Rigol DS1052E Wrong Reading at DC

[Mechatrommer]

IIRC i read something about this somewhere here, but cannot find the thread anymore.

last night i tinker with my Hantek 3x25 circuitry, and to make story short i checked (supposedly) 5V supply from PC USB using Rigol DS1052E, it read as steady 4.3V. sceptic, i opened up my Uni-T 71A to re-check the reading, it 3.96V! , more sceptic i tried to verify with my cheapo SUNWA DT-890G, and its tally with Uni-T at 3.96V reading. i re-check back with Rigol at different setting 2V/div (initially), and then 1V/div and then 500mV/div, still 4.3V reading from Rigol ok now who should i have to blame?

[Simon]

trust your meter, but that's low for a 5V supply

[Psi]

probe on x1/x10?

[Mechatrommer]

x10. i remember resetting it and re-compensate my Rigol probe. yes Simon correct, thats low 3.96V vs 5V, will check it back later, maybe the circuit bring the V down, i dont know. if i have to trust my meter, then my Rigol is wrong, is it normal or is it far off?

[scrat]

I'd trust the meter.
On the Rigol, such an error, especially if it is constant with voltage scale, is quite strange.
What does it show when you set the input GND coupled? If you set an offset (which is the position, IIRC, on the Rigol) does the voltage change? And if you swap the probes, or use a simple BNC cable?

[saturation]

The Rigol I have, unmodded, is at worst, 4% error consistently.  Before using it for any type of accuracy checks, you should self-cal the scope.  For probes, its better to be done x1, its DC anyway so frequency response isn't important, the voltage divider of x10 maybe not be precisely 10:1 ratio to make as accurate as possible, voltage measurements.

Self cal is best done everytime the weather changes: humidity, heat etc., particularly if your test environment has no tight climate control, you may have to do it several times during the workday.  You've nothing to lose since its all automatic, do it when you need to take a coffee break between work.

[Fraser]

Did't your DS1052E suffer a calibration/offset error during an attempted upgrade ?

I know it was recovered for your by a helpful member of this forum but did he have to use some default (best guess) calibration levels to replace those that had been overwritten in your DSO's calibration tables ? It might be worth asking him.

OT - I know the ICOM PCR1000 receiver can suffer this sort of problem. If the RF stage calibration EEPROM gets corrupted, and you don't have the original (unique) calibration settings backed-up, you have to load some settings that have been dowloaded from another PCR1000. They will not be exactly right but at least the receiver works again, but not as well as if the units own unique calibration settings were in the EEPROM.

[Mechatrommer]

@saturation mine is (4.3-3.96)/3.96 x 100% = 8.6% off!
Yes Aurora, you've dig well enough, this was a "crash recovered calibration data unit". Concerned, made some test after 30hours warm up, and twice the self calibration, probes compensated. The measurement to the Hantek circuit is still the same as (3) no change from last night, ie 27.4% off, but when testing other (AA and LA battery), i got satisfactory reading. so made me confused. how come the measurement is off at only certain value and correct at others? at a single frequency off 0Hz? does dso have calibration data at each volt level at each frequncy?

no. dmm      my rigol
1) 1.366V   1.3-1.4V    OK
2) 4.1V     4.1V        OK
3) 3.96V    4.3-4.5V    KO
4) 11.85V   11.8-12V    OK

and to Simon, i checked the USB issue. when the Hantek is disconnected, i checked the USB post is 5.1V, which should be within spec. but when the Hantek is connected, the voltage drop to 3.96. So i believe the Hantek is hogging the voltage down, a bug? (i checked the trace i was checking is indeed come directly from USB power line).

[bilko]

@saturation mine is (4.3-3.96)/3.96 x 100% = 27.4% off!
Yes Aurora, you've dig well enough, this was a "crash recovered calibration data unit". Concerned, made some test after 30hours warm up, and twice the self calibration, probes compensated. The measurement to the Hantek circuit is still the same as (3) no change from last night, ie 27.4% off, but when testing other (AA and LA battery), i got satisfactory reading. so made me confused. how come the measurement is off at only certain value and correct at others? at a single frequency off 0Hz? does dso have calibration data at each volt level at each frequncy?

no. dmm      my rigol
1) 1.366V   1.3-1.4V    OK
2) 4.1V     4.1V        OK
3) 3.96V    4.3-4.5V    KO
4) 11.85V   11.8-12V    OK

and to Simon, i checked the USB issue. when the Hantek is disconnected, i checked the USB post is 5.1V, which should be within spec. but when the Hantek is connected, the voltage drop to 3.96. So i believe the Hantek is hogging the voltage down, a bug? (i checked the trace i was checking is indeed come directly from USB power line).

Did you try using an isolated voltage source i.e. a battery 9v cell and compare the voltages ? The problem may be related to the earthing between the pieces of equipment. The scope is grounded, the meters I assume are not.

[Mechatrommer]

Did you try using an isolated voltage source i.e. a battery 9v cell and compare the voltages ? The problem may be related to the earthing between the pieces of equipment. The scope is grounded, the meters I assume are not.

all the later test are using battery (AA and lead acid) showing the correct voltage. but how come? the rigol is earthed, the hantek which is connected to pc which is earthed also, so they should be on the same ground? did i miss something?

[bilko]

Did you try using an isolated voltage source i.e. a battery 9v cell and compare the voltages ? The problem may be related to the earthing between the pieces of equipment. The scope is grounded, the meters I assume are not.

all the later test are using battery (AA and lead acid) showing the correct voltage. but how come? the rigol is earthed, the hantek which is connected to pc which is earthed also, so they should be on the same ground? did i miss something?

There could be some slight difference in earth potential or noise on the ground, probably requires further investigation. Do you confirm that errors only appear if you measure voltages on equipment that have a connection to earth ?

[grenert]

@saturation mine is (4.3-3.96)/3.96 x 100% = 27.4% off!

That's some funny math, there.

[Mechatrommer]

ok. i think i know now where to blame, its the usb cable used to power the hantek. somehow it deteriorated. changed to other usb cable then no more issue. the 3.96v usb anomaly also gone, now its back to 5V on the circuit. so my rigol, hantek and 2 other dmm may sleep in peace. thanx everybody for giving ideas. so next time, check your usb cable!

ps: i tested yachtronics idea, no. my ground within this room all have same potential. thanx now i confirmed also my room's ground is ok.
ps: opps grenert! i dont know what i was typing. maybe a curse. its 8.6% not 27% (corrected) sorry, still off saturation's standard 4%.
ps: to all 3x25 owner, dont be alarmed, i was not using the original cable came with it, they all got mixed up already.

[grenert]

Wow, did the bad USB cable look any different?  I have so many cheapo USB cables around, I wonder how many are failing...

[Mechatrommer]

Wow, did the bad USB cable look any different?  I have so many cheapo USB cables around, I wonder how many are failing...

no it looks quality one with metal (silver) braided shielding, but quite old, long time in storage, it got yellowish color to its transparent jacket. i also got that type connected to a printer, i'll check it later.

[Russel]

I once had a color printer at the office that would shift the colors of printouts. If I remember correctly cyan and black were fine but magenta and yellow shifted, more so toward the right side of the printout. I couldn't find anything wrong with the printer. It turned out that the USB cable was damaged, where it came straight out of the printer in the back, by being pushed against the wall. The printer worked fine with a new USB cable. That was one printer problem that had me scratching my head for a while.

[Fraser]

Be warned....

I have had several cases where external hard disk drives appeared to fail or would not start. In every case the cause was a USB cable, It would appear that with copper prices being high, some usb cables have an appallingly small cross sectional area of copper in the conductors, presenting significant resistance and voltage drop when current is drawn. The thin conductors also crack and become intermittant at the entry to the plug strain relief. USB hard disks really test cheap USB leads due to the current they draw on spin-up.... hence why some drives try to spin then give up and produce a hiccup effect that can be misinterpretted as a failed hard disk.

This can be a pig to diagnose in the field if you don't have a spare, good quality, USB lead to substitute for the suspect one.

[Mechatrommer]

This can be a pig to diagnose in the field if you don't have a spare, good quality, USB lead to substitute for the suspect one.

luckier if you suspect the cable to be the pig. i almost blame a $400 dso for that. with all this marketing/politic/economic bullshit (copper price) maybe i should thinking just DIY'ing the usb cable, to put extra diameter to it. good tips on HDD usb cable test. thanx

[Psi]

Do a calibrate and then test your 3.9v signal on 500mV / div and with the probe and scope set to x1
Also make sure your measuring the avg voltage and not peak or something

A while back I did some readings on my rigol to see just how much of an effect the 8 bit vertical resolution had
This was with a 5.00V signal

1x probe 50v/div = not possible 20v/div = not possible 10v/div = 5.4V = 8.3% error 5v/div = 4.99V = 0.2% error 2v/div = 5.08V = 1.6% error 1vdiv = 5.05V = 1% error

10x probe 50v/div = 6.63V = 32% error 20v/div = 6.00V = 20% error 10v/div = 5.73V = 14.6% error 5v/div = 5.02V = 0.4% error 2v/div = 5.01V = 0.2% error 1vdiv = 5.03V = 0.6% error

Also, i've seen problems where people have test equipment sitting on top of other test equipment which has a large 50hz transformer inside. The magnetic field causes all sorts of problems, ya just have to move it a bit further away and its fine. So that might be something to look at.

[Mechatrommer]

thanx psi.
1) how do you manage to read to 2 decimal places? i have to look carefully to estimate the 1st decimal on the graticule.
2) i can understand error when reading small value with big V/div. eg. reading 5V with 50V/div. for 50V/div. i will accept error around 1 to 2 tenth of div. ie for 50V/div, i will accept error around ±5 or worst ±10V.
3) fyi while studying data received by usb computer. rigol will send less than 8bit resolution data. ie on the screen what you are seeing is only around 200 byte valur resolution ie around 7.6bit only. not the famous 8bit.

[sigxcpu]

The DSO has measurements, you don't need to read the graticules

[saturation]

Its good practice to eyeball graticules, at least with the Rigol.  It measurements are based on the screen image, not internally, so its maximum accuracy is when the image is biggest, about 8 div peak to peak, and one full cycle.  When measuring far below its optimal setting, such as noise on a tracing that is 1/2 div or less, the manual graticule measurements are more accurate; you'll can then focus on it by adding more gain to isolate that section of the signal to get an accurate automatic measurement or use the manual cursors.

[Psi]

thanx psi.
1) how do you manage to read to 2 decimal places? i have to look carefully to estimate the 1st decimal on the graticule.

lol, i assumed you were measuring using the rigols measure functions. If you've never used them you should try them out.
You can tell it to measure all sorts of stuff (frequency and voltage in all sorts of ways, rms, peak, avg etc.) and it displays the value onscreen and in real time. As saturation says though, its only as good as the lcd resolution used, since its based on the image. Still more accurate than eyeballing the graduations in most situations though.

[Mechatrommer]

i usually use measure or cursor if i want to read frequency, min max etc or other more complicated reading, or maybe if the channel volt offset is not within screen. reading volt from graticule and hand calculation is faster for me without messing the screen with windows/lines etc.