TH2830 vs TH2832

[KungFuJosh]

That's all great theory, but we've already shown in practice that the meters we're discussing here can in fact accurately measure 100KΩ at 100kHz and 200kHz.

Those charts also point out that what I said is accurate. Using the correct test subjects at the correct test frequency was skipped. They only used a limited range of cap values that can be read in a broader frequency range. Most likely to reduce labor costs, or because whatever standard Tonghui was pretending to do didn't require anything beyond that. This gives a generalized result, without getting into more detailed specifics.

Attached here is the chart from the Sourcetronic manual for the 283X models.

[nctnico]

That's all great theory, but we've already shown in practice that the meters we're discussing here can in fact accurately measure 100KΩ at 100kHz and 200kHz.

Try to convince a calibration lab this statement is true.  In reality you have to be really careful to make claims about test equipment accuracy beyond the specification. It is better not to make such claims in order to prevent unpleasant surprises. The reason some values are not tested as part of the calibration is not due to laziness or saving work but to provide only meaningfull data. You can be sure that a 10k Ohm resistor measures within spec next year. Your 100k Ohm resistor may show an entirely different value... You won't be able to know for sure whether the resistor has changed or the meter has aged.

[KungFuJosh]

Try to convince a calibration lab this statement is true.  In reality you have to be really careful to make claims about test equipment accuracy beyond the specification.

Well, either way, we're talking about faulty firmware. The bad firmware had issues with 100KΩ across all the ranges. The closest measurements were actually in the 100kHz and 200kHz ranges. This isn't about discussing accuracy at those ranges in general. luudee's TH2832 did measure accurately in all ranges with his better firmware.

Can you read that ridiculous chart I posted? What does it say for accuracy of 100kΩ at 100kHz? It looks like it should still be in the 0.1 to 0.2% range to me. That's not too crappy.

[Martin72]

@nctnico:
I think his first concern is why he gets a plausible value with the original firmware and not with the other two.
Detached from the accuracy question.

[KungFuJosh]

@nctnico:
I think his first concern is why he gets a plausible value with the original firmware and not with the other two.
Detached from the accuracy question.

You are correct. Either way, within the limits in the chart I posted above, 100kΩ at 100kHz should be within a 0.1 to 0.2% margin of error. 100kΩ at 200kHz should be within a 0.25 to 0.35% margin of error. Device limitations is certainly not the issue.

[2N3055]

That's all great theory, but we've already shown in practice that the meters we're discussing here can in fact accurately measure 100KΩ at 100kHz and 200kHz.

Those charts also point out that what I said is accurate. Using the correct test subjects at the correct test frequency was skipped. They only used a limited range of cap values that can be read in a broader frequency range. Most likely to reduce labor costs, or because whatever standard Tonghui was pretending to do didn't require anything beyond that. This gives a generalized result, without getting into more detailed specifics.

Attached here is the chart from the Sourcetronic manual for the 283X models.

There is nothing funny about that chart.  As C or L start to approach a limit of zero or infinite ohm at the edges of chart, internal noise of instrument and accuracy of components (drift etc) start to be as large as measured component value.. And accuracy will drop off gradually, as it is shown..

To simplify it, it is about accurate measuring amplitude and phase of low micro(nano)volts and low micro(nano)amps.. Instruments cannot have unlimited dynamic range..

Middle of the chart is a sweet spot, Goldilocks area, hence best accuracy.. To the extremes, it worsens gradually..

[KungFuJosh]

Of course. Now let's look at collected data:

200 KHz: 100.35 KOhm
100 KHz: 100.33 KOhm
 10 KHz: 100.36 KOhm

The buggy ST2832 firmware failed at a lot more than just high freq.
ST2832 100KΩ:
100Hz = 49.5052
1kHz = 67.6866
10kHz = 78.2236
100kHz = 98.1972
200kHz = 91.5881
ST2832 100K DCR = 100.057
ST2832 10kΩ @1kHz fail = 6.78268

Same hardware, 100KΩ, switched back to TH2830 firmware:
100Hz = 100.013
1kHz = 100.013
10kHz = 100.005
100kHz = 99.8795

Assuming the worse option from the chart above (0.35) at 200kHz, that gives us a +/- range of 99,650Ω to 100,350Ω.

luudee's 200kHz reading stated 100.35 kΩ, which is as expected. So theory and practice agree. However, it being higher in all 3 ranges implies the resistor itself is high, and the meter accuracy is better than expected (though still within expected range).

The readings are goofy with the buggy firmware.

[Martin72]

This nomogram is just nicely plotted that...

...It´s too good to be true..

[luudee]

@Josh, the only thing that is buggy/not working/bad accuracy,  is when you flash 2832 FW in to 2830 HW.

Perhaps the CPLD is slightly different for the units ? 

When you stick with the original FW, you DO get proper readings, correct ?


BTW, the accuracy chart is exactly the same as in the Tonghui User manual.


Cheers,
luudee

[KungFuJosh]

@Josh, the only thing that is buggy/not working/bad accuracy,  is when you flash 2832 FW in to 2830 HW.

Perhaps the CPLD is slightly different for the units ? 

When you stick with the original FW, you DO get proper readings, correct ?


BTW, the accuracy chart is exactly the same as in the Tonghui User manual.


Cheers,
luudee

The ST2830 firmware also has similar issues to the ST2832. Both Sourcetronic firmwares are older than the Tonghui versions. The ST2830 firmware has less bugs compared to the 2832.

The hardware is identical between the units and likely the brands too. I'm sure all the user manual stuff is essentially the same besides the logos.

Actually, if @Martin72 could get pictures of the inside of his ST2830, that would be awesome...but I don't know if he can, since it's a work meter. I wonder if they have different silkscreens on the Sourcetronic versions or not. It's only 3 screws. 😉

[luudee]

The ST2830 firmware also has similar issues to the ST2832. Both Sourcetronic firmwares are older than the Tonghui versions. The ST2830 firmware has less bugs compared to the 2832.

The hardware is identical between the units and likely the brands too. I'm sure all the user manual stuff is essentially the same besides the logos.
...

@Josh, yes, fundamentally the HW is identical. BUT, the contents of the Altera MAX II CPLD may be different. At this point unknown.
The CPLD is clearly between the NXP MCU and the Analog box ....

The user manual, specifically covers all three models, TH283x (TH2830, TH2831, TH2832).  Where the differences
appear to be in the Frequency Ranges and number of points ....

Cheers,
luudee

[KungFuJosh]

@Josh, yes, fundamentally the HW is identical. BUT, the contents of the Altera MAX II CPLD may be different. At this point unknown.
The CPLD is clearly between the NXP MCU and the Analog box ....

The user manual, specifically covers all three models, TH283x (TH2830, TH2831, TH2832).  Where the differences
appear to be in the Frequency Ranges and number of points ....

I've already confirmed that there are no set limitations on any of the features, it's entirely controlled by the firmware. TH2831 is different hardware, that's an older device (EOL).

Running the patched firmware, I can type in any frequency I want, and it will work and register correctly on my DMM. DC bias, etc. Everything works. The only issues I've run into are firmware bugs.

This is from my test with the patched firmware:

[Martin72]

Hi,

I'm at work at the moment, all results later.
One thing:
97k Ohm at 100kHz...

[KungFuJosh]

97k Ohm at 100kHz...

Don't forget it needs to be warmed up for 30 minutes before correction. 😉

[mawyatt]

Had a few minutes and turned on the TH2830 and IM3536, let them warm up for ~45min and grabbed some 0603, 0805 and 1206 100K resistors. Had some standard 1%, and some 0.1% and a couple low TC (15 and 25ppm) types.

Edit: Remember to always let these instruments warm up with the measurement fixture attached 

Did an open/short cal with SMD fixtures, one on each instrument. Open was ~1206 spacing, short with plungers together under normal pressure. If we were doing a serious measurement, then the short would be done with a zero ohm SMD device of the same size as the DUT, and the open would be done with the plunger spacing same as DUT.

Both instruments agree well in Impedance Magnitude, Angle, Equivalent Parallel Resistance & Capacitance measured at 100KHz. The equivalent capacitance seems reasonable for these SMD devices (~0.05pF).

Anyway, the TH2830 is certainly a capable LCR meter and glad we have one on the bench to supplement the more expensive IM3536, a number of years past wise investment indeed

Edit: Will post results if interested but don't want to clutter up the thread. Should have done DCR measurements as well and compared with our KS34465A, but didn't 

Added per request, here's a few measurements including DCR and a KS34465A DCR measurement.

Setup 100 ohm output, 1V, 16 Avg, Slow at 100KHz.

Size                                  TH2830                                                                        IM3536

1206 1%             99.9877K -0.185 degrees, 99.9876K @ 0.05130pF      100.001K  -0.201 degrees, 100.011K @ 0.0559pF
0805 1%             100.164K -0.209             , 100.164K @ 0.05799pF      100.186K -0.247             , 100.185K @ 0.0685pF
0805 0.1% 15ppm 99,9706K -0.180           , 99.9709K @ 0.04988pF       99.9849K -0.218            , 99.9841K @ 0.0606pF
                            99.9712K DCR                                                         99.9830K DCR
        KS34465A                                       99.97951K DCR


Best,

[KungFuJosh]

Edit: Will post results if interested but don't want to clutter up the thread. Should have done DCR measurements as well and compared with our KS34465A, but didn't 

I'd love to see some comparison results.

No DCR comparison?? Shame! 🙄🤣

[KungFuJosh]

What inductors does anybody recommend for calibrationish testing? The previously posted sheet had values of 100uH, 1mH, 10mH, 100mH, 1H.

[nctnico]

Where it comes to uphacking, a few questions remain unanswered that need further investigation now it turns out uphacking an ST2830 to ST2832 is not working by just patching the device selection.

- Is the CPLD program different between TH2830, ST2830 and ST2832 firmware versions?
- What is the third data blob inside the firmware image?

[KungFuJosh]

I'm at work at the moment, all results later.

Did you get the version number? I'm very curious about that.

Thanks,
Josh

[mawyatt]

Edit: Will post results if interested but don't want to clutter up the thread. Should have done DCR measurements as well and compared with our KS34465A, but didn't 

I'd love to see some comparison results.

No DCR comparison?? Shame! 🙄🤣

Added above!!

Best,

[KungFuJosh]

Where it comes to uphacking, a few questions remain unanswered that need further investigation now it turns out uphacking an ST2830 to ST2832 is not working by just patching the device selection.

- Is the CPLD program different between TH2830, ST2830 and ST2832 firmware versions?
- What is the third data blob inside the firmware image?

Further investigation is needed, but we also need to be certain that the firmware we're using is reliable.

Here's what we know:
Both of the ST versions of the firmware we have show older version dates than the Tonghui version.
The newer Tonghui fw version (2021-10-14) was supplied to correct a similar bug to what I'm experiencing with both ST firmware versions.

My device sticker says TH2830:
Flashing ST2830 (2020-04-16) is mostly good, but has similar measurement bug that needs to be fixed.
Flashing ST2832 (2019-04-23) measurement bug is worse, and has issues recalling "previous state" but appears to recall correction. All advanced functionality appears to work as expected otherwise. Operationally appears fine besides those bugs.

It stands to reason that the Sourcetronic firmwares are outdated and contain similar bugs to what was experienced with the outdated Tonghui firmware, and the results agree with that. It's not unreasonable to presume that the oldest firmware is the least developed.

We either need a better TH2832 firmware, or newer/better Sourcetronic firmware.

[Martin72]

Don't forget it needs to be warmed up for 30 minutes before correction. 😉

It´s not the first time I´m working with it.
Made no difference but after open/short cal it "gains" one kiloohm more..
First a few pics, startup screen and then pics from the inside after loosing the three screws..
Seeing pictures or see it in real makes a difference - I was surprised about the good building quality.
If I hadn´t the hioki, I would buy one.
A little bit later the results from measuring on the dmm checker.

[KungFuJosh]

Made no difference but after open/short cal it "gains" one kiloohm more..
First a few pics, startup screen and then pics from the inside after loosing the three screws..
Seeing pictures or see it in real makes a difference - I was surprised about the good building quality.
If I hadn´t the hioki, I would buy one.
A little bit later the results from measuring on the dmm checker.

Hey, that 1kΩ is closer to the expected value. 😉

Agreed, excellent quality. As I expected, the hardware is identical, as is the silkscreen:

Edit: Did you get any photos that show the silkscreen better beneath the IEC inlet? It looks identical to mine, says VER1.9@2, but we can't see the year. Mine says VER1.9@2019.9. I'm curious what the year difference is between A6 and A7. Not that it really matters. 🤷

Edit again: lol, I just looked at luudee's photos, and your transformers are identical except ST added a nice sticker to yours.

Since your firmware version is the same, I think I'll reflash it and see if the bug is really there or not, or if I was going too quickly.

Edit again again: I think I may also try some THT 100KΩ resistors with a different fixture. My SMT fixture is the only off-brand one.

We really need some TH2832 firmware.

[Martin72]

Hi,

Can open it up again in 2.5 weeks...

Here the results of the resistance measure, worst case deviation is appx 1.4% when DCR is the reference.

[zrq]

I guess we need more in depth look at the firmwares.