SDS1104X-E and SAG1021 unwanted DC offset from USB powering

[HendriXML]

Hi,

I was trying to do some measurements using the dc-output of the SAG1021.

First I did a manual zeroing using a Fluke MM. (Also using the amp setting to have low imp. to check if that made any difference).

However my SDS1104X-E keeps measuring a 15 mV offset on all channels. Even with a direct BNC cable and 50 ohm terminator.

I got the idea that the non ground lead of the AWG leaks current trough the ground lead of the DSO (not using a terminator). And it does!! Its 1,5mA (while the AWG output is measured 0 V with the MM). When measuring the current trough the ground lead of the AWG it is 0 mA.

So I measured the voltage difference between those leads and it is 85,9 mV.

I guess because the AWG is powered by the DSO both circuit are not completely seperate, which results in this influence. But is it just me or is this kind of buggy?

Should it be possible to power the AWG from a normal USB adapter and have only the datalines to and from the DSO? Stripping 1 cm in the middle of a USB cable and disconnecting / injecting the power should be OK? Or am I missing something?

[tautech]

For the record can we have the current firmware versions in use ?
TIA

[HendriXML]

For me it’s getting a bit late. I’ll come back to it.

I measured the current between the 2 grounds and that is 36mA. So my guess is that the USB cable is dropping the voltage. This voltage is then also lost between the AWG and the DSO.

I’ve checked it, the difference between the grounds is indeed 85mV.

More investigation will follow!

[tautech]

BTW, new SAG1021 not long arrived have a bigger beefy USB cable, nearly 2x dia of the USB cable that comes with the scope............maybe Siglent changed the USB cable. 
Sorry I don't remember what came with the SAG1021 when I beta tested SDS1104X-E.

[HendriXML]

BTW, new SAG1021 not long arrived have a bigger beefy USB cable, nearly 2x dia of the USB cable that comes with the scope............maybe Siglent changed the USB cable. 
Sorry I don't remember what came with the SAG1021 when I beta tested SDS1104X-E.

I’ve tried several USB cables, all drop more than 70 mV. In that regard the BNC cable between the AWG and the DSO gives a much lower resistence, explaining the around 15mV drop. (Checked that)
With a beeeeefy power cable between the BNC ground terminals it went to 0,5 mV. (One could use the trigger out for that, but then an extra scope input is in use)

But the right way - I guess - would be to not have the USB power GND and signal GND to share the same wire. And thus using an external and floating USB powerline.

That brings us back to the questions regarding that. Is it ok to solve it that way?

[tautech]

I’ll take some pics of the new USB cable in the morning and do some measurements.

I’ll also check latest SAG units for FW # against the early SAG unit I have.

[HendriXML]

When I look at the wire USB layout, I see a GND which is seperate from the mantle.

In the DSO they’re tied together. In the AWG I measure about 1M ohm!

So the mantle is not used as a transfer of the “BNC ground”!

That seem very odd to me. There’s a way to seperate the wires, and they don’t use it. What makes it worst that the mantle is of lower resistance and it is not used.

I know one doesn’t want ground loops, but those GND’s will be connected when doing measurements. And now there’se a 70 mV (or 15 mV) voltage difference which affect the AWG output, but also all DSO measurements.

Maybe someone knows a valid reason for this design?

[HendriXML]

I’ll take some pics of the new USB cable in the morning and do some measurements.

I’ll also check latest SAG units for FW # against the early SAG unit I have.

The beefy part of the cable is prop the mantle, which is not used. Also my short power cable of I think 16 AWG (rated 32 Amp) still has a drop. Having car jumper cables between the devices is not the most elegant solution  .

[HendriXML]

Making a power injector like this is prop. the best solution. But will it work? We know now the GND power line needs to be connected to the DSO, but can the positive one be sourced from a USB adapter without any side effects?

[HendriXML]

I’ll take some pics of the new USB cable in the morning and do some measurements.

I’ll also check latest SAG units for FW # against the early SAG unit I have.

Thanks for looking into it! I found another cable, it dropped 37 mV, but that won’t stop the current flow between the probe shields.

Having the AWG usb powered is what is the real issue. I went medieval on a usb B cable and got only the connector. So one step closer to injecting the juice via another way. However I think it also needs a powersource which has for example no capacitance coupling to mains. In the Netherlands one of the mainswire is earths ground. So to my understanding if not taking precautions this might also affect the neg/gnd rail of the DSO/AWG.

So any suggestions from the more experienced would be great! What would be a good external usb power supply in this case?

[HendriXML]

I ruined another perfectly good USB cable in the name of science. It came with my PicoScope. It was the one with the lowest drop. The cable says 24 Awg (data) and 20 awg power, having opened it and exposed its veins and soldering my own power wires to them I can asure it was good quality 
So I ended with a Frankenstein cable, which even heat shrinks could not make less ugly, but functional it is.

I used my bench power supply to bring the AWG to life. Then I plugged it into the DSO. It got recognized normally. With output turned on it draws about 366 mA from the power supply.

The measured voltage drop at the scoop was still there, but lower: 7,2 mV. This when there’s a connection. Unconnected it went from 37 mV (good cable) to 14 mV.

So one might think my reasoning was wrong. But that’s not entirely true, there is still a shared wire (or trace) in the AWG. If one measures the GND voltage between what comes in the AWG and what comes out. That where the 14 mV is lost. This looks like wrong design to me. Why not split as soon as possible, there’s a metal case which could bring gnd from back to front 

The moral of the story is also it does not matter how fat the USB cable will be, those 14 mV are lost any way. And it is probably due to 365mA (thus 0.038 ohm), so not easily “short circuited”.

But shorting the gnd’s of the BNC connectors seems the only way to get it down to less than 0,5 mV.

I think a new Frankenstein project is about to arise. 2 T splitters and some jumper wire will do just fine.

[tautech]

My results with very recent ex-factory SAG1021.

SAG FW version:



Full SAG1021 hardware package with scope USB cable on right for reference



Measurements. (using pre-release SDS1104X-E SN#0012)
SAG1021 connected with supplied USB cable only.
Measurement between scope BNC (shells) and SAG1021 output BNC; 86mV (AWG OFF) ~50% more with AWG ON.

With supplied BNC cable attached between SAG1021 BNC and scope ch4: 8mV

[HendriXML]

Thanks for your efford! I’ve got the same firmware and hardware version.

I made the short circuit cable. I used 2x leads of loudspeaker wire of 2,5mm2 if I remember correctly. In those days copper was still affordable. 

The drop with a normal usb cable is 2.8 mV total. Between the connectors (wire site) its 0,2 mV. So the BNC connections drops 2.6 mV. That is a lot, but it is seems not be copper and not very thick/hollow. I used a lot of tin, and the copper wire penetrates one side of the T join some few mm. (I drilled out the centerpost off course  ) So that cannot be optimized much further. Repositioning the MM probe on the BNC connectors shows an increase or decease of voltage. That is quite funny to experience.

It also shows that those connectors are not made for “high currents”.

With my special cable the total drop is 1,3 mV (0,1 mV wire sites)

I measured the total drop between the unused BNC connectors. So I think you would endup with a bit more using that approach. But not if you used the scope off course.

1,3 mV is - I think - doable, but it is at the cost of having a more complex setup.

[HendriXML]

SAG1021 connected with supplied USB cable only. Measurement between scope BNC (shells) and SAG1021 output BNC; 86mV (AWG OFF) ~50% more with AWG ON.
With supplied BNC cable attached between SAG1021 BNC and scope ch4: 8mV

I think you incidentally uploaded one picture twice. So the measurement cannot be seen.

[HendriXML]

Full SAG1021 hardware package with scope USB cable on right for reference

I do not recognize the cable, there is only a small possibility that I’ve not taken the supplied cable in use. So It is probably an upgraded one. (Thicker but also longer?)

- correction I’ve got prop. the same one.

Mine says 28 awg/ip and 24 awg/2c.

So my special cable is better 

[HendriXML]

Is there a way to open the AWG without ending up with something ugly?

It would be nice to see the wiring....

[tautech]

I measured the total drop between the unused BNC connectors.

Likewise.

SAG1021 connected with supplied USB cable only. Measurement between scope BNC (shells) and SAG1021 output BNC; 86mV (AWG OFF) ~50% more with AWG ON.
With supplied BNC cable attached between SAG1021 BNC and scope ch4: 8mV

I think you incidentally uploaded one picture twice. So the measurement cannot be seen.

Measured with Fluke 15B.

Is there a way to open the AWG without ending up with something ugly?

It would be nice to see the wiring....

I'll not be doing a teardown on a new unit for sale. 

[HendriXML]

I'll not be doing a teardown on a new unit for sale. 

I wouldn’t have mind an improved one  .

[tinhead]

I'll not be doing a teardown on a new unit for sale. 

I wouldn’t have mind an improved one  .

i can post later today some pictures if you wish, i did already opened my SAG (and made bitstream / eeprom backup)

[HendriXML]

I'll not be doing a teardown on a new unit for sale. 

I wouldn’t have mind an improved one  .

i can post later today some pictures if you wish, i did already opened my SAG (and made bitstream / eeprom backup)

That would be great! I guess you open it by removing the rubber feets to access some screws?

Do you have any thoughts on tackling this issue? Essentially there will be some hundreds of mA going through the ground lead if used with earths ground points. That might result in unexpected behavior in many situations.

[tinhead]

honestly same results, offset 15mV, voltage 85mV between BNCs.

Sure, when i disconnect everything and do auto zero adjust with no load, the DC offset is something like 400uV only (measured with Yokogawa 7552). Of course on SDS i have still 15mV offset due to the USB issue. To get rid of that, i can of course connect SAG to CH1, with provided cable (high-z), and run auto zero adjust (as described in manual), then the offset goes down to 500uV-1mV (my BNC is bit wacky). Of course only on SDS display, in the reality there is still remaining 13-14mV offset (when measured with DMM). So yeah, one can calibrate for nice picture of SAG signal on SDS or for real use. Will check on weekend how this perform when using usb isolator and external power supply, open circuit calibrated.

Attached usb pic, other pics are on my 1drv (Siglent folder): https://1drv.ms/f/s!AiY46m4u993EgSg4ZGmBREL7uZRz

[HendriXML]

At least the current the AWG draws is stable, otherwise it would not “only be an offset”. 

Maybe it’s best to solder some powerwires at the point where the most current is drawn, or where the GND and neg V split (but I expect they don’t).

I continued with the measurements I wanted to do and got used to hook the AWG up to the power supply and using the short circuit cable.

The third channel of the PSU has only fixed V-settings which finally has some good usage now.

The “offset” is determined by the usb and whatever probing cables / awg cables are in play. They become parallel resistors each taking some of the 360 mA. So it might change during a session...while connecting and disconnecting earth leads.

I don’t want to think about such stuff while doing measurements, so I’m thinking about modding the thing when a better solution is found.

[HendriXML]

Attached usb pic, other pics are on my 1drv (Siglent folder): https://1drv.ms/f/s!AiY46m4u993EgSg4ZGmBREL7uZRz

Thanks!

Now I see where the screws are, it would be my second guess. And the reason I was afraid I might endup with something looking a bit more ugly. But outside ugliness  is better than inside ugliness  .

After those screws the PCB can slide out? Or is something attached to the front as well?

[HendriXML]

Maybe it’s best to solder some powerwires at the point where the most current is drawn, or where the GND and neg V split (but I expect they don’t).

Looking at photo’s (and seeing the ground planes) I think “the solution” always involves an external power source. However the V-neg should not start at the USB side, but at the BNC connectors. Just as the usb gnd wire. (Why not use that?)  That way the V-neg current path is “seperated” from the GND path.

This solution can be done without opening the AWG, just a BNC T-splitter with a good ground lead and a V-neg. followed by a V-pos injection and GND/V-neg break in the USB cable.

So a combination of my 2 Frankenstein projects!!!

[Rerouter]

The screws are hidden behind the face stickers. Not the feet. I have some pictures i can uplosd in about 6 hours