Issues connecting to eScribe, through a powered USB hub

[SSV]

Noticed one potential problem....

dna200 does NOT like to connect to eScribe through a powered USB hub.  This could present a potential problem when trying to set USB charge current thresholds and validate them, through device monitor.

Tried 2 different hubs....and a few different scenarios....scenario one is hub direct into MOBO usb port.  Scenario 2 is same hubs through an active USB extension cable....into same MOBO port.  eScribe rejected connection in all cases.

I noticed USB/serial is not handled through FTDI chipset (kudos to evolv for rejecting FTDIgate).

Could go into more details, but waiting to hear back first.....this is going to create a real potential issue when trying to calibrate and verify MAX USB current draw/thresholds.....I def don't want to try and pull an amp from ANY MOBO USB port (I don't care how fat the power traces are going to the port, anything above 420mA is against USB baseline specs....and can't be qualified or guaranteed reliable through 100% of potential USB scenarios).

Standing by with logs and more info if necessary....

[John]

We haven't had difficulty with powered hubs here, what brand and model is causing you trouble? The device won't violate USB spec if it sees data lines, so that isn't a great worry.

[SSV]

Ok, good to know.  I wanted to verify the current draw at MAX.  I tested the current draw externally on internal USB (mobo) port and it was within spec.  DNA200 monitor and on screen read 0.480 ish, but not wildly out of range.  One thing I did notice is that the board wouldn't enter charge display when connected to eScribe/monitor.

The hubs were Dlink and IOgear (I can verify model numbers if necessary).  The easier bet would be to test with a known working powered hub.  Could you tell me which hubs you guys tested?  I can narrow it down then.  It's possible this is 100% on my end.

So on the issue of sensing datalines, if a user mobo had a faulty dataline (for whatever reason) would the dna200 attempt to pull MAX current from the port?  This is what I wanted to verify.  I was thinking of building an inline shunt (into a USB cable) and hooking that up to a 2 amp powered hub and seeing if I could verify current draw under a potential fault condition.  

Thanks for the swift reply.  I have a few more potential issues, but will take them one at a time.

[John]

Do you have the powered hub connected to a PC? We've tested with amazon brand, sentry, ukonnect, pluggable, and a variety of generics. 

USB spec for a high power device is 500 milliamps (5 unit loads), and we tell the computer we will be drawing that much. If the hub can't supply that much or if other things are plugged into the same hub and it can't power all of them it could decline to connect. 

http://www.ti.com/lit/an/slyt118/slyt118.pdf if you don't care to slog through the actual USB docs.

Where do you get a value of 420 ma from?

The DNA 200 will in all cases limit current draw to keep the voltage at the port above 4.45 volts, so it won't overdraw or brown out a computer port regardless. If you want to do that test, simply purchase one of the moronic "charge only" USB cables that have d+ shorted to d- to emulate a charger.

I thought I had one that wasn't working for a second, but windows was just taking its time installing the hub. 

[SSV]

John said:

Do you have the powered hub connected to a PC? We've tested with amazon brand, sentry, ukonnect, pluggable, and a variety of generics. 

USB spec for a high power device is 500 milliamps (5 unit loads), and we tell the computer we will be drawing that much. If the hub can't supply that much or if other things are plugged into the same hub and it can't power all of them it could decline to connect. 

http://www.ti.com/lit/an/slyt118/slyt118.pdf if you don't care to slog through the actual USB docs.

Where do you get a value of 420 ma from?

Yes powered hubs were connected to the same port on the MOBO.....no connection to eScribe through them was allowed....

I measured the 420mA through a test cable I have built.....shunt inline....straight from MOBO output (standardized port)....virtually every consumer port i have tested has been around 420mA....with shunt losses nulled and full load.  V usually sags to around 4.4 ish V....unfortunately some manufacturers (quite a few actually) aren't building USB to spec.....I wanted to confirm various survival scenarios with the dna200 pulling MAX (under port fault condition).  

If I recall correctly USB max unit load = 100mA x 5 units?  Minus associated V droop....so 500mA under load is only possible if V potential remains 5V?

P.S. is there a way to force the dna200 unit to enter charge display, while connected to eScribe?

[John]

Milliamps is milliamps, voltage losses don't change it.

https://en.wikipedia.org/wiki/Kirchhoff's_circuit_laws

So I'm not entirely clear what you mean by "shunt losses nulled."

The DNA 200 and all responsible USB devices will limit current if the port can't hold voltage, but that doesn't change the 500 mA spec. 

There isn't a way to force charge display, but you can put charge parameters on the main display if you're interested. 

[SSV]

John said:

Milliamps is milliamps, voltage losses don't change it.

https://en.wikipedia.org/wiki/Kirchhoff's_circuit_laws

So I'm not entirely clear what you mean by "shunt losses nulled."

The DNA 200 and all responsible USB devices will limit current if the port can't hold voltage, but that doesn't change the 500 mA spec. 

There isn't a way to force charge display, but you can put charge parameters on the main display if you're interested. 

Quite familiar with Kichhoffs laws of potentials (I did at least make it through EE101)

I am talking about the specific relationship to USB potentials.  Correct me if I am wrong but USB max current is supposed to be limited with v droop?  I.E. 5v should equal max potential of 100mA x 5 units, but anything below (3.3v) should limit current slightly?  Not as a function of current potentials, but as a function of USB spec.  Isn't that a limiting factor in USB spec to avoid undue heat in traces?  500mA at lower V, if the load is dynamic= more total TDP....what I am trying to verify here is the correlation between device monitor readings and dna200 screen readings (vs actual draw with the shunt inserted on the low side of USB)....I am assuming they are sourced from the same variables in the MCU.....is there any interpolation loss going through device monitor that could account for an offset between device monitor and dna200 screen readings?


By shunt losses nulled I mean the entire measurement loop (probes etc) nulled against vREF (taking into account thermoelectric effects in probes etc).  I usually run the input stage of the DMM (dm3068) @ 10gOhm (so the burden from the DMM should be negligible).

[John]

Current law, not law of potentials (voltage law)

USB power delivery is a safe zone box. For a standard USB port, the box extends from 4.5v to 5.5v and from 0mA to 500mA. 

Outside the box the device should limit current to move back into the box. 

Realistically ports are built with enough output resistance that they typically droop to the minimum of the spec at the maximum current, but that is a designed behaviour, as it is the cheapest way to protect against short circuits. 

The heat generated in the wires is a function only of the current flowing through the wires and the resistance of the wire. I squared R, voltage doesn't enter into it. 

But I think I see what you're getting at. You want to see if our USB current measurement is accurate. If you have a good ammeter, just strip the VBus line out of a sacrificial cable and run it in series through your ammeter. The voltage supplied to the DNA 200 will be a little lower due to sense resistor loss inside your meter, but the current has only one path, so the current read by the meter will be exactly the current drawn and sensed by the 200. 

[SSV]

John said:

Current law, not law of potentials (voltage law)

USB power delivery is a safe zone box. For a standard USB port, the box extends from 4.5v to 5.5v and from 0mA to 500mA. 

Outside the box the device should limit current to move back into the box. 

Realistically ports are built with enough output resistance that they typically droop to the minimum of the spec at the maximum current, but that is a designed behaviour, as it is the cheapest way to protect against short circuits. 

The heat generated in the wires is a function only of the current flowing through the wires and the resistance of the wire. I squared R, voltage doesn't enter into it. 

But I think I see what you're getting at. You want to see if our USB current measurement is accurate. If you have a good ammeter, just strip the VBus line out of a sacrificial cable and run it in series through your ammeter. The voltage supplied to the DNA 200 will be a little lower due to sense resistor loss inside your meter, but the current has only one path, so the current read by the meter will be exactly the current drawn and sensed by the 200. 

I want to number one correlate device monitor with dna200 on-screen readings, vs external shunt (be it physical 1mV/A shunt, or the shunt in the dm3068).  I typically use an external shunt to try and avoid DMM burdens etc, but I see your point about port output resistance....chances are under a fault condition the limiting factor is the physical nature of the port? 

I have had some issues with other on-board USB charging devices blowing out MOBO usb ports (fault conditions)....and I want to verify that under a fault condition this wouldn't happen with the dna200 (via USB charging).  If we decide to package the chip with our wire profiles built in, I would like to offer customers a warranty and guarantee that charging is reliable.  This seems to be the number one complaint about the dna200 right now (which I don't agree with at all) is that it's limited to the internal 3s pack and on board charging.  I want to give any potential customers confidence, that the device is "safe" via USB.  One good way to do that is simulate a fault condition and see if I can get the DNA200 to pull more than 500mA from an industry spec USB port.

You'll have to please excuse any lack of knowledge I have on USB specifics, as it's not my field or forte' (I don't deal much with serial related devices....I am mostly in DC/DC hardware...and a piss poor software engineer....I know enough about serial protocols and coding to be annoying, but far from implementation level knowledge). 

P.S. even in DCv setting on the DMM there is slight burden from the meter....I am not sure it's relevant to the granularity of measurement here, but I typically try and null all uncertainty from a DUT.....overkill?  probably, but my background is in precision devices....so we have plenty of hardware here with overly fancy binding posts (IETlabs seems to have a large chunk of my bank account in their bank account). I'll strip down a few cables and give your test loop suggestions a try. 

P.P.S   If USB spec device (linear reg side) sees a 4.4V signal....it should limit current to move back into it's safe "bin" correct?

[SSV]

SSV said:

[QUOTE=John]Current law, not law of potentials (voltage law)

USB power delivery is a safe zone box. For a standard USB port, the box extends from 4.5v to 5.5v and from 0mA to 500mA. 

Outside the box the device should limit current to move back into the box. 

Realistically ports are built with enough output resistance that they typically droop to the minimum of the spec at the maximum current, but that is a designed behaviour, as it is the cheapest way to protect against short circuits. 

The heat generated in the wires is a function only of the current flowing through the wires and the resistance of the wire. I squared R, voltage doesn't enter into it. 

But I think I see what you're getting at. You want to see if our USB current measurement is accurate. If you have a good ammeter, just strip the VBus line out of a sacrificial cable and run it in series through your ammeter. The voltage supplied to the DNA 200 will be a little lower due to sense resistor loss inside your meter, but the current has only one path, so the current read by the meter will be exactly the current drawn and sensed by the 200. 

I want to number one correlate device monitor with dna200 on-screen readings, vs external shunt (be it physical 1mV/A shunt, or the shunt in the dm3068).  I typically use an external shunt to try and avoid DMM burdens etc, but I see your point about port output resistance....chances are under a fault condition the limiting factor is the physical nature of the port? 

I have had some issues with other on-board USB charging devices blowing out MOBO usb ports (fault conditions)....and I want to verify that under a fault condition this wouldn't happen with the dna200 (via USB charging).  If we decide to package the chip with our wire profiles built in, I would like to offer customers a warranty and guarantee that charging is reliable.  This seems to be the number one complaint about the dna200 right now (which I don't agree with at all) is that it's limited to the internal 3s pack and on board charging.  I want to give any potential customers confidence, that the device is "safe" via USB.  One good way to do that is simulate a fault condition and see if I can get the DNA200 to pull more than 500mA from an industry spec USB port (I typically see around 420mA when I do this test, but saw 480mA draw from the dna200 @ 4.4v).  This 4.4v vs 480mA I saw in device monitor was the trigger that set this question off for me....at 4.4v 480mA is still within tolerance? If 4.4V bins the V as "low" shouldn't the current draw be reduced?  Or is the controller fast enough to "bin" that low reading and say 480mA does not = 500mA, and is therefore an acceptable reduction of current for that scenario?

You'll have to please excuse any lack of knowledge I have on USB specifics, as it's not my field or forte' (I don't deal much with serial related devices....I am mostly in DC/DC hardware...and a piss poor software engineer....I know enough about serial protocols and coding to be annoying, but far from implementation level knowledge). 

P.S. even in DCv setting on the DMM there is slight burden from the meter....I am not sure it's relevant to the granularity of measurement here, but I typically try and null all uncertainty from a DUT.....overkill?  probably, but my background is in precision devices....so we have plenty of hardware here with overly fancy binding posts (IETlabs seems to have a large chunk of my bank account in their bank account). I'll strip down a few cables and give your test loop suggestions a try. 

P.P.S   If USB spec device (linear reg side) sees a 4.4V signal....it should limit current to move back into it's safe "bin" correct?[/QUOTE]

[John]

The "burden" of the shunt resistance as you call it will certainly depress the output voltage getting through the cable, but as you're trying to correlate current to current, I'd think that your meter is already compensating for any nanoamps lost through the measurement circuit. 

In terms of running a test to see if it will blow ports, that's totally valid. Attached to a data port, it is programmed to draw only 500 mA. What you want to try is using a "charge only" cable as those pretend to be a charger regardless of what the host device actually is. 

But absolutely, beat on it and report back what you get.

[SSV]

John said:

The "burden" of the shunt resistance as you call it will certainly depress the output voltage getting through the cable, but as you're trying to correlate current to current, I'd think that your meter is already compensating for any nanoamps lost through the measurement circuit. 

In terms of running a test to see if it will blow ports, that's totally valid. Attached to a data port, it is programmed to draw only 500 mA. What you want to try is using a "charge only" cable as those pretend to be a charger regardless of what the host device actually is. 

But absolutely, beat on it and report back what you get.

Will do, I'll grab some anemic "charge only" cables and see if I can pop something...I'll use your suggested method (just strip back the cable and put it into DC (I) side of the DMM). 

As far as burdens, I am referring to the burden voltage of the DMM itself....which is why I typically try and use the shunt into DCv.....but again I am typically measuring uA and such.  Probably not at all relevant to this test, but I do ALWAYS try to null any external uncertainty from the DUT.

http://zone.ni.com/reference/en-XX/help/370384N-01/dmm/burden_voltage/

http://www.fluke.com/fluke/uses/comunidad/fluke-news-plus/articlecategories/electrical/burdenvoltage

^----again probably overkill for this scenario, but I am creature of habit.....obviously the device monitor and dna200 screen are only reading out to mA, and I am looking at uA....but I do need to justify the price of those least significant digits on my bench meter...  I'll try running the inputs at both 10mOhm and 10gOhm and see if I can account for any uncertainty there...

P.S. having the ability to force charge mode in eScribe could be VERY useful....I'm sure that is low on the priority list right now, but would be nice to test screen layouts and correlation to external measurements + device monitor....

[protovapor]

The board switches to charge display mode normally while connected to escribe ...

[SSV]

protovapor said:

The board switches to charge display mode normally while connected to escribe ...

running latest versions of both eScribe and board FW and it my unit does not go into charge display mode when connected....the unit will charge, but display does not change over to secondary charge mode parameters....however it does when I disconnect from eScribe

[SSV]

[QUOTE=John]

There isn't a way to force charge display, but you can put charge parameters on the main display if you're interested. 

[/QUOTE]

John seems to be confirming what my test board is doing....@protovapor

[John]

Ah ha. Here's the confusion. 

The load you're measuring isn't resistive.

Burden voltage doesn't enter into it to measure the current accuracy of our charge sensor/display. Microamps or not. 

So I think we're on the same page now. Test away. 

[SSV]

John said:

Ah ha. Here's the confusion. 

The load you're measuring isn't resistive.

Burden voltage doesn't enter into it to measure the current accuracy of our charge sensor/display. Microamps or not. 

So I think we're on the same page now. Test away. 

ahhhh.....is the load purely capacitive (no ballast resistors on the input stage)? that def quenches my concerns....(I mean obviously any load is resistive, but if it's = to free air I see your point)

[protovapor]

[QUOTE=SSV][QUOTE=John]

There isn't a way to force charge display, but you can put charge parameters on the main display if you're interested. 

[/QUOTE]

John seems to be confirming what my test board is doing....@protovapor

[/QUOTE]

There is no command to immediately make it switch to that mode, however it will switch normally based on your time parameters. To my understanding, it switches after "active time" and "idle time" settings both elapse. If you set both of those to a very short time you can sort of get it to switch on demand.

[SSV]

[QUOTE=protovapor][QUOTE=SSV][QUOTE=John]

There isn't a way to force charge display, but you can put charge parameters on the main display if you're interested. 

[/QUOTE]

John seems to be confirming what my test board is doing....@protovapor

[/QUOTE]

There is no command to immediately make it switch to that mode, however it will switch normally based on your time parameters. To my understanding, it switches after "active time" and "idle time" settings both elapse. If you set both of those to a very short time you can sort of get it to switch on demand.[/QUOTE]

mine is not doing that.....can you test/verify that on one of your units?  If my test unit is connected to eScribe, it doesn't switch over to charge display at any point....I did play with times and nothing....*scratches head* 

[protovapor]

Yes verified on a bare board as well as on test mods. Do you have the screen set to be on (brightness >0) during charge?

[John]

I think he is using device monitor, so it sees the device as active. 

[protovapor]

It will switch even if being monitored... until it is fired of course

[SSV]

protovapor said:

Yes verified on a bare board as well as on test mods. Do you have the screen set to be on (brightness >0) during charge?

No, brightness was set on 50% for that scenario.....

I am going to try and reload the entire eScribe suite and FW......I am very puzzled why my unit won't go into charge display while connected, and another will.....

[SSV]

ok this might be a potential bug.....I just set the idle time to 10 seconds and it's catching now....I see charge stats onscreen AND via device monitor, and via my eternal test jig correlates those numbers....let me start ramping up idle times again and see where it stops behaving correctly....I did reload eScribe etc....will report back if the issue crops up again....

Thanks for the feedback @protovapor I can scratch this issue off the list for now....

[SSV]

idle time up to 30 seconds and still working fine....switching over to charge display after 30 seconds (timed it) and still showing stats correctly in device monitor, dna200 display and my external test gear....current draw is accurate (minus cabling losses)