TH2830 vs TH2832

[tv84]

Yup, hw ID is still the same.

SO, everything as expected. Your EEPROM/FRAM/whatever still has a HW version A7 but with that patched ST2832 FW, the Ax units behave as Cx units.

[KungFuJosh]

Yup, hw ID is still the same.

SO, everything as expected. Your EEPROM/FRAM/whatever still has a HW version A7 but with that patched ST2832 FW, the Ax units behave as Cx units.

Yes, unfortunately, something else is wrong. I can't complete short correction (open does complete). However, if I activate the frequency function in the correction menu, I can compensate for specific frequencies. 20Hz to about 24Hz works (though starts to get odd), then 25Hz through 1.5kHz give odd numbers.

For example:
25Hz short = 216.839kΩ
30Hz short = 130.555kΩ
50Hz short = 342.603kΩ
2kHz short = -168.641Ω
100Hz short = 178.500kΩ
200Hz short = 851.517kΩ
500Hz short = 19.5998MΩ
2.50kHz short = -213.862Ω
After that, it gets pretty normal:
2.51kHz short = 0.00915Ω
3.00kHz short = 0.00058Ω

[KungFuJosh]

Ummmm... Nevermind?

I reflashed ST2832 again, screwed around a lot more, and now it all works. 🤷

I think this fw has too many options or something. I was able to complete all the correction, and test high value caps again. 🤦

So I guess we're all good then!

[tv84]

Maybe it needed a flash with 2832 settings already in...

[KungFuJosh]

Maybe it needed a flash with 2832 settings already in...

Yeah, that's the ticket, yeah!

I'm like 80% sure it was a PEBKAC error. The 2832 fw has a lot more options, and ranges for things I guess. Though I don't know why I couldn't do short correction regardless before changing any settings.

I changed the signal source impedance from the default 100 to 30Ω. I also experimented with the Range option on the test screen. I've never played with that, I don't know if it was changeable in the 2830. The pdf says Range is only Auto, Hold, but clearly this version has more, unless that's a different range option. 🙄 Or maybe the ST firmware is fancier than the TH firmware? Or, again, me confusing something.

Anyway, I dunno if I need to have it recalibrated or not, it's pretty close for the test caps I tried. I'm assuming I need to read a different manual now.

Comparison:
2830 ref:
4.69238nF, 4.64197Ω esr (Range Auto?)
925.025uF, 0.02938Ω esr
2.16287mF, 0.02186Ω esr

2832:
4.68818nF, 10.8833Ω esr (Range 10Ω? Or 30Ω, I forget)
929.669uF, 0.02645Ω esr
2.16549mF, 0.01554Ω esr

[mawyatt]

Been following this, nice work tv84 and Josh

Were you able to get 200KHz? Also down to 20Hz?

The calibration is questionable, I guess you could make some measurements with the 2832 enabled, then revert back to the 2830 and see if they agree. You have a SMD fixture like we have, that would be a good way to do the capacitor comparisons, some quality COG/NPO caps might be worthwhile, even some good 0.1% low TC chip resistors. The Tonghui supplied Kelvin clips are pretty good, altho likely more chance of variability than the SMD fixture for caps.

One thing that just occurred is if you should do a recalibration of the Open and Short if you change back to the 2830 to verify a given set of measurements. Since apparently there isn't any physical changes, leaving the Short/Open cal alone might give a more "repeatable" result between firmware changes?

Anyway, nice safe cracking

Best,

 

[KungFuJosh]

Been following this, nice work tv84 and Josh

That was mostly tv84's skill, nctnico's insight, and me willing to screw my meter up. 😉

Were you able to get 200KHz? Also down to 20Hz?

Yes, full range in tiny steps if you want. I posted the 20Hz/20kHz screenshots previously. Also the DC bias voltage, and the DCR test voltage is also adjustable now. Instead of DCR test voltage being stuck at 1V, it can be anywhere from 50mV to 2V.

The calibration is questionable, I guess you could make some measurements with the 2832 enabled, then revert back to the 2830 and see if they agree. You have a SMD fixture like we have, that would be a good way to do the capacitor comparisons, some quality COG/NPO caps might be worthwhile, even some good 0.1% low TC chip resistors. The Tonghui supplied Kelvin clips are pretty good, altho likely more chance of variability than the SMD fixture for caps.

One thing that just occurred is if you should do a recalibration of the Open and Short if you change back to the 2830 to verify a given set of measurements. Since apparently there isn't any physical changes, leaving the Short/Open cal alone might give a more "repeatable" result between firmware changes?

I do like the idea of getting use out of the SMD fixture. However, it is mildly tedious switching back and forth between firmwares because all of the memory based settings are lost when switching between the two models.

My guess is that the short/open cal will also be lost, but I can do a test before and after running the corrections on the 2nd fw. My other guess is that if my cal data was lost, it likely won't exist for either fw anymore. I'll see how accurately those SMD toys test, and also compare with my ST42.

I added stuff for testing to my Mouser cart:



Edit: My previous cap tests weren't with the kelvin leads, I used the 26048A fixture. However, none of them were precision caps, and I kept them in a closed zip lock bag, so I assume not ideal.

[mawyatt]

Oops, forgot about Nick, so thanks to Nick as well

Think a 1206 or 0805 NPO/COG caps of 1000, 100 and 10pF might be good for some verification between firmware changes. Polystyrene & Mica are another very stable capacitor type, if you can find some (Surplus Sales Nebraska).

Anyway, well done by you, tv84 and Nick

Best,

[KungFuJosh]

Oops, forgot about Nick, so thanks to Nick as well

And Rudi too for that matter. Without his photos and stuff, this likely wouldn't have happened at all.

Think a 1206 or 0805 NPO/COG caps of 1000, 100 and 10pF might be good for some verification between firmware changes. Polystyrene & Mica are another very stable capacitor type, if you can find some (Surplus Sales Nebraska).

I have polystyrene and mica caps. I updated my mouser cart with more SMD crap. 😉

[mawyatt]

Ok, thanks to Rudi too

Another possible issue could be, does the TH2830 have factory calibration parameters for the "extra" frequency ranges that come with the TH2832?

Best,

[KungFuJosh]

Another possible issue could be, does the TH2830 have factory calibration parameters for the "extra" frequency ranges that come with the TH2832?

No idea. That was my initial concern when I was having issues with the new firmware, but that was likely just me having too many settings to mess with. I'll test stuff in different ranges and see what it looks like. Obviously certain things only work in certain ranges, so I'll have to figure that out too.

I don't know at what point they do the calibration in production. It's possible that they precalibrate before the final model is determined, but I don't know how likely that is.

The reason I had no doubt this would work is because most modern test equipment seems to be manufactured with a "highest possible profit margin" mentality. What separates one manufacturer from another is whether they also care at all about quality (which this manufacturer clearly does). That means the least variation between device versions as possible (avoid hardware changes if possible), with firmware crippling the "lesser" models being the most obvious choice to keep costs down. It's really smart, considering all they needed to differentiate one device from another was a sticker on the faceplate. I would also bet that the device IDs that separate any of these can be programmed via serial over USB on the back panel to keep it simple.

[mawyatt]

Our TH2830 came with a 6 page test cal report that was signed by the test tek & dated. It included all the measurement parameters including the range, reference device value, acceptance range, and actual measurements. The frequency range was just 100, 1K, 10K & 100KHz tho. There were 5 signal levels, 10 resistance ranges from 1 to 100K ohms, 7 capacitance ranges from 100pF to 100uF, at all the frequencies indicated, also DCR and Inductance ranges.

There is a paragraph C. DC Bias Level: ([Non]) indicated, so likely the same test procedure for the 2832, maybe with the added frequency ranges.

Anyway, all toll, 98 measurements recorded, indeed very comprehensive test data!!

Best,

[KungFuJosh]

Our TH2830 came with a 6 page test cal report that was signed by the test tek & dated.

Ha! Mine didn't come with anything but a "calibration certificate void if removed" sticker on the back...which I removed immediately. 🤣

I know it's not good enough for actual calibration, but I connected probe to probe with the SDM3055X-E and the bias voltage was accurate.

If you were going to get it recalibrated, where would you send it?

[Martin72]

Our TH2830 came with a 6 page test cal report that was signed by the test tek & dated. It included all the measurement parameters including the range, reference device value, acceptance range, and actual measurements. The frequency range was just 100, 1K, 10K & 100KHz tho. There were 5 signal levels, 10 resistance ranges from 1 to 100K ohms, 7 capacitance ranges from 100pF to 100uF, at all the frequencies indicated, also DCR and Inductance ranges.

There is a paragraph C. DC Bias Level: ([Non]) indicated, so likely the same test procedure for the 2832, maybe with the added frequency ranges.

Anyway, all toll, 98 measurements recorded, indeed very comprehensive test data!!

Best,

Hmmm...I would have to take a look at the Cal. reports from our ST2830 to see if this was done there in the same way for all areas.
In any case, your report was not created by Tonghui itself, they don't take that much trouble...

[luudee]

Regarding certificates, my unit also came with a 6 page calibration certificate ...

first two pages attached ...


luudee

[mawyatt]

In any case, your report was not created by Tonghui itself, they don't take that much trouble...

See below!! It came directly from Changzhou Tonghui Electronic Company Limited!! Even has a 7 digit document number, most comprehensive certificate of performance document we've come across.

Even include details on each reference used, which they use at least 30 individual reference devices. For example the 1nF reference @ 100KHz is 1.00060nF with DF 0.00010, and measured 1.00035nF, DF 0.00025, or @ 1KHz the reference 10.0100uF, DF 0.00370 measured 10.0091uF, DF 0.00370.

We're sure this is automated, but well done Tonghui for provided such detail, certainly builds one's confidence in this instrument's overall quality, and Tonghui instruments in general 

Best,

[mawyatt]

Regarding certificates, my unit also came with a 6 page calibration certificate ...

first two pages attached ...


luudee

Yep, that's the one we received, same test tek also!! Did your TH2832 have the capacitance measurements conducted at 200KHz?

Edit: Might be interesting to see how well these instruments did during factory verification. Here's the results from our instrument.

Best,

[mawyatt]

Our TH2830 came with a 6 page test cal report that was signed by the test tek & dated.

Ha! Mine didn't come with anything but a "calibration certificate void if removed" sticker on the back...which I removed immediately. 🤣

I know it's not good enough for actual calibration, but I connected probe to probe with the SDM3055X-E and the bias voltage was accurate.

If you were going to get it recalibrated, where would you send it?

Probably the usual Cal Labs here in US. Expect a hefty bill tho, detailed calibration, over these ranges, will be involved and expensive for sure.

BTW one of the reasons we recommended the TH2830 was the overall quality of this fine instrument. When we did our research a few years back, we ended up selecting this instrument based upon performance/cost and reliable/repeatable measurement expectations, and it has not disappointed

Best,

[Martin72]

See below!! It came directly from Changzhou Tonghui Electronic Company Limited!!

Exceptions confirm the rule..

[KungFuJosh]

Regarding certificates, my unit also came with a 6 page calibration certificate ...

first two pages attached ...

Could you please share more pages? I'd like to see what was tested in different ranges so I can do similar tests.

[KungFuJosh]

See below!! It came directly from Changzhou Tonghui Electronic Company Limited!!

Exceptions confirm the rule..

What exceptions? I'm the only one here that didn't get cal sheets from them, so I think that makes mine the exception. 😉

Actually, I'm curious to see what your ST cal sheet looks like. Did TH do the cal before sending to ST, or did ST do the cal after receiving it?

[KungFuJosh]

Probably the usual Cal Labs here in US. Expect a hefty bill tho

I've never sent anything for cal before, so I have no idea who is trustworthy. What is hefty? Over $200? More?

[Martin72]

Actually, I'm curious to see what your ST cal sheet looks like. Did TH do the cal before sending to ST, or did ST do the cal after receiving it?

I'll have to check the archives to see if there was a calibration record.
We have had the device calibrated externally since 2021, because otherwise we would not be allowed to use it.
Of course, it would have been great if we had also had such a protocol, which ludee and mawyatt had received, because then you can make a nice comparison...internal vs external according to ISO and DakkS standards.
I'll check tomorrow how many protocols are stored in our database.

[KungFuJosh]

SO, everything as expected.

Unfortunately, there's a bug/flaw in the patched firmware. Normally the meter remembers the last state of the device. For example, which func mode (Ls-Rs, Cs-Rs, etc.), speed, etc. With the patched fw, that state is lost after power cycling the meter.

I downgraded to the stock st2830 fw, and set func, speed (and avg), Vm/Im to on, and restarted. It remembered everything as expected.

Upgraded back to st2832patched fw, and those settings were lost. Redid those settings, and power cycled. All lost again.

Did compensating for the checksum require modifying something else? Or any other idea why this is happening?

Thanks,
Josh

[tv84]

SO, everything as expected.

Unfortunately, there's a bug/flaw in the patched firmware. Normally the meter remembers the last state of the device. For example, which func mode (Ls-Rs, Cs-Rs, etc.), speed, etc. With the patched fw, that state is lost after power cycling the meter.

I downgraded to the stock st2830 fw, and set func, speed (and avg), Vm/Im to on, and restarted. It remembered everything as expected.

Upgraded back to st2832patched fw, and those settings were lost. Redid those settings, and power cycled. All lost again.

Did compensating for the checksum require modifying something else? Or any other idea why this is happening?

Thanks,
Josh

The patch doesn't go there near. Maybe the info is stored as ST2830 (Ax) and when the software tries to retrieve it as ST2832, it fails.

That's why I say that a direct change in the FRAM/whatever from Ax to Cx would make all good without FW patchings.