10MHz GPSDO by "BH3SAP"

[KungFuJosh]

I got this 10MHz GPSDO in today, and it's pretty cool for the ~$60 I bought it for. Given the price, I wasn't expecting a lot, but it seems good to me so far.

It looked pretty stable on the SDS2504X HD:




But was much better on the SDG2122X:




The device doesn't come with instructions or a power adapter. It uses a 12V center positive DC adapter...they also didn't bother to note polarity anywhere.

Control with the single knob is a little weird at first. A single press will bring up the menu, and turning the knob counter-clockwise will scroll through some options. Press the knob again to change the selected setting...which appears to do nothing. 🤣

There's a frequency offset option which starts with 0.0Hz. I tried it at -10Hz, -100Hz, it set itself for some reason to 6543.6Hz, and none of them had any effect on the output frequency. Correcting it from the 6543.6Hz thankfully went faster than if it was actually set that high since spinning the knob takes forever.




The inside says "Made By BH3SAP" but that didn't help with any searches for more info. The attached board looks like an [edit] STM32 Blue Pill. If anybody is good with those, it might be a fun thing to mess with.




Thanks,
Josh

[squadchannel]

https://www.qrzcq.com/call/BH3SAP
BH3SAP seems to be Chinese. You might be able to find him if you search the forum somewhere.

[jasonquin]

BH3SAP is a Chinese call sign. This guy made some HAM radio stuff on Chinese internet.

And that is a STM32 Blue pill.

[KungFuJosh]

BH3SAP's design seems pretty cool. I was surprised by the stability.

This thread is the only thing that comes up searching for his callsign on this site. I'm curious if there's a firmware variation out there.

[daddygo]

Hello,

We bought two pieces for FM transmitters SFN sync... for test only now.
As long as you're measuring with a 1M ohm oscilloscope input impedance, you'll be just fine. (Not a nice sine wave, but acceptable, the 1PPS signal is not beautiful either, a bit noisy)

Once you plug it into a standard 1PPS/10MHz inputs (that expects you to use cca. 3Vpp and 50 ohm imp.) everything goes wrong.
This is more Chinese rubbish, pardon the expression...

I would suspect the source impedance of the 1PPS output being too high and when loaded with 50R input impedance load the level drops below valid threshold.

Br, D

[KungFuJosh]

We bought two pieces for FM transmitters SFN sync... for test only now.

What device did you buy? Your screenshots look nothing like what mine shows.

[daddygo]

Exactly like yours, only the GPS unit is different, uBlox Neo-6M...
What causes the problem is that the 1PPS output is directly connected to the SMA from the uBlox panel, no impedance control element

[KungFuJosh]

What causes the problem is that the 1PPS output is directly connected to the SMA from the uBlox panel, no impedance control element

Were your screenshots from the 1PPS output?

Either way, I don't think it's fair to call a $60 item "Chinese rubbish" when you can easily spend hundreds or thousands of dollars on one depending on the quality you need. I wouldn't be happy comparing a $400 oscilloscope to a $4000 oscilloscope either.

[daddygo]

Yes from 1PPS you can see this from the frequency 1Hz...
We bought the two units on AliExp for $210 e.g. 105 / pcs. and were refunded 75% after submitting the measurement results, we now have two "toy" GPSDO units.

You're lucky if you got it for $60, but it's not a serious tool 

We work with Trimble tools, but we were tempted to try this one as well, as it comes in a factory box and we box the Trimble stuff ourselves, like this on teh picture...

The impedance matching minimum requirement!
You can see on the pictures (attach.) that the PIN 1 of the GPS unit runs directly to the SMA connector, only one pulse LED with 150R more...

When you connect it to an $18 000 FM transmitter where everything is standard 3 - 3.5Vpp / 50R with the 1PPS mark drops to 1Vpp, the 10MHz also halves and you get this horrible distorted sine wave.

It's not the Chinese thing or origin that's the problem, the problem is that some people are selling it in bad faith

[KungFuJosh]

Of course it's a toy for $60! That's why I bought it. I was thoroughly impressed with it's performance because it's a $60 toy. I was also unwilling to spend the $$$ on a "real" GPSDO since I have absolutely no need for one.

You connected what you paid $25 for after discount, to an $18,000 piece of equipment, and weren't impressed with it. I'm not shocked by that. 😉😉

It almost goes without saying people sell stuff in bad faith, especially on AliExpress. Go buy a TinySA Ultra (which is a great tool for what it is). It's still on the toy end of the spectrum, and yet, lots of people sell mediocre clones in bad faith. I don't expect any of them, be them real or fake, to compete with a $1500 or $15,000 spectrum analyzer.

[daddygo]

I see you're a big fan of lab instruments, -I have been using them for 30 years 

Well, we'll stick with the Trimble or something like this:
https://tomverbeure.github.io/2023/07/09/TM4313-GPSDO-Teardown.html

 as it is also available for 80USD and this work properly and have impedance matching...

[KungFuJosh]

Stop it! I don't need another GPSDO! I'm not gonna buy it. I'm not gonna buy it. I'm not gonna buy it. 🤣🤣

Damn it, it even looks cooler.

[daddygo]

Ours is already on its way 

[KungFuJosh]

Seems like there are issues with the 1PPS on that one too: https://www.eevblog.com/forum/testgear/tm4313-gpsdo-strange-behavior-after-a-night-of-poor-reception/

I'll avoid the topic for now, since I really don't need one. I'll probably return the one I already have too.

I've seen some other variations in this lower budget range that have LCD displays and give more info, but I have to keep reminding myself, I have no use for this. 🤣

[nctnico]

BH3SAP's design seems pretty cool. I was surprised by the stability.

Did you reference your oscilloscope to a Cesium clock? If not, your measurement results mean very little (to put it mildly).

Best thing you can do in your own lab (without a Cesium clock) is to get 2 known-good GPSDOs and compare them to the GPSDO under test using a time interval counter (with better than 1ns resolution) over a period of 48 hours. You'll need to 3 take of those measurements. Then you can do a 3 cornered hat comparison and draw some kind of conclusion.

[daddygo]

Together with this divider it will work fine, I started to build it, have GERBER and everything, the author (Oleg) also has TM4313

https://olegkutkov.me/2021/04/10/1pps-square-clock-8-channel-distribution-amplifier/
https://olegkutkov.me/2021/02/28/1pps-distribution-circuit/

[nctnico]

Together with this divider it will work fine, I started to build it, have GERBER and everything, the author (Oleg) also has TM4313

https://olegkutkov.me/2021/04/10/1pps-square-clock-8-channel-distribution-amplifier/
https://olegkutkov.me/2021/02/28/1pps-distribution-circuit/

I don't like the way you split the signals at all. Better put resistors as close as possible to the chip driving the output amplifiers and match the impedances of each trace individually at least at 1 end. Otherwise you get a whole bunch of reflections.

Also, when dealing with 1PPS, termination is actually more important because it is a pulse. 10MHz is typically a sine wave which doesn't have trouble with reflections where it comes to deformation of the edges.

Another thing to keep an eye is adding input / output protection against ESD by adding some high-speed TVS diodes.

[flash2b]

Mine arrived today and has a purple PCB (date code 240617 so pretty new!) opposed to the green PCB (date code 231024) of the KungFuJosh

The GPS/Beidu module is a ATGM 336H-5N-31, where as the red module (on the PCB of daddygo above) is a NEO-6M.

Main controller ARM cortex STM32F103C8T6 (Blue-Pill, "Performance" version) with micro USB (the topic starter (KungFuJosh) has USB-C Blue-Pill version)

Result on the Racal-Dana 1998 (Option 04E)

Hope some new firmware arrives someday.

Did anyone attach an USB-TTL module to STM32, PA9  TX    —->   Rx of USB-TTL and  PA10  RX   —->   Tx of USB-TTL to see if there are things/text coming out from the controller ?

Manual from the QR code on the package
 GPSDO使用说明 英文.pdf (1348.16 kB - downloaded 14 times.)

[Gertjan]

Did anyone attach an USB-TTL module to STM32, PA9  TX    —->   Rx of USB-TTL and  PA10  RX   —->   Tx of USB-TTL to see if there are things/text coming out from the controller ?

Indeed a pity that this GPSDO does not have a serial interface with NMEA output....

It means no opportunity to see a lot of information that makes the use of a GPSDO more reliable, and makes it better to understand what is going on. Such as info on number of satellites received, their SNR., Azimuth and Elevation Graph, HDOP and VDOP) parameters etc.
See for an example of a NMEA monitor: VisualGPSView
It also precludes the use of Lady Heather.

Normally I would expect NMEA output on the TX terminal of the GPSDO module, but I do not know this ATGM 336H module.
Probably the STM32 also uses this port for communication. But as NMEA is a read-only data stream, it might be possible to read out the TX port in parallel. When the STM32 communicates only at startup with the GPSDO module (to set the module up), there might still be a clean NMEA stream at the TX output.

And, maybe, BH3SAP made the NMEA data available at an output pin of the STM32.
This, as well as the output of the GPS module, is easy to check with an oscilloscope.

Regards, Gertjan.

[flash2b]

I will do some probing on the ATGM 336H-5N-31 (datasheet attached) and the STM32 and let you know.

[KungFuJosh]

This was one of the reasons I returned mine. I couldn't set any offset, and when I tried, it set itself to this the next time I checked the offset. 🤷

[mianos]

I got one, it was something 10kHz out and could not be adjusted. As the above post, I sent it back. I wonder if they sent him mine.

[flash2b]

The bug only occurs when applying a negative offset, it does not occur applying to a positive offset. It is however a display bug since saving a negative offset does apply the value to the GPSDO. After re-powering, the saved value is not displayed correctly in the menu, but the saved value is applied correctly to the GPSDO. You only have to set and save the offset once.




The picture above shows the offset value displayed after re-powering. If you look closely to the value it is 65533 which is 3 less that 65536 which is the maximum value of 2^16 = 16 bit max. The bit16 is usually set for a negative value of a signed int so its value -3 which should be displayed -0.3 adding the decimal point after the 4th digit. Therefore I recon it is a display error. You should not save the value again since then the original is overwritten incorrectly.

[KungFuJosh]

The bug occurred on mine no matter what offset I tried, positive or negative. The offset also never had any effect at all on the frequency output.

[flash2b]

So your unit didn't apply the value to the output frequency of the GPSDO which seems a hardware failure. Your experience with the OFFSET seems the same as my unit has, but on positive and negative so maybe other firmware version with my unit.



I did enter the menu after PPb reached 0.0. (antenna needs to be connected and satellites need to be 'seen'). Once adjusted to an OFFSET value the PPb comes back with a + or - number and PWM will count up/down so that PPb will be 0.0 again. Its fiddly and not so easy as described in the manual, but the manual steps are described. The OFFSET function is not described so maybe the manual is for an older version. At the end the output is adjusted up or down according to the setting. Range is +/- 100Hz.