I've been working on a Heathkit IG-72 vacuum tube signal generator for fun, and as a learning experience. So far I have had much success. I've cleaned up all the switch contacts, installed NOS replacement tubes, replaced the electrolytic capacitors and all the 5% resistors (besides those in the attenuator). The wave output is now clean and stable across the full frequency range. However, I am not able to achieve the specified distortion figure of "Less than 0.1% from 20-20,000 Hz".
Through experimentation, I have found that the oscillator control is very sensitive, and only produces a clean signal in a small fraction of it's 600 Ohm range, less than 100 Ohm. Minimum distortion, about ~0.75 THD, is only achieved at the very lowest portion of the range, about 10 Ohms. Any lower and the signal drops out entirely. Any higher and distortion begins to rise. At this low end of the range, output voltage is reduced to about 1/10th of the expected ~3V.
This leads me to believe something is wrong with the circuit. I wouldn't expect the oscillator control pot to only provide clean output in such a small range, at the very least. At this point, I'm not sure where to turn.
I am getting what I take to be transformer whine at higher voltages. It varies in frequency/pitch as I adjust the controls. I think it's coming from the smaller of the two transformers.
Here is the manual: https://www.manualslib.com/manual/1854974/Daystrom-Heathkit-Ig-72.html
I did my distortion measurements using Room EQ Wizard through a Focusrite Scarlet.
Through experimentation, I have found that the oscillator control is very sensitive, and only produces a clean signal in a small fraction of it's 600 Ohm range, less than 100 Ohm. Minimum distortion, about ~0.75 THD, is only achieved at the very lowest portion of the range, about 10 Ohms. Any lower and the signal drops out entirely. Any higher and distortion begins to rise. At this low end of the range, output voltage is reduced to about 1/10th of the expected ~3V.
This leads me to believe something is wrong with the circuit. I wouldn't expect the oscillator control pot to only provide clean output in such a small range, at the very least. At this point, I'm not sure where to turn.
I am getting what I take to be transformer whine at higher voltages. It varies in frequency/pitch as I adjust the controls. I think it's coming from the smaller of the two transformers.
Here is the manual: https://www.manualslib.com/manual/1854974/Daystrom-Heathkit-Ig-72.html
I did my distortion measurements using Room EQ Wizard through a Focusrite Scarlet.
Last edited:
Ooh, good suggestion. I figured it was fine since it lit and dimmed properly.
I should note, I'm getting some whine (I assume transformer) at higher voltages. I think it's coming from the smaller of the two transformers.
I was going to suggest doing an analysis with Spice, but rayma reminded me the stability is done by the lamp. You may not find one identical enough so if it is good, don't mess with it. They never conducted enough to get hot and out-gas the filament.
Are you sure you are driving a load within the specs? That may have a big influence.
Fun to play with, but by todays standards of no real use in the lab. I just dumbed my HP generator and tossed my Heath years ago. A PC can be orders of magnitude cleaner.
It might be fun to build a modern op-amp based generator in the old Heath case. You should be able to get .001 pretty easy. .00001 gets harder, but is do-able.
Might I ask why you replaced the resistors?
Are you sure you are driving a load within the specs? That may have a big influence.
Fun to play with, but by todays standards of no real use in the lab. I just dumbed my HP generator and tossed my Heath years ago. A PC can be orders of magnitude cleaner.
It might be fun to build a modern op-amp based generator in the old Heath case. You should be able to get .001 pretty easy. .00001 gets harder, but is do-able.
Might I ask why you replaced the resistors?
I'll try sourcing a new lamp and see how that works out. I saw another discussion with people saying the lamp going bad could prevent the unit from reaching specified distortion levels.
My scope has an input impedance of 1M and my Scarlet has an input impedance of 60K, I think those both should be fine.
Yeah, my laptop can output a cleaner signal than this. Like I said, I'm doing this for the experience, and because, well, I don't work in a lab.
I replaced the resistors because I knew they were out of spec (after measuring a couple), and I was trying to fix issues with frequency selections being unstable.
My scope has an input impedance of 1M and my Scarlet has an input impedance of 60K, I think those both should be fine.
Yeah, my laptop can output a cleaner signal than this. Like I said, I'm doing this for the experience, and because, well, I don't work in a lab.
I replaced the resistors because I knew they were out of spec (after measuring a couple), and I was trying to fix issues with frequency selections being unstable.
That lamp must NOT light-up, ever. It is a 120V bulb and sees like 10V; incandescent light output falls-off real fast at low voltage.
Are your DC voltages reasonable?
You have replaced resistors? How about capacitors? That screen cap begs investigation or outright replacement. If it is bad you can have "good" DC voltages and terrible AC/Audio gain.
And yes there are several/many WRONG lamps which fit the socket. Any "spare" lamps in the shipping materials? These lamps used for stabilizers don't really go bad. H-Ps as old as yours may have tarnished lamp and socket, dud capacitor, even cracked tube heater (was a metal tube so took a while for the penny to drop), but not bad bulbs.
I’ve replaced the electrolytic capacitors. I did not replace the film/ceramic caps in the multiplier switch because they all tested in-spec. I also replaced the tubes.
If I turn the oscillator control above ~100 Ohm I get voltages that are higher than spec. It is these times that the bulb lights.
I am able to operate the unit with voltages at normal levels if I adjust the oscillator control between 10 and 100 Ohm. With that setting, the light does not illuminate.
If I turn the oscillator control above ~100 Ohm I get voltages that are higher than spec. It is these times that the bulb lights.
I am able to operate the unit with voltages at normal levels if I adjust the oscillator control between 10 and 100 Ohm. With that setting, the light does not illuminate.
The lamp should be very dim to not visible in normal operation. The higher you push the oscillator level pot, the more distortion you'll generate. Scrutiny of the screen cap is a good idea, as PRR suggests. You want all the "open-loop" gain the tube can provide.
The meter circuit is a non-linear network that may also contribute distortion; you could try turning the calibration pot to 0 to see if that helps distortion.
I believe you'll see some DC bias present at the top of the output level control. I believe this is undesirable, as it may allow DC loading at the output terminals to subtly change bias on the oscillator tube. (A proper generator shouldn't be providing any varying DC in its output, anyway.) I think an interesting experiment would insert an electrolytic cap at the top of the output level control to block DC from the oscillator.
P.S. At present without the proposed blocking cap, there is at least some DC voltage imposed across the lamp. This also not good for distortion performance.
Cheers!
The meter circuit is a non-linear network that may also contribute distortion; you could try turning the calibration pot to 0 to see if that helps distortion.
I believe you'll see some DC bias present at the top of the output level control. I believe this is undesirable, as it may allow DC loading at the output terminals to subtly change bias on the oscillator tube. (A proper generator shouldn't be providing any varying DC in its output, anyway.) I think an interesting experiment would insert an electrolytic cap at the top of the output level control to block DC from the oscillator.
P.S. At present without the proposed blocking cap, there is at least some DC voltage imposed across the lamp. This also not good for distortion performance.
Cheers!
Last edited:
I've done some more testing with your thoughts I mind. I observe that the lamp is illuminating at approximately 0.8v on the 10x multiplier (0.3v on the 1x and 100x positions, 0.05v on the 1000x position).
I checked, and there was a ~0.100-0.140VDC bias present. I added a capacitor as you said, and while it eliminated the DC, it caused the THD to double, from 0.75% to 1.5%. It also changed the output signal frequency (I used a 6.8uF cap)
Do you have a few 6AU6 and 6CL6 tubes to tube roll (one valve at a time) and then use REW to measure the different harmonic level changes? Make sure you aren't overloading the input to the Focusrite by changing the generator voltage selection and confirming no change in the HD% levels.
There's far less bias than I'd guessed. Probably not worth the trouble. The cap would need to be large. You want to ensure there is very little AC voltage present across the cap, as electrolytic's are distortion generators.
A revisit: if we're on the same sheet of music, you're observing about 120mV DC bias at the top of the Output Level control. From the schematic, this should be about the same voltage at the grid of the 6AU6, pin 1. But this is at odds with the Table of nominal voltages on page 17 of the manual. Maybe the tube is weak? Would you tabulate/report voltages actually seen in your unit, a la the table?
Sure, I can record the voltages I'm observing. The tubes are all fresh NOS tubes. I will measure and report back.
trobbins: I do not have any extra tubes. I threw out the originals when I swapped in the NOS tubes, and they were the only others of this type I had.
trobbins: I do not have any extra tubes. I threw out the originals when I swapped in the NOS tubes, and they were the only others of this type I had.
Last edited:
If curious about 120mV stray DC, leave the thing on for a week or more. Long-stored tubes need to re-getter their gas.
But I doubt 120mV DC matters here. (It does in VTVMs.)
I acknowledge ignorance about re-gettering phenomena.
If the tube recovers lost moxie, I'd expect the plate-cathode current to rise to values listed in Heath's table of nominal values. The reported bias voltage is only ~10% of what's indicated as nominal. A weak tube might explain OP's complaint of reluctance to oscillate and perhaps the disappointing distortion.
If the tube recovers lost moxie, I'd expect the plate-cathode current to rise to values listed in Heath's table of nominal values. The reported bias voltage is only ~10% of what's indicated as nominal. A weak tube might explain OP's complaint of reluctance to oscillate and perhaps the disappointing distortion.
The sensitivity of the output and the small range in which it produces a clean sine wave existed with the old tubes as well. I did not take distortion measurements with the old tubes. It will produce a clean sine wave at 3V output, however distortion is >=10% at this level.