Re-greening the Heathkit IG-18

[jackinnj]

Originally Posted by PB2

I've mentioned several times that this unit is very touchy. There are large disturbances in the output if I tap the circuit board and especially the lamp.

That's to be expected, let's say a tap to the case moves the potentiometer a tiny bit -- the frequency changes and the whole loop has to restabilize itself. A tap to the lamp changes it's impedance momentarily...

[PB2]

Yes, I see your point, and the gain control loop has a long time constant so that contributes to it taking a long time to settle. The feedback pot seems to have too much range and is also very sensitive. I'll look into reducing the range.

[PB2]

Anyone know of where I can find a copy of Bill Hewlett's Masters thesis online?
Here is his patent:
VARIABLE FREQUENCY OSCILLATION - Google Patents

It seems that the lamp requires DC bias in order to provide a stable operating point. Heath provided this through the DC offset via R5. Removing R5 in the simulation above with the lamp shows the issue. Wondering if there is a better lamp, probably better to just move to FET control as in the 239A.

[PB2]

I have found that various improvements often result in either HF oscillation, or squegging (motorboating) and it seems that the DC bias on the lamp helps to avoid the LF issue. I don't have a lot of experience using lamps in oscillator circuits but it seems that there are conflicting requirements. First, the lamp is required to maintain a gain of one around the loop in order to support oscillation. A short thermal time constant is required in order for it to reach thermal equilibrium in a reasonable amount of time. However, at LF it is possible for the thermal time constant to be so small that the temp varies between the peaks and zero crossings of the waveform leading to increased distortion. The resistance should very slowly adjust to maintain the gain of one as the amp gain changes due to variations in temp and Vcc over time but not (or minimally) over a single period of oscillation. This is why 4 bulbs are used for low distortion at LF as can be seen in Figure 40 of LT's AN43 and one bulb for mid and HF:
www.linear.com/docs/4134

I've not seen a detailed analysis of how the bulb should be biased for proper control in a low distortion oscillator. One reference suggested that it should just dimly glow. This makes sense since if the bulb runs hot then, with the large temperature differential to ambient, it will tend to cool much more at LF during the zero crossings.

I'd like to see a detailed analysis of the lamp biasing. It is non-linear and therefore not easy to solve.

[PB2]

I happen to have an HP 202C that I've not used in probably 15 years that I pulled out to take a look at. Such nice build quality! I did a google search and came across this HP Journal issue from 1956, that discusses the new balanced oscillator designs and includes some analysis showing that DC on the lamp is not a good thing - increases 2nd harmonics:
http://www.hpl.hp.com/hpjournal/pdfs...Fs/1956-02.pdf

Also discusses a zero output impedance tube amp stage employing positive feedback.

Found schematics for the 200C that is very similar to this unit and they use two 250V lamps in series.

[PB2]

The HP 202C looked so clean that I decided to power it up and believe it or not it works. Measured 21 V RMS on the two series connected 250V bulbs. Also measured about 4% THD at 1 kHz, looks like there are some sidebands that are probably due to hum in the signal. 60 Hz is about 80 dB down. There are not a lot of caps in this design, looks like all the electrolytics are in the power supply.

This is off topic, but I wanted to see where the lamps were biased relative to their rated voltage.

[jackinnj]

There's a Tektronix SG-505 "Parts" unit available on EBAY -- shouldn't be too difficult to fix up -- I have no affiliation with the seller: Tektronix SG-505 ULTRA Low Distortion Audio Oscillator | eBay

[richiem]

Thanks to a little prodding from a non-working GR 1562-A SLM calibrator, I needed to think about lamp characteristics more concretely. David did a nice job on the curve plotting for the 90MB, and I was led to think about renormalizing the curves.

My little page on lamps generalizes to basically every incandescent lamp and will help (I hope) folks to understand where in it's VI curve a lamp should best be used in an oscillator -- and it is not at the point where it is glowing, dimly or not. See here:

Lamps for stabilizing oscillators

[PB2]

Originally Posted by richiem

Thanks to a little prodding from a non-working GR 1562-A SLM calibrator, I needed to think about lamp characteristics more concretely. David did a nice job on the curve plotting for the 90MB, and I was led to think about renormalizing the curves.

My little page on lamps generalizes to basically every incandescent lamp and will help (I hope) folks to understand where in it's VI curve a lamp should best be used in an oscillator -- and it is not at the point where it is glowing, dimly or not. See here:

Lamps for stabilizing oscillators

Once again, nice work Dick,
Not sure were I read that about dimly glowing but I can see from your data that it is incorrect. I've been swamped with other work lately and I did not want to spend much time on the IG-18 anyway so not sure if I'll get back to it or not.

I'll probably build something from scratch, either the HP-239 or Bob C's state variable. Leaning toward the state variable given that it is also used in the early Audio Precision system. I'm also more interested in something with computer control for automated tests:
Audio Test Equipment - Computer Control

[mafalda]

There is a modelization of a lamp in the Intusoft Icap spice too.