Clifton Laboratories 7236 Clifton Road  Clifton VA 20124 tel: (703) 830 0368 fax: (703) 830 0711



To search within the Clifton Laboratories site, enter your search term below.


Software Updates
Softrock Lite 6.2
Adventures in Electronics and Radio
Elecraft K2 and K3 Transceivers


May 2011 Archives

18 May 2011

From 1922 to 1983, the Bell System Technical Journal presented technical papers from the Bell Laboratories staff. Many papers in network analysis, coding, and telecommunications were first presented in the BSTJ.

Alcatel/Lucent has now put the complete contents of the BSTJ on line at


15 May 2011

I've often wondered how accurately inductors can be wound following typical kit instructions, such as "wind 14 turns no. 24 AWG wire on a T50-1 core."

My curiosity never reached the level where I wanted to wind a couple dozen inductors and measure them, but recently I had an order for eight filters, each with two inductors of the same value. Hence, a ready made opportunity to collect data to partially answer my question.

The inductors have nominal values of 2.209uH and 8.943uH, representing 14 turns and 29 turns, respectively, wound with AWG No. 24 magnet wire on T50-1 (blue) powdered iron toroid cores.

The table below shows the measured value (3 MHz, using an HP 4192A LF Impedance meter) and the error from the target value. (The inductors were subsequently adjusted; these are the "as initially wound" values.) I've sorted the list from smallest to largest measured inductance.

uH Error uH Error
2.034 -7.92% 8.418 -5.87%
2.035 -7.88% 8.477 -5.21%
2.044 -7.47% 8.562 -4.26%
2.096 -5.12% 8.615 -3.67%
2.126 -3.76% 8.683 -2.91%
2.134 -3.40% 8.695 -2.77%
2.141 -3.08% 8.711 -2.59%
2.153 -2.54% 8.813 -1.45%
2.165 -1.99% 8.880 -0.70%
2.167 -1.90% 8.925 -0.20%
2.201 -0.36% 9.051 1.21%
2.217 0.36% 9.051 1.21%
2.223 0.63% 9.081 1.54%
2.226 0.77% 9.103 1.79%
2.229 0.91% 9.111 1.88%
2.254 2.04% 9.154 2.36%

The measured data shows a spread from worst to best of 10% (2uH part) and 8% (8uH part). This leads to the corollary question: how close must the value be to the target?

There's no automatic answer to this question, of course. If the inductor is to be tuned to resonance with a trimmer capacitor, ±10% or even ±20% may be perfectly adequate. When building filters with fixed capacitors, my target is ±0.5% for inductors. This may be overly stringent, but it's not all that more difficult to adjust an inductor to this tolerance than a looser one, such as ±1%. (I measure filter capacitors with a precision bridge and use the same ±0.5% error target.)

If we take ±0.5% as the acceptable tolerance for high accuracy filter work, then two of 16 2uH inductors are satisfactory as wound, and one of the 16 8uH parts are satisfactory. If we relax the tolerance to say ±2%, then about half the parts comply.

But, in the case of the typical kit, the assumption must be, at least for common kits, that the builder does not posses measuring equipment, and that the design must accommodate the expected range of "as built" values. Based on this limited data set, ±10% looks to be a generous estimate of inductance spread, and an argument can be made that perhaps ±5% is reasonable.

Note that the measured inductance values are not centered around the target value, rather both are skewed towards the low side. The reason for this related to the target inductance value and my choice of wire size and spacing. It turns out that in order to achieve the desired inductance, both inductors require the turns spacing to be quite tight, much tighter than normally found after winding. If the target inductance was a bit lower, then one might expect the range of values to be more symmetrically spread about the mean. But, in the general case, it cannot be assured that the required inductances will always be such that a "normal" wire spacing is required. In fact, it will rarely be the case.

One final observation. In theory, turns spacing does not change the inductance of a toroid inductor. In practice, however, inductance depends to some degree upon turns spacing; tighter spacing increases inductance and looser spacing decreases inductance.


08 May 2011

Yesterday evening, I scanned the 283.5-325 KHz DGPS frequencies using a Z1501 active antenna, a Watkins Johnson WJ8711A receiver, E-MU 0202 sound card and Spectrum Laboratory Software. (DGPS can be transmitted over a variety of frequencies, and at various data speeds, but for the purpose of this discussion, I'll use the term DGPS to only mean signals found in the 283.5-325 KHz band with 100 or 200 b/s data speed.)

DGPS is an acronym standing for Differential GPS, or Differential Global Positioning Service to expand the embedded acronym. DGPS stations transmit correction information to increase the accuracy of satellite-based GPS signals, using repurposed marine beacon and GWEN transmitter sites. GWEN, another acronym (Ground Wave Emergency Network), was a packet based military network operating in the US, over frequencies in the same range as now used for DGPS. GWEN was designed to survive a first strike nuclear attack and provide post-attack command and control communications.

An excellent reference on DGPS signals, from a DX'ing prospective, can be found at and a list of active stations, sorted by frequency, can be found at

DGPS is transmitted as either minimum shift keying (MSK), at a data rate of 100 b/s or 200 b/s. Since there's no human readable station ID or callsign transmitted, it's necessary to use decoding software. The first reference mentioned provides a good summary of available DGPS decoding software. I use Spectrum Laboratory, a free program written by DL4YHF, available at

My only hesitation in recommending Spectrum Laboratory relates to its complexity. I've used it for some time now and still find it difficult to operate at times. But, it's an excellent performing package and is worth the investment in time to understand how to navigate around the program.

To determine the transmitting station information, the "Ref ID" data string is used. This is a (as decoded) 3 digit number identifying, along with the frequency, the station location. The second reference provides a list of DGPS stations, with frequency, Ref ID, location and other information.

The image below shows typical DGPS data. Note that following the message type identifier (Msg9 and Msg3 in the image) is a RefID. For the purpose of identifying the station, we may disregard the rest of the decoded information.


I set up Spectrum Laboratory to simultaneously display:
  • Spectrum analyzer type display of receiver audio
  • Waterfall display of receiver audio
  • Decoded data at 100 b/s
  • Decoded data at 200 b/s

Of course, only one decoder will display current data, but with both 100 and 200 b/s decoders simultaneously functioning, there's no need to keep switching between the two possible data speeds when changing frequencies.

A small version of my display is presented below; click on the small image to see the full size version.



The result of my scan is tabulated below. 32 stations, three countries, and 19 or so states. I only looked at even KHz frequencies; many non-US stations operate on frequencies ending with 0.5 KHz, e.g. 286.5 KHz. None of these stations were audible during my scan, but it will be useful to park the receiver on these frequencies overnight, one night per possible frequency, to see what may be captured.
Frequency (KHz) Ref ID Speed Location State Country Date
286 8 200 Sandy Hook NJ USA 7-May-11
288 340 200 Cape Ray NL CAN 7-May-11
289 12 100 Driver VA USA 7-May-11
290 44 200 Penobscott ME USA 7-May-11
291 184 100 Hawk Run PA USA 7-May-11
292 192 100 Kensington SC USA 7-May-11
293 6 100 Moriches NY USA 7-May-11
294 196 100 New Bern NC USA 7-May-11
295 92 200 St. Mary's WV USA 7-May-11
296 312 200 St. Jean Richelieu QC CAN 7-May-11
297 136 200 Bobo MS USA 7-May-11
299 162 200 Sallisaw OK USA 7-May-11
300 318 200 Riveire du Loop QC CAN 7-May-11
301 58 200 Annapolis MD USA 7-May-11
303 46 100 Greensboro NC USA 7-May-11
304 218 200 Mequon WI USA 7-May-11
305 190 100 Dandridge TN USA 7-May-11
306 198 200 Acushnet MA USA 7-May-11
307 131 100 Hagerstown MD USA 7-May-11
308 484 200 Horta, Azores   AZR 8-May-11
309 170 200 Reedy Point DE USA 7-May-11
311 156 200 Rock Island IL USA 7-May-11
312 244 200 Tampa FL USA 7-May-11
313 320 200 Moise QC CAN 7-May-11
314 16 200 Card Sound FL USA 7-May-11
315 338 200 Cape Race NL CAN 7-May-11
316 42 100 Brunswick NAS ME USA 7-May-11
317 144 100 Hartsville TN USA 7-May-11
319 116 200 Detroit MI USA 7-May-11
320 160 200 Millers Ferry AL USA 7-May-11
322 118 100 Youngstown NY USA 7-May-11
324 94 200 Hudson Falls NY USA 7-May-11


03 May 2011

I recently was asked to build a custom filter that:

  • Passed 518 KHz (NAVTEX) with less than 3 dB loss
  • Rejected 567 KHz by 50 dB
  • Passed 1700 KHz with less than 3 dB loss

This is a challenging set of requirements; most importantly a standard band reject filter of reasonable order is not able to pass 518 KHz and meet the 567 KHz reject requirement. I should also add that since the 567 KHz signal is a medium wave broadcast station, a narrow notch exactly at 567 KHz isn't the optimum solution. To also notch the modulation products, the notch should be 50 dB wide ±9 KHz from the 567 KHz center frequency, so the rejection requirement should be understood to pass 518 KHz and reject 558 KHz by 50 dB.

The solution I provided consists of two filters in series. The first is a Z10100A medium wave notch filter. This is a filter I developed for a particular customer a couple years ago but have not made a standard availability product. Anyone interested in one of these filters should contact me by E-mail to discuss availability and pricing.

The Z10100A result is shown below. The ±10 KHz rejection depth is 50 dB and the  ±9 KHz depth is close to 60 dB. At 518 KHz, the notch filter has a loss of 1.5 dB.

The second filter is a modified Z10020B band reject filter, with 518 KHz attenuation target of 1 dB and a 1700 KHz target of 1 dB. There is some interaction when placing filters in series, so the Z10020B has to keep the 518 KHz attenuation low, accepting whatever additional attenuation is provided at 567 KHz. (In this case, the Z10020A adds about 10 dB extra attenuation at 567 KHz.)

The plot below shows the attenuation with both filters in series. All specifications are met and the 567 KHz notch at the deepest point is 75 dB. The ±10 KHz notch is a bit over 60 dB down.

Mechanically, both filters are mounted in identical size die cast boxes. One filter has a female N connector and the other has a male N connector so the filters may be connected to each other without adapters or a jumper cable. (The other connectors are UHF and BNC, as per the customer's request.)


02 May 2011

I've not been able to keep the Updates page updated recently because I've been busy on projects that I can't say much about. I'll try to do better, but there are some things that I work on for OEM customers that are confidential.

02 May 2011

A couple weeks ago, I bought a used book that must have been stored in a damp location, as it had a profound musty odor. Poking around the Internet revealed a fix; put the book in an closed container with a quantity of a strong desiccant, such as silica gel, to draw the moisture out of the pages.

Crystal cat litter is the bulk source of silica gel easiest to find. The crystals are similar in size to rock salt, and at least in the package I bought, have some blue de-odorizing crystals mixed with the white silica gel crystals. The de-odorizing crystals are useful in removing the  book smell.

For a closed container, I used a clear plastic bin with a locking lid, and placed about half a container of silica gel cat litter in the bottom. To keep the book off the silica gel, I used four PVC pipe caps as spacers.

I placed the bin in a position where it is exposed to the afternoon sun to aid in driving moisture out of the book.

After about two weeks, I can see a clear improvement - the musty odor is not 100% gone, but it's significantly improved from when the book arrived. I'll give it another week or so and call it done.