Rigol DM3061 6-1/2 Digit Multimeter

[nukie]

I don't like to spend money on shit like this so I took the liberty to post some of the pictures I copied over from the Chinese EE forums.

These parts no. were slightly sanded off but the author manage to decode
ADC: ADS1256, 24bit
VREF: ADR421, 2.5V

[nukie]

more

[Bored@Work]

OMFG?! Please tell me this is from a pre-production lot.

[alm]

With BAW. Some nasty rework and green wires going on. If a unit like this reached one of their customers, they should fire the monkey that's doing QC, and also the one who did the rework.

[nukie]

No it's not pre-production, poster's school laboratory bought it, and he/she took it apart. I am okay with the artistically positioned components  but all the flux residue there is going to cause some error in high impedance measurements. Pay peanuts get monkey. Actually, ~$700 is no peanuts, its very expensive POS!

The forum owner was later contacted via sms by China Rigol representative about the situation. They claimed that the units are engineering prototypes and offered the poster replacement units. The poster school laboratory bought two of these units and they have similar reworked condition.

Forum members don't agree with the representative. Engineering units should not have chips with numbers sanded and should not have landed in the poster's school lab in the first place.

If this is a unit sold in the market then the Rigol multimeter production group is definitely different from the oscilloscope group. The oscilloscope PCB manufacturing is a lot better and professional.

So now I am waiting to see the inner picture of Rigol power supply, someone who was sponsored by Rigol actually took the time to make a multi-part-star-wars-trilogy on how awesome the user interface is compared to a simple HP PSU. I am so impressed by how many times the reviewer can repeat the user interface is awesome the user interface is awesome the user interface is awesome till it drilled a hole in my mind.

By the way, this unit is discontinued replaced by a 5-1/2 digit meter.

[saturation]

Ouch.  I'm with alm and BAW.  This is what the tear downs are all about. 

The big question now is, did the unit work to spec?

With BAW. Some nasty rework and green wires going on. If a unit like this reached one of their customers, they should fire the monkey that's doing QC, and also the one who did the rework.

[nukie]

On another note, if you are a multimeter fan, you might have notice the white decade resistor divider network in picture no.5
Do not for one second think that it is a Caddock. It's actually made by Thunder Resistor who is notorious for packing normal through hole resistors into their epoxy case resistor divider. I'm pretty sure the T/C it will be a lot higher than Caddock 10ppm 1776 series.

[alm]

So the conclusion is that this series isn't worth the cost savings over equivalent Agilent or Fluke models? At least the DS1052E is much cheaper than Agilent/Lecroy/Tek, but the cost savings for Rigol bench meters over brand-name stuff are much smaller.

[nukie]

The construction of the input connector there is going to introduce a lot of thermal error for a 6 digit assuming the last numbers are usable.

It's not just the cost, with so many digits bench meter you want a heated zener to minimize TC to lowest possible since you dont have to worry about wearing down batteries. The top end meters on the market for eg. HP 3458A and the Fluke ehhem Datron 8508A both has a heated zener as the voltage reference. And of course 6 digits meters of this class are best served by the popular LM399. Examples are HP34401/10, Keithley 2000, Fluke 8846.

Older generation meters like the Fluke 8840 is based on Motorola SZA263 again heated zener. The 5 digit HP 3456A is based on the LM299.

I don't know why they choose to use FET based voltage reference probably because of cost but its not a wise choice. I doubt there is drift compensation built in to the firmware. I don't think so for a company this young because studying the character takes years and Fluke/HP sure did a good job not only in research and also documenting. Supporting circuitry for heated zener is much complicated, that probably cut some cost there. This clearly states the commitment of the company when it comes to the quality of their products. All I can say is that the products are made for users on a budget sacrificing performance.

5 digit on a budget? okay that's fine. But if you are after a 6 digit meter you sure know its going to cost more and you are probably going to use it for something critical, why would you want a shit class meter. No place for it in the market. 6th digit is a wank factor for Rigol.

If you want 6 digit from the East, you better off with Taiwan or Japan. There is a Taiwanese company that designed and made the Keithley 2001(which has not so good reputation). The other choice is Advantest. It's going to cost some money, because this Japanese company is one of the few who makes one of the 8 digit meter in this world, locally in Japan.

Btw LM399 is still being made and there are plenty of stock in the market.

[alm]

It's not just the cost, with so many digits bench meter you want a heated zener to minimize TC to lowest possible since you dont have to worry about wearing down batteries. The top end meters on the market for eg. HP 3458A and the Fluke ehhem Datron 8508A both has a heated zener as the voltage reference.

The HP/Agilent 3458A uses an LTZ1000, not sure about the Fluke. The Datron 1081 achieved 7.5 digits of resolution and a very good stability and tempco (better than LMx99-based meters) with just two pairs of two zeners with an opposite tempco.

And of course 6 digits meters of this class are best served by the popular LM399. Examples are HP34401/10, Keithley 2000, Fluke 8846.

Older generation meters like the Fluke 8840 is based on Motorola SZA263 again heated zener. The 5 digit HP 3456A is based on the LM299.

The HP 3456A is 6.5 digit, and has better stability than newer 6.5 digit models like the 34401A. Note that the LM199/299/399 are all the same part, just different temperature range and selection. LM199 is the best, and LM399 the worst, but the only one still in production. Of course if you select them, like Agilent and Fluke do, you can achieve the same result.

If you want 6 digit from the East, you better off with Taiwan or Japan. There is a Taiwanese company that designed and made the Keithley 2001(which has not so good reputation). The other choice is Advantest. It's going to cost some money, because this Japanese company is one of the few who makes one of the 8 digit meter in this world, locally in Japan.

I've heard decent things about Picotech, but no direct experience.

Btw LM399 is still being made and there are plenty of stock in the market.

I think LT still has them in production, National discontinued them.

[nukie]

Sorry I got mixed up.

What I meant was HP 3468A which is the 5.5 digit it has a LM299.

The HP 3456A/7A are selected LMX99.

[buffalo960]

May I ask what forum you found this on?

[nukie]

ourdev.cn use google chrome automatic language website translation

[MBY]

OMG! That actually explains something I long wondered!

I bought a DM3061 1.5 years ago at Conrad. The strange thing was that it was cheaper than DM305X, but with better specs. I got the impression that the DM306X was a bit outdated. I also found other Rigol bench DMM with a different name scheme.

Actual thread in combination with this thread https://www.eevblog.com/forum/index.php?topic=4750.0 gives a clue of why the "lower end" model was more expensive.

Anyhow, my DM3061 performs good, no problem at all. I have no reason to believe that it's our of specs. But the next best DMMs I have is "only" a 210 000 counts, 0.03% (TTi something "computing multimeter") and 100 000 counts, 0.0??% Fluke 45. They all agree on the digits they can display. But I really have no way to test the last digits on DM3061 (2 400 000 counts, claimed 0.0078% on 2 VDC).

Now I'm quite chocked!

But, when the DM3061 is not "on", the power button glows (fade in and out). I always assumed that it's keeping the heat up on stuff like voltage reference. But no heated ref on the pics? It is burning watts for nothing? Foolish of me not to understand why the DM3058 was more expensive!

[Bored@Work]

I bought a DM3061 1.5 years ago at Conrad.

Can you have a look inside your DM3061, and tell us if the build "quality" is the same as shown in the pictures here, please?

[MBY]

Yes, I was thinking of it. But currently the meter is connected and running a long-time logging session, so I cannot turn it of for quite some days to come. Also, as long as I don't open it up, I can live in happy denial and just assume that everything is okay inside there!

Well, of cause I must open this baby up! You have to wait a few days in anticipation!

[MBY]

Ok, my long time logging session is over!

I'm currently opening my DM3061 up. I shall shoot some photos but I can already say my unit doesn't look bad at all. I already identified the proper caddock resistor network and not seen any bad soldering or flux.

One question: How to identify the voltage reference? I see no metal TO-46 so I suppose it's in a SO8? I have higher resolution pictures available, so if someone wants a zoom in on a particular area, please tell. A couple of SOs are sanded, I assume they may be ADCs and voltage reference and such.

The bottom side of the PCB contains no components whatsoever. No remarks there.

I'm quite happy. I was worried for days that my trusted DM3061 was not so trustworthy, but I think its in the clear now.

Edit: Info from the utility menu:
Model: DM3061
Serial: DM3D115000887
Version: 03.12.00.03.09.00
CAL temp: 23C

I'm very happy with my DM3061 and I have used it heavily under my 1.5 years of ownership. I never turn my unit off on the back switch, only the "soft off" on the front panel. I use it for long time logging, and I like the stat functions.

I have only one big complaint: The null mode is somewhat retarded. I prefer a conventional "relative" mode. If I use the null mode, the DM3061 put itself in some sort of autoranging mode, even if you clearly has selected a fixed range. Its probably deliberately made this way, but I don't like it and I cannot really use it as a rel mode.

As I said before, I don't have any DMMs that can match the 2 400 000 counts, so I can only check the stability with my 100 000 counts Fluke 45 and my 210 000 counts Thurlby Thandar 1906 "computing multimeter" (thats TTi, is it not?). They all agree pretty much, but sometimes one of those three show a few LSB counts different. Hard to know witch is nearest the "reality" or most stable over time and temperature. But my faith in DM3061 is restored!

[alm]

The SO-8 IC with the sanded markings seems like a good candidate for a voltage reference. A quick search for each of the parts should give you the answer.

[MBY]

Yes, I guess you are right. Problem is, off course, how do you search for sanded parts? I tried to look with good light from different from angles, but I could not read any identifications.

Edit: I just realized I actually did a review of the DM3061 when I first bought mine (2010-04). Unfortunately in Swedish, but some forum member maybe can read it: http://elektronikforumet.com/forum/viewtopic.php?f=9&t=41945
(The link is to a Swedish electronics forum)