SDG

[@ray_gannon]

Thanks for the heads up on the TDC7200 (and descendants!) Looks very interesting

[@Minifig666]

Excited to see where this goes - have wanted to build a GPSDO for ages but never got round to doing the research, so thanks for sharing! I like the idea of having the control DAC and oscillator mounted externally at the end of a cable, that should make it a lot easier to insulate the whole oscillator package too.

[@19janiboy96]

Same here I'm in the middle of setting up my lab and a GPSDO seems very appealing to me

[@DanDrown]

For maximum NTP precision, a microcontroller with IEEE1588 timer support is useful. I've written an open source NTP server for the STM32F407 as well as the iMX RT1060 (Teensy 4.1) that uses IEEE1588 hardware timestamps for the NTP traffic as well as hardware capture of the PPS timestamps.

[@sdgelectronics]

I'll take a look at this - I was not aware of this standard, so very useful info. Thanks

[@markbutowski4487]

@Dan - Did you implement NTP or PTP?

[@DanDrown]

@@markbutowski4487 I implemented NTP. The peripherals were designed for PTP/IEEE1588

[@dwagner6]

I really appreciate this series as a jump-start to some personal projects of my own. Inspiring stuff!

[@ad2181]

Nice summary. Don't let the "tail(the gui) wage the dog(the time function)".

[@MisterBics]

R.I.P. Bob Pease

[@robatoto]

To the best of my understanding, the 1PPS signal is a digital signal generated by u-blox' digital circuit clocked from the internal 48 MHz clock. So its phase noise relative to the target 10 MHz signal will be horrible. It will drift back and forth within a fifth of a 10MHz cycle. Therefore I doubt precision can be improved by trying to better align with the 1PPS phase. The short-term stability of the OCXO is superior to the 1PSS signal. Only the long-term frequency of the 1PPS signal should be used to control the OCXO.

[@ChrisSmith-tc4df]

Much as with voltage references, a working surviving older OCXO specimen typically outperforms a newer one.

[@ad2181]

Have you considered using the AD9648 digital phase lock loop chip. Analog Devices has an video showing an wenzel oscillator locking to the Gps 1pps.

[@PapasDino]

Look forward to the project as it progresses...could be a nice replacement for my Z3801A and Rb standards at the bench! Thanks Steve!

[@simonbaxter8001]

2 words .... 'Leo Bodnar', unbelievable stability, low phase noise, quick time to fix and lock at 10Mhz. Just bought one, as the primary reference on my test gear, and the test on YT show it as being as accurate as a Rb standard.

[@twobob]

Quality as ever, personally enjoyed the Quadrature? wangle for the overclock. Sneaky. Cunning. Great stuff.

[@sdgelectronics]

It works very well, so I was quite happy with that solution - it's probably commonplace, but I didn't see it suggested when I was scouting around for information.

[@twobob]

​@@sdgelectronics You're educated but this was a nice round up of digital wangles like BPSK kzbin.info/www/bejne/npCalWCjZsqbpbs which one supposes is an identical application in another domain? Hope it helps

[@twobob]

@@sdgelectronics he uses 10 meg and a square. made me think of you. give or take an order of magnitude.

[@Valenorious]

Ok, maybe my shallow knowledge of the complexities involved will show. But is it not possible to generate a square wave from the GPS module in the Mhz range with an internal divider? Then divide that 48MHz (?) clock down to a lower Megahertz frequency, of which an odd harmonic (because it is a square wave) matches the 10MHz of the OCXO. Then adjust the OCXO input until it nulls the phase difference between them. So divide integerly from 48MHz (or whatever the max frequency the config allows) to 2 MHz, filter out the 5th harmonic and amplify it. Then compare both near 10MHz frequencies and make the OCXO match it's phase, or at least arrest the phase drift to zero.

[@FATCAP]

Hi !! Did you tried the TDC7200? I wonder if that IC can replace the 4046+RC filter, which, as you say, seems very temperature dependent.

[@JxH]

Funny. Mine just arrived today!! I was just testing out for the past hour. It's the usual little BG7TBL type.

[@sdgelectronics]

That's one of the units I have here, though it's the one with the LCD on the front. Seems to work well, though it has quite an old GPS receiver, so doesn't support many of the satellite systems in use today so I found the antenna needed good positioning.

[@t1d100]

I am not finding the next episode of this project? What is its episode number? If this is the last video, what is the status of the project? Thank you for your help.

[@sdgelectronics]

I ended up pausing the project - hopefully I will revisit it soon as I've just bought some nice displays to use for the front panel.

[@davidvb3754]

Any update over that project ?

[@SidneyCritic]

Would more mechanical isolation of the OCXO on the PCB help, ie, slots or a separate raised daughter board.

[@sdgelectronics]

I don't think the mechanical isolation is too much of a problem at 10 MHz. It's mainly thermal issues, but given it is a used oscillator, it may have been scrapped because it was behaving out of spec.

[@breedj1]

Nice video. I'm also building a gpsdo with the Tdc7200. In the past I have used a 4046 with a FET as a current source to get a linear slope when charging the capacitor. But that does not work well after I switched to 3.3V. I use a samd20 to do the counting and all the other work, no fpga needed for that. It can count and store a value in a register when the 1pps is detected without the interference of the cpu.

[@sdgelectronics]

I've had this project on the shelf for a while as I was starting to make things too complicated (which meant the firmware would never have been finished...). Thanks for your suggestion, I'll take a look when I revisit the design.

[@breedj1]

@@sdgelectronics my design will be posted as an open source project when finished. I'll let you know when I have done that so you can have a look at my solution.

[@BobHannent]

I just discovered this video and I'm gutted it stalled, would love to see more of what you described, especially the TD7200

[@hullinstruments]

Have you checked out the nice GPSDO and distribution amplifiers that were designed by that Japanese amateur radio guy? I’ve got to tinker with them quite a bit and they are super nice. They stand up very well to much more expensive pieces of gear. I bought a literal truckload of Nortel telecommunications stuff. Tractor truck load. Bunch of phone towers that were upgraded I guess. I ended up with over 20 Nortel Trimble gpstm modules. All mounted in massive cabinets with huge back planes and extremely nice but overdone powering system. Obviously not something I need for my lab. So I plan on converting a few of those for me and some buddies… And honestly think I’ll like them better than My rubidium standard. I think they will offer much more accuracy long-term and short term. I’ve been able to power a few of them outside of their back playing in cabinet. Not too complex. And luckily the Trimble of a Nised oscillator module and supporting PCB just pops right out on its own. Check out Tony albus on his KZbin channel. He’s got about a dozen videos regarding those GPSDO modules and distributions that were designed by that Japanese guy. They come in many configurations. But My favorite one has a 10MHz in and out… But also has its own standalone ovenized oscillator. In case you need a back up to the GPSDO. The OCXO in the distribution Aunt automatically disables when you feed the GPSDO into it. It switches over automatically. If my memory serves me correctly… It also has 5 MHz out, 1 MHz out, sign out, square out, and 1pps.... and I think 100 kHz out if memory serves me correctly. Hard to beat that for $100. And then the GPSDO is another hundred dollars so for around $200 you get both. Sweet deal in my opinion. If I didn’t have that stack of telecom equipment bought at scrap price… Then I would definitely be looking into keeping those GPSDO and Amps in my lab permanently I really enjoy your channel! I’ve been waiting months and months for someone to do a video on UVC LEDs. So I was so excited to see your video last week. I’ve done a lot with UVC LEDs and arrays over the past year. Very surprised some of the crazy samples I’ve been able to get my hands on. Some of them came from the big boy suppliers and companies who make them… And some from what I’m guessing are factory workers who are pocketing some of the high powered samples that weren’t on the market yet. A year ago I got my hands on some 50 mW 265 nm nikisso leds. A few months after that some of the 70 mW LG leds. And this was absolutely unheard of at the time. Sure there might have been a few articles published over the past few years about high powers … But there was absolutely nothing available that was over 20mw. At least not to the general public. The first thing I thought when I heard about Covid last February a year ago… Was that hopefully it would influence the UVCLED market. I know that’s a terrible thing to think. But no one took the pandemic this serious back then. So I guess things started evolving so quickly because of the pandemic… these factories were probably very hectic and I’m sure had many failed batches and experiments. Or chips that were rejected for some reason. I’ve always enjoyed buying military surplus, laser and medical equipment, and other optical obscurities from the Asian “deep web”. Not like the black market or anything but just websites that are unlisted unless somebody gives you a link. And it’s not very professional. You pretty much text back-and-forth with someone on WeChat, send the money, and pray that they hold up their end of the bargain. All while juggling a VPN and weird payment requests. But it’s been worth it in my opinion. The reason I think they were factory workers is because they made me swear I wouldn’t list these online for sale. And that if I put them on optics and laser forums… That I waited a while first. I was just happy and thankful that I got my hands on some neat things. Tell you what we’re really blow your mind… The first UVC LEDs I saw in person were about eight months ago… At a local Walmart. Yes Walmart. They were obviously cheap Chinese knock off‘s but they were real, ceramic package, gold plated, quartz window, UVC LEDs. They were at Walmarts all over the world, hidden in some of the sanitization items That began selling last summer when Covid really ramped up I have optical power meters, a few pieces of spectrometry gear, and other assorted optical equipment. Good Stuff I got cheap as broken or as untested… That I have repaired and had calibrated. So I was able to characterize and test all these LEDs. I’ve never worked with these wavelengths before..... but with a bit of tweaking and calibrating it wasn’t a big deal. And it sure was a lot of fun! Sorry to ramble on. Just haven’t had many folks to discuss UVC leds with over this past year and it’s been something I’ve spent countless hours on. If you want to hear more of my experiences… I have a long drawn out right up of some emails I’ve had back-and-forth with other optics/laser nerds over the past year. Regarding the evolving UVC technology, and the rapidly changing landscape and what we’ve learned about it. Not sure how much you’re in the optics and stuff. But you should definitely check out zenodilodon, Marco reps, tech ingredients on KZbin. Those three and several others have wonderful laser and optical content. I think Zenodilodon is my favorite laserist. That kid really knows his shit. And has zero fear breaking down, repairing, and realigning $30,000 lab lasers. And he’s super nice and helpful. Oh and I forgot les’ lab. He’s got an amazing newer channel and he’s already one of my favorite content creators. Been begging him to do a DIY spectrometer project because his attention to detail would really bring something new to the table and finally put a good quality and accurate DIY spectrometer on the open source market. Now if we can just get Marco wraps to ever finishes OSMU. (Open source measure unit)! But I might as well keep dreaming on that one 😂

[@danw1million]

Hi, how did you end up going with this project I the end, is there a pcb ?

[@djohnsto2]

Is there any way to get more than a 1PPS signal directly from the GPS? Like measure the modulation of the satellite's signal or something?

[@marcisberzins]

Up to few megahertz.

[@dwagner6]

What are you using to interface the TFT with the board to get the data display?

[@colydeane]

Very interesting project which I would love to build. Which PicoScope are you using?

[@stevengwilliam8096]

The TDC7200 could be a neat solution to any jitter on the 1PPS. Some of the non-timing variants of the GNSS chipsets can have quite a bit of jitter. So you can filter/average this out on the data from the TDC7200. Are you intending to bring out a 1PPS from the unit that is divided down from your 10MHz Osc but synchronous to the original GNSS 1PPS? So if you do lose sat signals, your 1PPS output is still locked to UTC as far as your holdover performance goes. Your GPSDO videos have been been spot on by the way. Really interesting.

[@sdgelectronics]

I haven't thought too much about bringing out the 1 pps. Until I started looking into my own design, I'd never really looked at that output so I'm not sure. Is there something it's commonly used for? Thanks

[@stevengwilliam8096]

@@sdgelectronics My background is broadcast radio. We use commercial GNSS disciplined oscillators to time DAB radio transmitters and so need both the 10MHz and 1PPS. The 1PPS is critical for maintaining the timing on single frequency networks. So for us, having a "clean" 1PPS output that is time aligned to UTC is pre-requisite and this has to hold true if your chosen GNSS constellation has issues and you are in holdover.

[@FATCAP]

​@@stevengwilliam8096Hi Steven, I know your comment is from 3 years ago, but you asked exactly the question I was going to do here. Do you know how can that be done? Because I understand the behaviour of Lars GPSDO and others but rarely they output a PPS, which is something I need to implement in my design. Any idea would be appreciated:)

[@George-gz5zm]

Isn't the TDC7201 the newer model of the TDC7200? Couldn't you put the 10Mhz oscillator into a comparator with a stable voltage reference as the threashold? Or is would that lead to too much inaccuracy on each rising edge? Great project Steve and nice explanation.

[@sdgelectronics]

There doesn't appear to be too many differences between the two devices. The TDC7201 is a small BGA package though which has the potential to complicate the PCB. The TDC7201 is indeed newer, though it looks like only by a year (2015 vs 2016).

[@George-gz5zm]

Hi Steve, what will be controlling the SPI protocol? FPGA or microcontroller??

[@leozendo3500]

Well done. Route to revolutionary

[@SQ5DBF]

Very useful video.

[@bertoid]

Rather than buy an oscillator in an oven, you can build your own. This Silicon chip project does that:- www.siliconchip.com.au/Issue/2018/October/GPS-synched+Frequency+Reference+Pt.1 The oscillator is 40MHz, and is controlled by a 24 bit DAC. The output is very finely controlled, effectively it divides the 40MHz +/-200Hz output control range into 13 million steps. The heater is a transistor, controlled by an 12 bit DAC, and monitored by a DS18B20, giving 0.0625 degree C resolution. (the firmware must implement the heater PID control) I'm working on a GPSDO based on this oven, but doing my own counter & firmware. The oven appears to work very nicely. My version of this oven is at the top left of my development rig:- imgur.com/4eahC0G Oven with cover removed:- imgur.com/U5VZyjq

[@NivagSwerdna]

My GPSDO case contains a RPI a la www.satsignal.eu/ntp/Raspberry-Pi-NTP.html it's probably the easiest way to get a proper NTPDd distribution. There is some jiggery pokery to get the PPS to work nicely on a level change GPIO but otherwise it is straightforward.

[@sdgelectronics]

I'll take a look at this project - thanks :)

[@balbes1165]

It is too complex and imprecise. Try linear approximation Tgps = a + b*Tquartz. BTW, 1MHz works without quantization errors with both 10MHz and 48MHz clock sources. So if one would use cheap MCU clocked from 10MHz temperture-compensated (e.g. ovenized) quartz and use it to time 1MHz coming from GPS, it would give more precision than one needs for tuning in no time. It is cheaper, simpler and faster solution.

[@k1mgy]

TI's TDC appears to be a knock off of the Acam TDC-GP1 (www.pmt-fl.com/tdc-time-to-digital-converters). I have quite a bit of experience with Acam's stuff. Your use of this in the application is worth considering.

[@Kosmonooit]

8:00 is that an Arduino thing?

[@gamerpaddy]

pro mini is a arduino thing

[@t1d100]

Excellent.

[@NivagSwerdna]

JLCPCB are so yesterday. ;o)

[@sdgelectronics]

I still use both, they each have their place and offer different advantages.

[@ad2181]

Correction AD9548 dpll chip.

[@AI7KTD]

Did you ever get the TDC7200 to work for this?

[@jenskaa4044]

Very Nice projekt which I could be interested in. Here are my comments: I think that you should make the PCB a Lot smaller, lot of free space arround. I need no display just some leds showing the state. I am not a PIC fan, I would prefer a Arm based mcu, ex, a STM32. You could also go for an ESP32, I do have development tools for both (PlatformIO). Using the ESP32 could make it possible to provide a web interface or similar instead of the display

[@sdgelectronics]

Yes, I plan for the PCB to be much more densely populated. The prototyping PCB was large just to allow easier access for probing and modifications. STM32 is an option, but I'm not familiar with the best toolchain to use. The ESP32 using Eclipse was not in any way smooth and I don't know if there is a dedicated IDE for the STM32. The nice thing about the PIC is MPLAB 'just works', and I can open a 15 year old project in the latest IDE and it'll still compile the same.

[@jenskaa4044]

@@sdgelectronics both ESP32 and the STM32 is supporter by PlatformIO. And both I think is supporting native environment CMSIS (STM) and IDF (ESP) but also Arduino STM can also be programmer in STMCUBE (free tool), or Atollic TrueSTUDIO (also free). I have been using the latter.

[@dieSpinnt]

​@@jenskaa4044 & @SDG Electronics Be aware that PlatformIO can blow up your code size and compile time to an unbearable amount( Of course, that depends on the platform you choose). For STM32 the CMSIS or libopencm3 build by your own Makefile/CMake backed up GCC tool-chain would be the professional and stable solution that we all can follow. I take such things like mbed or Arduino only for early prototyping. For an end-product this choice is amateurish and there is just too much bloat and ugliness. Btw. that anecdotal reference, that "MPLAB 'just works'" and even with 15 year old projects is a bit funny, when you try to sum up the alternatives you could(not) use in the past and now. I afford the freedom not to mention an IDE, because with the ARM-based solution I mentioned, you are free to choose what you like and are not tied to constricting manufacturer-specific proprietary systems. At all with only typing "make" or "make upload" you are good to go and get your code running on the chip. The smallest common denominator is a professional industrial build system that is robust and accessible to everyone. In the end, of course, you should use what you have the most experience with and use the tool and platform that is best tailored to the problem solution. A great project. I'm curious what will happen next:)

[@jenskaa4044]

@@sdgelectronics it would also be nice if the PCB supports more OCXO footprints

[@jenskaa4044]

@Mdmchannel Mplab is actually a very nice IDE, and it also contains a very comprehensive configution tool. Their free xc8 compiler is generating very bad code, this was the reason for me to abandon the platform.

[@TheBauwssss]

But why would you want this? What is it useful for? I honestly can't picture a genuine use

[@juniorocicat9573]

A lot of talk but do you ever produce something we can use, you seem to jump from one project to another.