I'm working on a reactance tube FM transmitter and while not completely finished it's far enough along I thought I'd share and solicit comments.

The basic Osc/Mod topology was plagiarized from an old RCA patent, which used a 6AN8, but I intend to eventually use 'dollar days' tubes, meaning they cost a buck each from on-line suppliers. That's where the 6CU8 comes in but a 6BL8 is what I had on hand and is used in the breadboard. The preamp is a 6AU6 but 'dollar days' candidates would be 6AG5, 6BC5, or 6CE5. Bias tweaking would likely be necessary with the alternates (I use 470 under the 6BL8 rather than the shown 820 for a 6CU8 and 15K under the 6AU6)



V1B is the reactance tube with V1A the oscillator. (The unlabeled resistor on C4 is a dummy to make spice work)

The 6AU6 preamp serves as both 75 uS pre-emphasis and to reduce modulator distortion via negative feedback from R2 sensing modulator current.

And the not so pretty breadboard



I would like to add an antenna winding but the thing radiates well enough for at least a 30 ft range as is. That's not a 'tested limit', simply how far away the radio I'm using is located.

To my ears the audio is excellent.

I intend to eventually mount everything in another UHF converter and breadboard power is coming from a Granco transformer out of one.



There's more detail on my web page, as well as some speculations on possibly doing a stereo MPX addon in the future.

http://flipperhome.dyndns.org/FM%20Stereo.htm

It's always interesting to see what other people come up with for FM transmitter designs.

Through experimentation with my "old band" transmitter, I found that routing a piece of bus wire about half way around the coil provided plenty of coupling without causing too much frequency shift caused by the antenna. It is grounded on one end and the other end leads to a BNC connector on the chassis apron. Using two 18" test leads in series as an antenna, coverage increased from 30' to well over 300' using my Radio Shack Patrolman (low-band VHF) radio.



For me, the trick for getting low distortion was to experiment with the value of the reactance tube cathode resistor. Getting the carrier to swing symmetrically around the center frequency was observed on my Hameg specturm analyzer as I changed it's value. I also discovered that it did not want to be bypassed.

Dave

Interesting design, and people will certainly like to experiment with this.

I'd stay far away from a 6CU8, it's one of the worst tubes ever built, right up there with the legendary 6GH8 as far as a high failure rate goes. Every TV we ever got in the shop that used 6CU8's they were always bad and a year later they would be bad again.

There are plenty of other choices for triode-pentode types that you can get cheap. You could even rectify the filament supply to get just a bit higher (DC) voltage and use series string tubes that no one wants like 3AU6 and 4BL8 or 5EA8, whatever you can find that has the right filament current to run in series.

Yeah, there are a number of ways to skin that cat. I was just fascinated with the reactance tube concept and wanted to try it.

I had attentively read your post and was going to try something similar but I'm wondering how you arrived at the half turn. I mean vs a full turn, or a smaller/larger radius, spacing from center (coupling), etc. Is there a particular ratio?

I'm also thinking of using a slug tuned coil and wonder how many turns for the secondary and where to place them.

I tried to find the most linear spot too but don't have a spectrum analyzer and, so, had to rely on the datasheet curves. That's why I used a different Rk under the 6BL8: it ran more cathode current, which increased the bias point.

Mine's different than yours, though. If I understood correctly yours is using inherent tube capacitance whereas mine is deriving it's reactance from a 90 degree phase coupled sample to the grid and, depending on the components, the resulting reactance can be either capacitive or inductive. I'm not sure what the result would be in removing the bypass but I would have thought it was moot at AF. Sure wouldn't hurt to try but I don't know how I could measure it.

Cathode AF distortion I can measure in Spice and then there is the noticable improvement in the listening with NFB.

I thank you for the tips.

Thank you.

Thank you, that's good to know. Or 'bad' to know Do you know why the 6CU8 failed so much? I mean, like, were they perhaps overdriven, which was not an uncommon issue with tube TVs. This circuit only runs about 5 mA at 120 V through them so dissipation is around a modest 600mW total, screen and plate combined.

One can use the 6BL8 or 6AN8 too, of course. They're only 4 or 5 bucks so it's not like we're trying to avoid a golden tube. The 'dollar days' thing was mainly a fun target.

The 6HG8 (not GH) was another 'dollar days' tube I thought about trying because it's gm is double that of the 6BL8 and I wondered how much that would increase modulation sensitivity. It pulls less heater power to boot. The 6EU8 is another 'dollar days' tube with good gm but neither of those had gm curves in my datasheets so I don't know where the best bias would be.

A 3AU6 and 4BL8 might just make it since heater voltage runs a bit high on these old UHF converter transformer windings but I'm not sure. As is, heater voltage is a hair over 7 V at 750 mA but the series connection would be less at 600 mA. Well, I think it ran 7.5 V with just the 450 mA 6BL8, which is already .25 V under the bogie 7.75 V for the 600 mA series case. On the other hand, it might not hurt to run them a tad low because this design isn't pulling anywhere near 'max' cathode current on them.

Rectifying for DC worries me because that loads the transformer more than straight AC. The power factor for half wave is 2.3, for full wave 1.4, and 750mA is already over what they normally powered. Although, the Granco transformer runs surprisingly cool and seems to be over rated. Plus, this circuit doesn't pull near as much B+ power as the original so some extra heater loading might not be a problem.

That's certainly something to think about.

I did, in fact, wind up using a slug tuned coil. The antenna pick-up coil was and it's position was done strictly though experimentation. I had previously built this circuit (for the present FM band) from an article published in Science and Mechanics magazine published in the mid '50's. using a single 12AT7. I think the title was "FM Phono Oscillator". A crystal cartridge was connected directly to the grid of the triode section used for modulation. At the time, I was amazed at the resulting coverage it provided. I don't think there were any "Part 15" rules regarding FM when the article was published. I built the circuit in the early '60's. I think there were probably some rules in place at that time for unlicensed radiation in the FM band but, being a sophomore in high school, I had no clue about FCC rules.

Dave

Well, I did a first shot try at a half turn antenna coil about .8", or so, diameter around the .5" coil, grounded on one end and the results were not promising. Frequency shifted a couple MHz, not a big deal, but deviation went to hell and it sounded like spittle.

Do those symptoms ring a bell? Got any ideas?

The oscillator circuit used for my oscillator is a Hartley while yours looks like a modified Colpitts. B+ is fed into the tap on the coil in my circuit. I think the impedance at this point is very low. The antenna pickup wire is near the tap so perhaps it's presence doesn't materially affect the frequency. The way I found the ideal placement was to watch the signal strength on a wave meter placed near the antenna and adjust pickup wire position for the highest reading.

I wonder if you might have better results coupling your antenna to the plate of the oscillator with a small-value cap.

Dave

It's my guess the 6CU8 like the 6GH8 was operated near their limits...

At least part of the problem for the 6GH8 was RCA couldn't build a good one... Warwick used mostly Sylvania tubes(6GH8 incl) in their versions of RCA CTC15 and we(at Sears) didn't have a lot of issues with those...

Tom

I can only remember the 6CU8 being used in a few RCA B&W sets for one or two model years back in the late 50's. Those sets were nothing but trouble, and very often the trouble was centered on the 6CU8 tube. The pentode was used as an IF amplifier, and the triode either a vert osc or sync separator. The pentode was always full of shorts, sometimes you could look inside the tube when the set was on and see sparks or the grid wires glowing. The triode was either down to about 10% emission or else the cathode was open. The 6GH8 as bad as it was, was still a much better tube than the 6CU8.

Yep- the antenna is loading the oscillator. Might want to put some sort of buffer amp between it and the antenna. Would recommend a cathode follower for low Z.

Scott Todd

A quickie test didn't load the osc, at least, but it wasn't easy to tell just how much it was radiating since the coil itself spews quite a bit.

Of course, just one shot at slapping a half turn on isn't really 'trying' either and a smaller one might work better. Unfortunately, my mechanical arrangement doesn't really cater to moving coils around seeking a 'sweet spot' but since I'm looking at replacing the thing with a slug tuned coil anyway I may hold off till I locate one and try that.

I found a candidate at Mouser but am doing a 'survey' of potential projects to see what else I might need because it seems ridiculous to spend 8 bucks of shipping just to get a 38 cent coil. That's why I did a quickie air wind in the first place

Don't get me wrong, I don't doubt your experience one bit. I was just wondering what 'the problem' might have been because TV sets were notorious for pushing tubes to, and past, their limits. Must admit, though, IF amp doesn't sound like a high stress job.

Of course, a repair shop is also a bit like a hospital; you don't see the 'healthy' ones very often

Thanks. Yeah, my blind shot is loading the osc but I'd like to avoid adding a tube.

Well, I think I have antenna coupling now. I say "think" because just the air coil radiates and without a field strength meter it's hard to tell how much there might be. All I can say for sure is if I put my scope probe near the coil I pick up signal and if I move it near the wet noodle antenna (not the final antenna intended) I also get signal, so 'something' gets to it. And, at least, there's no frequency shift nor does it flat kill the osc like before.

I wound a new coil on .3125" (ballpoint pen body) to up the turns a bit and put a separate same size single turn at the end spaced about the same as the others. Or, rather, it was to be a single turn but I miss judged the terminal block layout and, in bending it over, lost a fraction of the turn. I don't think it matters.

I'm still not sure why the other one didn't work but I never calculated the inductance of that large half turn so maybe it was too much.

The Toko slug tuned coil arrived today but they threw me a curve in that the 'square-ish' plastic form is molded around the coil and, for the moment anyway, I'm not sure how to add another turn to it for the antenna coupling. Or even if it would work because there's only 5 turns on it and a single turn would, therefore, be too much inductance, too low a turns ratio, even if I could get it on the form.

Btw, referring to our discussion in the MP3 player post, there must be something in the water over there because the Japanese aren't much better than the Chinese with documentation. In addition to the osc coil I ordered some 10 mH slug coils anticipating experiments on the multiplexer and, in their datasheet, Toko just doesn’t seem it worth mentioning what the pinout is and I didn't think much of it when ordering. I mean, a coil, two wires, but there's 5 pins on the blooming thing

Edit: Two turns, this time properly placed, seems to work even better. I'm not sure if phasing matters but I think I'm on the right track. With 'almost' 1 turn the scope probe on 50 ohm 1x saw 'something' but now it says "oh yes, there it is."

Okay, here's the current breadboard with the 5/16" coil and 2 turn antenna coupling, plus a couple other tweaks



You can see the two turn antenna coupling coaxially to the right of the primary coil with my blue wire wet noodle flying off the top.



R13 was increased to lower the pre-emphasis shelf a couple dB to reduce peaking above 15 kHz but I'm not sure it matters. The problem is pre-emphasis, with any FM transmitter, accentuates HF (it's purpose) but tends to create impulse noise on problematic source material, in particular if there's clipping in it. I'm trying to reduce that effect by limiting how much gain pre-emphasis continues to apply above 15 kHz and trying to avoid complicating things with a gaggle of filters.

Preamp maximum gain was increased (R11 and R12 decreased to 10k) to better accommodate cell phones and small MP3 players.

L1 is 10 uH, as opposed to the originally shown 2.2 uH, R2 is 470 and V1 shows the breadboard's 6BL8.

We have our first "Dollar Days" version using a 6AG5 for the preamp and a 6KE8 for the reactance tube/oscillator, both of which are "Dollar Days', 1 buck each, specials from online tube suppliers.

The 6KE8 not only has almost twice the 6BL8s gm but its voltage control range is only 1/3 as wide (note the lower, 220 ohm, R2), meaning the gm slope is dramatically steeper resulting in increased gain. So much so, estimated around 12-15 dB, I went back to the 100k input resistors and can still easily over drive the transmitter from a cell phone.

I also monitored my cell phone's FM radio output while driving the transmitter with 100 Hz to 10 KHz tones and equal amplitude in produced equal amplitude out so pre-emphasis works as advertised. 15 KHz was down 1 or 2 dB, which was expected due to the R13 shelf.

B+ voltages reflect how they roughly come out when using a Granco UHF converter transformer full wave rectified into 47uF with separate 5.6k/47uF filters for the preamp and reactance/oscillator sections, as previously posted.

Edit: corrected B+ on mod/osc section. Also, I don't think the gain adjustment is necessary anymore but it's still on my breadboard and might be useful for cell phones with large volume steps in order to fine tune deviation.
Edit: Moved C14 to across R6, getting it off B+ and on "RF ground."

Flipperhome wrote:
Edit: Two turns, this time properly placed, seems to work even better. I'm not sure if phasing matters but I think I'm on the right track. With 'almost' 1 turn the scope probe on 50 ohm 1x saw 'something' but now it says "oh yes, there it is."

So what's the verdict? How was the audio response? Any distortion? What do you suppose would be the possible ramifications if additional turns were attempted?

Audio response appears to be flat to 10KHz and down 2 dB at 15KHz. That is, assuming my cell phone FM radio does de-emphasis correctly.

I don't have a means to measure distortion but I don't see any. Spice indicates nominal distortion on R2 should be under .1% but I can't say for sure whether that directly translates to the received signal. What I hear sounds good, though.

2 turns seemed best. I tried three and, while it worked, there seemed to be no significant improvement in output but with some sensitizing of the oscillator to the antenna. That suggests to me that more would begin to load the oscillator.

Fantastic Flipperhome. I was curious as to how you came out with it, but it appears that you have a winner. Nice job! I wound mine up about 2.5 turns, but I didn't get a chance to try it out. When I powered up the xmitter afterwards, I was getting excessive current draw on the B+ (more than 40mA). The LED on the B+ blew before I had a chance to remove the power. Sacre bleu! Oh well, I there's always tomorrow.