Audio Generator

[TomCarlos]

I am using an Elenco GF8046 Function Generator.



It's ok... but I think I'd like to have a clean "Audio Generator" - sine wave only with low THD. Anyone have an recommendations for a unit under $150?

[km6xz]

That is a function generator so it will not be low distortion, that is not its design purpose. I have yet to find a better signal generator for a repair shop than the older Heath IG-5218. It is low distortion, accurate calibrated metered output, accurate decade frequency control, and an accurate step attenuator. I have had all types of lab generators and here, have two ultra-low distortion Sound Technology generators and a digital version but still use the IG-5218 as the default generator. I have had all the HP, GenRad, Tektronix, and even an Audio Precision System 2 dual domain but still sought out the Heath as the most practical generator for bench work since it has all the parameters of operation directly read off the front panel at a glance without connectioning any external devices: set a frequency, jump 1/10 or 10x with one switch position, yet can be directly set to a specific frequency with no frequency counter needed. The amplitude output is accurate on the db and voltage scales of the meter and the step attenuator drops or increases the level 10 db per click over a 70 db range plus a 0-10 ratio variable attenuator. Distortion is not high end hi-fi but is 0.016%Thd but can be improved with simple mods. It has high enough output to test speakers directly.
They go for $50-120 and worth every cent . I got most of mine at ham swap meets or the monthly test equipment swaps and flea-markets that were in the San Francisco Bay Area, for less than that but eBay has driven prices up and quality down.

[Weez]

+1 for the 5218. I have one, paid $100 on the 'bay a little over a year ago. Works like a charm, simple to operate, very accurate.

There are 5 listed right now, from 70 to 200 bucks. Here's the cheapest one (has some cosmetic issues, but most will, that I've seen):

Heathkit IG 5218 Sine Square Audio Signal Generator Vintage Test Equipment Works | eBay

[TomCarlos]

Thanks km6xz ....

I will check on the Heathkit once again. I remember seeing some articles on mods that help clean up the signal, etc. The Elenco generator has a loud fan and would benefit from a step attenuator. So I want to replace it.

Tom

[g1]

If you end up with the IG 5218 and are interested in performance mods: Enter the Six Characters to Begin Download
also, google "greening of the IG-18" (and re-greening for even further updates)

[km6xz]

The one on my bench have very few mods, really does not need any unless testing performance on higher end hi-fi gear. I added a active current source and my dropped from 0.016% Thd at 1khz to 0.0074%, all 2nd harmonic. The square wave converter stopped working a year ago but never bothered to open it since I never need square waves. I have an arb generator if I need any particular waveform. I have two Sound Technology 1700B's on the bench which are better but less convenient for one handed operation, they both read 0.00018% at 1khz which is excellent considering that is two way, the combination of the generator and analyzer residual.
It is surprising how accurate the frequency and level meter are.
If someone needs a very clean waveform, and have a 24 bit sound card, the free Visual Analyzer will give a lot of functions for the price....free. Visual Analyser details Look for the Beta, 2012 that has a mid 2013 update.
The only problem with using a sound card is the requirement to create an input attenuator to assure that the input is not overdriver. A simple resistor divider would do the job so the output of an amp is brought down to +/- 5volts. A 24 bit, 96k sound card is pretty cheap now and offers really good generator specs and analyzer DR. A 16 bit 44.1k card would do well enough for guitar amps.

[g1]

One question for km6xz, do you ever miss having a dial sweep, rather than a stepped frequency control? I know a dial sweep can be handy for finding cabinet rattles etc. that are frequency specific.

[Enzo]

Your function generator is good for that.

[km6xz]

One question for km6xz, do you ever miss having a dial sweep, rather than a stepped frequency control? I know a dial sweep can be handy for finding cabinet rattles etc. that are frequency specific.

No, sweep is not as much help as one would think since decade switches an get very close to peak or dip as the case may be, in course. For example, say a cab is being swept to use your case. You know it is a low or moderately low frequency you are interesting to detecting. Put the multiplier range to X10 and rotate the single most significant figure dial, which steps of 10 from 0 to 100, any resonance or frequency dependent artifact will be excited by close enough from that 100hz to 1khz sweep. If not found, select the X1 range and sweep from 10hz to 100hz. Once you find a peak or dip hone in on the exact frequency with the 0-9 hz dial. There is an additional 0-.9 hz variable dial. It takes far less time to do it than describe it. The advantage of the decade switches include high stability, high accuracy and high resetability. If you are tuning a guitar, and need a specific reference to a fraction of a Hz, just dial it in. It will be there each time. If you have another other need for a specific frequency, you do not have to use a frequency counter to figure out what frequency it actually generating. I had several HP and a GenRad variable generators in storage when I moved here but my first choice was the Heath for these reasons. Variable generators have a different harmonic content from one end of the dial to the other, a reason high precision generators like those in my Sound Technology 1700B's are decade frequency control also by using rows of push-buttons, still effective but not as fast to zero in on one unknown frequency as in the case you used of speaker cab resonance. I also have a digital analyzer by Sound Technology here but it is not as clean as the analog (0.003% Thd ) and harder to zero in on an unknown.
If there are a number of unknowns or where a real sweep is needed, like when trying to see the curves of a variable tone circuit, quickly, a spectrum analyzer with built-in tracking generator is the key. I have a HP3580A here that is my general purpose audio range spectrum analyzer that is used in parallel to the scope every time I look at a waveform to get more complete visual indication of the nature of the signal being observed. It has a tracking generator that has enough output to drive a speaker directly. Just put the scope/analyzer input probe on the speaker terminals or across a small series resistor, and sweep the entire audio range and see all the peaks and dips with direct readout of frequency and amplitude in 1/10 second or 1 second if you want more precision. You can do the exact same thing without expensive hardware like above, just use your computer's sound card and the program I gave the link to. It will be even more accurate in frequency, plus will give distortion and SNR sweep plots also.
So the short answer is, no, variable generators have more drawbacks than benefits compared to a IG-5218

[bkahuna]

If someone needs a very clean waveform, and have a 24 bit sound card, the free Visual Analyzer will give a lot of functions for the price....free. Visual Analyser details Look for the Beta, 2012 that has a mid 2013 update.

Along those same lines, I use Timo Esser's Test Tone Generator. Test Tone Generator Software for Windows
I bought it years ago, back when the only freebie generators around sounded like crap. I still like it better than most of the alternatives.

Another good freebie generator is part of Room EQ Wizard. REW Forum at Home Theater Forum and Systems - HomeTheaterShack.com

This is a pretty amazing tool for free. It's designed for Home Theater buffs, but can be used for just about any type of acoustic measurement.

[TomCarlos]

A friend let me borrow an Instek GAG-810 Audio Generator. Simple unit, seems to be a clean signal. Even though it has an attenuator, I might need to built another external attenuator box. I'm afraid of putting in a signal that is to strong. I'm not sure what would be a safe input level for most amps.

[Tom Phillips]

...I'm not sure what would be a safe input level for most amps.

50mV to 100mVrms is a good range to use for most guitar amp testing. More could still be "safe" but more than that makes it harder to set up the controls for testing and will more easily overload the front end stages of the amp.

[km6xz]

When testing an amp when it first gets to the bench, I use a few standard values so it is easy to see if there is an unexpected gain or distortion. I normally set the generator output to -30dbv, which is not too high for a high gain amp to see a relatively undistorted signal out output with preamp and master gain in the middle range, all pots are set mid point and channel set to clean. It is more revealing to work initially with the clean channel. If I am expecting an ultra high gain, -40dbv is a good starting point. If it is not clean there, it will not get any better with other control positions.
Last night a client brought 6 tube amps to be checked and he needed at least 3 for tomorrow. The all were tested while sitting on the heavy cart I use for moving them around. Three had high distortion in clean, and the other 3 were noisy or channel switching did not work. By having a reproducible test condition, it was an instantaneous diagnosis. The key to accurate fast diagnosis is being able to isolate data and be able to compare standard test conditions to experience or calculated expectations. The first amp plugged in showed 150ma AC lower current from the mains than expected in standby. That would explain the distortion that was expected when taken out of standby.....a power tube not heating. The second amp had expected wall mains, but sure had clean channel distortion, current limited on one have of the wave form, with high even harmonics: another tube. Third pulled expected mains current in standby but too low off standby, sure enough, bias putting tubes in class B-C...predicted open low side of bias pot. These 3 were finished in 30 minutes so he could take them with him before he finished his tea. If I did not have highly repeatable conditions, each would have taken longer to diagnose.
That is why I suggest having test gear that is more informative and stable, and setting it up to take a sampling of a great deal of simultaneous readings. A visual scan of the spectrum, mains current, mains voltage, time domain display(scope), input level etc, tells a great deal before even picking up a screwdriver. This is why I recommend the IG-5218 as the perfect pro audio test generator.