E-probe experiment (30m)

The Suburban Subharmonic Grabber uses a wider than 100Hz range.
Hence, occasionally a WSPR station makes it into the visual
(displayed) spectrum. When I did my E-probe test, under comparably
poor condx on 30m, RW6XC came up, low enough to nicely be seen
on the grabber spectrum. The interesting part, a visual spectrum
of a single station can be compared to the signal to noise ratio
determined by the WSPR software.

Have a look:
Date: 2009-12-09
Station monitored: RW6XC
Power (indicated): 5W
Distance: LN23AS -> JO22DA = 3059km

UTC...SNR
19:58 -20
19:46 -24
19:38 -20
19:34 -19
19:30 -15
19:26 -14
19:22 -12
19:16 -7
19:12 -8
19:08 -7
19:04 -11
19:02 -12
19:00 -19


The other part of the test was, to see if and how many
station are received by this minimal setup. Remember,
this is an E-probe feeding a homemade direct conversion
receiver...

Stations received during 24h using the E-probe:

Date.......UTC...Call...SNR.Loc...pwr.km.
2009-12-09.17:12.W1XP...-25.FN42fo.5.5546
2009-12-09.19:58.RW6XC..-20.LN23as.5.3059
2009-12-09.10:28.RA3ZSE.-19.KO80ws.5.2313
2009-12-09.13:40.OK2SAM..-9.JN99du.1.1008
2009-12-09.13:46.OE3EV..-12.JN88...5..981
2009-12-09.13:42.IQ4DJ..-12.JN54mq.1..955
2009-12-09.09:50.GM4KGK..+1.IO68ve.5..955
2009-12-09.11:32.OK2BUH..+3.JN89oo.5..945

So, second experiment, yes, you can receive stuff with absolute minimal gear...

indoor DCTL & WSPR on 30m

This is what my tiny subharmonic d.c.-rx heard when wired to the indoors DCTL.

Timestamp	Call	MHz	SNR	Drift	Grid	Pwr	Reporter	RGrid	km	az
2009-12-08 03:36	W3HH	10.140158	-24	0	EL89vb	1	PA1GSJ	JO22da	7276	42
2009-12-08 13:40	W1XP	10.140242	-17	0	FN42fo	5	PA1GSJ	JO22da	5546	51
2009-12-08 14:24	RW6XC	10.140147	-19	-1	LN23as	5	PA1GSJ	JO22da	3059	302
2009-12-08 09:46	I0/N2CQR	10.140152	-24	1	JN61fv	0.02	PA1GSJ	JO22da	1283	334
2009-12-08 08:54	F6EZP	10.140189	-20	0	IN93	2	PA1GSJ	JO22da	1029	21
2009-12-08 07:36	OK2SAM	10.140199	-23	0	JN99du	1	PA1GSJ	JO22da	1008	289
2009-12-08 07:30	OE3EV	10.140193	-6	0	JN88	5	PA1GSJ	JO22da	981	298
2009-12-08 07:42	IQ4DJ	10.140228	-5	1	JN54mq	1	PA1GSJ	JO22da	955	331
2009-12-08 09:44	GM4KGK	10.140281	-6	0	IO68ve	5	PA1GSJ	JO22da	955	131
2009-12-08 10:10	OE3WGW	10.140178	+9	-1	JN88cj	5	PA1GSJ	JO22da	937	300
2009-12-08 09:40	IK1ODO	10.140264	-19	0	JN35sa	20	PA1GSJ	JO22da	814	344
2009-12-08 10:34	DL5MHD	10.140212	-9	0	JN58xc	5	PA1GSJ	JO22da	699	311
2009-12-08 10:34	DG7RJ	10.140240	-23	0	JN58th	5	PA1GSJ	JO22da	665	311
2009-12-08 10:46	HB9TMW	10.140144	-25	1	JN36gq	5	PA1GSJ	JO22da	615	345

So folks, here you have it, you can receive weak signal living in a concrete building in a dens suburban environment.

In another experiment, I will use the E-probe, outside at my balcony, stay tuned for the data.

new low frequency ideas part 2 (RX)

There is another added bonus on CB crystals for low frequency operation. At least for the 136kHz band.

A CB-channel 22 TX crystal (27225kHz) would result in a frequency of 138.9kHz, just 1.1kHz off the upper band edge and 3.2kHz off the lower band edge.

Given the fact that the two last frequency divisions (see the earlier blog entry) resulting in 4 (two flipflops) it appears somewhat obvious to use just this crystal in a direct conversion SDR receiver.

new low frequency ideas (TX)

Working with dividers seem to have advantages to me, amongst: greater stability and easy digital design. The only question is, are there enough crystals in a close range?

Well, what about old CB crystals? And how would those get us onto the 136kHz band?

First things first!

Here is what I came up with and how I started the process:

There is an industrial crystal 27000kHz crystal, divide this by 196, get us to 137.76kHz, just 60Hz off the QRSS COA (center of activity).

How to divide by 196, well, that actually easier than I though:

1. 196/2=98
2. 98/2=49
3. 49=7*7

The recipe is hence, two counters to 7 in series (so that they multiply) and two FlipFlops. There we go, a 50% duty cycle with a cheap crystal in the middle of the QRSS range. Drift should not be an issue here, assume, the oscillator drifts by 1kHz, that would result in a 5Hz drift on long wave.

I think, there is even a canned oscillator of that frequency, making the design in particular simple.

And now to the added bonus: CB-XTALs!

I reach in my junk-box to see what crystals I got from old CB radios. Some calculation revealed the following:

1. 26600/196 = 135,71
2. 26610/196 = 135,77
3. 26620/196 = 135,82
4. 26630/196 = 135,87
5. 26650/196 = 135,97
6. 26660/196 = 136,02
7. 26670/196 = 136,07
8. 26680/196 = 136,12
9. 26780/196 = 136,63
10. 27000/196 = 137,76
11. 27005/196 = 137,78

This covers essentially the whole 136kHz band with junk crystals from CB radios. Even better, look at the channel count, to me that looks like I will be using the casing and the channel selector too ;-)

I missed a couple of frequencies, since I got no crystals for those in my junk box.

The internet discloses that there are RX crystals for the CB channels 8 to 40 ranging from 26600kHz to 26950kHz with 10kHz increments, a few channels skip however. 10kHz steps will result in 51Hz steps in the 136kHz band.

Additionally, there are TX crystals for CB radios. CB channels 1 to 4 (26965kHz to 27005kHz) would be interesting here.

One may consider to experiment with VXOs, since old CB radios have those anyway, ranging +/-5kHz, which would the cover essentially the whole band, with just a few gaps due to the lack of crystals.

The same principle could be serving for the 500kHz band. Here, one would require a division by 54, which would be two counters 9 x 3 = 27 and one flipflop => 54. CB TX crystals from upper channels could be used. Since I don't hold a licence to that band, I will not look any deeper into this. Nevertheless, I hope, that someone finds this useful.

80m results (06.12.2009)

Colin and I did some tests on 80m lately. The following spectrum shows the final minutes, before Colin QSYed. Receiver setup on my side: NASA Target HF3 + E-probe. The computer used for analysing is a Packard Bell "dot" netbook with an Intel Atom N270 processor.

The darker regions on the spectrum are due to a local (The Hague) sideband station pulling the AGC to the max. But still, there is some F1A (FSK-CW) coming through... The cheap'n easy setup still allows for detecting weak signals...

136kHz ideas

Some considerations to generation of frequencies in the region of the long wave band using counters and cheap crystals. The first number is the resulting qrg in Hz, the second number is the crystal frequency in MHz and the third number is the divider/counter.
Additionally, there are some approaches mixing different frequencies, I will not go into this in the present note.

The trick is to end the chain of dividers with a division by two, creating a 50% duty cycle.

I would like to point your interest to some frequencies for which ceramic resonators are available. Those could be nicely pulled, given the ratios, the stability of a ceramic resonators oscillator seems OK-ish to give some results still. Assume 1kHz drift @ 5.5MHz. This would resulting in a 25Hz drift @ 136kHz, not desirable, would allow for first tests however. It is clear that care should be taken to avoid drift all together.
For crystals, one could consider moderate super-VXOs.

At the end of this message, a band plan is attached. Some frequencies I listed are actually outside the band, consider this as a hint to SDR-receivers, in particular when the last division could be by 4.

There are two candidates for QRSS marked in red. However, both require a division by 13 :-(

Simple solution "ripple counters"

• 138550 = 4.433619 / 32 (=8x4=2x2x2x2x2)
• 138550 = 8.867238 / 64 (=8x2x4=2x2x2x2x2x2)

One decade counter + ripple counter or flipflops

• 136533 = 9.8304 / 72 (=9x2x2x2)
• 137500 = 22.000 / 160 (=10x8x2)
• 138240 = 22.1184 / 160 (=10x4x4=10x4x2x2)
• 136533 = 4.9152 / 36 (=9x2x2)
• 138888 = 5.000 / 36 (=9x4=9x2x2)
• 138888 = 10.000 / 72 (=9x2x2x2)
  
• 144444 = 5.200 / 36 (=9x4=9x2x2)
• 136533 = 6.5536 / 48 (=6x2x2x2)
• 136533 = 7.3728 / 54 (=9x3x2)
• 136533 = 14.7456 / 108 (=9x3x2x2)
  
• 136533 = 16.384 / 120 (=10x3x2x2)
• 136550 = 2.4579 / 18 (=9x2)
  
• 136533 = 24.576 / 180 (=10x9x2)
  
• 136533 = 3.2768 / 24 (=6x2x2)
  
• 136533 = 3.6864 / 27 (=9x3)
• 136533 = 4.096 / 30 (=5x3x2)

Two decade counters + ripple counter or flipflops

• 136364 = 3 / 22 (=11x2)
• 136364 = 6.000 / 44 (=11x2x2)
• 137675 = 3.579545 / 26 (=13x2)
• 137675 = 14.31818 / 104 (13x2x2x2)
  
• 137255 = 14.000 / 102 (=17x3x2)
• 136364 = 15.000 / 110 (=11x5x2)
• 136364 = 18.000 / 132 (=11x3x2x2)

Ceramic resonators

• 137500 = 5.50 / 40 (=10x2x2)
• 138889 = 5.00 / 36 (=9x4x2)
• 138889 = 10.00 / 72 (=9x8x2)
• 136389 = 4.91 / 36 (9x2x2)

Now, what to select, taking into account the band plan:

135.7 - 136.0 kHz

Station Tests and transatlantic reception window

136.0 - 137.4 kHz

Telegraphy

137.4 - 137.6 kHz

Non-Telegraphy digital modes

137.6 - 137.8 kHz

Very slow telegraphy centred on 137.7 kHz

Grabber news: WSPR2 works

The new version of WSPR allows to set a RX BFO other than 1.5kHz. The usable range stops at 3kHz, just enough to use the subharmonic direct conversion receiver for spotting WSPR signals.

Assume the following: The mid-qrg of the WSPR band on 30m is 10140.2kHz. With a "BFO" setting of 1.5kHz, that gives the famous 10138.7kHz. The subharmonic receiver's LO runs on about 5068.8kHz (depending on the individual canned oscillator). Double that, it will give 10137.6kHz, resulting in a "BFO" frequency of 2.6kHz to reach the WSPR-band.

My resulting LO frequency is a about 69Hz further off, here comes a correction in, one has to measure first. For me, it results in a "Dial Frequency" of 10.137531MHz and a "Rx BFO" of 2669Hz.

Good news for WSPR-users, I will be now monitoring whenever the grabber is set to 30m.