uni-t ut139c true rms failure or bad triggering

[kokodin]

Hello, my first topic, post and all that jazz, and stright away a question.
I have bought new uni-t 139c true rms multimeter for the true rms feature, i wanted to measure non-sinusoidal AC voltage of around 5-8V rms at 50Hz. It is a crt tv tube heater voltage driven by synchronization returning lines, and it should be set to 6,3V rms acording to service manual with a true rms multimeter, or rather "effective value voltmeter of non-sinusoidal waveforms"as the service manual calls it.
Normal non rms meter measure about 5-5,5V AC, but that is totaly unaccurate. I have 2 nan rms meters and they read about 0,7V apart, But service manual says, that with non rms meter i should calculate it by 1,15 to get +-5% accuracy, whitch is about +-0,3V

My new ut139c reads either 0,006V at 15,6KHz 50% duty cucle or somwhere between 3 and 10V at 50Hz 49, 37 or 27 % duty cycle at random, but it stick to one value at the time. I turn it off, and back on and it is the same story again, either 0.006V or something random, so i have absolutly no faith in its readings.
What is even more strange i can read a value of this voltage only when i use uni-t probes, when i solder test leads directly to the test points i get 0,006V every time.
Normal AC voltages from a transformer or mains voltage is displayed normaly but i had also a little bit of a wierd experience when i was trying to measure secundaty voltage on a small transformer.
it was fine on auto range, but it was 0.006 on manualy set range, and transformer was outputing about 8V 500mA


So the question is, did i get a foulty meter, and if this is the case how to support my claim for the guarantee. Or is this meter that badly shielded or calibrated, so it picks up interference trash and nothing can be done.

What do you think?

edit ..................
i don't know how should i describe better my measured voltage, so i did the first best thing i could do, I took a photo of my oscilloscope screen while probing this voltage. It shuld be about 22Vpp with about 25-30% duty cycle sawtooth, frequency however might be correct, because it is bacically 60us like on the screen. (that would also calculate corectly for 625 lines at 25 frames per secund of pal color system)
But why isn't my meter doing my math for me?

edit2***********
ok i think i know the reason.
As i mentioned before i was sure, that my tv crt heater was powered by returning lines, but i mistook lines with frames so i was convinced it should be 50Hz from vertical deflection, and it wasent , because it is driven by horizontal deflection at 16500hz.
The meter rms bandwidth ends at 1000hz, so i have 16 x higher signal bandwidth . The meter flaw is debunked and results from all my meters are flawed as well.
The question is now, Why uni-t did not make "out of range" notification , based on measured frequency? It would be a nice feature an a meter with limited range of rms measurments.
The secund question is, would an analog meter work any better?

[joeqsmith]

Hey, I thought I had seen you post about this here.   The scope shot helps and really, this is what I would be using to make this sort of measurement. 

What I can do is replicate your waveform and then run the meters I have left across it.  The really cheap analog meter you had asked about and my UT61E are both changed too much to give you any sort of meaningful data. 

[joeqsmith]

Roughly the same freq.  Amplitude is a little lacking with this amplifier but close enough to give you some idea what you are asking.
And again, just to be clear I would be using a scope.  All of the meters shown are TRMS and in working order (at least for this measurement).

For those that have watched any of my later reviews, these results should not be too surprising.

EEVBLOG branded Brymen BM235 - 0.000
HIOKI DT4252 - 0.024
AMPROBE AM530 - 0.441
Fluke 115 - 1.367
Extech EX540 - 1.7404
Brymen BM869S - 3.5370
TPI 194II - 3.5815
UNI-T UT181A  - 3.5850, This meter has a modified PCB but response should be close to stock meter
And just to throw it in the mix, the highly modified UNI-T UT61E which in no way represents a factory stock meter. 3.267

It is a little strange about your UNI-T.  I never replaced the 139C after I damaged the second one.   

[kokodin]

When i tested my uni-t 139c with squarewave  generator, it read ac voltage up to about 1kHz and after 1kHz it rolled down .
For some reason i could read square wave hrequency and duty cycle only, when my meter was ac coupled through a capacitor. Othrwise it read 0 Hz and 0% duty cycle. I figured it out only ahter probing my diy 555 generator with a scope. My generator was dc coulped to the 0V rail, (PWM DC situation) so i testet the meter with another PWM suply, my 555 and transistor based model train controller unit.
With this thing meter was only showing dc voltage, no ac , no frequency, no duty cycle. Ac coupled trough a capacitor, it is reading all those values.


After work i post what i "designed" for analog meter input bufer. And as  posted on youtube my waveform might be vertically fliped
Thanks for help

[joeqsmith]

You are correct.  The 139C requires a zero cross to measure frequency and duty cycle.  I have seen this with other meters as well so if you decide you want a better meter, you may want to ask if you can't find what you need from the manuals.   

[kokodin]

Glad to read it is not a broken unit... darn, i almost make a pun.
I don't know if my hobby needs would justifies another meter purchase. i almost have 4 counting two uni-t, one chinese chepie (but in a rubber holster) and this analog one.

uni-t aside for a moment here is my "circuit" that i "designed". In theory should work as an opamp with gain of 1, input impedance limited to 3mohm, and adjustable ground level . should work as a dc or ac coupled bufer, and it is not referenced to any outside voltage other that measured signal, and shouldn't put much load on the crt filement like few kohm analog meter would.
I know it is a super basic block, cookie cutout schematic, but who said that anything must be complicated.
If the input and output waveforms would match, analog meter would be able to measure this voltage to some degree of accuracy

[macboy]

A feature to look for when buying a True RMS meter is that ability to measure "AC+DC". This is the True true-RMS value, as it will include both DC and AC components of the signal. Many meters include DC blocking for the AC input, so they will not include the contribution of the DC offset in the reading. Equally frustrating is when a meter does not include DC blocking for AC readings, so if you expect/want it to be there, you can end up with bad data. You need to get to know your meter ... does it measure AC+DC or not? My B&K 5360 handheld has a "AC+DC" mode for AC ranges which you must select manually, otherwise you get the AC component only. My Keithley 2001 has selectable "DC Coupling" (which measures AC+DC) or "AC Coupling" (AC component only). My Keithley 199 can't do AC+DC at all ... one of its few major shortcomings.

To compensate for lack of AC+DC mode, you can usually measure each DC and AC separately then add them as the Root-sum-of-squares. The meter's auto-ranging can become confused when presented with a large DC offset on AC, or a large AC signal on DC mode, so you should manually select the range for both DC and AC modes (and maybe one range higher than you think is necessary, to start with). Again, get to know your meter by creating such a signal yourself then measuring it to see if you/it can do this. A function generator that can put a DC offset on the signal works great for this task. e.g. a 4 V_RMS AC signal with 3 VDC offset is actually 5 V_RMS. (i.e. sqrt(3² + 4²))

[joeqsmith]

I would just hook the scope up and call it a day. I guess you could build a good front end for a bad meter or a whole meter? As you saw from the data I collected, it may not be so simple.  Good luck with the repair.

[kokodin]

I would just hook the scope up and call it a day

A good calibrated dso might do the trick :] Calculation by hand and paper would also work, but much longer, and surprisingly what i calculated match the readings from my ut201 meter of 5,15V. And i dont remember how i did it :] (Vpp*0,707)/3 i think. if we conclude, that there is no negative swing  and the peak is positive that is, but that would make more sense for PWM and it matches ac measurment from non rms meter :] (meter impedance is listed as 10mohm//<100pf with bandwidth of 40-1000hz)

Great picture, :] i love those diagrams
i might think of building the meter, there is one analog meter specyfied in tv service manual, polish domestic product, and it is still at value, because used and working ones cost about the same as new ut61E (and they disapear quickly from ebay equivalent site, even broken ones). i need only one range and i have the scale and schematics for that meter, discreet transistor schematic :], the problem is, i would have to calculate the reading by 1.14 and it would be -+5% Value for the other meter. Who knows how close my would be :].

macboy
In general it is a closed loop, transformer secondary winding goes trough an adjustment coil, trough the filement of the crt and back to the transformer closing the loop. One side of this secondary winding i conected to 53V rail, the other side is not connected to anything else, so it is floating at 53V but there is no dc component in the signal, there is no connection to ground or any other voltage in this loop at any point. Anyway meter on a dc range show 4mV, on ac 5,15V (i belive it is average value, because the ut201 in not rms meter, it is a clamp though).
The only problem is, that this signal is non sinusoidal, transformer is driven by return lines from the deflection coil. Almost like pulse driven transformer.