EEVBlog 121GW Discussion thread

[npelov]

The BLE reflects the display. The display was fine when the SD card log showed non sense. Also BLE sends at a rate independent of SD cart log interval. I feel like the bug is just in the SD card write routines. I selected 10 seconds  for the last try - logging both SD card and android app. I logged for a few hours and the DMM froze. Just before freezing it looked like this:


The second display shows 5dE04 ... while it was still recording. It should be showing the record number. The last record number in the SD log is 1140. Last voltage 10.345V (you can see on the android app 10.190V), last power 17.03VA, note that it's 16.74 on the display.  So the log stopped a while ago. I was measuring 12V/9Ah battery discharge on a 20W car bulb, so the power gets lower with the voltage.

Then it frose like this and only turn off fixed it:

The SD card log was good this time, but it was for just few hours at 10 second interval.

The android app save button didn't work. I guess it has too many records.

[joeqsmith]

I am not that tech savy where I need one of those cell phone things, so I can't replicate what you are seeing with the Android.   

If you had a Windows tablet with BLE, you could give that a try.   You could also try using the BLE without the SD card and see if it hangs the same way. 

Personally, if I were needing to run long tests like this and couldn't trust the equipment, I would start looking for something better. 

[npelov]

The hanging part has nothing to do with using the SD card and BLE at the same time. It happened before, but I don't remember the circumstances. I think it was again on VA, but I'm not sure.

You are right about the logging equipment, but the whole idea of buying this DMM is using it for it's special features - BT, SD logging, low burden voltage, and they all don't work reliably.

[joeqsmith]

The hanging part has nothing to do with using the SD card and BLE at the same time. It happened before, but I don't remember the circumstances. I think it was again on VA, but I'm not sure.

You are right about the logging equipment, but the whole idea of buying this DMM is using it for it's special features - BT, SD logging, low burden voltage, and they all don't work reliably.

On the lock up, I would have to take your word for it as I have not ran any tests like this. 

Beyond the EEVBLOG brand, I didn't see a reason to buy one.  There have been a couple of years of posts about the problems with the meter,  several in regards to logging.    There have been countless firmware releases.   I'm not sure why anyone today would have confidence in the product.   

There is virtually no support outside of the people who own them.  Even simple questions seem to go unanswered.   

Think of the years it took to build the EEVBLOG brand.   I think it was sold cheap and damaged from making this deal.  My advice is for you to learn from this.   You could waste a lot of time playing around with this meter, or you could get something that fits your needs and spend your time working on your projects.   

Outside of offering to run a similar test using a full blown PC to log the data using custom software,  I am not sure what else I can do for you.   Again, this seems to be reliable but until a few months ago, I assumed they has the SD card storage sorted out.   

You should be aware that there are some major problems with the VA mode.  This was seen with the prototype and 2+ years later has yet to be addressed.   

 

[1anX]

Fair call Joe!

I bought one of the meters simply because it was EEVBlog branded and wanted to support Dave in his quest. I enjoy Dave's blog and forums, but I wont be buying another one of his products as the meter has not been supported in any meaningful way to rectify its numerous issues. Probably because it needs hardware revisions and re-manufacturing.

The meter does a few things well, but is not the reliable meter I turn to use for my electronics projects. The meter as you point out has not evolved much at all over the time it has been available and the firmware has not fixed the issues.

I cant call the meter a "lemon" because it does function, just not in the way I expected. The meter IMO just did not deliver on Dave's pre-production hype and the manufacturing faults did not help build confidence. I rate the EEVBlog 121GW the "lime" award, as its left a sour, lemon like taste in my mouth.

Would I buy a revised and re-manufactured version of the 121GW? I'm afraid not, as once bitten twice shy, applies for me and this product.

I would like to thank Joe Smith for all his testing of this meter and providing an unbiased data set for any potential 121GW meter buyer to use as a reference outside of the manufactures specs.

[npelov]

You should be aware that there are some major problems with the VA mode.  This was seen with the prototype and 2+ years later has yet to be addressed.   

If you are talking about the adding of burden voltage  - that's not a problem. Also it's only a problem if you want to measure load power. I am measuring power supply (in my case battery) power. Every measuring instrument has imperfections and as long as you know about them - you can workaround them. The problem however when it starts freezing - that's not a normal DMM flow, it's a bug.

But yes, I agree that reversing the current probes in VA mode should be in the manual - it's a nice idea how to measure load power better (not power out of the supply).

[CDaniel]

As I said before they underestimated the firmware complexity ... anyway for +2 years  it's impossible that more than 2 guys worked at this  , or maybe one just stood and watched the other 

[joeqsmith]

You should be aware that there are some major problems with the VA mode.  This was seen with the prototype and 2+ years later has yet to be addressed.   

If you are talking about the adding of burden voltage  - that's not a problem. Also it's only a problem if you want to measure load power. I am measuring power supply (in my case battery) power. Every measuring instrument has imperfections and as long as you know about them - you can workaround them. The problem however when it starts freezing - that's not a normal DMM flow, it's a bug.

But yes, I agree that reversing the current probes in VA mode should be in the manual - it's a nice idea how to measure load power better (not power out of the supply).

If this were true, you would not be posting here for help.   I'm not aware of any other workarounds beyond the one I mentioned.  You could wait for the next firmware, or the one after that, or after that.......   Looking at the release dates, it seems they are winding down.    With David2 now out of the picture, I doubt you will see much effort being put into the application side.    Good luck.

[joeqsmith]

As I said before they underestimated the firmware complexity ... anyway for +2 years  it's impossible that more than 2 guys worked at this  , or maybe one just stood and watched the other 

It's not just the firmware.   The meter is less robust now than it was with the original front end design.  Sadly, I think a few of us were aware of that before I even ran the test.  Maybe they finally have a handle on the switch but honestly, I doubt it.    These are basic things that I would expect the designers to have sorted out before the first meter was ever produced.   

[EEVblog]

With David2 now out of the picture, I doubt you will see much effort being put into the application side.

That was never the plan. We got the app to a good refined working point and released it as fully open source software. If people want to improve upon it and add features they can, but we never had any plans to extend it ourselves beyond what it already is.

[joeqsmith]

With David2 now out of the picture, I doubt you will see much effort being put into the application side.

That was never the plan. We got the app to a good refined working point and released it as fully open source software. If people want to improve upon it and add features they can, but we never had any plans to extend it ourselves beyond what it already is.

From your comment, it seems as the development had stopped prior to David2 leaving.  The MS store shows a release date of 2/28/2018.  When I made that video showing how poorly it worked with my touchscreen tablet, I had the latest software.    At least I now know not to expect any further improvements.   

[Brumby]

Why criticise the original plan?  It's not as if Dave kept it secret.  Besides, as Dave has stated, software isn't his forte.

All we need is for others who are better equipped to develop the App to take up the challenge.  Then things can improve.

[npelov]

The app is not the biggest deal. If the firmware is working fine there will be someone to develop an app. If the firmware is not working there is no point of developing an app.

If this were true, you would not be posting here for help.

I'm not posting here for help. I'm reporting bugs. I payed for the DMM because I think it's a great idea. I still want it, so I'm doing what I can to help fixing the bugs. It doesn't help that they didn't open source the firmware. The problem is that the more bugs, the less Dave sells the DMM -> the less interest the Korean guys have in fixing the bugs.

Here is my observations on the 500mA current offset:
When I noticed the offset it was quite big - 0.65mA and it was slowly drifting down. It took about 2-3 minutes to drop down to 0. Now here is the strange part. When I move the probe to mA/uA socket the decimal place moves, but the offset stays in the same place on the screen - 0.65mA becomes 0.0065mA (that's the 5mA range I guess). And while it was drifting down I moved the probe between the A and mA/uA socket - the last two digits were almost the same and drifting down with the same speed - when it was 0.20 mA it became 0.0020mA on the other socket. It looks like this offset is closer to the ADC (or the firmware) rather than the front end.
The problem is that it eventually dropped down to zero and no matter what I did I couldn't make it rise (including turning off, switching ranges etc.).

One strange thing was that I moved the probes to A while it was still on 12V battery (switched to volts). Because the beep takes a second to start it took me 2-3 seconds to remove it. So I thought the fuse is gone. I guess there is some mosfet switching going on because the fuse if fine (Like Dave explained in the low burden video). Like I said this dmm was a great idea, but I think the wrong company manufactured it.

[joeqsmith]

Why criticise the original plan?  It's not as if Dave kept it secret.  Besides, as Dave has stated, software isn't his forte.

All we need is for others who are better equipped to develop the App to take up the challenge.  Then things can improve.

All I am saying is that a little feedback goes a long way.  If the software was the final release and there wasn't going to be any further updates, change your MS store to reflect that.  If you're selling old revision of the meter when newer ones have been released, change your store to make a note of it.   

The app is not the biggest deal. If the firmware is working fine there will be someone to develop an app. If the firmware is not working there is no point of developing an app.

Does this mean if the hardware is not working, there is no point of developing firmware?

If this were true, you would not be posting here for help.

I'm not posting here for help. I'm reporting bugs.

I thought in your first post,

Is there any way to log data with this DMM in a stable way?

you were asking for help.   My bad.  There is a thread to report bugs. 

[e0ne199]

i am starting to think that every new firmware released depends on how many 121GW are sold to the customers...i always believe that 121GW is actually a very good product... the real problem lies only in the firmware which is still not open source...if the source code is given to this forum members i believe that 99% of the bugs will be fixed asap

anyway i also have the logging problem, either using android app or using SD card....for android, when i use the logging feature for about 30 mins through bluetooth the app suddenly freezes and stops logging the data although the 121GW still runs with no problems.... as for logging using SD card, the problem is the same as what several forum members here mentioned before

[dcac]

https://www.eevblog.com/forum/blog/eevblog-1157-transistor-zener-clamp-circuit/msg2043871/#msg2043871
https://www.eevblog.com/forum/blog/eevblog-1157-transistor-zener-clamp-circuit/msg2037994/#msg2037994

Here are Dr Franks’s posts he made just after Dave released his video discussing the advantage of the double transistor zener type clamp, and that this now would be implemented in 121gw.

Now there’s nothing wrong with Dave’s video as such, I think he does a great job presenting it. But it’s very hard to see how Dr Frank could have made it any clearer that this clamp solution would not work as transients at +/- 25V now could reach U9 the HEF4053. Although it of course depends on exactly how you implemented this new clamp, U9 would have a very hard time handling these voltages. There were also quite a few other members in that thread voicing their concerns.

Still the clamp change went ahead and new HW revision of 121gw released, and it seems it‘s this revision Joe received with the two production meters he recently bought. And anyone following Joe’s videos knows exactly how well this new clamp worked.

Anyway, I guess this also is an example of the (broken) “information age” we now live in, but still it’s things like this that's really frustrating as a 121gw owner, seeing perfectly valid and detailed feedback not really be taking into account.

[Dr. Frank]

https://www.eevblog.com/forum/blog/eevblog-1157-transistor-zener-clamp-circuit/msg2043871/#msg2043871
https://www.eevblog.com/forum/blog/eevblog-1157-transistor-zener-clamp-circuit/msg2037994/#msg2037994

Here are Dr Franks’s posts he made just after Dave released his video discussing the advantage of the double transistor zener type clamp, and that this now would be implemented in 121gw.

Now there’s nothing wrong with Dave’s video as such, I think he does a great job presenting it. But it’s very hard to see how Dr Frank could have made it any clearer that this clamp solution would not work as transients at +/- 25V now could reach U9 the HEF4053. Although it of course depends on exactly how you implemented this new clamp, U9 would have a very hard time handling these voltages. There were also quite a few other members in that thread voicing their concerns.

Still the clamp change went ahead and new HW revision of 121gw released, and it seems it‘s this revision Joe received with the two production meters he recently bought. And anyone following Joe’s videos knows exactly how well this new clamp worked.

Anyway, I guess this also is an example of the (broken) “information age” we now live in, but still it’s things like this that's really frustrating as a 121gw owner, seeing perfectly valid and detailed feedback not really be taking into account.

In case that Dave really has implemented the FLUKE clamp as described in his videos, it's quite obvious, that this would fail.. and I also think that I have explained it reasonably well...but no feedback from Dave so far.

On the other hand, I would have expected that Joe might have re-engineered and analysed the actual protection circuit, either for confirmation or disproof of my findings .. instead he just did his brute-force / destructive tests on the 121GW.
Again, I'm no real friend of these tests, as they do not really comply to the usual test standards, therefore I also do not share his totally negative conclusions in full depth.

His findings about the faulty over-voltage / HV warning indication I fully support.

When I participated on the Kickstarter campaign, I was never expecting to get a fully professional and perfect DMM, so I'm not disappointed at this point in time at all.
I simply call it my 'Dave-O-Meter', so I have to admit that I also bought it, because it's from Dave and it's got the eevblog logo on it.

But in daily use, it's got several very practical and very well working features, like data logging via BT and on battery operation, which the other of my much higher grade DMMs don't have, so I'm really appreciating this instrument. 

I've found and reported many bugs on this instrument, so please don't call me a fan-boy, like I also found many bugs as well in the past on professional equipment like the 3458A or the 344465A.
Maybe some guys here shouldn't be too super-critical, because they simply had too high expectancy upfront.

Frank

[joeqsmith]

Maybe some guys here shouldn't be too super-critical, because they simply had too high expectancy upfront.

The feedback I often receive is something to the effect:

We been watching Dave's videos for years.  He is very critical of every meter he looks at, depending on the brand.  We are expecting his meter to not have the problems he rants about in his videos.   

Of course the problem with this is that they see what he has released and I think the question is, why is it now alright to sell a product that you would normally not spend the time to review. 

On the other hand, I would have expected that Joe might have re-engineered and analysed the actual protection circuit, either for confirmation or disproof of my findings .. instead he just did his brute-force / destructive tests on the 121GW.

I had invested a fair amount of time looking at the prototype.  It seems that most of what I pointed out was ignored.  It really made no sense to revisit how to improve it yet again as based on past history, I already knew what the end result would be. 

[dcac]

But Dave did change things, after it became clear that using 4007 clamp diodes with leakage in uA range wasn’t a good idea if you want to measure resistance using 20nA as reference and you don’t want the measurement to drift too much with temperature changes.

[joeqsmith]

If you looked at those last couple of posts of that UL listed TPI meter I showed, there are no MOVs and the meter was easily damaged.   I assume they don't need all the MOVs and fluff for UL be cause the meter does not need to be robust.  It needs to be safe.    This is the difference between what Frank is mentioning with following the standards.   There are no robustness standards so rather than follow the safety standards, I branched out on my own to look at their robustness.     

In the 121, it's anyone's guess what their goals were for robustness.  I am guessing something along the lines of UNI-Tish was good enough.  But they include the MOVs and basic parts.   

So while they did make a change, they didn't consider the impact it would have on how robust the meter was.  Or maybe they did and that was a very low priority.  Your guess is as good as mine.  One thing is certain is that it's now on par with many UNI-T products I have looked at. 

[CDaniel]

For 50 ohm range I thought that thermistors are the cause for the last 2 digits drift ... if I heated with my finger  ...  then I replace them with wires and surprise , the drift was the same . At the ridiculous low current for measurement and hi-gain , every solder joint in the signal path is generating enough thermovoltage .
I don't say it's the same for mV range but could be .

[dcac]

If you looked at those last couple of posts of that UL listed TPI meter I showed, there are no MOVs and the meter was easily damaged.   I assume they don't need all the MOVs and fluff for UL be cause the meter does not need to be robust.  It needs to be safe.    This is the difference between what Frank is mentioning with following the standards.   There are no robustness standards so rather than follow the safety standards, I branched out on my own to look at their robustness.     

In the 121, it's anyone's guess what their goals were for robustness.  I am guessing something along the lines of UNI-Tish was good enough.  But they include the MOVs and basic parts.   

So while they did make a change, they didn't consider the impact it would have on how robust the meter was.  Or maybe they did and that was a very low priority.  Your guess is as good as mine.  One thing is certain is that it's now on par with many UNI-T products I have looked at.

As we know UEi already changed the design again, now using transistors with a VEB0 at about 16V and that will clamp close to this voltage instead of 25V as your production 1-2 probably had. And as manufactures usually aren't changing their design just for the fun of it, they at least seems to be trying for some level of ‘robustness’.

[joeqsmith]

If you looked at those last couple of posts of that UL listed TPI meter I showed, there are no MOVs and the meter was easily damaged.   I assume they don't need all the MOVs and fluff for UL be cause the meter does not need to be robust.  It needs to be safe.    This is the difference between what Frank is mentioning with following the standards.   There are no robustness standards so rather than follow the safety standards, I branched out on my own to look at their robustness.     

In the 121, it's anyone's guess what their goals were for robustness.  I am guessing something along the lines of UNI-Tish was good enough.  But they include the MOVs and basic parts.   

So while they did make a change, they didn't consider the impact it would have on how robust the meter was.  Or maybe they did and that was a very low priority.  Your guess is as good as mine.  One thing is certain is that it's now on par with many UNI-T products I have looked at.

As we know UEi already changed the design again, now using transistors with a VEB0 at about 16V and that will clamp close to this voltage instead of 25V as your production 1-2 probably had. And as manufactures usually aren't changing their design just for the fun of it, they at least seems to be trying for some level of ‘robustness’.

I did offer to run one of the new designs.   I have a feeling there is not a lot of confidence and honestly, I think for good reason.   I test both +/-.   Clamps are not a stead fast number.   Again, the reason I went to the mux they have in there today.   The margins are very tight but it could be protected if that were a goal.   

[dcac]

A basic robustness test I’ve seen Dave use a couple of times is to put mains 240VAC into whatever mode the mode switch can be placed in, perhaps for a couple of seconds or so. I do my damndest trying to avoid doing a “mistake” like that. But still not an unreasonable test to do and see if the meter survives. So in this scenario the meters clamps, PTC’s and other input series resistors have to cope with the stress resulting from about 340V peak to peak, there can of course be some nasty overlayed transients on top of that but unless you’re monitoring the VAC with a scope you have no clue if this is the case.

Anyway, I can’t really remember if Dave also did this test on the 121gw, but I think so, in its original configuration anyway with the 4007‘s. But has he also done it with the new clamp configuration i.e. to show that it’s just as robust as it was before. If it isn't perhaps mentioning this in the next user manual revision is a good idea.

[EEVblog]

A basic robustness test I’ve seen Dave use a couple of times is to put mains 240VAC into whatever mode the mode switch can be placed in, perhaps for a couple of seconds or so. I do my damndest trying to avoid doing a “mistake” like that. But still not an unreasonable test to do and see if the meter survives. So in this scenario the meters clamps, PTC’s and other input series resistors have to cope with the stress resulting from about 340V peak to peak, there can of course be some nasty overlayed transients on top of that but unless you’re monitoring the VAC with a scope you have no clue if this is the case.
Anyway, I can’t really remember if Dave also did this test on the 121gw, but I think so, in its original configuration anyway with the 4007‘s. But has he also done it with the new clamp configuration i.e. to show that it’s just as robust as it was before. If it isn't perhaps mentioning this in the next user manual revision is a good idea.

Yes, done that (245VAC) countless times on every design variant.
I also do +/-1100V DC and 1100V AC as well, but both of my high voltage supplies will go into current limit mode when the meter is on the ohms range.
Survives all this just fine.