Cheap GPSDO Buying Help 2024?

[fourfathom]

The LEA-M8F (if I recall correctly) has a 30.72MHz output and a programmable output.
[...]

This could be useful info, thanks!  I've just begun to search for a solution.

[Leo Bodnar]

We have just introduced NMEA data output to LBE-1420 GPS locked clock source which is a revised version of mini GPS clock.

After I am happy that there are no side-effects to the main operation of the synthesizer we would introduce the same functionality to the mini GPS clock and the two port GPS clock.

Sorry, it took some time to implement it.

Leo

No, the Bodnar does not deliver NMEA data, and no, you are not alone on wishing it did.  In one of my applications I use a cheap GPS chip/module on my board to get NMEA as well as 1pps, and an external Bodnar for the reference clock.  I will eventually try to combine these functions, but for now using the Bodnar saves me the time and effort of implementing a GPDSO.

Is anyone aware of an inexpensive GPS chip that simultaneously does NMEA, pps, and high-frequency clock for a GPSDO reference?  This doesn't have to be time-nut quality, just good enough.

[jjoonathan]

Sweet! Order placed.

[PCB.Wiz]

Yes, I'm involved peripherally with some designs that run the clock out pin way up in frequency (to drive a GPSDO), but in addition to the frequency reference I also need the 1pps time reference.  Or *some* absolute time reference that's more precise than the arrival of an NMEA sentence. 

What frequencies do you need ?
The choice of CMOS counters/dividers is fading a bit over time, but you can still buy HC40103 /N  and even HC4059 dividers.
Those are simple to apply, with no coding, but the PCB area increases if you use multiple packages, and the prices of even moderate MSI parts are inching upwards.

Small MCUs are another option: their peripherals are getting smarter and not much code is needed to configure a 16-20 pin part as a multiple output hardware divider.

[fourfathom]

We have just introduced NMEA data output to LBE-1420 GPS locked clock source which is a revised version of mini GPS clock.

After I am happy that there are no side-effects to the main operation of the synthesizer we would introduce the same functionality to the mini GPS clock and the two port GPS clock.

Sorry, it took some time to implement it.

Leo

Very nice!  Do the NMEA sentences show up on the USB/serial port, and if so, how do you switch between configuration mode and NMEA mode?  Or are there two virtual ports?
When will this show up on the Mini and Dual units?  Is a firmware upgrade to existing mini's possible?

[Leo Bodnar]

Very nice!  Do the NMEA sentences show up on the USB/serial port, and if so, how do you switch between configuration mode and NMEA mode?  Or are there two virtual ports?
When will this show up on the Mini and Dual units?  Is a firmware upgrade to existing mini's possible?

Yes, the firmware upgrade adds CDC interface to the existing HID interface. NMEA messages flow through virtual COM port and HID configuration works like before, independently of NMEA.
I would say in a month or two both Mini and Dual will have the same functionality upgrade.
Thanks to all, Leo

[Leo Bodnar]

Small MCUs are another option: their peripherals are getting smarter and not much code is needed to configure a 16-20 pin part as a multiple output hardware divider.

tvb has been making his since the dawn of time http://www.leapsecond.com/pic/picdiv.htm

[NT0Z]

Update from OP:

I replaced the MAX4016 and the 74ACT244 -- reasonably monumental achievements for me as someone who has never done any SMD work previously -- and the GPSDO is behaving exactly as it did, save for the fact that freezing or tapping the 74ACT244 no longer introduces visible glitches on the scope or prompts brief periods of outputting a working 10-MHz ref signal.

Summary: My unusual BG7TBL-style GPSDO, which worked for 6 years, recently stopped outputting a 10-MHz ref signal, even though Lady Heather notes that the RX is working nicely, time functions, too, as well as OCXO lock/control, etc. Everything but the ONLY thing I require: a 10-MHz signal to discipline my freq counter.

The end of the ref output chain looks like this: MAX4016 op-amp, which feeds a 74ACT244 distribution or gating chip, which outputs 10-MHz ref signals, one sine, one square, through separate SMA connectors (just before the connectors there are a couple of SMD caps, probably some kind of low-pass filter?).

I have a 10-MHz "almost perfect sine wave" signal on the input pin of the MAX4016, but NO output to the 74ACT244. I replaced the MAX4016, no change, but when I tapped on the 74ACT244, I could "glitch" the 10-MHz signal. I replaced the 74ACT244. No change, but now tapping does nothing, which probably indicated a bad 74ACT244?

Lady Heather suggests that the rest of the unit is working fine. I have a 10-MHz signal on the input pin of the MAX4016, and on the "shoulders" of the 1k resistors there.

Thought 1: Acquire the signal through a small capacitor and feed it through a MMIC amplifier on a small PCB and, once buffered and isolated, pass it to the input of a distribution amp through a separate BNC connector for use in downstream devices (with controlled amplitude).

Thought 2: Acquire the existing signal as above, amplify it with a MMIC, and then reintroduce it to the caps in the output network, right near the sine wave SMA output connector, thus leapfrogging the goofy parts and supplying an amplified and buffered 10-MHz signal to the output. Looks like any dc present on the amplified signal would not be "fed back" into the 74ACT244 because there's a series, dc-blocking cap inline. This method might take advantage of the small network of caps just before the SMA output connector for harmonic filtering. ??

I could also buy a new cheap GPSDO unit as discussed above, but fixing and hacking this thing is now driving me to either fix it, or fix it to death.

Any thoughts?

BTW: I don't have a schematic for this unusual unit, but I will upload a pic or two later tonight.

As always, thanks,

--TG

[Leo Bodnar]

I have a 10-MHz "almost perfect sine wave" signal on the input pin of the MAX4016, but NO output to the 74ACT244. I replaced the MAX4016, no change, but when I tapped on the 74ACT244, I could "glitch" the 10-MHz signal. I replaced the 74ACT244. No change, but now tapping does nothing, which probably indicated a bad 74ACT244?

This sounds like a bad solder joint rather than a failed part.
Try resoldering all the connections downstream of the OCXO output — including SMT joints.
If this does not help, try replacing the passives — there is more chance that they have failed rather than ICs.
Leo

[NT0Z]

I can give this a try, but the risk of me killing the unit increases, as the system is comprised of three stacked PCBs held together by stand-offs and fragile ribbon cables. But I guess in for a penny, in for a pound...

I don't yet get how a signal can be present on the input pin of an op-amp without anything coming out of the same op-amp if that op-amp isn't defective. But apparently it can, cuz it's still that way even with the new chip.

Thanks,

Kirk, NT0Z

[NT0Z]

Pix promised by the OP:

PCB View: The SOIC8 at left if the MAX4016 opamp. The SOIC20 is the 74ACT244. Pin 2 of the opamp (top, second from right) is the input. Pin 1 is the output. The "hot" sides of the two 1k resistors just "north" of the opamp connect to the input pin. The 10-MHz waveform you see in the scope shot appears there.

Junk on scope: This janky, random waveform appears on the output pin of the opamp, which connects to the input of the 74ACT244 via a series cap. It also appears everywhere else downstream, right to the SMA output connectors.

Both chips have been replaced.

Thanks,

--NT0Z

[Leo Bodnar]

Check DC bias on pins 2 and 3 of the opamp.
It's possible that the trimpot or one of the 1k or 2k resistors are faulty.
Leo

[dietert1]

If your scope image can be trusted, you have about 25 mVss, which is a very small input signal. So the fault may be upstream. With that small input signal, output will appear only after readjustment of the potentiometer and it will be jittery.

Regards, Dieter

[NT0Z]

Hey, Leo and the gang,

Ironically, I'm trying to fix this GPSDO so I don't have to buy one of Leo's lovely units!

The pair of 1k resistors seems good. I measure 550 ohms between pin 2 and GND, and a pair of 1k resistors in parallel should be in that ballpark, right?

The 25 mV is while using a 10X probe, so 250 mV in actuality? My old Rigol scope doesn't seem to know what the probe setting is.

Is there a trimpot in there somewhere? Not the big blue precision pot that nets the OCXO, right?

I will measure/probe the other pins on that opamp.

Thanks.

--NT0Z

[dietert1]

250 mVss sounds better. The potentiometer sits next to the two 1 KOhm resistors. It's small.

[NT0Z]

Dang -- I didn't even know that was a potentiometer!

Any idea what it might do?

--NT0Z

[dietert1]

It is used to adjust the discrimination voltage of the amplifier. When it's off level, the 10 MHz outputs get stuck at logical 0 or 1.

[NT0Z]

Okay:

That mini trimpot is very twitchy and subject to being "bumped." I rotated it through its range, then gave it a drop of fader lube, then rotated it some more, and then I gave it a taste of the hot air station.

After cooling down I was able to find a fragile setting that results in this output from the square and sine wave ports. When doing so you can see what Lady Heather says.

For most of the pot's rotational range there is no coherent output on the scope. Shown here is the "best" setting, but the square wave isn't quite square, and the sine wave is a bit goofy, although the oscillator stops/locks at 10.0000 MHz (after drifting upward for a few hertz after every cold start or reset), which is the decimal precision limit of the counter in my old Rigol scope (never calibrated by me).

For various "nearby" pot settings the waveforms get weird double humps (lower side) and look really goofy, with the counter displaying freqs between 5 and 11 MHz (I assume because of the weird wave shapes or zero/trigger crossings?).

Left as shown, the waveforms have remained locked for a day or so.

On the MAX4016 opamp: When there's no coherent output on the scope, pin 8 (dc in) and pin 3 (non-inverting input) show 5 VDC. When the trim pot is rotated to produce the display as shown, and there is coherent output on pin 1, pin 3 shows 2.5 VDC -- half of the dc input. I think this is normal.

My problem all along probably was related to the trim pot, which seems to measure 500 ohms from top to bottom, although I don't know if any other resistances are involved. The coherent output is produced near the halfway point on the trimpot.

The pot is probably bad...and replacing it seems VERY difficult. I can barely see the three obscured solder pads. I could barely get the SOICs replaced.

What if I supply my own (separate) dc voltage to pin 3 with a real, large trim pot (through a 10k resistor to limit current)?

Can the existing waveforms be used to discipline my HP 5328A freq counter if I adjust the reference signal for proper input strength?

What if I connect the outputs to a distribution amp with adjustable gain?

I am learning a lot through this torturous process, which is good. But I'd really like to come through this with a working GPSDO.

Thanks again for your help,

--NT0Z

[dietert1]

The blue output waveform indicates a higher resistance termination, if you are using a 2 m cable between generator and scope. If you terminate the cable correctly at the scope end, the waveform will look clean. Amplitude will be about 2 V - half of the generator voltage.
If the HP 5328A external timebase input includes correct termination it will work.

Regards, Dieter

[NT0Z]

Dieter,

You are indeed an officer AND a gentleman! And perhaps a psychic!

New pic shows the dual-trace scope with 50-ohm terminating resistors "teed" onto each input with the requisite 2-meter-long BNC cables.

This looks usable! Sorry Mr. Bodnar -- no cash flow for you (just yet)!  Without replacing the SMD trimpot, I don't expect that this unit will work forever...

With help from many on this list I have learned a lot throughout this repair process. I learned how to replace SMD chips without killing the chips or the DUT. That's something that's been on my list for about 20 years. I learned that SMD trimpots exist -- who knew? I learned the basics of my DSO, and about potential termination issues. And more.

After learning about the discriminator level trimpot late in the game, it's my chips probably didn't need replacing, but it was for a good cause.

I will connect this to the HP 5328A later tonight, where I expect that things will work as normal.

If not, I'll be back. 

Thanks, everyone, for all the help. In the modern era most mentoring is distributed via the interwebs.

Regards,

--NT0Z

[EvgenyG]

Hi, Leo. Just received the new gen LBE-1420 unit, very happy customer. I've done some extensive research comparing the specs against the bigger A brand units and cheap no-name units and decided that it is absolutely worth extra $100. it is hard to beat that phase noise figure at that price point. I think most people don't understand the importance of phase noise. Good to know that there is NMEA support now, I was actually going to ask. Perhaps worth adding that to the product description page as well as brief information what different LED states mean.
Also wanted to note that the build quality is amazing.

[dietert1]

In order to get best results, the antenna makes a big difference independent of the GPSDO device. I mean a good antenna installed on the roof with a full free sky view can be difficult to obtain and will by far exceed the cost of the device. A geodesic antenna meant for measurements can be US$ 500 or more. One also needs to think about a good RF cable if it needs more than 5 or 10 m. Or place the GPS receiver next to or inside the antenna.
Once you use a patch antenna inside a building, the GPSDO device won't make a big difference.
Leo Bodnar's stuff is excellent but the LBE-1420 isn't exactly a GPSDO. Instead of adding some fanout for the 10 MHz it adds a wide band signal generator 1 Hz to 1.1 GHz. It can certainly be configured to output 10 MHz with low phase noise and one could try a splitter to sync two or three devices. Don't know if one can run two devices from one antenna, in case one needs 10 MHz and 1 MHz.

Regards, Dieter

Edit: Ours looks like this. Will replace it by a GNSS antenna.

[kevin.gibbs]

Does your frequency meter measure the period? If so, you can feed the 1PPS signal from the GPS receiver directly to the frequency meter input. You can also enable averaging (to filter receiver jitter). This should already be sufficient for your accuracy.

You can also take the signal from the MAX4016 input and use it.

[NT0Z]

Kevin -- thanks.

My HP counter can measure period, but I have never done so or made period measurements with any other devices.

After abusing the trimmer pot in the GPSDO (who knew that was a trimpot?) I have the sine and square wave output squared away as shown in the dual-trace photo. I am in the process of building a 10-MHz WWV receiver (TRF-style) so I can compare my GPSDO signals (on the scope) with those from WWV. After wrenching on the GPSDO and replacing two SOIC chips, I want to have another NIST-reasonable reference to make sure everything's okay.

I have a "TCXO in a box" that's been reading 1.2-Hz low at 10 MHz for several months. At least it's stable! In the interim I acquired a pair of inexpensive 10-MHz OCXOs, and if my counter is indeed disciplined correctly, each is reading 1.2 to 1.3 Hz high (no more adjustment range). One of the OCXOs will go into an insulated box and pretend to be a rubidium standard.

Using the WWV receiver I can feed a 10-MHz signal into the counter or into the scope that's 1 Hz or better at 10-MHz (avoiding dawn and dusk as appropriate) to at least verify that my GPSDO is locked onto the correct frequency...

I will also try the "zero beat" method over the weekend.

Regards,

--NT0Z

[SCSKITS]

Any of the inexpensive GPS modules should be able to produce better than 1Hz at 10MHz.
I ran a NEO-M9N on a MIKROE-3922 board (about $39 at Mouser) at 4MHz and got about +/-0.75E-8. At 10MHz, +/-1Hz would be +/-1E-7
A simple multiplier like the PT7C4511 set to 2.5 would give a 10MHz output.
The PTC4511 is in an 8 pin SOIC for easy soldering.
The frequency difference would be multiplied up also (by 2.5)  to less than +/-2E-8 assuming the PLL filter in the multiplier chip had little effect on the error. 
It looks like the 4MHz is getting better over time, I would have to run for a longer time to see just how stable it becomes.
If running the GPS module alone at 10MHz, the error will be greater than with the 4MHz output followed by the multiplier.
I have a test board that has the multiplier followed by a divide function that can be bypassed, I will try to get a run with that maybe over a longer time.

Another option is the LEA-M8F. It is more expensive but is one of the UBLOX timing modules and should do better than the NEO-M9N.
The 30.72MHz output is about +/-1E-9 right after startup, I did not run for a long time. I did not try the programmable output yet.
Starting with 30.72MHz output, you would need a more sophisticated multiplier divider such as the SI5328C which also requires a micro to initialize the part.
The SI-5328C is also difficult to solder for a DIY project.

ed