Confused about Leo Bodnar "Mini Precision GPS Reference Clock" grounding

[wobbly]

I have a Leo Bodnar Mini Precision GPS Reference Clock ("GPSDO" from now on).
https://www.leobodnar.com/shop/index.php?main_page=product_info&cPath=107&products_id=301

For a hobby project I thought I'd make a simple circuit that takes its square wave output and divide it down to provide a 1 pulse per second signal.

My first thought was about the GPSDO's chassis - is it floating or not?  I expected it to be Earth Ground referenced when I power it from my PC's USB port.

1. Measuring my PC first: from the mains Earth pin to the USB connector outer shell (of the USB A port) reads about 20 Ohms or so (considering cabling and whatnot sounds reasonable to me).
2. Then the USB cable I'm using - is fully shielded from one end to the other.
3. Measuring the GPSDO's Mini USB B connector outer shield to the GPSDO's Aluminium enclosure reads 1.5 Megohms!

I wasn't expecting that!  The GPSDO's Mini USB B connector outer shield is touching the aluminium enclosure (there is no apparent attempt to pull the metal away from the connector to provide electrical clearance).  I guess the Aluminium oxide is preventing contact?!

If another kind EEVBlog Forum user that owns one of these devices could try the same measurement (#3 above) and let me know what they read I'd be grateful.

To my mind, if the Mini USB B is touching the case of the device, then it's probably intended that the case is sharing the same voltage reference point.  But if the DC resistance between connector shell and case is stable in the Megohms range that also means that there is no attempt to connect them on the internal PCB either.  A contradiction?

What's going on here?  Is my unit faulty?

(Yeah I know I should be asking Leo Bodnar but my warranty is expired so I can't expect any help from them).

Any guidance / ridicule welcome.

[J-R]

Leo Bodnar support is good, I'm sure they would respond to your questions no matter the warranty.

I have the dual-output model and it has a similar issue where I'm guessing the metal case has a small amount of coating/oxidation.  If you are a bit more aggressive with the probing you will be able to punch through it.  So the case and all the grounds are connected, at least on my unit.

[wobbly]

I've emailed Leo Bodnar's tech support, will report back once I receive a response.

Meanwhile I've attached a picture showing what I mean...

The screw at "A" is my reference point.
"B" (SMA connector outer part) measures zero ohms.
"C" measures megohms.

[807]

The USB shell on my mini reads even higher. Approx 500 MegOhms (2 nanoSiemens).

[wobbly]

Haven't heard back from LB tech support, so I'll give them another week.  But in the meantime I'll crack it open and see if there are any clues inside that might lead me one way or the other.

For comparison I found IMSAI Guy's video #984 (pause as soon as the video loads for a decent image of the board)...
https://youtu.be/p-NUnCnOHHI?t=669

My unit:
Front

Back


Circled in green is a convenient solder link point for connecting the USB connector shield to chassis ground!  It is left open from the factory, suggesting that the intention is a fully floating chassis.  Good enough for me, so I'll leave it alone.

I notice that IMSAI Guy's unit has two EMI shielding cans over the UBLOX GPS system and what appears to be (to my layman's eye) a somewhat thermally isolated TCXO.  My unit doesn't have those cans, which is not relevant to my thread, but it's interesting anyway.

While my unit was in bits, I took a fine metal file to the aperture for the USB connector, widening it by a tiny amount.  Then I wrapped the USB connector with a layer of Kapton tape to prevent it touching the aluminium end plate.

Reassembled it and now the USB-to-chassis resistance is open circuit as far as my DMM is concerned (compared to 1 megohm or more before).

So at least I've removed the uncertainty of the USB touching the chassis.  I'll run it over the weekend and see if I get any different results (I suspect not, but we'll see).

[Solder_Junkie]

I run my LB GPSDO from a plug top (wall wart) PSU, so it is isolated from ground connections.

SJ

[Phil1977]

What about the GND-pin of the USB-socket?

In your PC it is for sure connected to PC-GND which means also to PE. It´d be strange if it would be galvanically isolated in the GPSDO, at least I can't see any components that are needed for this.

I suppose that LB also discussed if both sides of the shielding of a USB-cable should be connected to GND or if it´s better to ground it only one-sided. There are reasons to do this, especially if you have RF-connectors on the same device.

[blackdog]

Hi

I took some pictures of my miniGPS from Bodnar.

I with me the housing is not on the ground of the electronics.
This too puzzles me...

The two covers are connected to the SMA connectors though.

The outside of the USB connector is not connected to anything.

It is difficult to measure, because of the thick oxide layer of the aluminum case.
For these measurements I used very sharp probes from “Probe Master”
And a very sharp knife to poke through the oxide layers.

Looook Ma, no connection!
Test SMA connector to one of the screws securing the left cover.
And the same goes for the SMA connector.


.

I would love to hear what Bodnar gives for responce.

Thanks and regards,
Bram

[wobbly]

What about the GND-pin of the USB-socket?

In your PC it is for sure connected to PC-GND which means also to PE. It´d be strange if it would be galvanically isolated in the GPSDO, at least I can't see any components that are needed for this.

I suppose that LB also discussed if both sides of the shielding of a USB-cable should be connected to GND or if it´s better to ground it only one-sided. There are reasons to do this, especially if you have RF-connectors on the same device.

You're right.  Yes, the GND pin is indeed connected to the chassis.  Good point about the cable shielding at one end.

Okay, that makes more sense now, thank you.

I run my LB GPSDO from a plug top (wall wart) PSU, so it is isolated from ground connections.

I'd be interested to know if there's a performance difference (noise / jitter etc) between a GPSDO running from an earth-isolated supply compared to an earth-referenced one.

I'll have a play later on, not that I really have the gear to measure such things!

[Solder_Junkie]

I'd be interested to know if there's a performance difference (noise / jitter etc) between a GPSDO running from an earth-isolated supply compared to an earth-referenced one.

I'll have a play later on, not that I really have the gear to measure such things!

The tool to measure noise/jitter is a phase frequency analyser, I have a TinyPFA, which is a Nano VNA H4 with different firmware (easily reversed to convert back to a VNA). You use it with a very stable reference, such as a rubidium oscillator.

My own measurements show no difference between running a Leo Bodnar on a "wall wart" and from the USB socket of a PC. I have a dual BNC Leo Bodnar and a later LBE-1420 GPSDO, neither are what I would term "low jitter", it depends on how much jitter your equipment can tolerate. Both LB units produce jitter up to 4 parts in 10^10 in my measurements. See attached image, which shows a LB unit in green and my home made GPSDO (from the Lars Walenius design elsewhere on the EEV Blog forum). The home made version is much lower in terms of jitter, but takes a lot longer to lock and is limited to 10 MHz, it depends on what use you have for a GPSDO.

SJ

[wobbly]

Leo Bodnar tech support responded today.

They said the USB shield is supposed to be floating, and the chassis is grounded through pin 5.

Seems strange to me that they would let the USB connector rub on the metal like that though 

Mystery solved.