DIY GPSDO starter project advice sought

What sort of frequency accuracy do you need?

A surplus rubidium will be awfully good. A typical trim range is +-1 PPB, so it will be pretty good as long as it locks.

Johnny B Good wrote

If you search ebay for GPDSO you find many examples, cheapest one I see on page one is a 51$ bid with free shipping:

Initially, the raw, jittery 10MHz straight out of the GPS module would do me for starters, assuming I'm not underestimating the effect of the jitter. I'm just trying to calibrate the 0.1ppm 50MHz TCXO I've installed into an FY6600 DDS AWG as an upgrade to the 50ppm commodity 50MHz XO chip it had originally been cursed with.

As I understand it, the lamp life of rubidium oscillators is limited (about 20 years I understand) so a surplus (presumably 2nd hand pull) is a bit of a risky investment. The GPSDO option looks to be my best one so far, especially if I can obtain almost instant gratification, assuming the jitter doesn't represent an intractable problem to my calibrating TCXOs against the 10MHz output of the GPS module.

If nothing else, the experience of building a GPSDO will be an education in itself. It's one thing to read about the problems of short term jitter on the GPS module's 1PPS output (regardless of the actual frequency chosen) but you can't properly appreciate the dry statistics until you've experienced the practical effects for yourself. I'm sure I'll obtain a better understanding of the jitter problem soon enough once I have a GPS module in my hands to play with. :-)

Am 26.01.19 um 16:35 schrieb Johnny B Good:

Is that you who has triggered the current thread on the time nuts mail list?

regards, Gerhard

Not guilty! Indeed, there's not a Cat in Hell's chance of my joining a mailing list - I draw the line at signing up to fora. I think the EEVBlog forum is only the second one I've ever signed up to over the past two decades (and is the only one I'm currently active in (FSVo "active").

I've only posted my query here in the SED NG so far. It seemed a more likely source for advice and guidance with regard to purchasing a suitable GPS module. I may well land up posting this query as a new thread in the EEVBlog forum to get a 'second opinion' (there doesn't seem to be that much 'cross pollination' between SED and EEVBlog).

I haven't looked recently, but when I bought mine a used Rb was about $100. At that price you could buy two, and swap out the first one if it fails.

Yup. Beating it against a rubidium will be amusing as well.

Cheers

Phil Hobbs

(whose first job out of astronomy school was designing most of the time/frequency electronics for the first civilian DBS system)

The time-nuts list is very well-behaved and centered around the technical content. There are no tracking cookies etc on a mail list and you can use a throw-away address to subscribe if paranoid. If you have Ghostery you can see how web sites track you.

Even social content is minimized, some recent stories about the early days at HP are an exception to the rule. There are some olds farts who spent their lives at HP; the list owner did the frequency multipliers in the HP Cesium clocks, among other things IIRC.

You may look into the archives; this week someone asked exactly your question.

In comparison, s.e.d. is a whorehouse.

Cheers, Gerhard

Thanks for the advice, Gerhard.

I'll have another look at their archives and see if I can identify the relevant thread (and work out whether it's ordered top down or bottom up). A quick scan to find the start of the thread I was looking at left me confused, possibly on account of the top posting nature of emails, contrary to the common practice in usenet postings. Hopefully, I'll figure it out and track down the relevant thread.

It is an interesting PLL, deriving a super stable 10 MHz clock from a nasty wandering jitterey 1 PPS pulse.

That sounds very similar to the NB3N502 PLL multiplier chip I mentioned in my op. I've still not been able to track down anything relevant to my query in the archives though. Perhaps a bit of 'spoon feeding' maybe? :-)

Also, I've not had any answers with regard to the "Ublox-NEO-M8N GPS navigation signal amplify module for arduino Rasppery PI" that I was thinking of buying. On the face of it, it looks just the item I need (and at the right sort of price).

The only reservations I have (possibly needless) are that it's a generic no name unit with no further vendor supplied data other than that it's meant to interface to Arduino, RasPi and PCs, has an external, powered, antenna socket (and from the single image offered, includes a ceramic plate antenna stuck to its underside) as well as a micro-usb socket.

I'm assuming (ASS-U-ME danger here) that the vendor doesn't really need to offer any more information than he's given on the basis that the target buyers will be more than familiar with hooking it up to their Arduino/RasPi/PC based GPSDO project and the data sheets are freely available from Ublox's web site to "Make programming a snap".

Since nobody seems to have bothered following the eBay link to cast an expert eye over my GPS module prospect to offer an opinion on the wisdom of such a purchase, I'm still none the wiser as to whether to place the order or just hold out for something more suitable. It looks like I'll have to post my query into the EEVBlog T&M forum in the hope of getting a more concrete answer after all.

Usually, if you're after an opinion on something, a usenet posting is a good starting point since usenet is, in general, an endless supply of nothing but opinions, especially in the matter of purchasing recommendations. Here, in SED, I find myself somewhat at a loss over the seeming absence of expert opinion.

Never mind, life is so short and 21 quid so little an investment in proving whether one's instincts for recognising a bargain can still be trusted, I may as well place my order. At least I've not had any advice to the contrary in my choice of GPS module.

This GPSDO looks like it it has similar specs. "GPS Variable Frequency Reference" I'll probably be building something similar if my existing GPSDO again goes into temporary insanity mode. One of the local hams built one and claims excellent stability and performance, but I haven't seen it yet.

I checked out those clearskyinstitute links and what struck me just now is how much time and money he could save if he were to build a MK II "GPS Variable Frequency Reference" based on a Feeltech FY6600 or 6800 AWG to deal with the DDS part of the project. :-)

Anyway, since I didn't receive any warnings against purchasing that Neo- M8N based Arduino/RasPi adapter board, I went ahead and ordered one last

However, afaiui, I should be able to program it via the usb port to give me 10MHz on the PPS pin. I don't need the nano straight away and I'm planning on buying some of those ultra low jitter NB3N502 PLL Clock Multiplier chips so I can reprogram it to give me a 1 or 2MHz output without the jitter you see from dividing a 48MHz TCXO clock signal down to 10MHz.

All the fancy satellite numbers/quality statistics display and all such frippery can wait for now. A minimal 10MHz GPS clock signal will be a vast improvement over what I'm reduced to relying now.

Incidentally, my inspiration came from this chap's instructional youtube videos here:

I'd think that this should work for your purposes, although I'm not sure I'd use the GPS receiver in its 10 MHz output mode for this purpose.

For the initial calibration, the easiest approach would probably be to run the 10 MHz signal into the "external reference" input of a good frequency counter, and use this to count the 50 MHz signal. Depending on the counter's maximum gate time, this should let you calibrate and trim to within a fraction of 1 Hz.

Once you're there, you can go to a phase-oriented approach - switch the GPS receiver to 1 PPS (so that its PLL is out of the picture), use this as the capture/trigger input to a 'scope, capture the 50 MHz signal, and look at the relative phase drift over time.

I did something very much like this, to check and calibrate a surplus rubidium standard, using a modified VE2ZAZ GPSDO as the long-term standard (I rewrote the firmware to use a PID algorithm).

The VE2ZAZ design is an FLL, not a PLL, and it tends to wander back and forth a bit... with the current PID tuning parameters, it drifts back and forth by a fraction of one cycle, over a period of 20-30 minutes or so. If the rubidium is mis-trimmed there will also be a systematic drift of the phases, visible over a period of hours.

I used a Rigol o'scope to capture the relative phases of the two 10 MHz signals, once a second or so, and then plotted the phase differences using gnuplot. I tweaked the rubidium fine-adjust pot until the systematic drift was trimmed out, leaving just the periodic back-and-forth drift.

A bit so. What you'd want is to buy from a seller who will test the lamp output level and give you an honest report on the signal voltage (most of the surplus rubidium oscillators I've seen make this available) so you can compare it to the spec and get a sense for the amount of lifetime remaining in the lamp.

I don't run my own rubidium standard all the time... I just turn it on and give it an hour to warm up, when I want to check calibration on my other instruments. It's definitely more stable (over the period of minutes to a few hours) than the GPSDO.

The raw 1PPS signal has quite a bit of short-term jitter and drift - it's just in the Nature Of The Beast. I don't know what sort of filtering time constants are present inside the Neo receivers, to damp this out when they're used in "drive a PLL" mode - I suspect the answer amounts to "not much filtering".

So, to get down into the parts-per-billion-of-error range, you'll need to build a "proper GPSDO", with an ovenized (ideally) TCVXO, and some sort of hardware/firmware/software controller that causes the oscillator to track the long-term average rate of the GPS.

But, in order to really appreciate it, you need to have a more stable reference to compare it against (>>grin

Yup... very probably the case.

That's what I found when I bought a Neo, to replace the old (and died-for-some-reason) Motorola GPS module I was using in my GPSDO.

I haven't tried to use the PLL in mine (all I wanted was an as-good-as-practical 1PPS signal) but I had no trouble hooking up the module, or using its various outputs.

True. However, the article was intended for publication in QEX magazine: which is the ham radio experimenters journal. Articles are not intended to be fully documented construction articles and are often work in progress. In other words, and exchange of ideas, not products. A device does not need to be economical, optimal, practical, and possibly even safe. Just some ideas which may or may not fit well with your intended application. I kinda like the idea of having a display with tiny worms (satellite tracks) crawling around the screen.

If you want to save some time but not money, just buy something on eBay and you're done: Just $110 from Smog Angeles. If you don't mind burning time and building something truly cheap, just build or buy a 10MHz OCXO, which will probably be accurate and stable enough for most applications.

Thanks. Looks interesting. I'll view those videos tonite.