Hi all,
I know this is a topic discussed over and over here and in many other places, but I seem to experience some specific issues which I am not sure how to handle.
Here's what I've done to the IG-18 (based on all of the literature I could find out there):
Radu.
I know this is a topic discussed over and over here and in many other places, but I seem to experience some specific issues which I am not sure how to handle.
Here's what I've done to the IG-18 (based on all of the literature I could find out there):
- PS upgraded to using LM317T
- PS caps replaced with Nichi/PW 1000uF (all other PS caps upgraded also)
- meter buffer based on 2N3904 per Dick Moore
- R12 replaced with cascode JFET CCS (PN4393/2N5486) adjusted for 3.5mA
- placed a switch on/off for the square wave in the back of the chassis
- R3 = 100k
- C5 = 100uF
- 220uF cap over the meter directly on its legs
- tried cascode JFET CCS (J202/2N5486) for R4 (adjusted for about 1.1mA), but experienced instability (the sine is pulsating); I assume some caps strategically placed would cure this, but I'm not sure how to do it
- I just acquired my first distortion analyzer - an HP 334a - but I don't seem to be under 0.6% distortion with this after the mods... (to be noted, before any work on this the sine was pulsating like nuts which I assumed was due to issues with the PS. A check of the PS caps showed they were in surprising good shape, though). This is magnitudes over where I thought I am, obviously. The HP is either mis-calibrated (though I'm learning this is unlikely), and though I measured a signal generator app I have on my iPhone and got 0.3% distortion which sounds pretty reasonable;
- I could see the waveform cleaning up on my scope as I was doing the mods steps... (though the square wave has never been pretty). But there may be some recent degradation of the lower leg of the sine which may be giving these high distortion results... I will try to get pictures of the scope depicting this and post here;
- When I started the distortion measurements I inadvertently shorted the output of the generator (repeatedly, as "the sleep of reason produces monsters"). The output went back every time to the same AC voltage, but I am thinking that maybe I pushed the output transistor pair over their jagged edge... If yes, I was planning to get a pair of MPSA05/55 in there as replacements.
- Switching off the square wave stage degrades the sine wave noticeably (!). This is extremely strange as all sources seem to indicate this should dramatically clean it up. I suspect a thorough readjustment is needed to get the improvements in line but I am pretty thrown off by this. What gives?
- Could anyone describe in plain language how the grounding mods are supposed to be performed? I couldn't understand this from any of the articles I read. I guess a quick description of how to check whether the frequency adjustment vernier zeros out would also be nice...
Radu.
Lamp should not be lit -- if it is, the feedback pot is misadjusted. You don't say how you are turning the square-wave circuit off. If you are breaking the supply line, then some other part of the circxuitry could be cut off as well -- just check the routing. The square wave section ground trace should be cut away from the sine-wave section, with a separate ground line going to chassis ground. The sine wave ground board trace(s) should go to the chassis ground of the 0.047uF cap which is also the ground for the sine-wave output jack. Hope this help.
The "motorboating" of the sine output indicates misadjustment of the feedback pot too, plus possible misadjustment of the bias pot. When amplitude stability is poor and the lamp is dimly lit, there is not sufficient resistance change in the lamp to control the output. The capacitors in the bridged-T filter circuit should be checked with a capacitance meter -- absolute accuracy is not as important as matching, so the cap meter doesn't have to be expensive. I got a micro-based one from China for under $30 that is also a transistor/diode/fet tester -- very handy.
The "motorboating" of the sine output indicates misadjustment of the feedback pot too, plus possible misadjustment of the bias pot. When amplitude stability is poor and the lamp is dimly lit, there is not sufficient resistance change in the lamp to control the output. The capacitors in the bridged-T filter circuit should be checked with a capacitance meter -- absolute accuracy is not as important as matching, so the cap meter doesn't have to be expensive. I got a micro-based one from China for under $30 that is also a transistor/diode/fet tester -- very handy.
Dick - thank you very much for pitching in!
I didn't yet get a chance, but I am planning to:
1. check all voltages again. They were off when I first fired this up, but after doing the mods they all fell nicely into place. I'm curious to see if things are still in whack now that I have reasons to believe distortion is way off.
2. as you suggest, re-adjust everything per manual.
An answer - (pictures illustrate this) as B+ for the trigger schmitt and the sine part are neatly separated in one trace, I just carved open the trace and inserted a switch for the trigger schmitt. The ground trace, though, is way more convoluted, which is why I was unable to sort out the ground 'fixing' easily.
Also as promised, I enclose some pictures of my work. Maybe some of my decisions are provoking the issues. I enclose both the initial (dual) CCS arrangement [third picture] - which I thought, due to too long wiring, may trigger the issues, and so I dumped the R4 CCS and shortened the R12 CCS legs to a minimum [fourth picture].
Thank you again,
Radu.
I didn't yet get a chance, but I am planning to:
1. check all voltages again. They were off when I first fired this up, but after doing the mods they all fell nicely into place. I'm curious to see if things are still in whack now that I have reasons to believe distortion is way off.
2. as you suggest, re-adjust everything per manual.
An answer - (pictures illustrate this) as B+ for the trigger schmitt and the sine part are neatly separated in one trace, I just carved open the trace and inserted a switch for the trigger schmitt. The ground trace, though, is way more convoluted, which is why I was unable to sort out the ground 'fixing' easily.
Also as promised, I enclose some pictures of my work. Maybe some of my decisions are provoking the issues. I enclose both the initial (dual) CCS arrangement [third picture] - which I thought, due to too long wiring, may trigger the issues, and so I dumped the R4 CCS and shortened the R12 CCS legs to a minimum [fourth picture].
Thank you again,
Radu.
Attachments
Last edited:
You're going to have to figure out how to separate the sine and square grounds. It is a pain, but it can be done. I would be careful about trying to get too much gain in the diff pair stage and keep wiring for that stage as short as possible. Most of the gain of this amp, like most diff input, three stage amps, comes from the second stage, but trying to get overall gain too high compromises the total gain bandwidth, which leads to instability. Single stage CCSs work just fine -- compounding is not needed or desirable. In order to keep the oscillator working at 100kHz or a bit more, the bandwidth of the 2nd stage is crucial -- trying for too much gain there just makes a mess; but the tradeoff is higher ditortion. I would also be careful that you not try to get too much supply voltage out of the regulator circuit -- make sure it has plenty of headroom -- say at least 4-5 volts, maybe more depending on total current draw under load and the regulator itself. Most folks will not need the full 10VRMS output from the oscillator, and lowering the supply output will reduce output swing as well, but overall performance will improve.
Last edited:
I will be getting back to this tonight, and the first thing I want to go to is fixing the grounding. So here's what my plan seems to be:
1. Cut the traces in such a way that I isolate the sine from the square. This may be a bit easier than I initially thought, actually. Not a total pain.
2. As it stands right now, the PS PCB runs a ground wire (BLK) to point C on the PCB, which is now the square ground trace. Leave it. Make sure square jack ground remains connected to square PCB ground;
3. Run a wire from sine PCB ground to sine output jack ground (if not connected already... sorry, I'm at work and don't have this in front of me, just the manuals and my notes, but want to have the plan clarified for tonight). I think this ground is currently crossing over to square PCB ground.
(Q: does this mean that the sine ground is run now through the chassis to provide power to the sine oscillator? In other words, I'm unsure how the sine oscillator is being grounded to be power supplied. I seem to be missing one required connection...)
Is this a good summary per your recollection of steps?
Thanks!
Radu.
1. Cut the traces in such a way that I isolate the sine from the square. This may be a bit easier than I initially thought, actually. Not a total pain.
2. As it stands right now, the PS PCB runs a ground wire (BLK) to point C on the PCB, which is now the square ground trace. Leave it. Make sure square jack ground remains connected to square PCB ground;
3. Run a wire from sine PCB ground to sine output jack ground (if not connected already... sorry, I'm at work and don't have this in front of me, just the manuals and my notes, but want to have the plan clarified for tonight). I think this ground is currently crossing over to square PCB ground.
(Q: does this mean that the sine ground is run now through the chassis to provide power to the sine oscillator? In other words, I'm unsure how the sine oscillator is being grounded to be power supplied. I seem to be missing one required connection...)
Is this a good summary per your recollection of steps?
Thanks!
Radu.
Dick - I may be crazy, but my IG-18 floats the ground. The output ground (banana jack) of either waveform is not actual (chassis) ground.
The 0.047uF goes right in between the sine output ground (triangle on the schematic) and the actual, chassis, ground (ground symbol on the schematic). If this were true: "The sine wave ground board trace(s) should go to the chassis ground of the 0.047uF cap which is also the ground for the sine-wave output jack," the 0.047uF cap would be shorted.
Also, if I cut the trace carrying power ground to the sine ground trace, and do everything else mentioned in my steps above, the sine oscillator would not be powered.
Additionally, the ground lug in between the boards (connecting to the chassis and also to the trace that goes around the board but connects to no parts or anything) has no connection to the ground trace of either waveform. I even checked for continuity.
What am I missing?...
Thanks for bearing with me on this.
Radu.
The 0.047uF goes right in between the sine output ground (triangle on the schematic) and the actual, chassis, ground (ground symbol on the schematic). If this were true: "The sine wave ground board trace(s) should go to the chassis ground of the 0.047uF cap which is also the ground for the sine-wave output jack," the 0.047uF cap would be shorted.
Also, if I cut the trace carrying power ground to the sine ground trace, and do everything else mentioned in my steps above, the sine oscillator would not be powered.
Additionally, the ground lug in between the boards (connecting to the chassis and also to the trace that goes around the board but connects to no parts or anything) has no connection to the ground trace of either waveform. I even checked for continuity.
What am I missing?...
Thanks for bearing with me on this.
Radu.
Ok, so here's where I'm at:
1. Cut the traces such that sine and square trace grounds are apart. The only real difficulty was getting R16s ground leg to connect to square as opposed to sine, which it was closer to. Had to run a wire.
2. Ran appropriate wires from each of those newly separated grounds to respective output jack grounds & attenuators grounds (output jack to attenuator wiring was there and I left it as is).
Now, I still am confused on the next two steps (and thank you VERY MUCH for patiently bearing with me through this).
1. If I were to connect the sine output jack / attenuator to the chassis ground, I'd short the 0.047uF... Is that what I am supposed to do? (do I have an unusual kit? the wiring, as far as I can tell, was done per manual at initial assembly) Because, from your description, it sounds like those points are electrically connected, but in my kit they're not. They are separated by the 0.047uF cap. I have no problem dropping it altogether, but I want to make sure it's required.
2. It sounds like I need to connect square trace ground with chassis ground at the ground lug in between PCBs. Is this a fact?
3. If I do the above, the chassis ground is not floating anymore, but electrically connected with the power supply ground. Is this also intended?
Thank you very much, Dick, and if I could get itemized responses above it would be much appreciated.
Radu.
1. Cut the traces such that sine and square trace grounds are apart. The only real difficulty was getting R16s ground leg to connect to square as opposed to sine, which it was closer to. Had to run a wire.
2. Ran appropriate wires from each of those newly separated grounds to respective output jack grounds & attenuators grounds (output jack to attenuator wiring was there and I left it as is).
Now, I still am confused on the next two steps (and thank you VERY MUCH for patiently bearing with me through this).
1. If I were to connect the sine output jack / attenuator to the chassis ground, I'd short the 0.047uF... Is that what I am supposed to do? (do I have an unusual kit? the wiring, as far as I can tell, was done per manual at initial assembly) Because, from your description, it sounds like those points are electrically connected, but in my kit they're not. They are separated by the 0.047uF cap. I have no problem dropping it altogether, but I want to make sure it's required.
2. It sounds like I need to connect square trace ground with chassis ground at the ground lug in between PCBs. Is this a fact?
3. If I do the above, the chassis ground is not floating anymore, but electrically connected with the power supply ground. Is this also intended?
Thank you very much, Dick, and if I could get itemized responses above it would be much appreciated.
Radu.
Okay, it depends on how far into a mod you're going to go.
1) I removed the 600 ohm load switch as a load switch, and used it as a switch for floating ground or chassis ground. Some equipment will want the oscillator floating and some will do better with it grounded. So wire the sine output and attenuator grounds to the cap until you decide how far to go. Se my revised attenuator circuit on my webpage for IG-18 #1. IG-18 #1
2) Yes.
3) The chassis ground is always connected to the power ground; the board grounds do too in the Heath setup -- the 47nF cap just floats the sine output ground off DC ground
for connected equipment, but keeps an AC ground.
Keep working -- good job!
1) I removed the 600 ohm load switch as a load switch, and used it as a switch for floating ground or chassis ground. Some equipment will want the oscillator floating and some will do better with it grounded. So wire the sine output and attenuator grounds to the cap until you decide how far to go. Se my revised attenuator circuit on my webpage for IG-18 #1. IG-18 #1
2) Yes.
3) The chassis ground is always connected to the power ground; the board grounds do too in the Heath setup -- the 47nF cap just floats the sine output ground off DC ground
for connected equipment, but keeps an AC ground.
Keep working -- good job!
Alright. I have some progress and maybe a key to why I couldn't figure out what it was I needed to do.
Firstly - I completed this (grounded to chassis the square trace ground; star-connected all sine things to load switch chassis ground - and for now eliminated the 47nF cap). Fired it up and it looks fine on scope (no distortion measurements yet...). Still motorboating if I switch off the square stage, but no adjustments have been completed after much of this work. I had it before, and adjustments cured it.
Secondly - not sure how chassis grounding was supposed to be done by design (I almost read the entire manual at this point, but I have to close-read the parts referring to grounding), but mine for sure was not connected to power supply ground. Maybe this is why I had some additional trouble understanding what needs to get done.
I guess if this passes re-adjustment and reasonable distortion figures, I may endeavor in making the load switch a "floating ground" switch. And maybe make some further tweaks.
Decent distortion is what I'm after, though. I will post here when I manage to get some readings.
Thanks for al of your help, Dick!
Radu.
Firstly - I completed this (grounded to chassis the square trace ground; star-connected all sine things to load switch chassis ground - and for now eliminated the 47nF cap). Fired it up and it looks fine on scope (no distortion measurements yet...). Still motorboating if I switch off the square stage, but no adjustments have been completed after much of this work. I had it before, and adjustments cured it.
Secondly - not sure how chassis grounding was supposed to be done by design (I almost read the entire manual at this point, but I have to close-read the parts referring to grounding), but mine for sure was not connected to power supply ground. Maybe this is why I had some additional trouble understanding what needs to get done.
I guess if this passes re-adjustment and reasonable distortion figures, I may endeavor in making the load switch a "floating ground" switch. And maybe make some further tweaks.
Decent distortion is what I'm after, though. I will post here when I manage to get some readings.
Thanks for al of your help, Dick!
Radu.
Having had a bit of time to think about all this grounding thing - I guess the strategy here is to insulate the sine wave generator from the square as much as possible (chassis resistance), while providing a very short path for square wave generator ground to earth ground (= AC ground)... Is that a good summary?
And secondly - is adjustment procedure from the original manual a good guideline to follow? I have read your notes on adjustments, but just wondering if the manual's systematic approach is still valid, to be followed by fine tuning per your notes.
(besides, I am unsure how the adjustment to minimize distortion is supposed to be performed, having in mind how sensitive and borderline finicky a classic distortion analyzer such as the 334A can be.)
Thanks!
Radu.
And secondly - is adjustment procedure from the original manual a good guideline to follow? I have read your notes on adjustments, but just wondering if the manual's systematic approach is still valid, to be followed by fine tuning per your notes.
(besides, I am unsure how the adjustment to minimize distortion is supposed to be performed, having in mind how sensitive and borderline finicky a classic distortion analyzer such as the 334A can be.)
Thanks!
Radu.
(sorry for the multiple postings...)
well, maybe not.
50kHz distortion figures:
- 0.1V level out: 0.042%
- 10V level out: 0.073%
Tried 100kHz, but it seems it's too jolty to measure.
90kHz:
- 0.1V level out: 0.056%
- 10V level out: 0.09-0.095%, though not too stable between the two values.
It looks like upwards of 90kHz and... 5V things start going south. At least in what regards the level, this replicates what you have been observing.
Overall, it doesn't sound bad at all. HF stability still a concern.
Radu.
well, maybe not.
50kHz distortion figures:
- 0.1V level out: 0.042%
- 10V level out: 0.073%
Tried 100kHz, but it seems it's too jolty to measure.
90kHz:
- 0.1V level out: 0.056%
- 10V level out: 0.09-0.095%, though not too stable between the two values.
It looks like upwards of 90kHz and... 5V things start going south. At least in what regards the level, this replicates what you have been observing.
Overall, it doesn't sound bad at all. HF stability still a concern.
Radu.
You're getting decent performance for this oscillator. Tuning cap values are critical at 100kHz, and a frequency counter is useful for checking the decade frequency steps its best if they are within a couple percent or less of the decade step values, based on the parts for 1kHz. However, the 5uF cap is the bear to change if it's bigger than the others, so I just matched the others to it, by building up paralleled caps to get values within 1% or so of the decade values.
The bias control setting used by Heath works fine, since it sets the amp up for max output at the given supply level. The distortion pot adjust is best done with the 334 -- the two adjustments interact a bit, so do them a couple of times.
The 334 is not in the least finicky about auto tuning once you manually tune for a notch of say 20dB -- then the auto tune just does its thing IF it's working correctly and the lamps and photocells are all good. You are correct that manual tuning with low distortion gear is essentially impossible. Note that the residual distortion of the typical 334 is about 0.01% at 1kHz -- these things were mostly used in radio and telecom, where truly low distortion was not on the table. They work very well for tube gear, but then so did the old manual tuned 330B series.
I'm interested that switching off the square wave section upsets the sine oscillator -- is the square generator/Schmitt trigger working correctly?
The bias control setting used by Heath works fine, since it sets the amp up for max output at the given supply level. The distortion pot adjust is best done with the 334 -- the two adjustments interact a bit, so do them a couple of times.
The 334 is not in the least finicky about auto tuning once you manually tune for a notch of say 20dB -- then the auto tune just does its thing IF it's working correctly and the lamps and photocells are all good. You are correct that manual tuning with low distortion gear is essentially impossible. Note that the residual distortion of the typical 334 is about 0.01% at 1kHz -- these things were mostly used in radio and telecom, where truly low distortion was not on the table. They work very well for tube gear, but then so did the old manual tuned 330B series.
I'm interested that switching off the square wave section upsets the sine oscillator -- is the square generator/Schmitt trigger working correctly?
distortion results and further info on the T-S
Yes, Sir! This starts getting where it was supposed to go. As soon as I get energy for a more mechanical modification, I will more the transformer out of the chassis. Though, I must say, switching off the high pass 400Hz filter only increases distortion by the tiniest of amounts, so maybe the 60Hz bleed from the transformer is not all that sizable.
The square on/off behavior is weird to say the least, though. I include a pic off of the scope (sorry for the quality). Not too handsome, I'd say.
Voltages on the transistors from that section are kind of all over the place in relation with what's spec'd on the schematic, but those latter ones go with certain positions of the adjustments pots and I wasn't looking forward to lose the adjustment... None, though, (in the T-S section) show proper Vbe of 0.65V or near that. At the same time, as the grounding now insulates them, and as the T-S seems to place a beneficial load on the output of the sine section (the only explanation I can come up with as to why sine looks better with the T-S on...), I'm tempted to just leave it alone. The other option is to switch it off and readjust everything, so that not only motorboating would stop, but the whole sine section would fall into a new (hopefully even quieter) place.
Should one critical avenue to follow be to adjust feedback for less output? I'm thinking this may also help with all of the instability issues I am still noticing.
Radu.
Yes, Sir! This starts getting where it was supposed to go. As soon as I get energy for a more mechanical modification, I will more the transformer out of the chassis. Though, I must say, switching off the high pass 400Hz filter only increases distortion by the tiniest of amounts, so maybe the 60Hz bleed from the transformer is not all that sizable.
The square on/off behavior is weird to say the least, though. I include a pic off of the scope (sorry for the quality). Not too handsome, I'd say.
Voltages on the transistors from that section are kind of all over the place in relation with what's spec'd on the schematic, but those latter ones go with certain positions of the adjustments pots and I wasn't looking forward to lose the adjustment... None, though, (in the T-S section) show proper Vbe of 0.65V or near that. At the same time, as the grounding now insulates them, and as the T-S seems to place a beneficial load on the output of the sine section (the only explanation I can come up with as to why sine looks better with the T-S on...), I'm tempted to just leave it alone. The other option is to switch it off and readjust everything, so that not only motorboating would stop, but the whole sine section would fall into a new (hopefully even quieter) place.
Should one critical avenue to follow be to adjust feedback for less output? I'm thinking this may also help with all of the instability issues I am still noticing.
Radu.
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.