Blitzortung.org Forum

Full Version: Low Pass Filter problems
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
Hi,

I have bought 3 off digital filter ICs but have one problem and a question.

E-Field Problem
When I select the LP filter on the E-field I just see loads of 50Hz noise.  Clearly there is an open circuit on the LP filter output, but I have been all round the chip and all connections are okay.  I am thinking it is further on in the circuit. 

Has anyone had this problem and could suggest where the open circuit might be? Alternatively I can trace the signal if I had a circuit and layout diagram.  I am an electronics engineer and have access to a full workshop with lots of test kit.

H-field question
I installed LP filters on H1 and H2, they are working great and set to 16kHz, they cut my noise from 60mV to 16mV.  However, I expected the thresholds to reduce, but they haven't, so whilst I have really low noise, I actually detect less signals as the signal peaks have been smoothed a little and now don't make the thresholds.

I am running in auto mode, should I be in manual and are there any suggested settings?

Thanks

Robin
Schematics can be found under Hardware Descriptions on the Compendium page of the BO website.
(2016-10-20, 19:54)DrRobin Wrote: [ -> ]H-field question
I installed LP filters on H1 and H2, they are working great and set to 16kHz, they cut my noise from 60mV to 16mV.  However, I expected the thresholds to reduce, but they haven't, so whilst I have really low noise, I actually detect less signals as the signal peaks have been smoothed a little and now don't make the thresholds.

I am running in auto mode, should I be in manual and are there any suggested settings?

Hi Robin,

after adding 2 ICs to the H-field section I started in manual mode setting different values and settings followed always with hours of observation about the results. I never found a combination which was better than automatic setting. Once I disabled noise adaption, set to manual much lower values  (I have also very low noise here) but after some time I have seen that the threshold range was set automatically complete out of the range of the signal from one of the antennas.
So finally after a lot of experimenting I moved back to default and automatic mode. Mostly I'm on page 2 or 1 or maximum 3 refering to "Efficieny" and I'm happy with this.

If you also start with experiments take your time and maybe you have more luck. I'm interested in results!

Cheers

Hartmut
Thanks Andrew,
I have found the schematics.  It looks like a problem with one of the multiplexers or signal lines when the filter is selected on the E channel.  I will have to trace through with a scope probe.  

Hartmut
This morning I made a note of the gain settings whilst running under 'automatic' and then adjusted the gains using manual mode on the channels I have LP filters fitted.  I am now running much higher gain on the two channels with LP filters, the noise has come back up a bit, but signal to noise has gone up a lot.

There is quite a lot of activity around North Africa and southern Greece just now and my two H-field antennas with LP filters are much more sensitive and picking up these strokes much better since making the change.

I am going to see how well it runs today and then look at thresholds.  After which I will document my findings.

Robin
(2016-10-21, 08:21)DrRobin Wrote: [ -> ]I have found the schematics.  It looks like a problem with one of the multiplexers or signal lines when the filter is selected on the E channel.  I will have to trace through with a scope probe. 

Does "been all around the chip" mean a visual inspection or a reflow of each pin with fresh flux? Apart from the chip I missed actually soldering a complete side (Doh!) I had "fun" with one of the others, every pin looked well flowed and connected but a reflow cured the problem so obviously something wasn't properly soldered. I have a feeling that there is some form of protective substance over the bare solder or plated areas as putting the through hole stuff on didn't "take" as quickly as I'd expect.

As for settings that's just getting a balance between the conflicting variables: interference, noise floor, bandwidth (faster rise time), filter delay, etc. For the last couple of days I've been in manual with minimal filtering (54 kHz on Ch1 (null to Anthorn), 19kHz on Ch2 and 54 kHz on the E-field. It was working quite well. Note "was" I've fiddled with antenna orientation and given it a supply earth connection today and it doesn't appear to be working as well as it was.  B-(
Hi Dave,

Reflowed. I hand soldered them in and all of the joints are A1. I have over 40 years experience of soldering so you get to know a good joint. I reckon the open circuit is one of the components soldered at supply and just never spotted/tested before.

I didn't put a filter IC in H-3 and if I select the LC filter on this channel then it just shows no signal, not 50Hz noise so it must be an input open circuit, which has no pull down. My first guess is the second (o/p) multiplexer, but tracing through with a scope will pin point it. Sadly I am away for the weekend so it might have to wait until next week.

However, the H-field channels with LP filters and manual settings for gain are working really well.

I noticed your station working really well yesterday, picking up at long range, I would re-align the antennas back to where they were.

Robin
He, he, I've been wielding a soldering iron for a similar period of time. I know what you mean about a good joint and why I commented on the bare solder or gold pads not "taking" quite as I've come to expect. At least it's a very simple bit of circuitry, not many places for it to have failed. My problem is the tiny size of this SMD stuff and eyesight that is almost fixed focus at 6".

I need to get my head around the alignment of the antennas, Ch1 I'll stick with aiming the null for minimum Anthorn/Skelton. Ch2 will have a compromise between getting swamped by Athorn on one lobe whilst having the other facing the middle of the SE quadrant.

The station may have been doing well yesterday but that was pretty much just on the Ch1 loop. The Ch2 one was facing NE and was rarely sent, Ch3 Horizontal was sent more often! The E-field about as often as Ch2. Since today's antenna fiddle and ground connection I set a single LP filter in Ch2 at 19 kHz to use the filter slope to bring down the combination of Athorn and Skelton then switched to Automatic. It's fiddled with the gains and now all four channels are sent more often than not. It's currently seeing the strikes from the storms off the coast of Spain south of Barcelona (1000 miles) and between the heel of Italy and Greece (1500 miles).
(2016-10-21, 16:34)allsorts Wrote: [ -> ]
(2016-10-21, 08:21)DrRobin Wrote: [ -> ]I have found the schematics.  It looks like a problem with one of the multiplexers or signal lines when the filter is selected on the E channel.  I will have to trace through with a scope probe. 

Does "been all around the chip" mean a visual inspection or a reflow of each pin with fresh flux? Apart from the chip I missed actually soldering a complete side (Doh!) I had "fun" with one of the others, every pin looked well flowed and connected but a reflow cured the problem so obviously something wasn't properly soldered. I have a feeling that there is some form of protective substance over the bare solder or plated areas as putting the through hole stuff on didn't "take" as quickly as I'd expect.

As for settings that's just getting a balance between the conflicting variables: interference, noise floor, bandwidth (faster rise time), filter delay, etc. For the last couple of days I've been in manual with minimal filtering (54 kHz on Ch1 (null to Anthorn), 19kHz on Ch2 and 54 kHz on the E-field. It was working quite well. Note "was" I've fiddled with antenna orientation and given it a supply earth connection today and it doesn't appear to be working as well as it was.  B-(

which chip and component had the welding problem?
In my case the optional, SMD, solder on yourself, filter chips, nothing wrong with the factory mounted stuff here. DrRobin has fitted the optional chips as well but one channel, when the LP filter is enabled, is not working properly. He's confident of the soldering so suspects a board fault but hasn't had time to fault find it.
Hi,

I have performed a lot more testing and the E-field is actually ringing/unstable when signals arrive. I swapped the LTC chip out into another channel, the problem moved. I also desoldered one of the good LTCs and put it in the E-field, this works okay, so it looks like I have a faulty chip.

I will need to get a replacement and possibly a fourth for H3.

Robin
Fault moving, know good fitted in original faulty position working, doesn't leave many options other than a duff chip. Not good at not far under £10 a pop.
Hi,

New LTC1569 arrived today and fitted, problem with the E-field is fixed.

I have however, discovered another problem, if the input voltage drops a bit, which it does with the extra load of the filter ICs the E-field channel rings at about 1.5kHz. The other channels seem fine.

A measured input of 4.8V or higher (as shown on the status page) seems okay, but below this (4.6V) I see the ringing. My iPad PSU sits at 4.8V, I found another USB psu, which is about as quiet, which gives 5.2V indicated. I also have a linear psu I made about 35 years ago with a fixed 5V output and variable 7 - 15V. Unfortunately the fixed 5V only gives 4.7V under load and suspect the 7V will be too high. Pity because the size of the smoothing cap on this PSU will make it ultra quiet. Perhaps I will pull it apart and modify it.

With all three channels filtered at 17kHz, the only noise I see is a spike at 45kHz which could be switching noise from the psu?

Robin
Hi,

I now have a very low noise system, here is what has been changed:

1) LTC1569 filter ICs fitted to both H-field antennas and E-field antenna, set to 16kHz to cut Anthorn/Skelton to manageable levels.
2) iPad PSU swapped out for an 'industrial switcher' 5V, 25W with low noise output.
3) DC power cable passed through a ferrite before the miniUSB plug.
4) Ali case all bolted back up. This made more of a difference at killing the 45kHz than I had imagined.

So now I have very low noise on all three channels, even though only one ferrite has it's null position towards Anthorn. The result is that my Stroke rate has gone up nicely, but the locating rate has gone down quite a bit. I think I might be detecting lighning by reflection from the US/Caribbean over-night when the hop distance is just right?

Robin
Or perhaps, it is just that over-night here is their afternoon/evening and there is more lightning activity?
(2016-11-02, 18:05)DrRobin Wrote: [ -> ]Or perhaps, it is just that over-night here is their afternoon/evening and there is more lightning activity?

Maybe. lightningmaps.org might be persuaded to give a suitable regional based activity plot.

Glad you've got your system sorted out. If memory serves something around 45 kHz occurs when the controller and PSU don't get on with each other. If you can see it with LP filters enabled and set to < 20 kHz it isn't getting through them, they really are very steep and deep. I see no sign of MSF on 60 kHz on any channel with a LP filter set to <20 kHz, if fact 54 kHz is enough to have pretty much the same effect. And that'll be on a channel with the best part of 100 mV of p-p 60 kHz present, same channel has 200 mV p-p of 19.6/22.1, unfiltered.

I know have Ch1 loop with it's null aimed to get minimum Anthorn/Skelton, which means it's lobes are almost exactly N/S. Gain is normally 2560 with 50 - 60 mV "noise". Ch2 loop looking NW/SE as far as it can go W/E without Anthorn/Skelton being so F***ing loud they constantly trigger the system, that means about 200 mV p-p unfiltered. I filter this channel at 17 kHz which brings that down to the 50 - 50 mV range with a gain of 4000. This is the only channel I have a filter in.

The system seems to be working very well. Yesterday there were many in the western end of Iran and the area where Turkmenistan, Uzbekistan and Kazakhstan all meet the far side of the Caspian Sea. That's 2,700+ miles (4300+ km). See attached:
  • Time range: from 2016-11-01 00:00:00 UTC to 2016-11-02 00:00:00 UTC
  • Count: More than 1000 Strokes found
  • Oldest: 2016-11-01 10:40:57 UTC
  • Newest: 2016-11-01 23:58:31 UTC
Participated strikes are marked with a large square blob with a light center, just detected are small round blobs. The shaded/unshaded is the boundary of the 3000 mile search radius. That map is a capture from my install of MyBlitzortung which I think only has strike information relating to my station.

I'm not sure I believe the few participated and other strikes that show up south of the equator off the coast of Congo/Angola! They are 4,200 miles (6,700 km) away... Is there a way to verify them?