Monday 15 September 2014

LF Regen - Experimental Building Notes

Courtesy Roelof Bakker PAØRDT

Anyone experimenting with the LF regen circuits described by Mike Tuggle and Roeloff Bakker in my recent  "Tuggle-Style LF Regens" blog, might be interested in some helpful 'building notes' posted to the Yahoo ndblist by noted NDB/BCB DXer and exceptional RF guru, Steve Ratzlaff, AA7U of La Grande, Oregon.


" Thanks to Mike Tuggle for his "1AD" (one active device) article using a dual gate Mosfet as the active device, and to Roelof Bakker's article about his own version of the 1AD I decided to build one too. I wanted to keep to the true 1AD configuration (no extra amplification) if possible. Mike uses very sensitive soundpowered headphones from the World War II era but those are very scarce to find in usable conditions these days. I was hoping to find an alternative set of phones that could be used but still not have to add an amplifier stage. I believe I have found a source of phones that can be currently purchased that are sensitive enough for the LF 1AD receiver.

Mike uses a Siemens 1216 Mosfet that he obtained from junked TV UHF tuners, but that particular transistor is no longer available. My first thing was to purchase a variety of Mosfets of other types and test them on the bench to see if they might be suitable. I bought one or two of all the Mosfet types available from this non-eBay US seller, Scott Lowe. (He used to sell on eBay but now only sells through his own webpage. And while checking the URL I notice it's now listed as Mike Lowe instead of the previous Scott Lowe--I don't know why the change.)

So I got the dual gate Mosfet types 40673, BF981, 3SK88, MFE201, BF960, BF961, BF998, 3N211 to try. When I mentioned this project to Mike Tuggle he was extremely generous to send me one of his rare Siemens 1216 Mosfets to test along with some other Mosfets he'd been saving for years. Thanks Mike! I bought five BF966 Mosfets from an eBay seller in Italy, so I was able to test the same type Roelof is using too. (The eBay seller of the BF966's seller name is "radionova22" and current item number for five BF966's is 331079146309, for $1.00 plus shipping ($10 for me in Oregon). He shipped immediately, and it took about 10-12 days for me to receive the small padded envelope. My one experience with this seller was 100% positive.) Note that almost all, if not all, Mosfets are diode-protected on the gates, and are more or less insensitive to handling static damage. But most sellers will ship the Mosfet in a static-protective wrapping. The Italian eBayer does not do this. But the couple BF966's I tried in my sensitivity test worked fine, thus were not damaged by any lack of antistatic shielding. I always use antistatic methods if I build something with Mosfets or anything that might be sensitive to static damage, and I recommend you do the same. At the minimum that means sit perfectly still while you're handling the Mosfet, don't shuffle your feet. Try to touch a grounded object before handling the device. In the US that could be as simple as a wire attached to the third wire/ground on the AC outlet powering your soldering iron that you can touch first. More elaborate would be an antistatic mat and an antistatic wrist strap all connected to the third wire ground of the AC outlet. Most modern soldering irons use a 3-wire AC cord with the actual metal of the soldering iron connected to the third wire ground, but I don't recommend you try to touch the very hot metal soldering iron part to obtain your antistatic ground!

I built the LF regen toroid version that Roelof used, and used the diode source bias that Mike and Roelof used and used my existing USI soundpowered phones (also now rare to find) with the Bogen T725 audio transformer and tested each Mosfet on the bench for sensitivity. I used a calibrated signal generator at mid-band, 300 kHz, reducing level to the faintest level I could detect. Initially I used Roelof's Gate 2 separate bias circuit, but soon found that I got the same sensitivity with no separate Gate 2 adjustment needed (using Mike's circuit of tying both gates together) and that's what I used for all the sensitivity tests. What I found was that all the Mosfets were more or less equally sensitive, within a dB or two, if they would work in the circuit. Not every Mosfet would work, that is would give proper regeneration when the Regen control was varied, and detect a signal--a few types did not work at all. Perhaps they would work with separate Gate 2 bias but I didn't try that to see if they might work. The ones that did not work were the 3SK88, MFE201 and the 3N211 made by Siliconix. Mike's selection of Mosfets included a 3N211 made by Texas Instruments, and oddly that one worked fine and was equally as sensitive as the other Mosfets that did work. So the good news is if you select a Mosfet from the ones I tried that worked, then you don't have to worry about one Mosfet being more sensitive than another. That list is the BF960, BF961, BF966, BF981, BF998, 40673 (and Mike's own "unobtainable" Siemens 1216). Note the BF998 is a surface mount part, very tiny, and will require you to use magnification and some ingenuity in order to solder wires to it. I was able to solder thin #30 silverplated wire stripped from #30 wirewrap wire but it took a number of attempts to get the three wires on (both gates are tied together and both gates are on the same side of the package making it slightly easier). Using my sensitive USI phones I was able to get -135 dBm sensitivity--that was the faintest signal I could hear, for my sensitivity test for the Mosfets.

Steve's Testbed Regen
A eBay seller in the San Francisco, California area, Ming, who some of us buy our Litz wire from, also advertises some Chinese soundpowered phones and one regular magnetic headphone. Since he's been advertising them for years he apparently has a ready supply of them from China, and I bought a pair of each of his phones to try with the idea that maybe these phones would prove sensitive enough to use in the 1AD LF set without having to add a stage of amplification. I did another bench sensitivity test similar to the Mosfet test. I compared the sensitivity of my own USI phones with these phones. As for the Mosfet test, the USI phones gave -135 dBm sensitivity. Ming's non-soundpowered 4400 ohm phones gave -125 dBm; Ming's soundpowered EH-3 phones gave -120 dBm; Ming's soundpowered EH-5 phones gave the same -120 dBm sensitivity. (Ming told me he sometimes can get one other model of Chinese soundpowered phones, but he doesn't advertise them since he only gets limited quantities. He sold me a set of them to try, Model SH-091A which tested at -130 dBm. Perhaps you could ask Ming if he has any currently available, and if so, what the price is.)

Perhaps I should say something about the units I'm using for sensitivity. "DBm" stands for "decibels referred to one milliwatt across 50 ohms". 50 ohms is the standard impedance for test equipment in non-telecommunications industry (75 ohms is the standard for telecommunications). The larger the number, in negative terms, the lower the signal from the generator to give the sensitivity in my tests. So, -135 dBm is 5 decibels lower (for my USI phones) than the next most sensitive phones, -130 dBm for the SH-091A phones. For power measurements in decibels, every 3 decibels (dB) change, the power is cut in half (or doubled if going up in power). So ideally for the 1AD you want to use the phones that are the most sensitive.
But in the real world of LF, the external noise is high even if you happen to live in a location of low AC noise, and my most recent tests have been to see if one of Ming's readily available phones would work about as well as my more sensitive USI phones.

Something else to consider is one needs to use an audio transformer with soundpowered phones, in order to match their relatively low impedance to the higher impedance of the output signal at the drain of the Mosfet. That's what the Bogen T725 transformer that Mike uses does, and the transformer that Roelof uses. But one of the phones Ming offers is a non-soundpowered type that already has relatively high impedance, the 4400 ohms phones. Using that, no audio transformer is needed, which simplifies the 1AD circuit, as long as these 4400 ohms phones are sensitive enough with the 1AD. I've been testing this 1AD configuration for the past 5 days, on real NDB signals from my antenna and believe that these phones are sensitive enough. So using them, one takes the output at the Mosfet drain and 0.001 uF bypass capacitor directly to one of the 4400 ohms phones leads and the other phones lead goes to the +9 volts (where the top of the Bogen T725 transformer would normally go).

I've been using my active whip antenna with my 1AD and that works very well. Roelof uses his standard Mini-Whip with his own 1AD. Mike uses a high impedance wire antenna for all his 1AD listening and has his own tuner he's developed to feed his 1AD consisting of a series inductor and variable capacitor with the antenna link on the 1AD. The inductor (about 2600 microhenries) has one end to the antenna and the other end to the 1AD antenna link with the other end of the antenna link to the series variable capacitor (about 1000 pF) with the other side of the capacitor to ground.

Scott Lowe's webpage has a bunch of electronics parts stuff. My own 1AD is currently using a custom two-section high capacity miniature polyvaricon variable capacitor that he offers. He has a knob to go with it as a separate item. I'm using the switched capacitor arrangement that Roelof shows in his diagram and the same fixed capacitors. The idea is to separate the NDB band into two sections and to reduce how far up in frequency the tuning goes. If you use a similar capacity variable capacitor, the two fixed values (300 pF and 100 pF) work well. You want some overlap in tuning for the two switched sections. My own 1AD has about 8-10 kHz overlap. You don't want "underlap" which means the lower section stops, say at 310 kHz, and the upper section starts at say, 315 kHz, which would leave a gap of 5 kHz where you couldn't tune the radio in that section. The Scott Lowe polyvaricon is labeled as "dual 335 pF and 20 pF plastic variable capacitor" that I'm using. He includes a sheet for the pinouts of the various sections. We only want to use the two 335 pF sections. I find this capacitor is (just barely) adequate for use in the 1AD, with the dual section switching arrangement. By that I mean I can turn the shaft and get enough resolution to smoothly tune in fine increments without jumping over frequencies. A larger knob than is offered would help, but the "larger knob" option that Scott Lowe offers (it's not really a large diameter knob!) works. It takes some careful tuning with thumb and forefinger but I don't have any major problems tuning. Of course a "proper" much larger variable capacitor, especially with a vernier drive, would be far more preferable (like Mike and Roelof use). But I'm suggesting that one could build a pretty inexpensive LF 1AD radio using readily available parts, using Ming's 4400 ohms phones which don't require an audio transformer.

I used toroid that Roelof gives in his article, FT140-61 with the specified turns. That worked fine. I had a smaller FT114-61 toroid I used and that worked equally well. In North America both these can be obtained from Diz at (He also ships internationally.) I used #24 wire for the main coil on the FT140-61. For the FT114-61 I used smaller #30 silverplated wirewrap wire, with 88 turns, about 9 feet of wire. Regen winding 5 turns; antenna winding two turns. For both, I wound the Regen winding next to the ground end of the primary winding. Of course proper phasing is needed for the Regen winding.

Ming's eBay items can be found by searching for his seller name "mkmak222" in the advanced search options under "seller". The 4400 ohms phones I bought are called "NOS Military 4400 ohm Headset for Crystal radio, ham". The picture shows a banana plug on the end of the cord. He has another 4400 ohm headset also listed but I didn't try that--it has a phone jack on the cord--I don't know if it's the same sensitivity as this one. Current eBay item number for the one I tried is 161409812818; price is $24.99 plus shipping. (He also has a very good ferrite audio transformer with a wider range of taps than the Bogen T725, if you're using soundpowered phones.) Of course you can see all the items he sells by clicking on the "See other items" on any of his eBay ads.

I live about one mile from my local NDB, 296 LGD. It only causes some problems from extremely strong signal strength (which isn't adjustable on the 1AD) within about +/-25 kHz or less. Once the 1AD is set to proper regeneration the Q of the tuning inductor is high enough that outside that tuning range LGD isn't a problem. It can still be heard sometimes way down in the background until you tune far enough away. Even tuned very close to it, just the actual headphone level is the main problem. I have a BCB station about 6 miles away which is quite a bit stronger in actual signal level. That causes some mixing products here and there in the NDB range, but those are on some even 10 kHz frequencies. If you're in an urban area with a bunch of AMBC stations, perhaps you'd have more problems with the 1AD or may need to think about how to maybe add a lowpass filter in front of the set.

One last comment, about winding lots of turns on a toroid. I use a homemade stiff cardboard bobbin that I wind the wire on then use the bobbin to wind the actual toroid turns. This makes for fast winding where lots of turns are needed compared to pulling a long length of wire through the toroid core for each turn, without the bobbin. The bobbin I used is about 4" long x 1/2" wide with a 1/4" notch cut in each end where the wire is wound. (The bobbin width of course has to be narrower than the inside diameter of the toroid and the length is generally several inches or more, depending on size of toroid and length of wire/number of turns to be wound.)

Good luck with making your own 1AD if you decide to do so; I've tried to give some ideas on how you can buy current products that will work for a true 1AD configuration. Please let ndblist know of your results if you build one! "

Steve adds that so far, in just a few nights of casual listening (and under not so good conditions) he has logged 104 NDB's using a small homebrew active whip!

Thanks for all of the great tips Steve.

No comments: