Topic: My # 671 Blitzortung Unit Quality --->Goldilocks?

[schwab]

The number of USA Blitzortung Network participants is growing and so is the quality of the the nationwide lightning detection network.

How does any given participant best measure the quality of lightning detection given our dependence on one another's signals?

Some units have a very high number of strikes, others are low and my unit #671 is usually somewhere in the middle....i.e. Goldilocks??

How can we as individuals tweak our systems to maximize performance of our units and therefore the USA network?

I just do not know what measurement of all the data is the most important.

I'll appreciate any advice!

[miraculon]

The number of USA Blitzortung Network participants is growing and so is the quality of the the nationwide lightning detection network.

How does any given participant best measure the quality of lightning detection given our dependence on one another's signals?

Some units have a very high number of strikes, others are low and my unit #671 is usually somewhere in the middle....i.e. Goldilocks??

How can we as individuals tweak our systems to maximize performance of our units and therefore the USA network?

I just do not know what measurement of all the data is the most important.

I'll appreciate any advice!

This is all a compromise between 'interference mode' during frequent/intense lightning and range. If you have worked this out, I would say that your "Goldilocks" settings are pretty good.

Otherwise you will wind up chasing gain settings. With the "Green" boards taking the receiver box apart to reset the jumpers is too much of a chore. That is one advantage of RED, you can change the gain with software settings. I just wish that they would get auto gain control up and running. If my ambient noise is low, I can crank up the gain and get phenomenal range. If a neighbor turns on the interference source (whatever it is), I have to back down.

By the way, the waveform plots on LightningMaps.org now say "PCB GREEN". Your terminology has caught on!

Greg H.

[dfroula]

I like to use the "Strike Ratio" as a measure of performance. This is the ratio of the strikes your station was involved in computing to total detected strikes of the system. To make matters confusing, the strike ratio at various distances from your station is called "Efficiency" on blitzortung.org and Strokes/hr Ratio on lightningmaps.org.

Whatever it's called, it seems the best overall indicator of system performance.

The next best is what is called "Efficiency" on lightningmaps.org. This is the square root of the product of your strike ratio and signal ratio. (Signal ratio is the ratio of received station detections versus total detected system strikes). This number factors in false hits from noise as a measure of efficiency. However, it can also give a lower score if your station is picking up real strikes that can't be located because not enough other stations see them. I don't like the metric as much for that reason, as you can be penalized in the score for really good or really bad operation. I don't think the signals ratio matters much, if you're still contributing to 99% of all detections!

Confusing, no?

Best,

Don

[schwab]

It's been about two months since I first started this thread regarding any given Blitzortung station's quality.

It remains confusing in my opinion.

It seems that the number of signals detected is important for quality by Blitzortung.com even though there's just no way all those signals are actually leading to strike detection or "effecitivity"(needs spelling change).

http://www.blitzortung.org/Webpages/index.php?lang=en&page=4&region=3&subpage_0=30&sort_column=5000_km_60_m_e&sort_order=desc

On the other hand Lightningmaps.org has strokes/h and signals/h and  efficiency ratings.

http://www.lightningmaps.org/blitzortung/america/?bo_page=statistics&bo_show=network&lang=en&bo_station_id=22

The RED 10.3 systems appear to be detecting more signals than the GREEN 6.8 systems

However, it is not clear to me if detecting more signals via RED 10.3 systems is actually leading to better lightning detection.

In addition the lightning strikes in any given area is always changing as do the the "quality" measures on these sites.

Any user opinions/recommendations regarding how to best statistically measure quality of any given Blitzortung.org system?

Before I try to tweek my system I will await that measure in order to compare my system to my colleagues on the network.

I am trying to achieve that Goldilocks system...ie. "Just Right" - not too sensitive - not too insensitive.

[dfroula]

Use the 24 Hour average strike ratio on lightning maps.org as your basic performance  metric. Set your gains high enough so that only thunderstorm activity within 300 miles of your station location forces you into interference mode. The overall lightning activity in the continent will affect this somewhat. Avoid the temptation to ride the gain control continuously to get a small increase in the ratios or to stay out of normal interference mode.  This just confuses your estimation of system performance. Set it and leave it a few days before making any changes.

If overall lightning activity is low there can be large swings in the strike ratios as lightning activity changes location. Meaningful adjustments need to be made when there is a moderate to high level of afternoon activity with normal early morning minima.

Don

[Cutty Sark Sailor]

I can only say that my area NWS office contacted me about a stroke inquiry recently. The forecaster (a friend) commented that it appeared our system had detected strokes their system missed. Indeed, I/We had recorded those strokes, and in one instance our deviation was 0.3km, the rest within a couple.  I found the strikes almost immediately and provided him with a google image, waveform, and time / location of one, just to be a SA Braggart, and suggested he check with the farmer the stroke likely affected for more confirmation  
Since then, I see several hits a week on my Blitrzortung pages from a couple of NWS offices... which only proves that no one is 100% and all resources are valuable. And each station is an additional resource.

For example: Today is a slow day, sferics-wise. I was riding a high efficiency on LD until all those cells fired up in Colorado on the Western Range, and in Eastern Utah. Man, lots of activity. My efficiency almost immediately dropped. They're west of me, harder to detect in a similar latitude line, and stacked up behind each other (My theory of 'Thor's Hamma Slamma Jamma" is working overtime for this direct W-E line.) But the rest of the net is getting 'em!  That's the beauty of this thing! And the reason we need more stations!

A very generalized assumption: From what I gather, expectations for any LD detector are between,say for e.g. 30-50%, depending on conditions such as time of day, etc, and any detection efficiency over 50-60% in distances >500-800km for any given station of whatever type, in whatever network, is outstanding, and well above average.

[dfroula]

In the absence of any local noise issues, I tend to ignore the efficiency ratios (ratio of detected system strokes to total signals received by a single station). I am quite confident I am detecting real lightning that never gets correlated with other stations. I have read a few papers on commercial systems that claim the average efficiency ratio for a system averages about 33%.

The ratio of strikes my station contributes to, to the total number of detected system strikes seems a much better estimate of my individual station performance.

Don

I can only say that my area NWS office contacted me about a stroke inquiry recently. The forecaster (a friend) commented that it appeared our system had detected strokes their system missed. Indeed, I/We had recorded those strokes, and in one instance our deviation was 0.3km, the rest within a couple.  I found the strikes almost immediately and provided him with a google image, waveform, and time / location of one, just to be a SA Braggart, and suggested he check with the farmer the stroke likely affected for more confirmation  
Since then, I see several hits a week on my Blitrzortung pages from a couple of NWS offices... which only proves that no one is 100% and all resources are valuable. And each station is an additional resource.

For example: Today is a slow day, sferics-wise. I was riding a high efficiency on LD until all those cells fired up in Colorado on the Western Range, and in Eastern Utah. Man, lots of activity. My efficiency almost immediately dropped. They're west of me, harder to detect in a similar latitude line, and stacked up behind each other (My theory of 'Thor's Hamma Slamma Jamma" is working overtime for this direct W-E line.) But the rest of the net is getting 'em!  That's the beauty of this thing! And the reason we need more stations!

A very generalized assumption: From what I gather, expectations for any LD detector are between,say for e.g. 30-50%, depending on conditions such as time of day, etc, and any detection efficiency over 50-60% in distances >500-800km for any given station of whatever type, in whatever network, is outstanding, and well above average.

[scarecrow93]

It's been about two months since I first started this thread regarding any given Blitzortung station's quality.

It remains confusing in my opinion.

It seems that the number of signals detected is important for quality by Blitzortung.com even though there's just no way all those signals are actually leading to strike detection or "effecitivity"(needs spelling change).

http://www.blitzortung.org/Webpages/index.php?lang=en&page=4&region=3&subpage_0=30&sort_column=5000_km_60_m_e&sort_order=desc

On the other hand Lightningmaps.org has strokes/h and signals/h and  efficiency ratings.

http://www.lightningmaps.org/blitzortung/america/?bo_page=statistics&bo_show=network&lang=en&bo_station_id=22

The RED 10.3 systems appear to be detecting more signals than the GREEN 6.8 systems

However, it is not clear to me if detecting more signals via RED 10.3 systems is actually leading to better lightning detection.

In addition the lightning strikes in any given area is always changing as do the the "quality" measures on these sites.

Any user opinions/recommendations regarding how to best statistically measure quality of any given Blitzortung.org system?

Before I try to tweek my system I will await that measure in order to compare my system to my colleagues on the network.

I am trying to achieve that Goldilocks system...ie. "Just Right" - not too sensitive - not too insensitive.

I'm sure you're seeing that each one of us has a reason for their measurement methodology.  FWIW...here's my 2 cents.

I'm currently still slowly, methodically, testing my RED station.  I've noticed, at least with my setup, the best quality range I get with my system is in the neighborhood of 2000km.  No matter how high I set the gain to get to the "Long Range" Effectivity of 5000km on the chart(http://www.blitzortung.org/Webpages/index.php?lang=en&page=4&region=3&subpage_0=30), it only results in more signals sent with no increase in detection. 

To determine how well my site is doing, I use the Overview tab at the top of the Blitz page (http://www.blitzortung.org/Webpages/index.php?lang=en&page=1&region=3&subpage_0=30), set it to only display strikes in the last 20 minutes and the counter boxes.  I then toggle on/off my individual station and see how well my station is doing versus the strikes detected by the network.  That's how I came to determine my ideal range.

As far as monitoring, I use the charts on lightningmaps.org (http://www.lightningmaps.org/blitzortung/america/index.php?bo_page=statistics&bo_region=dist2000&lang=en).  I just select "max 2000km" and use the numbers displayed on the top left of the chart to determine my effectiveness for that range.  I set it for "min 2000km" to see how it does outside of my target range.

My testing objective is 90%+ detection inside of 2000km with minimal erroneous signals sent to the server.  My definition of "Goldilocks".   

Lance
Huntingtown, MD

[schwab]

Our Upper Midwest USA area has had very active lightning in the past few days.

I am still struggling with understanding my 6.8 or GREEN system's lightning strike location quality vs. the new 10.3 or RED systems' quality.

In general the 10.3 RED systems are detecting many more signals and strokes...up to 5-10 fold more.

The best data demonstrating this may be the Livonia, Michagan station that has both 6.8 and 10.3 systems running simultaneously.

Doesn't the network server require a minimum number of signals from different stations to locate a lightning strike?

How can a select few 10.3 Red stations with the highest signal and stroke counts also have the highest "effectivity" and "efficiency" when no other stations on the network approach similar signal and stroke counts? 

These high count stations just cannot be contributing to locating many more lightning strikes can they?

In my opinion the 10.3 RED stations are definitely sending 5-10 fold more signals to the network server, however, many of those signals must be "non-locating" signals since the other stations on the network are sending many, many fewer simultaneous signals.

The Blitzortung and LightningMaps maps of lightning strikes are very accurate in my experience BUT both the Blitzortung "effectivity" and LightningMaps "efficiency" in lightning strike detection, in my opinion, need review and possibly new formulas as measures of quality.

[miraculon]

The best data demonstrating this may be the Livonia, Michagan station that has both 6.8 and 10.3 systems running simultaneously.

Two "words of caution" about my station.

• The 6.8USB is more susceptible to noise than the "Red" system. I suspect that this is due to location. The "Green" antenna/amp is in an upstairs closet on the north side of the house. The "Red" is in the garage rafters on the south side. Due to this noise, I have needed to adjust my threshold (I modified the circuit) to get the system out of interference mode during moderate noise. This will have the effect of making the system less sensitive. Sometimes I get severe noise that I cannot even adjust out with maximum comparator thresholds. There are times that both detectors pick up the same noise. I suspect that a local hospital is running their MRI, or there is some other large industrial motor drive in operation.
• The other thing is that the antennas are different. "Green" has ferrite rod antennas, I believe that they are 20cm. The "Red" has 1M loops with 9 turns of wire in each. Even though I added the damping resistors and opened up the filter bandwidth on Green, the "Red" waveforms look like they have a faster response than "Green" which look more rounded in comparison.

I have some suspicions, but no data, that there is a phase delay in the "Green" system. I don't know if it is due to circuitry or the ferrite rod antennas.
I notice that there are strikes that the "Green" will "participate" in and the "Red" will not, and vice-versa. Some large strikes will show up in both.
I also tend to get a lot of "signals", but few strikes. I see a lot of non-noise yellow LED activity without being participated with other stations on the "Green".

Greg H.

[Jumpin Joe]

I just switched, yesterday, from the ferrite rods antenna's that Egon sells to the 12" Ball Antenna that Lance "scarecrow93", Huntingtown, MD created.  My percentages went from an average of 30-40 to high 80's into the 90's.

I believe the reason we are seeing significant differences with the RED systems is the antenna that's being used. I bet if everyone responds with what they are using, a pattern will develop pretty quickly for the most part.

Just my two cents worth.

Regards,

Joe

[Cutty Sark Sailor]

I just switched, yesterday, from the ferrite rods antenna's that Egon sells to the 12" Ball Antenna that Lance "scarecrow93", Huntingtown, MD created.  My percentages went from an average of 30-40 to high 80's into the 90's.

I believe the reason we are seeing significant differences with the RED systems is the antenna that's being used. I bet if everyone responds with what they are using, a pattern will develop pretty quickly for the most part.

Just my two cents worth.

Regards,

Joe

Ok: Station 689 Frankfort KY... As I type this, both channels are at 10*8 at 120mv. I generally run 10*10* at 120mv threshold, and at night 5*5 at 120mv or possibly 4*8. I'll go to 10*16 one one or both channels if conditions are quite around here. When we get more western stations online, I probably won't go that high... let them grab the strokes.  
Using home made DonF ferrites, 6 rod (300mm) as opposed to his original 5 rod (250mm)... unshielded.  Noise is better, interference a bit more correctable, while the gain is about 2-3 times that of the 120mm kit ferrites I had to run shielded. Lower noise floor may be because of thinner rod and more windings. Even a 'new' ???? intermittent interference signal around 55-56KHZ is controllable below threshold, while still getting adequate gain.  

Because of some failing sodium lights I can't get power company to replace,   I have to cut gain back at night, but still get more 'efficiency' than I did with the other ferrites, and definitely the flat panels picked up everything. That is, assuming I remember to back the gain down.... if I don't, I'll be cut off large parts of the night and dip well below 40-60% on LR. Other wise, were in good shape on 'efficiencies' most of the time...whatever the heck that says.  

I'm considering winding a couple of 350mm or 400 mm as soon as my wrist heals from winding those two a few weeks ago. I've retired my big flat panels, and the kit 120mm ferrites. These homemades just work too well. Maybe I got lucky  

Oh, an aside... I suspect many of the 'higher performing'   Reds are centrally located, also. They're kinda destined to pick up more signals in 360° touching base with more stations....

Need More Stations!!!!

Cheers!
Mike

[W3DRM]

...snip...
Because of some failing sodium lights I can't get power company to replace,   I have to cut gain back at night, but still get more 'efficiency' than I did with the other ferrites, and definitely the flat panels picked up everything. That is, assuming I remember to back the gain down.... if I don't, I'll be cut off large parts of the night and dip well below 40-60% on LR. Other wise, were in good shape on 'efficiencies' most of the time...whatever the heck that says.  

Mike,

Contact your power company and tell them that if they don't correct the noisy lamp problem, that you will make an official report to the FCC (after the Government gets back in business of course...) regarding their lights generating interference.

That has worked here in Nevada several times. All it took was a letter to NVEnergy stating they had equipment that was interfering with radio reception. They immediately responded and sent a crew out to make the corrections. Emails and phone don't seem to work as well as sending them a letter.

You mileage may vary of course but it's worth a shot to see what kind of response you get.

The FCC has an interference reporting section specifically for these kinds of issues. Do you hear any interference from the lights on your AM radio (try tuning off-station). If so, that's even more evidence there is a problem.

[Cutty Sark Sailor]

Oh, I'm not pushing, right now. Several reasons. Yep, interference on AM... and I'm aware of the FCC stuff...there's 4 of 'em causing it -- all old. There's more to the interference story, though. PM me if you want to hear it... A friend says I oughta wait until it's quite cold, freezing, and they completely fail, THEN ... 

[scarecrow93]

I just switched, yesterday, from the ferrite rods antenna's that Egon sells to the 12" Ball Antenna that Lance "scarecrow93", Huntingtown, MD created.  My percentages went from an average of 30-40 to high 80's into the 90's.

I believe the reason we are seeing significant differences with the RED systems is the antenna that's being used. I bet if everyone responds with what they are using, a pattern will develop pretty quickly for the most part.

Just my two cents worth.

Regards,

Joe

Ok: Station 689 Frankfort KY... As I type this, both channels are at 10*8 at 120mv. I generally run 10*10* at 120mv threshold, and at night 5*5 at 120mv or possibly 4*8. I'll go to 10*16 one one or both channels if conditions are quite around here. When we get more western stations online, I probably won't go that high... let them grab the strokes.  
Using home made DonF ferrites, 6 rod (300mm) as opposed to his original 5 rod (250mm)... unshielded.  Noise is better, interference a bit more correctable, while the gain is about 2-3 times that of the 120mm kit ferrites I had to run shielded. Lower noise floor may be because of thinner rod and more windings. Even a 'new' ???? intermittent interference signal around 55-56KHZ is controllable below threshold, while still getting adequate gain.  

Because of some failing sodium lights I can't get power company to replace,   I have to cut gain back at night, but still get more 'efficiency' than I did with the other ferrites, and definitely the flat panels picked up everything. That is, assuming I remember to back the gain down.... if I don't, I'll be cut off large parts of the night and dip well below 40-60% on LR. Other wise, were in good shape on 'efficiencies' most of the time...whatever the heck that says.  

I'm considering winding a couple of 350mm or 400 mm as soon as my wrist heals from winding those two a few weeks ago. I've retired my big flat panels, and the kit 120mm ferrites. These homemades just work too well. Maybe I got lucky  

Oh, an aside... I suspect many of the 'higher performing'   Reds are centrally located, also. They're kinda destined to pick up more signals in 360° touching base with more stations....

Need More Stations!!!!

Cheers!
Mike

I'm usually running my 12 inch ball and have tested that pretty well.

I quickly built an 8 inch ball prototype tonite (the Ocho) that I was going to build over the winter.  Today and tomorrow's setup was too tempting to let pass.  Might get a little quick test done the next couple of days.

I'm planning on building your version of DonF's ferrite antenna (6 core) over the winter.  I have all the mats for it but not the time.  I'll be shielding my version 'cause EMF is a bear around here.

Regards,
Lance
Huntingtown, MD
Station 724

[schwab]

I decided to try increasing my 6.8 Green system's gain from 70 to 111 to see if it improves detection.

Change was made to Unit #671 at 10:15 am Central Daylight Time or 15:15 Universal Time.

[schwab]

After going into a sustained interference mode without any nearby lightning I returned my unshielded 12cm Ferrite 6.8 Green system's gain back to 70 from 111 last night October 9th ~10pm Central Daylight Time or 3:15am Universal Time October 10th.

[schwab]

Blitzortung Website Bug?

When my Green 6.8 unit went into interference mode it did not show up as being in interference mode on the Blitzortung participant list.