. Is this a kit that you sell?

[TimFox]

The 2N2188 was a Ge PNP.  Back in the day, PNPs tended to be Ge, while NPNs tended to be Si.  Probably, any general purpose Si PNP would work, like a 2N2907.  Look at the circuit conditions to verify ratings.
Similarly, what is the actual voltage across the FET?  There is only +32V applied to that part of the circuit, and the quiescent voltage across the FET is about 14V.  The parameter to compare for a substitute is Idss, the zero-bias drain current.  It must be higher than the actual drain current, so that the FET operates with reverse gate bias.  Note that the original device is a P-channel JFET.  This is left as an exercise for the reader.