Handheld meter robustness testing

[EEVblog]

I don't think YouTube is the place for video sharing any more. It's too commercialized and too unfriendly to individual content creators. Maybe better to consider Vimeo or Daily Motion?

Only a fool would move to another platform from Youtube. If you do then watch your existing audience drop to close to zero, and have almost zero views through search and related videos, absolutely guaranteed.

[kalel]

I don't think YouTube is the place for video sharing any more. It's too commercialized and too unfriendly to individual content creators. Maybe better to consider Vimeo or Daily Motion?

Only a fool would move to another platform from Youtube. If you do then watch your existing audience drop to close to zero, and have almost zero views through search and related videos, absolutely guaranteed.

One additional problem with e.g. Vimeo is that maybe you will have to pay. Not the storage limit, the limit in megabytes is possibly reasonable for a starting channel. It is up to 500 MB per week, and if I understand that right, that is possibly enough for one video per week, but I'm not sure what they consider commercial content and if electronic channels would fit there.

Businesses may not use Basic or Plus accounts to host videos. If you want to upload commercial videos, you must use Vimeo PRO or Business. Commercial content includes:

Videos promoting or representing a for-profit business or brand
Videos containing any form of advertising
Videos hosted on behalf of a business (i.e., uploaded to Vimeo and embedded on your company’s website)
Product demos and tutorials
There are exceptions, however! If you are an independent production company, author, artist, or non-profit, you may use any account type (Basic, Plus, PRO, or Business) to showcase your work.

I'm guessing basic is fine, considering the exceptions? But then, what if a channel receives a product for review (and many a do, a company sends something to review) they might have to change the plan in order to feature that video. I'm not sure.

[joeqsmith]

Regarding robustness, now that I am waiting for a rebranded an860b+ and this video is more important for me than before:
...
(I don't remember details so I am watching it again, although it will take a bit).

I wonder, does the larger PCB comparing to the small version make a difference? There's more space, but I don't know if it does provide better clearances.

If the goal is to make something robust and CAT IV rated, yes having more space would make the job easier.  It does not mean that a large meter is going to be more robust than a smaller one or meet a higher standard.   The UT181A is physically a big meter and died with one strike of the stupid little grill starter.   Again, the layout needed some help.  Strange as UNI-T makes some of Danaher's products which are certified to meet the EMC standards.  The TPI194 is the same, really large but was damaged with the AC line applied.   So having all that room does not mean the designers don't have to still do their jobs. 

Look at the Brymen BM235, Fluke 101 and 107.  These are all fairly small meters and at least from my testing, some of the electrically most robust meters out there. 

The meter you referenced made all the way to 4KV.  It had some pretty bad design problems where it would give false readings.  But so can the $850 Gossen meter I have.   More money does not guarantee a good design.    I did ask Fungus to check his new one but I don't believe he ever did.   

[rsjsouza]

Joe, excellent video. Interesting to see how the clamps and the thermistor are safe and sound while the meter is on fumes. Perhaps they are protecting the 3V battery?

[joeqsmith]

Joe, excellent video. Interesting to see how the clamps and the thermistor are safe and sound while the meter is on fumes. Perhaps they are protecting the 3V battery?

In the case of this meter along with the 8002 you are correct.   These special designers put the PTC and clamp remain safe and sound behind the rotary switch.   The switch has to take the full voltage which obviously it can't.   Most of the meters I look, if they even have a clamp, it's  located before the switch and limit the voltage to a few KV.  Sometimes they design the switch to handle this voltage, sometimes we get a light show.

Still no word on the 8008.  I do plan to run the thermal testing on these two meters (8002/8) for those of you who asked.

[rsjsouza]

Joe, excellent video. Interesting to see how the clamps and the thermistor are safe and sound while the meter is on fumes. Perhaps they are protecting the 3V battery?

In the case of this meter along with the 8002 you are correct.   These special designers put the PTC and clamp remain safe and sound behind the rotary switch.   The switch has to take the full voltage which obviously it can't.   Most of the meters I look, if they even have a clamp, it's  located before the switch and limit the voltage to a few KV.  Sometimes they design the switch to handle this voltage, sometimes we get a light show.

Oh, I see... It may be the same special designer that did the work on a meter that had the 10A fuse on the V input (I can't recall the model/brand).

[joeqsmith]

Joe, excellent video. Interesting to see how the clamps and the thermistor are safe and sound while the meter is on fumes. Perhaps they are protecting the 3V battery?

In the case of this meter along with the 8002 you are correct.   These special designers put the PTC and clamp remain safe and sound behind the rotary switch.   The switch has to take the full voltage which obviously it can't.   Most of the meters I look, if they even have a clamp, it's  located before the switch and limit the voltage to a few KV.  Sometimes they design the switch to handle this voltage, sometimes we get a light show.

Oh, I see... It may be the same special designer that did the work on a meter that had the 10A fuse on the V input (I can't recall the model/brand).

I really need to proof read     I have seen a few meters with fuses in-line with the V-input.  That last RadioShack meter was like that.  But 10A??!!  Assuming they actually shared the 10A with the V, they had to switch it out.  Hard to believe that was done with the rotary switch.   They can get away with it in the mA ranges, but 10A, wow!   At least it had a fuse!   

[evava]

I do plan to run the thermal testing on these two meters (8002/8) for those of you who asked.

And we have no objection to the 80 deg C span ! (as Dave had) 

But, I would be interested in how it is with the alleged thermocouples whose influence prevails allegedly (as Dave mentioned) - could possibly someone (maybe even from Metrology section?) explain that in depth and confirm or disprove Joe's results?
Still, there is data of UT181a which does not seem to be affected by that alleged significant thermocouples influence. Could that influence be mittigated by suitable(proper) material of it's input posts in relation to the connecting conductors?

All that could be prevented (IMHO) if Joe had performed these termal testing with input posts shorted, so no different temperature and voltage potentials on the leads.

[alm]

See the first few pages of this Keithley publication (basically an excerpt from the excellent Keithley Low Level Measurements handbook) for a good discussion of thermoelectric EMFs a.k.a. the Seebeck effect.

The basics are explained in the attached figure (taken from the low level measurements handbook). If you have to dissimilar metals A and B, for example A is a piece of copper wire and B is oxidation on that copper wire, than each A-B and B-A junction generates a voltage that is approximately linear with temperature. This is how thermocouples work (hence the name thermocouple effect). As long as T1 and T2 are the same temperature, you will have two voltage sources with opposing voltages in series, so they will cancel out. But if T1 is 10 K higher than T2, then the AB junction may have a voltage that is 10 mV higher than the BA junction, resulting in a net voltage of 10 mV at the inputs of the meter. Note that these metal junctions may be inside or outside the meter.

One way to prevent this is to insure that the entire circuit is isothermal. This may not be easy if it is dissipating power somewhere. The other way is to choose your materials for low Seebeck coefficient. Copper oxide is pretty much worst case, hence why I gave that as an example. But you can't really expect your average DMM to have low thermal EMF connections. A possible location would be the leads in and out of the thermal chamber. If one of them has a shorter length outside the thermal chamber, the connector may be hotter/colder than the one with a longer length outside the chamber.

Add to this that testing near the bottom of the range exaggerates any offset errors and hides gain errors. So basically I agree with Dave that a better measurement would near full scale and not on the lowest range. Something like 1 V or 5 V might work. Sure, at 1 mV differences between the meters are larger, but are these differences meaningful and indicative of better real-world performance? Or are they just coincidence?

[joeqsmith]

I do plan to run the thermal testing on these two meters (8002/8) for those of you who asked.

And we have no objection to the 80 deg C span ! (as Dave had) 

But, I would be interested in how it is with the alleged thermocouples whose influence prevails allegedly (as Dave mentioned) - could possibly someone (maybe even from Metrology section?) explain that in depth and confirm or disprove Joe's results?
Still, there is data of UT181a which does not seem to be affected by that alleged significant thermocouples influence. Could that influence be mittigated by suitable(proper) material of it's input posts in relation to the connecting conductors?

All that could be prevented (IMHO) if Joe had performed these termal testing with input posts shorted, so no different temperature and voltage potentials on the leads.

Just for reference, the following link is to the original posts:
https://www.eevblog.com/forum/testgear/new-eevblog-branded-multimeter-coming/225/

I agree that the thermocouple effects are present and that to get a better picture of what the drift is doing that we need to look at several points.  There was also the concern of running them outside of their specified temperature ranges and how doing so invalidates the test.    I think the last point was they should be tested in the V range rather than the mV and near the upper limit of the range.   

Then there is the a question of how the data should be presented. One way would be to compare the overall change as a percentage of reading.  I have been showing the total difference between the min/max temperatures over the total change in temp.  The later has people questioning if the drift is linear (which I doubt) and again would point to needing more data. 

Based on one member's testing,  I have ran a few of the logging meters with a fixed temperature with the inputs shorted over an extended period.  I use a thermistor for feedback, a fairly large fan and resistors to control the temperature of a double insulated box.


I would suggest when you comment, consider that meters could range from the free harbor freight to the overprized Gossen.  Some meters may not even read into the mV let alone uVs.   Test times are a premium.  You may not want to invest a week looking at the drift of a free meter.   

[kalel]

I would suggest when you comment, consider that meters could range from the free harbor freight to the overprized Gossen.  Some meters may not even read into the mV let alone uVs.   Test times are a premium.  You may not want to invest a week looking at the drift of a free meter.

Of course. While more data is always better for comparison, for any testing that includes many meters, an automated system would be needed (something to minimize human time needed), and that would have costs associated. You're not doing this commercially, so building something too costly would be an investment that can't be returned.

[evava]

Thank you for your answer, Alm.

So, if I understand it well, Joe measured (on 1mV level at 80 deg C span) mainly thermocouples voltages, because half of his leads(and connections) were on (about) 25 deg C, while other half of his leads(and connections) was at changing temperature from -20 deg +60 deg.

Why no one had mentioned that before, after all everyone here watch his tests?
Why you all had been waiting on Dave and bad looking test of his 121GW, then suddenly everyone declares Joe's thermal testing for invalid?
Why did we wake up so late?
No one takes seriously testing of some UNI-T or something?
BM235 test went out well, so test was at that time good?
121GW test went out bad, so test is suddenly bad?

I hope I dont get ban for it...

[alm]

Not necessarily. If the meter input is at 80°C, and the voltage source is at 25°C then you might have a copper-tin-copper junction at the meter (+ terminal) at 80°C, one at the voltage source at 25°C (+ terminal), another one at 25°C at the voltage source (- terminal), and another one at the meter (- terminal) at 80°C. If you do the math, then it should all cancel out. The linked Keithley document discusses this. But at these temperature differentials, it is easy to see how for example one of the jacks at the voltage source might be a few degrees warmer because it is closer to the thermal chamber. Same goes for the internal components of the meter (the input jacks may be cooler due to the cables drawing heat away).

I am not claiming that it is only the Seebeck effect and that Joe's measurements at 1 mV are worthless, but I am not convinced that the Seebeck effect is negligible either. It is somewhat suspect to me that meters that showed no difference at 1 V suddenly showed large changes at 1 mV.

As for why I did not bring this up before, because I do not regularly watch Joe's videos. I am not that interested in handheld DMMs. I only hang around for the occasional discussion of testing methods .

I would suggest when you comment, consider that meters could range from the free harbor freight to the overprized Gossen.  Some meters may not even read into the mV let alone uVs.   Test times are a premium.  You may not want to invest a week looking at the drift of a free meter.   

In my opinion testing at the lowest range has several problems (including Seebeck, ignoring gain errors and variation in the lowest possible range). I would suggest picking a single voltage in the 1 V - 10 V range and sticking to this. If interested, you could do multiple points in a range for a single meter at multiple temperatures to study linearity in both the voltage and temperature range, but doing that for dozens of meters is insane in my opinion. One thing I think is important because differences in the 1 V range often seemed in the order of 1 LSD is to indicate the +/- 1 LSD uncertainty. So if one meter went from 1.000 V to 1.000 V, and another went from 1.000 V to 1.001 V, then I would say the first one had 0% +/- 0.1% drift, and the second 0.1% +/- 0.1% drift.

[mikeys]

BM235 test went out well, so test was at that time good?
121GW test went out bad, so test is suddenly bad?

I hope I dont get ban for it...

The 121GW is a prototype unit and Dave said his intent was to send one to Joe for him to see how much it took to blow up. Testing isn't representative of the final product and though I see Joe's side of the story as he states many times that it's not a finished product and the results aren't to be taken to heart, people might make their minds up about it before it is even finished, which wouldn't be fair to Dave's efforts to make a good mid-ranged meter nor to the actual product itself.

[evava]

For Dave's effort - we all thank him of course.
For this forum also.
His new 121GW is going to be nice meter and good selling no doubt, no matter what happens.

I did some "temperature measurements" myself (unpublicable - in the car in the sun  ), and it seemed to me that (IMHO) meters have some internal (software-firmware) compensation for common temperature range (up to about 45 deg C), because meters are very stable in that range.
I consider that as "cheating".
That is why I welcome Joe's tests, they are out of this internal (software-firmware) compensation and then meter really shows what has got inside.
Therefore, (IMHO) those non-linearities.

P.S. LCD displays of the meters do not like hot+sun - do not ask how I know! Fortunately it is almost reversible.

[joeqsmith]

....
121GW test went out bad, so test is suddenly bad?

I hope I dont get ban for it...

Yes, I believe some of what you wrote is true.  I often run into people trying to justify the levels that meters fail the transient testing as well.    I wouldn't be too concerned about getting dinged for asking questions.  It would be better to discuss the merits of running these tests and how best to run them, regardless of the meter being testing.  I am open for suggestions on how to run them.  Again, I have nothing to gain from it.  To me it's just data.   

It is somewhat suspect to me that meters that showed no difference at 1 V suddenly showed large changes at 1 mV.

As for why I did not bring this up before, because I do not regularly watch Joe's videos. I am not that interested in handheld DMMs. I only hang around for the occasional discussion of testing methods .

First, thanks for the bit of feedback. I doubt we will see too many people weighing in on this topic.     

I see a meter buff as someone who collects meters and treats them with great care.  They display them in their show cases to impress their friends perhaps changing batteries once a year and cleaning them.      

I'm a little lost on what data are you referring to.  Which meters? 

Anyway, I don't completely disagree with what you and other have wrote.   I have always assumed the offset error would dominate from the little data I collected early on.   I can't prove it because I don't collect this data.   I agree about running them shorted for the offset.  However, we really want both gain and offset.  For gain pick a level that would not OR on any of the meters I have in their mV range.  I see no reason to run them through all of the attenuators to see how each resistor effects the gain.   I also can't see changing the range switch during the test.   Drop the temps to 0-40 to try and assume the drift is somewhat linear.  Stay with the two temperatures, so four data points total.  Use the same set of cables for all the meters on both leads, keeping everything symmetrical to minimize these errors.  We are not looking for nV errors anyway.   For presenting the data, I would just keep the raw data.   Ignoring the free HF meter, I could use 250mV as a test point for gain.  I would stay with the hour dwell times after the chamber reached the set point.

[alm]

I'm a little lost on what data are you referring to.  Which meters? 

The only data I could quickly find on your channel: the data shown here. Sorry if this is outdated data, I could not find it in your spreadsheet or in another easily searchable video. Now that I look at it again, actually none of the meters that perform not very well with the 1 mV tested are in your 1 V data set, so my statement about them performing much worse is not completely true. I guess the most striking difference is the BM869S which performs pretty well in the 1 V tests but mediocre in the 1 mV test.

However, we really want both gain and offset.  For gain pick a level that would not OR on any of the meters I have in their mV range.  I see no reason to run them through all of the attenuators to see how each resistor effects the gain.   I also can't see changing the range switch during the test.   Drop the temps to 0-40 to try and assume the drift is somewhat linear.  Stay with the two temperatures, so four data points total.  Use the same set of cables for all the meters on both leads, keeping everything symmetrical to minimize these errors.  We are not looking for nV errors anyway.   For presenting the data, I would just keep the raw data.   Ignoring the free HF meter, I could use 250mV as a test point for gain.  I would stay with the hour dwell times after the chamber reached the set point.

I agree with not changing the range switch during the test. I am curious why the mV range. As you say, the divider should not have much of an effect either way. I guess there could be differences in tempco between the two legs of the divider, but that does not really seems worth worrying about and is not feasible testing.

So as far as the rest of the front-end and the ADC are concerned, there is pretty much no difference between shorted inputs at the mV range or the 100 V range. Why not pick a range that is less error prone, like the range that includes 1 V? I would also guess that for most users the lowest and highest ranges are the least used, so picking a middle range makes sense. Are you deliberately testing for thermoelectric voltages inside the meters?

I guess if you wanted spectacular videos you would test at 1000 V and see how they deal with condensation .

[floobydust]

For ambient temperature tests, the chinese complain about AN8002 having 100ppm/°C reference drift ala internal TL431 so that was behind the move to a dedicated external reference IC in the AN8008 ICL8069DCZR rated 0.01%/°C

Another test to do at the same time is the cold-junction compensation, it seems to be a MCU internal diode with the same cal factors recycled for every unit. I'm not sure how well that works.

Canadian winters running a DMM even at -20°C the test leads get very hard and brittle, the LCD display fluid starts to freeze up, alkaline batteries crap out too.
I think 0°C is the lowest any DMM is good for.

[EEVblog]

The 121GW is a prototype unit and Dave said his intent was to send one to Joe for him to see how much it took to blow up.

It wasn't even for that, it was never meant to be a formal test, it was just for fun because everyone was asking for it, look at the poll in this thread! Also Joe had done one or more videos publicly asking for me to send one, so I (foolishly?) sent one before it was finished.

[2N3055]

....
I did some "temperature measurements" myself (unpublicable - in the car in the sun  ), and it seemed to me that (IMHO) meters have some internal (software-firmware) compensation for common temperature range (up to about 45 deg C), because meters are very stable in that range.
I consider that as "cheating".
....
P.S. LCD displays of the meters do not like hot+sun - do not ask how I know! Fortunately it is almost reversible.

Nonlinearities in tempco come firstly from reference chip alone. And also all kinds of compensations.Bandgap reference is basically tempco compensation circuit.
Software compensation (lookup table or polynomial or whatnot) is advanced technique I doubt would be used in cheap meter..  And all of them are compensated for minimum error at normal work temperatures, at the expense of larger errors where nobody cares..  It's not cheating, it's clever engineering..

Canadian winters running a DMM even at -20°C the test leads get very hard and brittle, the LCD display fluid starts to freeze up, alkaline batteries crap out too.
I think 0°C is the lowest any DMM is good for.

This.
I don't do measurements in Siberia at -55°C. Nor I do measurements at 60°C. When temps reach 40° it becomes unsafe for people to work. I would never use my meters that I payed for at those temps, for fear of being damaged or at least being knocked out of specs.. Batteries leak, or explode (Samsung Galaxy7 ?), LCD gets damaged, plastic gets brittle or soft, so it breaks or deforms.....

It is unneeded to test outside of temp specs of device. It doesn't matter what it shows.. You can't trust it, and if you plan to do it on a regular basis it might damage your equipment. Manufacturer clearly told you , when you bought it, in specifications, that if you measure something while at 60°C meter might start showing elephants instead of Ohms, for all they care.
Unless it is a special meter specified for those temps..

I really like that Joe tries to find breaking point for voltage resilience. That is useful information. If meter gets killed by grill starter, yeah, that means your carpet can kill it.. That is something that is very likely gonna fail in service.. Also overvoltage events are unpredictable and dangerous.. So all info that shows a bit more about that is nice. And also Joe doesn't test outside of specs when testing overvoltage.. If anything, he tests at energies that are much smaller than IEC testing standards prescribe.

Of course I'm not trying to patronize anybody to tell them what to do. I like Joe's videos, and he might chose to make some that have entertainment value more so than engineering one.. That is fine.

I'm just saying that these numbers do NOT have real-world practical meaning, as some viewers seem to think.
I assure you, Joe knows that very well, too.. But he is allowed to have a bit of fun in his own private time, and that is what he does.

I just think it's a shame that he seems to have access to real thermal chamber, and is more than qualified to make real, useful thermal characterization for equipment.
So he could do it that way and finally we would be able to see some numbers manufacturers are hiding like snake is hiding legs...

But of course, it is a serious, time consuming project and I understand if he just don't have time to do it. And that is also fine with me..

[3db]

I don't get why the test of Dave's meter is causing such a fuss.
All Joe did was shove some meters in a temperature chamber same as he did with the UT-61E and some others before.
He has always said that Dave's meter is NOT the final version.
Whilst the huge temperature range that Joe used in the tests isn't really the best way to assess the temp co of the meters
it does show how they survive the experience.

3DB
 

[2N3055]

I don't get why the test of Dave's meter is causing such a fuss.
All Joe did was shove some meters in a temperature chamber same as he did with the UT-61E and some others before.
He has always said that Dave's meter is NOT the final version.
Whilst the huge temperature range that Joe used in the tests isn't really the best way to assess the temp co of the meters
it does show how they survive the experience.

3DB

I can't speak for everyone else, but I couldn't care less about Dave's meter.. If it happens to be very good meter I'll might buy it, support it and recommend it to all my friends. If it comes out to be not so good, I will say that out loud too.
I respect Dave and his work, but it's not religion.. I don't worship anything or anybody, and Dave is not exception to that.

It's not about Dave or his meter.. If anything, Dave's meter did well compared to two other more expensive meters.

It is about test that has no technical value, to any meter out there..
It is entertainment and fine as such. Problem is that many people started to take is as  valuable info.. It is not.

Anything tested outside it's specified parameters is outside it's parameters.  It has no useful information. It's just entertainment, good fun. Nothing else.
Trying to extrapolate conclusions about quality, tempco and such, that will have practical value to anybody is at best naive...

So there you are, my explanation, for my reaction...

[joeqsmith]

The only data I could quickly find on your channel: the data shown here. Sorry if this is outdated data, I could not find it in your spreadsheet or in another easily searchable video. Now that I look at it again, actually none of the meters that perform not very well with the 1 mV tested are in your 1 V data set, so my statement about them performing much worse is not completely true. I guess the most striking difference is the BM869S which performs pretty well in the 1 V tests but mediocre in the 1 mV test.

I wondered if the original data is what you were asking about.  I never made the actual spreadsheet available like I have with the transient testing.   The video you linked and a few posts are about the only places you will find it.  That data set was up to date as far as what I did with those meters. 

I agree with not changing the range switch during the test. I am curious why the mV range. As you say, the divider should not have much of an effect either way. I guess there could be differences in tempco between the two legs of the divider, but that does not really seems worth worrying about and is not feasible testing.

So as far as the rest of the front-end and the ADC are concerned, there is pretty much no difference between shorted inputs at the mV range or the 100 V range. Why not pick a range that is less error prone, like the range that includes 1 V? I would also guess that for most users the lowest and highest ranges are the least used, so picking a middle range makes sense. Are you deliberately testing for thermoelectric voltages inside the meters?

Yes I was deliberately looking for everything causing the drift down in the muck.   It appears the 8002 is marked 0-40C.  The meter that failed to run at all at -10C (TPI) is also rated for 0-40.  -10 is nothing.  I've done work many a time below these temperatures.  Granted, 10% accuracy would be fine but the meter needs to at least work.  This is partly why I pushed the temps even lower. 

I guess if you wanted spectacular videos you would test at 1000 V and see how they deal with condensation .

Seeing my 1000 V supplies only put out a mA or so, it shouldn't be a problem.   Maybe just run connect them to the 220AC 50A line, drop the temp to -40, then open the door and wait.. 

[joeqsmith]

The 121GW is a prototype unit and Dave said his intent was to send one to Joe for him to see how much it took to blow up.

It wasn't even for that, it was never meant to be a formal test, it was just for fun because everyone was asking for it, look at the poll in this thread! Also Joe had done one or more videos publicly asking for me to send one, so I (foolishly?) sent one before it was finished.

Sorry you feel that way Dave.  I would say if anyone was the fool, it was me.  There is little I can do at this point beyond not mentioning the meter and pulling down the videos.  If you like I can remove the last one as well.   Your call.   

[joeqsmith]

I'm just saying that these numbers do NOT have real-world practical meaning, as some viewers seem to think.
I assure you, Joe knows that very well, too.. But he is allowed to have a bit of fun in his own private time, and that is what he does.

I just think it's a shame that he seems to have access to real thermal chamber, and is more than qualified to make real, useful thermal characterization for equipment.
So he could do it that way and finally we would be able to see some numbers manufacturers are hiding like snake is hiding legs...

But of course, it is a serious, time consuming project and I understand if he just don't have time to do it. And that is also fine with me..

Anyone who has ever done thermal testing knows it can take a fair amount of time (weeks, months).   With the handhelds, even if I wanted to run a complex test, the data would have to be collected manually or possibly using multiple cameras inside the chamber.  A lot to invest for some cheap meters.   I'll leave that to the next person.   

So we are back to what test could be ran in the shortest time that yields the most useful information.   Again, I am open to suggestions and have posted a proposal that Alm has weighed in on.  Feel free to do the same.