Heath IG-18 IG-5218

I had a chance to go back into my IG-18 to replace the original 500 Ohm stability pot on the Morrey board which was a single-turn open track carbon comp unit. It had always been very drifty. The 500 Ohm 15 turn cermet that replaced it allowed me to get about 10 dB reduction in second and third and much faster settling time due to better adjustment capability. Stability drift seems to have disappeared.

I also compared the physical construction of the lamps. The Eiko 90MB has a slightly overall shorter filament construction than the original Syl 90V which has longer loops. Having said that I don't think there's a closer replacement.

While adjusting the aforementioned trimmer I did disassemble the Eiko from its base and paralleled it to the Sylvania. Combined, the lamps had the expected shorter time constant and amplitude stability seem to converge faster. Unfortunately the overall output was reduced and I eventually removed the Eiko.

It makes me wonder if two lamps in parallel, each having a slightly different thermal mass but similar current voltage/current characteristics might provide a dual time-constant leveler. I've seen his done with CdS cells in compressor/limiters.
 
@mediatechnology -- I don't think you can get enough spread in thermal mass to make it work well, unless you use an amp (emitter follower?) to drive each one separately so that sufficient current is available to get them going a little. If the driving amps have gain, you'd be able to play with the gain through each branch -- a lot of work, I think, but maybe fruitful....

@thaumaturge -- I ended up with a total of a bit over $3K in mine after replacing some parts and having Agilent cal it. No tchump change, but I just never think very much about the meter anymore -- I just take the readings at face value and move on. Nice.
 
@thaumaturge -- I ended up with a total of a bit over $3K in mine after replacing some parts and having Agilent cal it. No tchump change, but I just never think very much about the meter anymore -- I just take the readings at face value and move on. Nice.

Don't want to threadjack, maybe we could start a 3458 thread... But briefly, I repaired maybe half a dozen of them. Generally replacing broken input mounts on the lame plastic front panel. But one had a real problem with the RMS converter. It always irked me that Agilent would wrap such a lame package around such a superb piece of technology.
Doc
 
I've had a couple on the bench and have seen a couple more in other places, and have not seen problems with the panel assembly. Mine needed a new display board and a new A5 ROM board (with the new snap-on backup batteries) -- the old board had the Dallas ROMs with real old batteries and they were fading fast. Mine is overdue for a cal, but that won't happen until next year some time.
 
DO NOT hard bump your terminal cluster. That entire assembly is held to the front panel by one screw anchored into a little plastic disk that shatters. Can be repaired of course, just a PITA. As to your cal... you can get away with that at 4ppm a year DC drift. It's the 100ppm absolute on the AC from 10hz to 10Mhz that I find the most mind boggling.
Doc
 
I have finally done some work on my IG-5218. I replaced all of the capacitors and measured the distortion with my uncalibrated HP 332A. I measured 0.06% for the unmodified IG-5218. I installed the meter buffer from Williamson's Greening article. I adjusted the bias for lowest distortion using my distortion meter. The distortion is down to 0.015% which is quite close to the lowest my distortion meter can measure.

Jim
 
I think your IG-5218 is doing quite well - fixing the meter with a buffer is almost the most important mod that you can do easily. You will need a better analyzer to see any further real improvement from anything else you might choose to try.

I would say that #2 in importance is fixing the ground runs, which pretty much entirely eliminates spiking from the square-wave generator.

Good luck with further work.
 
Thanks

I am hoping to improve it a bit more, but your right my HP 332 is not quite up to the task of measuring this. I was looking for an excuse to put the Active Twin-T notch filter on my list of things to do. I did notice that getting the distortion down was very dependant on the bias setting, so after fixing the grounding scheme and building the Active Twin-T notch filter, I will look more closely at the differential pair.

Jim
 
Is it possible to get really low numbers when we are dealing with caps and switches? Would one not want to follow the oscillator with narrow bandpass? How about a more modern VCO/AGC?

Need to dust my old one off. I normally use my PC as it is lower distortion, but twitting knobs has its place too. I also have an old HP 409 because nothing beets a bid dial to zero in on a room resonance.
 
@tvrgeek --- The Tek SG505 uses switches, as does the KH 4402B, etc. At harmonic product levels below (or well below) -120dBu (not even talking noise), you might also ask if wire and solder affect things... and humidity and barometric pressure -- I am serious. Nothing is above a careful look; but some things just take a LOT of time, and time is precious, so a good whack from the intuition genie is worth a lot more than I used to think.

Williams' famous 10kHz oscillator has the huge advantage of being single-frequency. If you're willing to live with that constraint, many things get easier to control.

Anything following the oscillator stage has to be incredibly low in both distortion and noise; I've seen passive LC bandpass circuits that actually made thing much worse, and some that did actually improve the signal, but that was back in the day before I got interested in really low distortions below 1 part per million, so I can't generalize about using either active or passive bandpasses to improve distortion at really low levels.
 
See what Steve Lafferty has done with putting the very fine HP339A oscillator into the IG-18/5218: Modifications: Heathkit IG-18 Audio Generator

For performance results, see my build here: http://www.moorepage.net/IG-18-3.html

I think you could easily lay out a PCB from his work, or just use perfboard and wire as he did.
This post by Dick Moore was dated 2013. Sadly, Dick passed away in 2015. However, in 2016, his wife Robin graciously allowed Tronola (Stephen H. Lafferty) to preserve his website, Moorepage, where you can find more information.
 
Well, maybe we could organize a group buy.
Say re-engineer the board, to reflect the current Opamps that need less components on the board (as per Dick’s experiments) and adding a more modern power supply section and have it made. I am not competent enough to design the board in one of the CADs like Eagle, but I could manage logistics, have a number made and then ship them out to the group. Since it’s not a huge amount of money I can advance it. I am located inItaly and can handle easily all EU shipments.
Besides, I am a bit lost in all versions of the Tronola hosted IG18 improvements and maybe this way a definite choice can be made!
What do you guys think?