Tekway DST1102B Vertical amp drift

[bilko]

I Have a feature with my Tekway DST1102B scope.
When switched on from cold, channel 1 trace has drifted. It appears to settle down after about 30 minutes and goes to the calibrated zero point.

If CH1 is switched from AC to DC coupling it makes no difference, but when coupling is set to ground the trace goes to zero.

It is repeatable, CH2 doesn't do this it is always at zero from cold CH2 does exactly the same.
If I press invert the trace goes positive by the same amount

Any ideas what could be causing it, reminds me of the old valve scopes ?

Forgot to mention, there are no probes connected

[bilko]

Switching the probe multiplier makes no difference, the range switch does. I assume then that the problem is before the range selection. I also have a Rigol, this certainly does not exhibit these characteristics. Do I have a faulty scope ?

[tinhead]

rule 1 - you should always use warmed-up measurment device.
rule 2 - if you don't like to wait run self-calibration on cold device, the drift will be gone for next 25 mins ... but no, alway use rule 1

honestly, if you power up the DSO for 5 mminutes to quick measure something you can alway change manualy the position.
If you work daily, you can power up and wait until the DSO warmed-up (actually everything need warm-up time)

[bilko]

@tinhead, thanks for the reply.
The scope has been on for about two hours, I've done three calibrations. Each time I calibrate it is Ok for a while. I can't do any auto measurements because the trace keeps drifting.

Does your scope do the same ?
This feature does not exist on my Rigol scope

[tinhead]

When switched on from cold, channel 1 trace has drifted. It appears to settle down after about 30 minutes and goes to the calibrated zero point.

...

The scope has been on for about two hours, I've done three calibrations. Each time I calibrate it is Ok for a while. I can't do any auto measurements because the trace keeps drifting.

...

so which statement is true? 

Normally, if you did calibrated once by room temperature, the next time you boot up (if the room temp is same) the DSO
will be "calibrated" after let say 5 mins.

Exception - the room temperature changed, so if you open window in winter and in your room rapidly droped the DSO
will start to drift - if you close the windows wt will drift again in a different direction.

Drift after two hrs warm-up with constant room temperature = broken DSO.

[bilko]

Room temperature is constant, 23 deg C
I take your point. The scope does seem to drift more than I am used to. The trace drifts back to calibration very slowly, more than 5 minutes maybe 20 minutes. Its like waiting for a big capacitor to get to its terminal voltage, exponentially.

This is only noticeable on minimum voltage settings 20mV with a x10 multiplier. I suppose the real error is in the order of a few millivolts, but this gets multiplied up when doing auto measurements.

I assume that if I leave it switched on for over two hours the drift becomes insignificant.
It does appear to be quite temperature sensitive, more its own internal temperature rather than ambient.

Could there be a problem with PSU voltages ?
If I select ground in the coupling menu, the trace always goes to zero. Do you know if the input is actually grounded or if the firmware just draws a line at zero level ?

Its an annoying feature, I am just trying to find out if it is common or if I have a defect with the scope

[tinhead]

Could there be a problem with PSU voltages ?
If I select ground in the coupling menu, the trace always goes to zero. Do you know if the input is actually grounded or if the firmware just draws a line at zero level ?

the input itself will be not grounded, but the first opamp will be disabled, so it is more or less equal to grounded input.
This drift is common for Tekway/Hantek, on other devices you will see maybe less drift but the measured data
will be still crap. You have always to wait until device properly warmed-up, how long depends on device/manufacturer, generally spoken
something between 10-15 mins is good enough for fast measurments, 30mins for exact measurments.

[bilko]

OK, I've done a quick test.
The scope was settled and stable after being powered on for the last two hours.
I switched it off for 5 minutes to allow it to cool down.
When I switched back on, the trace had gone negative by 20mV on the grid.
Within 5 minutes of being powered up it was back up to zero.
The room temperature was constant 23 deg C.

Measuring the scope temperature with an infra-red thermometer, the case temperature is not uniform. There are hot spots just where the handle is (above the PSU), the temperature here was around 53 deg C  (30 deg C over ambient), the rest of the case is around 32 deg C.

The measurement board is at the bottom of the case so should not be significantly affected by the PSU temperature. Problem is that there is a large temperature gradient within the scope housing. The Rigol scope I have is the DS1052E, it is fan cooled so that may explain why there is a difference. It probably stabilises a lot quicker with the forced air flow.

Thanks for your input and help

[Mechatrommer]

if you want best without waiting to warm up time, put the unit in temp controlled chamber and set the temp to the circuit operation temp after 30 minutes outdoor. so next time boot, its already warmed up.

[bilko]

or just leave it switched on 

[saturation]

Many electronics that operate at a noticeably warm temperature are at its best with a warm up period, to settle components.  Its not that it won't function, it just won't be at its best for a precision instrument.  Not just scopes but even plasma TVs and LCD TVs are preferred to be warmed up before making calibration adjustments.

Yes, some metrological instruments are rarely to never shut off and placed in climate controlled rooms, all for optimal response.

FWIW, the Rigol 1000e manual claims calibration should be done frequently, not once in blue moon, if you require accurate voltage levels.  I calibrate most every time I work, since its just a matter of hitting a few buttons before going for coffee.

or just leave it switched on  

[am2pgs]

My Hantek (equivalent to tekway) does the same , but more noticable on one channel. I always thought it is more or less normal for this class of scopes and due to offset voltage/bias currents on the input stage.

Based on what Tinhead says on "Input Grounding", one has to assume it is the first op-amp that causes the offset. I take it that it is fairly high power/bandwidth op-amp, so I would expect it to heat up internally quite a bit.

So when the unit is switched off afte having warmed up, even for five minutes, the internal temperature of the op-amp will fall rapidly to the temperature of sourrounding areas.

[Mechatrommer]

FWIW, the Rigol 1000e manual claims calibration should be done frequently, not once in blue moon

i dont know. this is what i got from rigol_DS1000_series_manual.pdf that i have

"The recommended calibration interval is one year"

[saturation]

User’s Guide for DS1000E, DS1000D Series
page 2-84
Self-Calibration
The Self-Calibration adjusts the internal circuitry to get the best accuracy. Use these functions to calibrate the vertical and horizontal systems. For maximum accuracy at any time, run this calibration if the ambient temperature changes by 5°C or more.
Disconnect any probes or cables from all channel inputs, otherwise failure or damage to the oscilloscope may occur. Press Utility?Self-Cal to enter the self-calibration interface as follows. Press “RUN/STOP” key to start the operation and press “AUTO” key to exit the system.
Figure 2- 116
Self calibration
NOTE: The oscilloscope must have been working or warm-up at least 30-minutes before running self-calibration to get best accuracy.


The 'other' calibration which is an annual one described on page 5-7, is the NIST style traceable calibration, if one thinks this is necessary or required by your operation.

FWIW, the Rigol 1000e manual claims calibration should be done frequently, not once in blue moon

i dont know. this is what i got from rigol_DS1000_series_manual.pdf that i have

"The recommended calibration interval is one year"

[Mechatrommer]

well, that 5°C difference... may means what you said could be everyday or every hour i just afraid wearing it off much frequently. ymmv.

[saturation]

Its mostly for best accuracy.  5C is 42F, you really only need to do this when the season changes.  In the USA, room temperature in the winter is about 68F, but in the summer with air conditioning, its about 75-80F, time for a calibration.

Even in the hot Pacific, at night its about 75-80F, heat of noontime about 120F, still within spec for calibration, but its easy enough to do without thinking even if everyday.

well, that 5°C difference... may means what you said could be everyday or every hour i just afraid wearing it off much frequently. ymmv.

[alm]

5°C may be 42°F, but 5°C difference is only 9°F difference. Gotta love standardized units .

[bilko]

Anybody care to post a screen shot of their offset at power up from cold ?
I think that my scope offset is excessive and want to compare it with others. No other scope I have owned shows this level of offset. Its either poor quality components, bad design or faulty scope.

[saturation]

You're right alm, that's what I meant but forget the next part.  So going from 68F->75-80F, think calibration at season change, with climate control in rooms.  But in hot climates like Pacific islands, and even with climate control, mid-day is so hot, internal room temperatures can rise from 9F on a daily basis.

5°C may be 42°F, but 5°C difference is only 9°F difference. Gotta love standardized units .

[tinhead]

Anybody care to post a screen shot of their offset at power up from cold ?
I think that my scope offset is excessive and want to compare it with others.

sure, see attached cold and after 5 mins. Note: my DSO is having small fan, without fan i would have to wait max. 10 minutes to get same result (just because temp. drift will continue without air-flow and ambient temp. dependant).

No other scope I have owned shows this level of offset.

really ? As already said, every device need warm-up, some are "smart enough" to display almost no offset (where still the data is not really valid - just a software trick), some displaying much less (due better hadrware optimization - which cost money) and some just
don't care (as HanTekway) becuase you have always to wait until warm-up completted to get as much as possible valid data.

Its either poor quality components, bad design or faulty scope.

what do you think, which part of the scope is creating this offset ? And now surprise, the input stage is exact the same as Rigol is using,
so why there is no "such big" visible offset on Rigol ? Magic power suply?, magic resistors ? No, just a software trick, it is not that hard to calculate
the drift and add some correction to the firmware (which still didn't means the measured data is valid as this is only calculated correction).

On the other side, the offset is visible on 20mV/div and a littlebit on 100mV/div within first 5-10 minutes.
Every EE who need to "exact measure" (what a joke, DSO is not for exact measurments) with such low volt/div will definitely wait until
DSO, test circuit, power supply, bench multimeter, signal generator and so on are warmed-up.

[bilko]

On the other side, the offset is visible on 20mV/div and a littlebit on 100mV/div within first 5-10 minutes.
Every EE who need to "exact measure" (what a joke, DSO is not for exact measurments) with such low volt/div will definitely wait until
DSO, test circuit, power supply, bench multimeter, signal generator and so on are warmed-up.

@tinhead
Thanks ever so much for the screen dumps. I needed a comparison. Its not the drift, its the amount of drift that is the problem. 26mV on 20mv/div at ambient temperature of 23 degC. It appears that this is about normal for Tekway. My scope takes 40 minutes to go to zero. The problem could well have been faulty / wrong spec / wrong value components in the input stage. Without a comparison it is difficult to tell. You have answered my questions, Thanks for your comments

[tinhead]

at ambient temperature of 23 degC ... takes 40 minutes to go to zero.

that's definitely too long, it would be not a bit deal for me as most my daily used devices anyway powered up the whole day but honestly
i would not trust input circuit/measurment if the DSO need that long to show "zero".

[dimlow]

This happens on my Rigol too, but does settle down quite quickly. Its starts off about -20mV after about 3 mins its offset is 0mV to about 1mV. Chanel 2 seems to vary more than channel 1.

[rf-loop]

New HW Hantek DSO5102B

(test started with selfcal what have done totally in different temp becouse machine burn in tests are going on. This do not mean anything becouse this test is only for look change)

Before start scope was more around 3 hours powr off and roon temp around 22 C whole test time. Scope is on the table alone so that outside temp do not change markable.

Vertical settings: DC, 1x, 2mV/div,
20MHz limit off, Average 64, realtime.
50 ohm terminator in input.

Look only change, not absolute value.

Time min; display (mean) mV

0;2.24
1;2.32
2;2.24
7;1.68
10;1.36
15;1.12
20;1.04
25;0.96
30;0.88
35;0.80
55;0.72
120;0.64 - 0.72

But it must also think that it is nearly only vertical shift. Not gain error what is much more severe problem if it drift.

partially this is becouse big part of cooling is working as free convection trough outside case air holes. So inside case it takes long time it goes slowly higher and higher temp after it reach balance between heating and cooling.

Front end (whole front end box inside main board box) final temp is quite high.

If there is fan it do not continue warming so high and balance is reached more fast.

same time I look how my one analog tektronix drift.

2mV/div it drift total in same time 0,3mV

If there is fan in Hantek I believe that drift is less and it reach maximum drift more fast.
This is price about fanless... read noiselless scope.

yes it can also design much more better for handling thermal things without fan, but, construction is then not so cheap. In production there is lot of other costs than only components and material itself. (it need be so that it need minimum amount of human work hours)

I really do not understand why they design to run it as high temp as it is. maybe thermal design is not so wanted work... all want write code... mechanical and thermal design so that it is also clever in manufacturing needs lot of experience and knowledge...it need real professionals. Example in Tektronix and HP they have done lot of goods and bads and they have learn slowly in tens of years and still it seems be difficult sometimes....
(I think all know Tek 2400 series hybrids extremely critical thermal design... just as most bad what engineer can do. But who care... with name can sell... and who cares... equipments useable time is going more and more short. Years ago minimum is 20years. Now many company try design that machies fails just after warranty time. In some Tektronix if fan stop... scope die after half hour. There is not any watch dog for fan or thermal harakiri. After 5 euro fan stop... you just loose 10000. And this is Tektronix and some before years state of art oscilloscopes.

There are many components around 50 - 70celsius in normal temperature. What if use in 40 celsius enviroment and also if scope is place where air can not freely run so as free convection need. (yes we can do high exhaustion pipe from top of scope and do free air convection flow more powerful... yeah... put 2m high 100mm pipe...

[bilko]

partially this is becouse big part of cooling is working as free convection trough outside case air holes. So inside case it takes long time it goes slowly higher and higher temp after it reach balance between heating and cooling.

Thanks, for tests and analysis. I agree, internal temperature although in spec is too high. This causes excessive drift and shortens the life of the components. Solution is internal fan to equalise the temperature in the case and to eliminate some of the heat.

Installing the tiny heatsinks on the chips although useful has marginal benefit. there is a lot of thermal stress designed into this scope.

[dimlow]

This interested me so i made a little animated gif of the warm up on the rigol. Each frame of this anim was taken at 1 minute intervals. the first few frames i had made the mistake of not measuring channel 2. It took about 16 minutes for the scope to fully settle down.

EDIT: More detail, first i let the scope warm up for about 2 hours, I then self calibrated the scope and set up my channels to 20mV/Div and 200ns/Div. Switched the scope off for 5 hours. Then the test was made by saving a bitmap every minute until the scope settled to 0 offset. Ambient Temp was 20 Deg C

[rf-loop]

Scopes are individuals.

Alll these 6 randomly selected units have been power off enough to stabilize to room temp as it is... around 22 now. All these started exactly same time. (time on window is related to time when I save it to USB) with same settings as can see in pics. In this test all inputs are open.

Scopes are on the same table. Scopes are NOT selfcalibrated before tests. (yes they are selfcalibrated but this time only randomly warmed and random temp.. selfcal have only do for "check".

Also I think this random sample is not so that that there are most good and most bad  units... I think there can find more bad and more good if can select from big lot of scopes.

Captured time is only ~11min but most of drift happend in this time and it also give enough information what is going on.
Do not look zero line. (look only amount of drift)

All these 6 units are newest HW and  serial numbers are near each others.

(do not worry, same is with Agilent and Tektronix also, but not just this amount what is in most bad unit....maybe quality variations are not so big... and of course then with all others and finally also of course in these cheap oscilloscopes.)

If people have good luck.... this is lottery:

[saturation]

Thanks for actually making photos and animations of this.  Its a reason to warm up any device before using it for precision work  and that modern instruments still require it.