Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Eloran frequency standard

Status
Not open for further replies.

dr pepper

Well-Known Member
Most Helpful Member
A while back I posted some comments on a frequency standard I built using Msf, the Uk version of Wwvb.
The machine works well, and is as accurate as could be expected, after sundown accuracy drops off a little due to ionosphere reflections, and can be a uS or so.
Loran c used to transmit short (200uS) pulses of 100kc, and the rx just 'listened' to the first few cycles of each pulse, the principle reason being that the first few cycles were not affected by atmospheric reflections, and therefore gave very accurate timings & good position fixes.
In the Uk loran has been revived and is now transmitted again, and is very accurate, so I'm toying with building a better frequency standard, not being as affected by atmospherics I should be able to make an improvement on accuracy, or rather be able to use the thing at night time.
I found someone on a Uk forum that is also working on this, he uses a microcontroller, samples the signal & sync's the micro's clock using the osccal register, pretty clever.
I found lots of info on loran c, but not so much on eloran, eloran I believe to be compatible with loran c receivers, the 100kc carrier on loran c was not spec'd to be accurate, eloran I think is however I'd like to find some spec on this.
 
Ok, not so many have messed with this then.
My first attempt was to bung an amp on a tuned loopstick to see if I can get loran, and I can easily, I get a nice large signal.
Next up, using the 100kc waveform is going to be tricky without high speed stuff, so I'm going to put together a simple trf receiver using a zn414 and a tuned loopstick to pick up loran, this will give me just the envelopes of the loran pulses, the timing between 1/2 way up the rising edge and the peak of one envelope and the same on the next is precisely 1mS, so using a processor with a basic A to D I could easily generate 1000 interrupts per second, and use this to control a dac which in turn controls an vcoxo, in fact it wouldnt be hard to generate 1000 pulses per sec with a analogue circuit.
Using the 1/2 way point on the rising edge removes a lot of skywave time/phase distortions and makes an accurate signal as used for navigation, this might work better with a dc coupled rf amp, the zn414 has internal ac coupling, but I'll give it a go and see I dont think there will be a lot of difference over a few degrees C.
 
Last edited:
i donno if this fits the scope here . . . but i doubt or wonder the x-tal osc being operated at their resonant frequency with the circuits the web can find
  • if you start pre phase excite 'em ? then how you tell if their at resonance (the CMOS osc. circuits are basic simplistic with no t° or any other compensation nor nifty sense/control features) ??? -- i assume all such X-Tals work at below their resonance
  • i mean there is a contradictory -- you need to give the X-tal the energy to keep it oscillating by that disturbing its phase/operation -- by simply electrically reading it's signal you extract off the energy and disturb it's operation/phase
  • the only trivia here i can see is to map/chart the energy/frequency dE.in dE.out for the extremely stable X-Tal driver apparatus, find it's "accurate" op. point by comparing it to external more accurate clock . . . then preset the ESxtD to that -- why we need an x-tal here when the stable oscillator can be set accurate(/stable osc. f.) and by some other "condition" isolated physical system
  • more likely the question is -- if the X-Tal res. is 4MHz and in simple CMOS osc. driver it is most stable @ say 3.9988MHz -- then this fact? is never pointed out in any tech. documentation of devices using that clock ???
 
Not sure I get you on those points, but I'll make these statements.

A while ago I built a standard using a oscillator that didnt have quite enough gain to oscillate, its a 60kc xtal, then 60kc pulses from Msf were received amplified and loosely coupled to this 'oscillator', the xtal rang and bridged the gaps in the 60khz pulses of the rf signal by continuing to ring, this was then used to phase lock to a ocxo by dividing the ocxo to 60kc.
It worked but skywave reflecions messes up the Rf signal from Msf and causes more than 1 cycle error at 10Mc at night or worse on sundown.

The skywave reflection if you are reasonably close (in the same country) is later than a few uS single or double skip, so if you use only the first few cycles, or the first part of the rf envelope then skywaves dont have much effect on the signal, so if you sync an oscillator on the the first few cycles, or rectify the signal and create a pulse as soon as the signal is detected you get a very accurate signal without skywave phase delays, using the rf wave obviously gives 100kc, using the envelope gives you 1000 pulses/sec as loran transmits pulses with 1mS between.
It just so happens loran syncs the rising through zero point on the third cycle of each pulse with utc time, this zero cross is 1/2 of max amplitude, I think they did this to easily make simple receivers accurate.
My receiver misses the first pulse or so in the noise on the floor, so detecting it as soon as I can gets me close to this ideal 3 rd zero cross.

Using a micro I can ignore spurious pulses effectively, meaning accuracies of less than a microsecond is practical, pro receivers manage 200 picoseconds, as accurate as that cesium resonator on the side of the hill at Anthorn.

If I were to use digital signal processing techniques I'd be able to completely recover all of the signal as its repetitive however I dont really have the technology or time to do that.
 
Last edited:
The skywave reflection if you are reasonably close (in the same country) is later than a few uS single or double skip, so if you use only the first few cycles, or the first part of the rf envelope then skywaves dont have much effect on the signal, so if you sync an oscillator on the the first few cycles, or rectify the signal and create a pulse as soon as the signal is detected you get a very accurate signal without skywave phase delays, using the rf wave obviously gives 100kc, using the envelope gives you 1000 pulses/sec as loran transmits pulses with 1mS between.
First, I am excited to hear about ELORAN and will follow your progress with interest.
Second, what you describe is how I tackled WWVB. First, I set the clocks (nearest second) with the whole signal. Then I simply sync with the first state change.

I am about 6 milli-light-seconds from Fort Collins (WWVB) and debated whether to set my clocks to Fort Collins' time or correct for the time difference in Cleveland. ;)

John
 
That would done right be the most accurate, only the first few cycles are not affected too much by propagation.

After swotting how loran combats propagation effects I realized this could be done with Msf, our version of Wwvb, allthough I suspect ours is different, Msf commutates the carrier, whereas Wwvb I believe is either phase modulated or carrier amplitude keyed, dont spose that matters for your app.
So a 1 second pulse could be derived from Msf more accurately than just syncing an oscillator to the 60kc carrier like I'm doing with my previous project.

I used my Gps frequency source and my usb 'scope to get an image of the loran signal from my crude loopstick front end on the bench, you can see there is some visible envelope distortion, the tone burst is sposed to have a nice raindrop profile like it does in the memory buffer display at the top, but in real life zoomed in you can see its all messed up.
Even with the loopstick in a 'quiet' area its still messed up, loran receivers at least later more accurate ones must use multiple samples and digital filtering to get around this phase distortion.
I had issues with feedback derived oscillations, so I need to build the preamp into a screen, I have some old tube radio If tranny screens, they'll do.

Loran.png


The issue here is as you can see the loopstick rings at 100kc, I would have liked to detect the loran signal by timing zero crosses, the first 2 5uS crosses denoting the first invomming wave, but with the loopstick ringing at its tuned freq I cant do this as theres always some 100kc signal, so maybe a whip antenna would be better or an untuned loop like the trask.
 
Last edited:
Ok I demodulated the loran signal, amplified it then stuffed it into a comparator, so now I get digital pulses on every loran toneburst.
I tried some idea of using software to detect the edges and make timing corections, but with a clock speed of 16mc thats not going to work.
So I'm going to use a software/hardware approach, maybe a gated Pll, compare divided down 1kc pulses from an xtal osc to the loran pulses when they occur at 1000uS apart, then disable the Pll loop filter input when the pulses are not 1000uS, so I only phase lock on the 8 pulse tonebursts (which is pretty much what loran is) and wait inbetween.
This just requires say a 'duino to 'know' where abouts the master & slave pulsetrain sequences are and switch a Pll enable o/p on & off just before & just after the 8 tonebursts, the fact the master has a 2mS gap on its last pulse should make that easier, and the master & slave delays are constant.
 
Loran plot.png
Ok then, learned a bit with 'duino's and the internal timers.
Using timer1 configured to capture mode on pin 8, and being clocked by an external clock on pin 5 I did the following graph using the 'duino's serial plotter, the 'clock' didnt come from an ocxo, for now its comming from my sig gen set to 100kc, and the capture i/p is connected to the 1mS output pulses from the loran receiver.

What your looking at on the plot is the counts between capture or loran pulses, you can see the 1mS pulses, and the 20mS pulses between master & slave as well as the repetition time or as loran call it the Gri which shows on the plot just under 40mS.
The end design will have a 10Mc vcoxo instead of the sign gen at 100Kc, maybe it will be possible to use all the time periods to sync to not just the 1mS ones, I'm going to use a long variable so max count time will be 10e6 / whatever the max value is for an unsigned long, so maybe even the long durations will be useable.
Its hoofing down right now so the signal is particularly noisy, as you can see on the plot...the rain's just subsided and the signals much better so it is affected by weather.
The drops to zero is due to my inability to get a schmitt trigger to work, I'll address that more next time.
 
Last edited:
I'm having difficulty with reception.
Because the loopstick is tuned and a fairly high Q it rings, this causes delays, and also instead of a signal suddenly appearing it kinda slowly appears from zero to max amplitude, well actually not zero, there is noise present all the time at the loopsticks resonant frequency.
So I think I need an E field antenna rather than B field, and a means of selectivity that doesnt cause to many delays.
So I'm thinking divide a frequency down from the xtal osc and use that to form a superhet receiver, maybe I can get close to 455kc, perhaps 500kc and use standard coils, then hopefully I can detect the incoming signal on the first wave.
 
When I was building my WWVB set, most of the work (i.e., all that I could find) used magnetic loop antennas -- either regular loop or shielded loop.

Here is a nice reference from Bertus aac: https://web.archive.org/web/20050504190334/http://www.longwave.de/smlloop.pdf
There are many more, including on the ARRL site that I do not have free access to. The antenna I eventually used was copied from here: https://www.febo.com/time-freq/wwvb/antenna/

I just used PVC tubing for my support:
upload_2017-9-14_10-0-14.png



And, here's the tuning box:
upload_2017-9-14_9-58-36.png


John
 
I'd not considered an untuned mag loop thats interesting.
The bertus design has a Q of less than 1, sounds pretty good for what I'm looking for I'll give that a try, and I might try and improve the noise situation by adding a centre tap and configure the op amp as an instrumentation amp.
I get the fact that a low input impedance amp counteracts inductance in the loop, however I cant help but think this is operating like a folded dipole with many folds, but if it works, it works.
Come to think of it I recall marine loran antennas being a double loop, one 90 degrees rotated to the other looking down the vertical plane, the 2 probably just for non directionality.
Thanks for that.
 
I am more familiar with ADF receivers. The magnetic loop is directional and the "sense" antenna (E field) is used to resolve the 180° ambiguity based on phase. For my purpose, I didn't need that resolved. Hence, I just aligned the plane of the antenna with the known direction to Fort Collins, Colorado (location of WWVB).
 
Ok maybe the 2 loops 90 degrees rotated were not loran then.
Just had a look at some old loran antenna pics, 2 loops like a football with a vertical on top.
Yep I just need the single loop, might use tri rated insulated, or might use magnet wire, I think I have a bucket lid somewhere to use as a former.
I found i got some fake Lm358's a while back, just tried some Ne5532's I got at the same time, no surprise these are fake too, I htink they are 4558's.
 
Last edited:
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top