Ordered me a new GPS

PPS matters…





Both Garmin 18x, both RS-232 connected.

Same location…both in my home.

As you can see the 5PPS is by far better.

The 1PPS is connected to a 4core Intel CPU and the 5PPS to a 2core Intel even at lower speed.

The cloud conditions are terrible at the moment, for both.

The scale isn’t the same, too bad, as I was testing drivers and settings yesterday.

It will be better as soon as I have all settings figured out, still testing…but time is good enough.

@stevesommars as for the USB PPS1 U-blox, you tested it on an Raspberry, I did run it on a fast Intel. Also, I used the PPS kernel module with SHM-connection.

This is not so bad as you describe. In fact, I ran with it for a long time before I bought the first Garmin.

Intel CPU’s can be intterupted while ARM CPU’s can’t, as such they are not as good in handling PPS. As USB isn’t interrupt driven either, it’s 2 sides that do not know interrupts.
It’s for that reason that I do not recommend those CPU’s for PPS driven stuff.

This was one of the reasons that Sparc’s failed at such tasks. ARM is a bit better, but not as good as Intel and AMD, interrupting is a blessing for NTP.
It used to be bad, when CPU’s where single core, as it could halt the CPU, but today it’s not possible.


Correct me if I’m wrong, but as I understand it, the GPS satellites send out a PPS signal every 1 second. Anything greater is just the receiver chipset sending out a PPS pulse according to its own internal clock. You don’t get a more precise PPS signal per say, just a greater number of pusles to average with.

This thread here on EEVblog talks about just that.

The 3.3V is for the chipset itself. The 5V is for the BiasT (voltage/power to the antenna) if you have a GPS antenna with a built in LNA. You don’t have to supply 5V if you have a passive antenna, but that would require a tuned antenna to have a completely unobstructed view of the sky, as well as a very short coaxial run (direct connect is best) to prevent excessive signal loss. You could get a cheap 5V to 3.3V regulator, they are only a couple of bucks on Amazon for a number of them.

A cheap Serial to USB adapter can work for connecting to a PC for the timing messages, though I am unsure how you would work the PPS signal in that case as I’ve only built mine for a Raspberry Pi.

There was someone who built a USB/serial adapter board for a close to plug-and-play setup with these timing boards. The circuit board plans are still out there so you could make one yourself, but I can’t find it right now. I do recall I found them when I was looking for the pinout of these boards, but it seemed overly complicated for something that really only needs a couple of wires to have it work.

No, that wouldn’t work. Then the receiver would get multiple such signals, and with different timing, depending on the satellites’ and receiver’s relative positions/distances.
Rather, the pulse is always generated locally based on timing recovered from multiple satellite signals. Time is one of the unknowns that the receiver needs to solve for to get its position.


Sorry but that isn’t quite true.
The Garmin receivers are very good at receiving the signal without being outdoors.

Both my receivers are indoor, one is below a woorden roof, no problem at all.
The other is behind a stone wall and it receives fine too.

Most GPS signals are transmitted arround 1GHz give or take a few 100MHz, they can penetrate walls and other materials. Just like your 900MHz cell-phone does.

The only problem is that people insulate their houses today with gold-shielded-doubleglass, glass-wool with alu-foil, steel enforced concrete…etc…

That creates a cage of Faraday…and signals won’t be able to get to your receiver.

You do not need a clear sky-view, it helps, but it’s not a must.

1GHz signals will pass most constructions unless they block signals :ok_hand:

You can check this with your cellphone, if it can find your location indoors, it will work for GPS-locked-time too…as your phone is far worse then e.g. a Garmin 18x receiver.

You only need 1 sat for time keeping, not 3 or 4 for positioning, 1 is enough if you have other clocks that will tell what the time should be.

I could be wrong, but that is how I understand it, and so far it works for me.

If I’m wrong, please correct me…anyone…people only learn from making errors and I’m no exception :slight_smile:

IIRC GPS L1 Band Freq is about 1575 MHz (1.575 GHz) with a bandthwidth around 20MHz.
There is also L2 & L5 Band. Both with lower frequency. I haven’t seen an L2 receiver in the public or payable for private people. L5 isn’t used so far.
Same with the Glonas L Band.
L1 is in ther upper region ~ 1.6GHz like the GPS L1
L2 is in the lower region ~1.2GHz

More information on the ESA site :slight_smile:
GNSS signal - Navipedia (at the botton further links to GPS, Glonas, Galileo)

1 Like

Like I said, give or take a few 100Mhz…

Still they penetrate pretty good.

Even GSM’s do at 1.8/1.9GHz…

Mpegger was talking about a u-blox receiver connected via coax to an antenna – you really want an “active” antenna which has a bandpass filter and amplifier in that situation.

The Garmin pucks have the receiver built into the same enclosure as the antenna, so they may not need an active antenna with a tiny distance from antenna to receiver, and they have power if needed.

Also I bet the cheap GPS chips in modern smartphones have far more simultaneous channels than the 12 of the Garmin GPS 18x line. I don’t think it makes sense to compare receivers designed 15+ years apart.

1 Like

“You only need 1 sat for time keeping, not 3 or 4 for positioning, 1 is enough if you have other clocks that will tell what the time should be.”

My understanding differs. Most low cost GPS gear requires 4 Satellites for position and time; this is an ongoing requirement.

More expensive chipsets such as the Ublox f9T have explicit timing features. They can achieve near nanosecond time accuracy. Careful antenna positioning, clear sky, low multipath are other considerations.
I believe one first does a “self=survey”, where the GPStiming chipset attempts to accurately determine its location using all available satellites. This position is stored in memory; at this point only one satellite is needed for time recovery. An accurate position known from other sources can instead be loaded into the timing chipset.

gpsd has some self-survey support. See ubxtool You’d probably need ublox specific documentation too.

I’ve never used a GPS timing chipset. It would be overkill for my NTP work.

1 Like

Why? As 3D lock isn’t needed for time.
You just need 1 feed that gives time, that is enough.

For position you need more, as they calculate time-difference and position of the sat.

But we do not need position. More sats make it more accurate, but we just need 1.

Position and time are 2 different matters, think like DCF77…1 source,
Sat is not different. But if you have e.g. 5 and they know the position, it could be a lot more accurate…not saying it is…could be.

But all you need is 1 sat to feed you time. Accuracy could be a discussion, but then you are in very low ns-teritorry…if it’s even that good. Don’t know.

GPS sats give time from just 100km or so, DCF77 for me is 1000km (direct, via reflections can be 5000km!). Think about it. Still pretty accurate.


aplus is correct, GPS L1 is 1575Mhz, as I’ve built a couple of antennas for GPS L1 signals both with and without LNAs (Low Noise Amplifier) to test with. I have a 4 lobe “Cloverleaf” antenna and a 15 turn Helical antenna that I’ve been playing around with and plan on using once I setup my NTP servers in a more permanent location. Both perform as expected for thier designs, though I wish I had a NanoVNA to tune them for the best performance.

And GPS signals just don’t compare to cell phone signals or cell phone GPS, especially when cell phone towers 1) transmit at much higher power compared to GPS, and 2) are usually in a very dense layout, with many towers located within a small area, specifically for the purpose of making sure no matter where the cell phone is, it will have a signal. Plus, cell phone location works through a combination of GPS and the local cell towers, with modern cell phone OSs also using other signals such as known wifi and bluetooth signal locations to estimate a cell phones location even without a GPS signal. With a working network connection the cell phone can also update the GPS almanac for faster signal aquisition. I test my antennas in a pretty bad location, so bad that even my cell phone will lose it’s connection to the cell phone network. GPS on the cell phone? It’s dead too in this location, yet the cell phone can still get a good estimate of it’s location just by the local WiFi and bluetooth devices around, good enough to follow me around.

GPS at the low power it transmits at, by the time the signal has penetrated the atmosphere and reached ground level where a GPS reciever would likely be, is already around the -130dBm level at the antenna, when the surrounding noise level is around -120dBm to -110dBm (or higher in dense cities). Compare that to a cell phone signal which (depending on your location to the tower) can be easily as high as -20dBm, well above any noise floor. Indoors, that cell phone signal is probably still around -80dBm to -70dBm. The GPS signal? Good luck with that as the signal would be all but completely gone. At that point, it’s up to having a good designed, high gain antenna, good filtering, good LNA, and a very sensitive GPS receiver, and even then, the reception would be poor. I have some old Garmin GPS units, and in my test location, even with a modern puck style GPS antenna with it’s built in LNA, these old Garmin GPSs can’t get a lock. 9th gen uBlox chips with a tiny fingernail sized patch antenna with no LNA at the same location, it’ll get a lock and “see” enough satelites that it’s able to maintain and keep that lock. Seeing as how a modern cell phone would be using a modern GPS chipset, it’s no wonder just how they can still get a GPS signal in your average indoor locations.

As stevesommars pointed out, GPS timing chipsets can still provide time even when only 1 satellite is being received, and will still output time & PPS, not to mention that thier PPS signal should also be more accurate then non-timing chipsets, at least for uBlox chipsets that is the case. uBlox timing chipsets are also programmed to prefer satellites directly overhead (less time travel through atmosphere, less chance of multi-path signal, stronger signal) over satellites near the horizon, since the assumption is that these more expensive chipsets would be used in a situation where a proper external antenna would be mounted with a clear view of the sky. Non-timing GPS chipsets may not provide a timing output if it doesn’t have a position lock, plus the PPS signal isn’t as accurate.


The u-blox timing receivers have a TCXO, or temperature-compensated crystal oscillator. While the real time-nuts prefer an OCXO, or oven-controlled, which uses resistive heating to maintain a constant temperature, that draws significant power and is certainly overkill for NTP. The TXCO means the u-blox should be able to provide great PPS even with only intermittent GNSS reception, A.K.A. frequency holdover.


[quote=“davehart, post:32, topic:3251, full:true”]

Ahem, DOCXO is preferred. OCXO is for the masses. :beer:

Couple of problems with this explanation. First, it’s a direct in line transmission, as such it doesn’t need much power. I mean, I have been on a hill in Belgium with a 1W 70cm transmitter and talked to a repeater 120KM away, no problem at all.
There are also distance records at 2.4GHz wifi over 80Km an more with standard 100mW.

My example is a simple porto with a coil as antenna…not much dB gain, rather -6dB or more, still doing 120Km.

Anyway, receivers do -150dB noise-levels, so a problem can be a city as you say.
However, the GPS signal is right-rotating-polarized. Did you ever consider that in a city there is too much reflection causing it to rotate the opposite direction, and as such you miss reception?

Loads of studies have been done to make GPS-receivers more city-proof by adding a second receiver but connect it to a left-rotating-antenna.
No idea what antenna you build, but did you consider the rotation-problem in city-area’s?

I live in a rural area with no buildings that will reflect the signal, just some trees that weaken the signal a bit, yet it’s still then strong enough to penetrate a stone wall.
Even my GSM can locate GPS signals indoor, and I live in a concrete/brick building.

This is my 18x 1PPS indoor behind a brick wall…it works fine and the clouds are filled with water at this moment.

GPS sat’s do not fly very high, so there isn’t much power needed to reach the ground.

Plenty sat’s, sorry. My 18x 5PPS is even better under a wooden roof :slight_smile:


Longwave AM transmissions can go for thousands of miles, at 1/10 the power or even less then FM radio transmissions that pump out 10kW or more, yet the FM transmissions only reaches between 100-250 miles. The Apollo and Voyager ships use less wattage then current L1 GPS satellie transmissions, surely at the distances those space ships were at it should be impossible for thier transmissions to even reach Earth seeing how weak the L1 signal is at sea level, yet even now that Voyager has left the Heliosphere NASA is still communicating with and receiving data from it.

You’re comparing different radio frequency technologies (again) as if they are the same or even comparable to begin with, which they are not (line-of-sight or not).

Multipathing or reflected signals, whichever you prefer to call it, is not a desirable signal in GPS as it introduces timing errors which of course, leads to position errors or no position at all. It’s the reason why having a matching circular polarized antenna on the receiver to increase gain of the desirable signal, and reject those undesirable reflected signals before any amplification or receiving occurs is important. It’s why I made my own antennas to use to increase gain while rejecting multipathing, because of the environment I’m located at. And yes, they are RHCP.

Proper antenna design is (just one) key to help receive the signals in the first place. As the saying goes, “garbage in, garbage out”, which is what you get with a poor antenna.

I have no idea what “studies” those are, nor have I’ve ever come across any, but I do know of other advancements in GPS technologies to help improve reception in dense cities and even indoor locations, such as Japan using GPS retransmissions in certain buildings, as well as having an additional geostationary satellite to augment GPS reliability. There’s also the L5 GPS band that would require an additional antenna or a multiband antenna to properly receive the 1175MHz signal it transmits at, which by design of the signal and the frequency it’s at, has better penetration and is much less susceptible to multipathing (so no need for LHCP antennas). Is that the additional antenna you might be referring to?

Or maybe you’re referring to diversity antenna arrays and a diversity receiver? These have been around for decades and are (were?) used in specific situations (usually scientific studies) in which an array of RHCP and LHCP antennas are connected to a diversity receiver that switches between the two types depending on the SNR. I don’t recall why or how such as system was used with GPS, but it had to do with studying sea level and wave heights in the ocean. Not exactly something that someone with a cellphone in the middle of the city needs.

GPS receivers have come a long way over the decades. As I mentioned previously, I have some older Garmins, I’m talking about late 90’s, early 2000’s units, that cannot even compare in GPS receive performance to a modern GPS, even ones from just a decade ago. And those GPS from just a decade ago? They can barely compare to GPS receivers released in the last couple of years. So yes, being able to receive a GPS signal indoors is indeed possible, but that isn’t because of the signal, which has remained virtually the same since the first GPS satellite was up in space. It’s because of the leaps and bounds in receiving technology that we have had over the years. Better and more precise antenna designs, better filtering, better amplifiers, better receivers, all that combined make it possible for a modern surface mount 4mmx4mm GPS receiver chip capable of pulling that GPS signal out of the depths of practically nothing.

Those trees around you? My Garmin Streetpilot would lose satellite signal. Hell, my body leaning over it on my motorcycle, was enough to lose signal…


The Garmin GPS 18x LVC was introduced in March 2008 if Amazon’s current listing for it [1] is to be believed. It works with GPS only, no Galileo, GLONASS, nor BeiDou, and has only 12 receiver channels, as his screenshot showed. I realize you don’t need a great number of channels for timing, but more channels may allow a better time solution for non-timing receivers as there are typically more than 12 satellites in view.

I’m tempted by the $35 “ublox LEA-M8T-0-10 HUAWEI GPS Module” listings on eBay which all seem to ship from China, so delivery might take a while, and are mostly used, likely pulls from cellular base stations which need a great frequency reference. One seller [2] is offering new in box at that price and when I made an offer the shipping estimate was 9 days. At that price I’m more willing to try my skills at bodging together a useful reference clock for use with either a PCI serial card or a low-jitter USB 3 serial adapter. For people confident in their hardware skills, coming up with an 8-pin header to connect those boards to ports and some kind of enclosure, I bet they could turn around and sell complete units with USB A or B for power and DB-9 serial for $100-$200 which would need just a LNA GPS antenna to provide a good reference clock for NTP software.

[1] https://www.amazon.com/Garmin-18x-LVC-Navigator-Unit/dp/B0016O3T7A
[2] U-BLOX ublox LEA-M8T-0-10 HUAWEI GPS Module New | eBay

Dave, why don’t you get an industrial NUC PC instead? The have native RS-232 connectors and very energy efficient CPU’s.
Typical they use the same as an RPi4 but have SSD and run on 2.5A 12V adapters.

For example: https://www.amazon.de/-/nl/dp/B0BB1D22J3/ref=sr_1_5

But if you look a bit around you can find them even cheaper.

These run standard Linux of Windows, whatever you prefer :slight_smile:

Mine is such too. Else if you use a normal PC, look on the motherboard, most motherboards still have an RS-232 header, then you only need the cable to get it outside.


My Garmin bike GPS has no problems receiving the GPS, even indoors, but it takes a while for it to lock.

What bide do you ride? Mine is a 2000 VFR800Fi, with 1 tooth less front sprocket, makes it ride like a GP-bike :slight_smile:
I’m near the German border, so I need acceleration and topspeed, it maxes at 231KM/h on my GPS :rofl:

You wouldn’t be sad with the purchase, getting what once was a near $300 timing chip for almost 1/10 the price. The chip alone is still listed on digikey and mouser for around $250. The performance to a regular 8th gen ublox GPS navigation module (non-timing) is also noticeable. Offset and Jitter are regularly in the single digit microseconds (0.001ms) on the timing module, while the navigation module has Offset/Jitter in the 20-40 microsecond range (0.02-0.04ms).

And as far as packaging it up for sale, you’d have some competition out there (not from me) as I have come across some for sale which as obviously some kind of SOC board with the timing module for around $150 here and there.

I found that link to an adapter board as well. https://partiallystapled.com/hardware/pps_piggy/


At the time with that old Garmin, I had a CBR 600 F4. Took both touring around when visiting family over 1000 miles away. Didn’t have enough space even with the expansion “RAM” to store half the maps needed except for the major highways and main roads. And of course, no navigation on old units. You had to do everything manually with waypoints just like a paper map, or sitting at a PC to view actual maps and create a route.