Well, I've been able to get the inductuner to 'do something'. Turns out they snuck a resistor to case/ground inside two of the sections and, of course, that would be the ones I first tried. So I snipped one end loose (too tight to safely get the other) and, after tapping way down on the air coil, was able to tune roughly a 5 MHz range on the low end of the FM band using one section (of the three it has).

Audio quality was weak and dismal but, to tack the inductuner in, I was using 7 inch flying leads and that can't be good for stray capacitance, wire inductance, and osc loading. Oops, come to think of it, after snipping the resistor loose I neglected to put the inductuner case back on Well, that probably didn't help things either.

It does suggest, though, that when the osc is mounted in the chassis with more rational lead lengths the inductuner should be able to give tuning over a portion of the FM band. So now we have three tuning options, air coil squeeze/spread, the "PenTune," and inductuner, with the first two known to work even on the lax breadboad layout.

I don't see how negative feedback would affect the linearity of the reactance part of the circuit since the feedback is not derived from the resulting transmitted audio.

Dave

The amount of injected reactance is proportional to gm, which is plate current per grid volt. Unfortunately, gm vs grid volts has a semi 'S' shape, it deviates from linear on each end.

Now, the reason we care for gm to be linear is because our 'input' signal to the 6KE8 is a voltage on the grid, and gm tells us how much current that produces, but it is plate current that 'does the work'. It's the injected reactance and if plate current were linear then things would be fine.

So, instead of just slapping volts on the grid and letting plate current fall where it may, due to the gm curve, what if we had a way to directly produce the desired plate current?

Rk senses plate current, albeit a bit indirectly because the cathode also has screen current. That's why I mentioned the assumption that screen current follows plate current by a fixed ratio but, assuming that holds, negative feedback from Rk forces plate current to be linear relative to the 6AG5 input signal. Signal on the 6KE8 grid will be 'distorted' in inverse to the gm curve but we don't care about it's grid volts anymore because we force plate current to follow the transmitter input.

I agree with you that since negative feedback is not taken from transmitted audio it can't take care of 'whatever' might happen in the reactance circuit, like say the phase network shifting due to grid input impedance, but it should improve the gm curve.

Flipperhome wrote: "It does suggest, though, that when the osc is mounted in the chassis with more rational lead lengths the inductuner should be able to give tuning over a portion of the FM band."


I know that you are using another means of freq adjustment (as opposed to the air variable cap that I am using), but I found that the specific physical placement of the air core coil in my transmitter along with the coil lead arrangement will affect tank resonance significantly.

Hmm. Since there's a 1nF capacitor bypassing the 220 ohm cathode resistor, it is essentially a short circuit for anything above 1Mhz, so it's not sensing plate RF current. Since the injected reactance depends only on the RF voltage at the plate (and the AF plus DC bias at the grid), I don't think you can consider the cathode resistor to be providing any negative feedback in the normal sense of the word. What's happening is that the cathode resistor is controlling DC bias, and is shifting operation to a point on the gm curve which is more linear. It's not changing the gm characteristic, just the location on the gm curve where the thing is operating.

The capacitor across the cathode resistor is too small to bypass the audio signals, so it will reduce the AF current at the plate, but that shouldn't have any effect on the reactance injected back into the oscillator, and it doesn't produce any AF voltage on the plate, because the plate load is a 10µH inductor which is a dead short at audio frequencies.

I may be overlooking something, but the only audio effect that I can think of, is that the cathode resistor would limit AF current drawn from the power supply.

One point that was brought up in one of the earlier posts related to the difference between a triode and a pentode as a reactance tube. I mentioned that the pentode is easier to analyse. This is because, in a triode, the transconductance varies with both grid bias and plate voltage. The definition of transconductance is "The change in plate current caused by a change in grid voltage, when the plate voltage is held constant." That's tough to do with a triode. It doesn't mean that a triode won't work as a reactance tube, but it means that the published gm curves won't be much help in predicting behavior. With a pentode, it's easier; the plate current is relatively independent of plate voltage, as long as you keep the screen voltage constant, so the gm characteristic is fixed, and except for normal production variations, should follow the published curves. If you wanted to linearize the gm characteristic, you would have to unbypass the RF at the cathode, in order to sense the RF current.

There is also another possibility. You can try different screen voltages to see how they affect linearity.

I see that the cathode resistor value has changed over several circuit changes from 820 to 470 to 220. This seems to agree with what I've seen in the curves.

Now, I've got a couple of questions, based on some preliminary and rather crude circuit model that I've set up. It appears that the 5-45pF trimmer has very little effect on frequency. Is that true?

Secondly, it appears that the optimum grid bias for best frequency linearity would be around -0.63 volts. Due to grid contact potential which is in the range of -0.5 volt, the required cathode voltage may be reduced by that amount putting it in the range of 0.13 volt. What voltage do you measure at the cathode of the reactance tube?
And between grid and cathode (assuming you have a high impedance voltmeter)?

Your parenthetical is the point. The amount of reactance is proportional to the current set by AF plus DC bias at the grid. RF operates 'at' the bias point.

You seem to be focused on just Rk but that simply sets the idle bias. AF then shifts bias to modulate the frequency. If the pentode had a perfectly linear gm curve then when grid volts increased 50% cathode current would also increase 50%, but the gm curve is not linerar so it doesn't. However, negative feedback senses cathode current and 'adjusts' grid volts to get the 50% increase in cathode current.

It affects injected reactance by setting the current. That's how the modulation works. I.E. As you noted, there is no 'AF voltage' so how does grid AF effect modulation? By changing the amount of injected current. Normally that would according to grid volts times gm, but with NFB grid volts are dynamically adjusted so that cathode current follows AF.

True. It acts as a peaker.

I have a VTVM but don't trust it's calibration as it seems to read high. My DVM says cathode is about 1.25 V, the VTVM says 1.5 V. I believe the DVM. The VTVM says grid to cathode is 80mV, but since it reads high it's probably a bit lower.

I should clarify my previous answer to this. According to the patent I plagiarized it is, indeed, intended to provide a 1.5 MHz adjustment range. However, during my observations it also seemed to affect symmetry which, in turn, affects amplitude. So, rather than use it as 'tuning' I adjust for best symmetry, which is easiest to tell by seeing amplitude peak and the reason for my sloppy term 'peaking'.

Now, that poor thing has to go from one side of the breadboard to the other so it could be the wave shaping I see is aggravated by stray coupling and perhaps it would be better behaved with a proper layout but on the breadboard I discounted it as a means of tuning.

Btw, I changed the 1nF to 47pF to pad the span down (upper end would kill the osc) and do see about a 2 MHz range.

In my simple triode circuit, the plate voltage does remain almost constant (at audio frequencies) since the modulator plate it is connected directly to the oscillator plate. The source for plate voltage is the oscillator coil fed by the 47 ohm B+ decoupling resistor thus providing a very low impedance plate load. Seems like carrier deviation is controlled only by the change in interelectrode capacitance within the modulator tube caused by a change in grid voltage. I've never seen curves for this phenomenon.

I think there is enough stray cathode-to-filament capacitance to provide RF bypass to ground while the cathode resistor sets the bias for relatively linear operation and provides some degenerative feedback at audio frequencies.

Dave

http://www.ebay.com/itm/New-L-C-Inducta ... 3f0f208749

For $40.00, this LC meter that I picked up from this seller on eBay does a pretty good job on measuring very low capacitances and inductances which is great for high freq work. I was testing out this instrument out last Sunday for well over an hour, and I just couldn't find any reason why I would want to return it. The banana plugs & leads that came with it fit sort of loosely into the sockets, but it doesn't seem to matter.

Thanks. I put it in my watch list.
Mine is an ancient Tenma 72-370 and I may try giving it a thorough cleaning before buying another although, even if working fine, it's range is not as good as that one.
I ran across an interesting 'bare' one. http://www.ebay.com/itm/L-C-F-Inductance-Capacitance-precision-meter-0-01pf-/220822951938?pt=LH_DefaultDomain_0&hash=item336a12d802
Looks like the same thing without enclosure except, strangely, the "HI-C" button is missing.

From what I could gather from that seller's feedback profile, at least a couple of people who previously purchased it were happy with it. It looks like a decent meter if you don't mind it not having the Hi C, but we're only talking about $27.00. These meters all have one thing in common though; they're all from Hong Kong or China.

I located the last tank coil for the "New Band FM Transmitter" a little too close to the air variable tuning capacitor (the rotor was brushing up against the pickup coil) so I made up "yet" another coil today (with slightly longer leads). Yesterday, I did measure the old coil at 0.114 uH with the LC200A so I adjusted the new coil accordingly (with the LC200A) before soldering it in. Sure enough, the oscillator came up on the FM band as before so no further coil adjustment was necessary.

I do not have an antenna connected yet to the xmitter, but the signal can easily reach the next room. The audio has some distortion so I'll have to look into that next.

Turns out there's two versions. The 'S' doesn't have HiC but the 'A' version does. 'Best deal' was, I think, about 3 bucks more for the 'A'.

Congratulation, that sounds more like it. Dave can probably help you on the distortion end as that's his circuit.

Without any audio processing (compression, peak limiting, peak clipping) it is very easy to overmodulate and go beyond the + / - 75 kHz deviation swing. To complicate matters, the processing threshold needs to follow the pre-emphasis curve. Without processing, the audio will sound very weak compared to other stations on the dial before distortion occurs. If you use source material that has been pre-processed (like most internet streams) better results will be achieved. Digital audio on the internet seems to need processing similar to FM to keep the S/N ratio high without running into the digital brick wall. I use the various free channels from RadioIO.com with good results.

If your distortion is occurring at all modulation levels, perhaps an adjustment in the value of the modulator's cathode resistor is needed.

Dave

I don't know. Maybe it's the bit rates they're using or they just want it to 'sound like' broadcast but there's no reason why digital streaming, per see, needs compressing. But then, neither does CD and you see what they did to that medium.
You might like to try (Winamp) Shoutcast as well. All of those are 'free' with up to 320KBPS MP3 and AAC.

Thanks for the reply and the info Dave! I'll keep those in mind. One thing that I noticed about this transmitter is that the oscillator actually has a full 5-10 minutes warm up time upon power up before the transmitter is up to speed. I'm listening to CD music from the transmitter right now on the FM band of my '51 Philco AM/FM/Phone console, and fiddling a bit with the volume & pre-emphasis knobs.

I thought I'd get back on the Inductuner idea and upon closer inspection, well, it won't work, at least not unmodified in my circuit.

The sections are not really 'a variable inductor' but two concentric variable inductors with external connections to 'the same end' of each and the other end of the two connected together, plus going to ground in 2 sections (although that can be easily disconnected in the Mallory). The problem is, if you try to use the two as 'a variable inductor' from the two available connections they're 180 out of phase with each other. It's like winding a coil with half the turns in one direction and the other half in the opposite.

I've been trying to find some way to mechanically hack them into something more useful but the physical layout makes that almost, if not actually, impossible.

Does anyone have an idea?

Well, I have the beginning of an idea. I've been able to solder a wire onto the brads holding the wiper contact onto the rotor shaft and I get continuity from the 'inductor' ring to my wire. I don't think it would work, though, to just leave enough wire to 'wrap around' the shaft as it turns, especially since it would act as a 'floppy' inductor turn.

But, as it turns out, there's a brassy looking bushing pressed into the rotor, so it rotates with the rotor. It also, conveniently, protrudes about 3/64" from the rotor, with a lip on the outer edge, and I have successfully soldered the wiper wire onto the outside of the bushing lip.

Now, that diameter might be small enough let a semi short 'wrap around' wire just hang off it but I was thinking of using the bushing as a sliding contact.

So, here's the question, does anyone have an idea for what could be used to fabricate a 'springy' contact to ride in the bushing groove?

Here's a first attempt at some kind of 'wiper' connection. I couldn't find anything really 'good' lying around in junk parts so I fashioned a 'hook' wiper from 20 gauge magnet wire. I don't think it's robust enough but it's survived some rotations and at least tests the concept.Turns out the 'brassy bushing' is more than just a bushing. When 'pressed in' it locks the rotor to the shaft but it can be loosened and the position shifted. That might turn out to be convenient but it's also irritating when it comes loose and upon 'fixing' it I found the wiper being 'uncurled' because it would snag, so the curve is elongated past the bearing surface and has a 'Little Bo Peep' staff reverse curl of the tip to present a smooth surface rather than the butt end of the wire. The hole it goes through was drilled by dremel.

Tension is by spring, connected by cord because I was concerned a metal spring would affect inductance, even if technically unterminated, plus, as implemented, its secured to the chassis and would ground short the wiper.

Well, it took a long time, partly because I got diverted off into a host of other things, but I think I may have finally figured out something.

My current understanding is your circuit isn't 'technically' a reactance tube but, rather, operates on the Miller effect and I had a bit of an epiphany there too. I mean, I always knew of the Miller capacitance but it was just 'something to remember', an 'equation', and I didn't have a good conceptualization of it.

Well, perhaps because I posted the Twin Coupled amp design, so it was still smoldering in my head, but, well, I use a garter bias on the output tubes and since they're riding on top of the cathode OPT the 'signal ground side' of the grid leaks have, instead of signal ground, an equal amplitude anti-phase signal on them (the reverse of a bootstrap), so the effective grid leak impedance is half the component value.

Well, ding ding ding that's the resistor equivalent to Miller capacitance ding ding ding By golly, it isn't so 'mysterious' after all and nothing actually 'changes'. It's just that there's a signal 'on the other end' and in phase it's a bootstrap but anti-phase it's 'the Miller effect'.

Back to your circuit, it isn't that the interelectrode capacitance changes, per see, but the gain of the tube does. I.E. the Miller capacitance is the plate-grid interelectrode capacitance times 1+Av, and the gain changes as bias changes. (note that in my garter bias circuit the cathode gain is roughly 1 and 1+1 is 2, which is why the grid leak impedance is roughly half).

Like you I had wondered what 'the curve' for that capacitance was, such as you can get for a varactor, but there isn't a (separate) 'curve' because nothing (not already plotted) is actually 'changing'.

Yep. For it to be a true reactance tube circuit, there should be an RC or RL phase shift network connected between plate and grid to give a phase shifted voltage at the grid relative to the plate. Since plate current is equal to transconductance times grid volts, the plate current will then be out of phase with the plate voltage. Hence, it behaves as a reactance.

Since transconductance varies with grid bias, changing the grid bias will change the amount of phase shift at the plate, and therefore the reactance seen at the plate also changes. So, to determine linearity, you need to look at the tube's transconductance vs grid volts curves.