UT61E drift and recalibration

[Nuno_pt]

cuban8,

Nice catch, do that or only replace the 2K pot with an more stable one, but one with 5% tolerance and 10ppm/C° will cost about the same as the DMM.
So increasing / adding / replacing R16 for another more precise and lowering / replacing VR1, should give more precise and stable readings in Vdc.

[cuban8]

Actually, tolerance is basically irrelevant for both VR1 and R16 - you're trimming this resistance anyway. Tempco is more important, but since VR1 and R16 form a voltage divider with R15, this resistor also needs to be changed for a better one if you really want to improve things.

[Nuno_pt]

Ok, since we are one this, the schematic that I've list VR1 = 1K and R 15 / 16 = 9K09, we already know that VR1 is 2K, so the best is to take out the resistances and measure them and then replace both of them, for new values if you change the VR1 value, I think that Vdc will've better stability and accuracy with this mod.

Now I'm looking at the capacitance, for what I can see on the ES51922 data, R1k (pin 49) and R9k (pin 48) require an precise 1k and 9k resistors, they have 909k (R17) and 9k09 (R19) on the same schematic, and R18 (1k) is this in series / parallel with the other two, for better accuracy should this three be remove and place some precise ones with low % and low ppm/C°?

If you change that would it be better to change also VR3 (10k pot), R33 (100k) and R36 (200k x 4)?

Does VR3 adjust all ranges of capacitance or just one?

EDIT:For adding Capacitance below.

From what I can see on the ES51922 data, page 11 parag. 1.5:

C1 is adjust by itself for 22.000nF range (this should be an precise cap, % and ppm/C°), for more precise readings and stability.
VR3 adjust only 220.00nF = That is C2
10k adjust 2.2000nF = That is C3 (On my schematic there is one 100k resistor between C2 and C3, connected to pin 9 = BUFFH)

The ranges from 22.000µF to 220.00mF are all adjust by the ratio of the pins R9k/R1k, (R17,18,19 network) the more precise this network more precise the ridings on this ranges.

Is my reading correct on this?

[stephenlm324]

Although fitting the LT1790 improved the temp stability it did not cure the problem. I have stuck a thermocouple on the PCB under the main chip and tested with a 2V reference over 18 to 28 degrees C. Without the LT1790 the display count changes by 10, doing a good impression of a thermometer. After the LT1790 was fitted this dropped to a count of 6 over the same temp change.
To get rid of the last of the temp drift I have resorted to a thermistor (Vishay NTCLG100E2104JB Farnell 116-4823) in series with 1Mohm across R16 (9.09k on the schematic but measured at 7.55k). This gives a count of 2 drift at the moment, I need to fiddle more, dropping the 1M to 947k should do the trick. This compensation could be used without the LT1790 but the series resistor would need to be 780k ohm (calculated) to get more compensation, probably cheaper as well. This will of course mess up the calibration but no problem if the horrible 2k pot is still in.
Temp stability of the resistors in the potential divider should not be an issue, so long as they are all the same coefficient they will go up and down together with temp, and the voltage should stay the same.
Big problem is that 2k multiturn, the data sheet for this type of pot states 1% error with vibration which is 20ohm. I replaced with a 50ohm and a couple of resistors in series (there wasn't one standard value so used two). Vibration susceptibility now 0.5ohm and much easier to adjust. To figure out what resistor/s are required desolder the pot, measure the resistance with another meter and subtract 25ohm so that it's in the middle of the travel on 50ohm replacement.
The good news is that the resistance range does not drift significantly with temp trimming R34 will not help.

Thanks to Nuno_pt for spotting the mistake, now corrected.

[Nuno_pt]

Setephenlm384,

Thanks for the follow up on the update about the temp stability, seem that we are starting to get somewhere even without external Vref, the big must in here is always replace the 2k trimpot for one more stable / lower value, to get the precision and stability.
Like you point out if we can get it stable at 0 / 1 count at the 18° to 28°C  temp range that will prove that this meter could sold from factory without being more expensive, just by trimming better the components.

Could you leave the meter on in that temp range for about 24h or more and see what is the drift after that time with external Vref and without only with the resistors arrangement, one thing that I read about the references it's that you've to burn it for about 300 hours or more to make them age, then redo the calibration.

As for the resistance I don't see nothing to improve the ranges that uses R34 as an divider.

I'm looking now at the capacitance side, like I describe in the early post, after looking an little better, R19 is not present, only R17 (9k09) and R18 (1k).

On the 1M and the thermistor in series shouldn't be R16 instead of R1.

[Nuno_pt]

Another Chinese Vref mod see here ( http://www.crystalradio.cn/thread-398876-1-1.html ).

Would be nice if someone made the translation.

[markman]

Translated:

http://translate.google.com/translate?sl=auto&tl=en&js=n&prev=_t&hl=en&ie=UTF-8&u=http%3A%2F%2Fwww.crystalradio.cn%2Fthread-398876-1-1.html

I think the LT1790 might still be a better match with specs maybe slightly worse but current consumption also less.   

So...  If one substitutes all the key resistors and adds a reference, then this might become a stable meter over time??   Maybe while we're add it, we add some circuitry so it can do low-voltage current measurements?  Is there anything significantly better for $60 + $25 in resistors + $10 in reference parts?

Does anyone have a schematic for this meter?

[Nuno_pt]

Hi markman,

The schematic that I've is in attach, is the one that is on this forum, but be careful that this DMM has suffer many revisions.

R16 is 7.55k will on schematic is 9k09, the other is R19 is not there and so one.

About the low current measurement is an problem in all multimeters, some more than others, and is problem from the shunt resistor, so there is nothing that we can do about.

Only attenuate by Dave's µCurrent.

[stephenlm324]

I have finished fiddling my temp compensation. The meter now reads correct from 18 to 33 deg C briefly displaying +1 count.
I suspect it is the main chip that is temp sensitive, certainly the on chip Vref is off so it's a good bet that the rest of the chip is off.
Anyone considering adding the external Vref might consider using LT1790BCS6-1.25 which has a typical temp coefficient of 12ppm they are £2.70 GBP from Farnell, half what I paid for the top spec and I still had to add the compensation.
I have seen no sign of any long term drift. All of the heating and cooling I have done would accelerate this but it always goes back once the temperature settles. Any reports of long term instability should be put down to vibration and temp changes as these are having a much bigger effect.

[Nuno_pt]

Stephenlm384,

Can you post some pictures and the schematic of your mod, so the mod will be better understand by all, (1 picture worth 1000 words).

Glad that you got it stable at ~1 count, that is pretty impressive, now we've a stable DMM from 18 to 33°C.

Now that we've the stability problem solve, maybe we can address the calibration on the capacitance side, if that will be worthy of the time, what you think?


I've just found this from another Chinese mod, they are calibrating the resistance side, with trimmers.

See link ( http://bbs1.38hot.net/thread-6945-1-1.html )

[Lightages]

I am curious. What are the costs for the external ref mod and the other parts? It seems to me that this is turning a relatively inexpensive meter into something more expensive and then also needs to be recalibrated after the mods to be of any use. Is it worth it? If it is just for the fun of it then yes of course it is worth it but as far as being of real value?

[stephenlm324]

As the details for the external Vref are listed earlier I have left that bit out. The 780k ohm resistor (calculated value without external Vref) would be 680k + 100k in series and would probably need adjusting to match each meter, but it would be a good starting point. With the external Vref I am using 947k but again this would probably need adjusting for each meter. If you can get hold of the thermistor without paying a fortune for postage it would be a very cheap way of improving the meter.

[don.r]

Would these thermistors serve? http://www.ebay.ca/itm/10-pcs-Thermistor-Temperature-Sensor-NTC-MF58-3950-B-100K-ohm-5-DIY-New-/121252422483?pt=LH_DefaultDomain_0&hash=item1c3b352353&_uhb=1

They have a Beta of 3950K vs 3977K for the Vishays you got. They are also 5%.

[Nuno_pt]

It should be ok.

See also the research that was made regarding Ohm and Cap.

[don.r]

As the details for the external Vref are listed earlier I have left that bit out. The 780k ohm resistor (calculated value without external Vref) would be 680k + 100k in series and would probably need adjusting to match each meter, but it would be a good starting point. With the external Vref I am using 947k but again this would probably need adjusting for each meter. If you can get hold of the thermistor without paying a fortune for postage it would be a very cheap way of improving the meter.

Would 50R be sufficient for VR1? I was thinking of using a 200R to get an order of magnitude reduction in vibration deviation but still allow for a wider adjustment over time.

[Nuno_pt]

See the comments from stephenlm324 on the previous pages, he switch with one of 50Ohm, and a network of two resistors.

The more course the pot (Ohm) more drift it will have, most have 1%, it's more drift on 200Ohm that on 100 or 50Ohm.

[don.r]

As I understand it, its not the drift on VR1 that is the issue but more the vibration sensitivity which seems pretty constant across all these pots at around 1%. So instead of 20R of change we get 2R for a 200R pot or 0.5R for a 50R pot. My question to Stephen, I guess, is why 50R and not another value?

[mcinque]

Maybe the complete schematics could be useful to find what exactly what that VR does. Hope it helps.
Found on http://www.frankshospitalworkshop.com/electronics/diy-voltage_reference.html

[don.r]

I am curious. What are the costs for the external ref mod and the other parts? It seems to me that this is turning a relatively inexpensive meter into something more expensive and then also needs to be recalibrated after the mods to be of any use. Is it worth it? If it is just for the fun of it then yes of course it is worth it but as far as being of real value?

As no one has replied to this yet and I am in the process of applying several of the mods here is the cost.

LT1790ACS-1.25 - $5 (I got two for that price but that is the going rate)
caps/res - 10c total
NTC thermistor - 20c (10 for $2)
200ohm (or 50ohm) genuine Bourns pot - $2. (Chinese knockoff - 50c)

So all together less than $10... probably a lot less if you check your junk box or shop around.

[Lightages]

Hmmmm, so now the "bargain" UT61E goes from being a $60 meter to being a $70 meter, and still no proof of long term drift benefit, has very poor input protection, and needs to be re-calibrated. Don't forget the time to do the mod in the equation. I am not so sure I would recommend this meter anymore. I hope those who have taken the time to do this actually benefit from it. If you want the benefit to last a while, I would look at adding the input protection too. It won't necessarily make it safer for you but might prevent the meter from dying prematurely.

[Monkeh]

$5 of GDTs fixes the input protection (really, the fuses are fine. Really. You're not measuring 800VDC current with this meter. If you are, just think again.).

I think it's still a pretty capable meter for the price, but like any tool it has its limits and they must be known.

[don.r]

I'll be adding a couple of MOVs when I do the VRef mod. I don't normally measure anything over 150V so a 300V MOV should do it. Long term drift will only be known over the course of time maybe a year after the mod is done. It only takes a few minutes to do the work as all the pads are there. I will say that it really is only a 2200 count meter without the mods and you could pick up good 6000 count for the same money. I have another mod in mind however that is only capable with this meter. I'll let you know what it is when I have finished it.

[sorin]

I will say that it really is only a 2200 count meter without the mods and you could pick up good 6000 count for the same money.

Which model do you refer?

$5 of GDTs fixes the input protection

What you mean with "GDTs" ?

[don.r]

An Amprobe AM-270 for high voltage use or a used Fluke.

GDT = Gas Discharge Tube. You can use those or spark gaps or MOVs. Spark gaps and GDT operate at 1KV and above usually and I have no plans to take this meter anywhere near there so I'll stick to a 275V MOV (clamps at 600VDC).

[Nuno_pt]

Hi all,

I think that was some tutorial about changing the PTC that he brings with GDT or MOV, but can't find it.

The long term stability can only be view after an year after the Vref mod, against a calibrated source, DMMCheck Plus after the annual calibration, or something better.

Trimming the resistance side it's much more trouble since it controls V + Ohm, you've to mess with all the resistors in that chain + R34.
The 100Ohm rage it's adjust by itself, just put a lower value.

Still studying the capacitance side.

You can spend maybe another $40 in mods, the time doesn't count it's just for fun, so I think that for $80, 4.5 digits with 22.000 counts but more precise and stable, and maybe better on Ohm and pF, it's hard to beat for the price.

You can always by a HP 3478A, or others for around $150, but not portable, maybe not much space on the lab, etc...