6.5digit multimeters: Siglent SDM3065X vs fluke 8845A

[Elmar]

So I'm about to buy a 6.5 digit multimeter.

I'm planning to use it for measuring highly efficient DC-DC converters in the following range:
Vin:40-120V
Vout:140V
Iin:0.5-7A
Iout:0.14-6A

Preferably I'd get a multimeter with an integrated multiplexer so i can measure all 4 values at the same time, but it seems like that doesn't exist in my ranges and I'd have to make an external multiplexer myself, I'd probably make a setup with 4 relais, 2 shunts and a nucleo board to control the relays using a comupter. (I don't have the budget for buying 4 bench multimeters btw) The external shunts might be inaccurate but this is ok becasue i can callibrate them against each other and find the relative error in resistance between both of them making my efficiency measurements accurate even though i have inaccurate specced shunts. In order to control the multiplexer and save the measurements I'd probably write a labview program.

I'd like to be able to have a resolution of 0.01%, and an accuracy of around 0.10% on the  efficiency measurements, so desiring around 0.0025% resolution / 0.025% accuracy on each 4 of the individual measurements. So it seems reasonable to go for a 6.5 digit multimeter.

I'm now considering the following options (but open for other suggestions):

Siglent SDM3065X
fluke 8845A

And I'm having a really hard time deciding.

Seems like I'd have to use the 1000V range on the fluke for the voltage measurements, whereas the siglent has a 220V range giving an extra digit there.

for the current measurement I probably end up using a 10mOhm shunt and the 100mV range on the fluke, or a 20mOhm shunt and the 200mV range with the siglent.

So I went through the effort of calculating the worst case measurement errors on the overall efficiency measurement, taking into account the 4 measurements. I used the specified errors for 1 year after callibration and being within +/-5 degree from the calibration temperature.  i made this google sheet

Than i got this:

Iin	0.5A	1A	2A	3A	5A	7A
max error siglent	0.159%	0.090%	0.056%	0.044%	0.035%	0.031%
max error fluke	0.247%	0.142%	0.090%	0.072%	0.058%	0.052%

So the siglent seems to be slightly more accurate on paper, the inaccuracy at lower input voltages / currents seems to be slightly worse than what i desired, but i can deal with this, and even fix it by taking a larger shunt for those measurements.

I'm able to get the fluke for EUR1075 , the siglent for EUR820 including VAT.

So on paper the siglent seems more accurate, it has a nicer display and than it seems to be slightly cheaper.

But than pretty much everyone is telling me that they would probably go for the fluke because they are known to make really good multimeters.

So now I'm trying to see if anyone else here could give me some thoughts on this, or maybe there is someone that has experience using both multimeters and can share some of the experiences.

Thanks a lot

[Mr. Scram]

I'd go for the Fluke. You know they are realistic or even conservative with their specifications and have a proven track record. You know support is decent too, which has been debated in the case of Siglent. The resale value of the Fluke is also better, bringing the TCO closer.

[Elmar]

Thanks for the advice! I'm indeed feeling like even though the specs of the siglent seem better on paper this might not actually be the case in practice. I'm also scared because i have no good way too calibrate / check them.

In the meantime i also found 2 good second hand options:

Keithley 2000   €475   2nd hands not callibrated
rohde schwarz HM8112-3   €675   2nd hands callibrated early 2018

i got excited about the rohde and schwarz until i read a review where it was metioned someone had it drift out of specs within a year, the fact that it's not produced anymore also got my suspicous, but on paper it seems 2x more accurate than the siglent, i added it to the spreadsheet.

[Performa01]

I don't have either of the two instruments, even though I've thought about buying a Fluke 8846A for quite a while - but ended up getting something different in the end.

I do have experience with the older Fluke bench meters like 8840/8842A and even though I think the additional measurement ranges on the latter are rather useless because short-term stability and noise aren't that great, their long-term stability (hence also accuracy) is much better than specified. Like so many reputable brands, Fluke today isn't what it used to be in my book, so I'm not sure if my experience with earlier products would still hold true with their recent offerings.

I've only played a little with some early Siglent SDM3065X and my impression was good, but that cannot replace a proper test, not to mention practical long-term experience.

Yet I think you should know that there is a rather nice scanner option for the SDM3065, you just need to order it together with the meter because it has to be fitted in the factory:

https://www.siglenteu.com/accessory/sc1016/

[tszaboo]

There is also Agilent 34970A, which is a 6.5 digit DMM with switch cards to measure more values. I would throw in that into the comparison. Sometimes it is very cheap on ebay, since it is "not a DMM".

[JohnG]

There is also Agilent 34970A, which is a 6.5 digit DMM with switch cards to measure more values. I would throw in that into the comparison. Sometimes it is very cheap on ebay, since it is "not a DMM".

I highly recommend this option. You will also want to get a 34901A card with 20 channels + 2 1A channels. The 20 channels can be used up to 300V, and also for temperature measurements with thermocouples, RTDs, thermistors, etc. You can also get an old free version of the Agilent benchlink data logger software, which will allow you a good bit of flexibility. Temperature measurements should not be overlooked - they are useful information when measuring efficiency. These systems are tried and true.

In the US, you can get this combo for ~$1200 refurbished: http://www.bellnw.com/manufacturer/Agilent/34970A.htm

John

[PA4TIM]

Calibrated is only useful if you have the full history and cal-rapports. I would not buy a used one unless you have the recent history.

I have a Prema t7,5 digit meter, that was very stable until a few months ago in the 10V range. It is still very close, actual even better as it was, but this is a very bad sign.
If I calibrate it (and adjust it software-matig) You will think you have a great meter. But without a history and before/after rapport you know nothing about the problem it has.
I bought my pride and joy, a Keithley 2000 about 8 years ago new. Straight from Keitley and fresh calibrated.  I only calibrate it a few times a year (calibrating is only comparing it and noting the measured values compared to a standard) I never dit any form of adjusting. Is it 100% accurate ? No, but I know the small deviations and how it behaves and it is stable as a rock.

Keithley comes with a new 6,5 digit meter. If all goes as promised I get one for a review/test (on loan basis)  If I like it I will probably buy one. The 2000 is my #1 everyday meter.

you can not use the filtering if you multiplex.

Fluke make very good and stable meters. I like the 8846A spec/function wise. It was my second choice when I bought the 2000.
I have a Siglent scopemeter. The firmware is a bit buggy, it was well within specs, but I only use the scope function (it triggers like a dream). I would not buy a 6,5 digit from them.  Maybe they turn out to be good on the long run, maybe not. Time will tell.

[nctnico]

I'm wondering why the Keysight 34461A is missing from the list. It is much more modern compared to the Fluke (which I also looked at when looking for a 6.5 digit meter). If you automate the setup you can probably measure both voltage and current by switching between the voltage and current ranges on one meter.

I wouldn't worry so much about absolute calibration anyway. If you compare the meters you are using in the setup to eachother then you can get relatively high accuracy without needing very precise meters becaue the result you are after is relative anyway. Given your requirements you can probably get away with 4.5 digit bench meters as well.

[tautech]

I have a Siglent scopemeter. The firmware is a bit buggy, it was well within specs, but I only use the scope function (it triggers like a dream).

Yes and they're an old design now but we know Siglent are working on new firmware for them.
How much they'll fix/change is unsure.....maybe a new GUI which SHS models really need.

I would not buy a 6,5 digit from them.  Maybe they turn out to be good on the long run, maybe not. Time will tell.

Until you've used one you really wouldn't know as they're a modern bit of kit. IIRC LM399 ref and as the OP has identified well spec'ed and accurate.
Add the SC1016 option and 12 multi-purpose + 4 current channels become available.
https://siglenteu-qwavztc8hvq2w.stackpathdns.com/wp-content/uploads/2017/12/SC1016_Datasheet_DS60030-E02A.pdf


As Nico has mentioned a lower res DMM could be a valid option for the OP and at a subsequent lower price.
The SDM3055 (5.5 digit) can be optioned with SC1016 too.

[TheSteve]

Depending on the price in your country you might want to consider the new Keithley DMM6500 - it looks like it could be the value leader in top brand 6.5 digit meters.

[nctnico]

By the way: given the voltages and currents I would be careful using relay cards. The relays are likely not rated for the currents at those voltages. General purpose relays are very crappy at dealing with large DC currents at several tens of Volts. That means using external shunts or dedicated meters.

[tautech]

By the way: given the voltages and currents I would be careful using relay cards. The relays are likely not rated for the currents at those voltages. General purpose relays are very crappy at dealing with large DC currents at several tens of Volts. That means using external shunts or dedicated meters.

Exactly.
The OP stated:

So the siglent seems to be slightly more accurate on paper, the inaccuracy at lower input voltages / currents seems to be slightly worse than what i desired, but i can deal with this, and even fix it by taking a larger shunt for those measurements.

[Elmar]

By the way: given the voltages and currents I would be careful using relay cards. The relays are likely not rated for the currents at those voltages. General purpose relays are very crappy at dealing with large DC currents at several tens of Volts. That means using external shunts or dedicated meters.

yeah i was planning on using external shunts and switching the voltage measured across, so the relays would'nt switch any current.

Depending on the price in your country you might want to consider the new Keithley DMM6500 - it looks like it could be the value leader in top brand 6.5 digit meters.

will consider

There is also Agilent 34970A, which is a 6.5 digit DMM with switch cards to measure more values. I would throw in that into the comparison. Sometimes it is very cheap on ebay, since it is "not a DMM".

Agilent 34970 seems like a nice option now as it would save me the trouble of making a multiplexer as it has this internally, i also like that is can do thermocouples. I managed to find a 2nd hand set locally for 1146 EUR including VAT, a new "6.5dig Siglent SDM3065X with 12-channel multiplexer" comes at EUR1134, not sure what i would prefer between those two. 

also, on ebay i see alot of way cheaper adds for a Agilent 34970 but they all say "not tested, no warranty, sold as is" so that scares me a little.

Yet I think you should know that there is a rather nice scanner option for the SDM3065, you just need to order it together with the meter because it has to be fitted in the factory:

https://www.siglenteu.com/accessory/sc1016/

The siglent switchcard specs confuse me as it's both saying:
maximum DC voltage 110V
maximum switching voltage 220VDC

I would need that 220VDC rating but now i'm not sure if that's ok

I wouldn't worry so much about absolute calibration anyway. If you compare the meters you are using in the setup to eachother then you can get relatively high accuracy without needing very precise meters becaue the result you are after is relative anyway. Given your requirements you can probably get away with 4.5 digit bench meters as well.

this is good feedback, I'm starting too get less worried about accuracy as I'm indeed only interested in a relative measurement.

[PA4TIM]

tautech: Yes and they're an old design now but we know Siglent are working on new firmware for them.
How much they'll fix/change is unsure.....maybe a new GUI which SHS models really need.

I hope so, I really like it and use it almost as much as my 350 MHz HMO-3522 Hameg(R&S), it is nice build and  deserves  better firmware. Now sometimes weird things happen and the whole GUI is not finished and not very consistent. The DMM part is, how do I say it friendly    , better as nothing.

The fact that they release a scope that is over 1K euro with  firmware that looks like a beta version,  a soldered glass fuse for the 10A range and a continuity function with a reaction speed af a geriatric turtle is not a goods sign if you want to sell more high end stuff.

[tautech]

Yet I think you should know that there is a rather nice scanner option for the SDM3065, you just need to order it together with the meter because it has to be fitted in the factory:

https://www.siglenteu.com/accessory/sc1016/

The siglent switchcard specs confuse me as it's both saying:
maximum DC voltage 110V
maximum switching voltage 220VDC

I would need that 220VDC rating but now i'm not sure if that's ok

You're being wisely cautious. 
For SC1016 the physical clearances are not great and the relays quite small which I strongly suspect are governing the max ratings.
AFAIK you need to use the SC1016 or the DMM for measurements and can't use both together where you could instead use the DMM for the voltage measurement and SC1016 for the other measurements. 
I need to check this ^ in my demo unit to make sure.

However there still is the SDM3k Dual measurement functionality you could use.

[JohnG]

A couple more things on the Agilent 34970A:
1. You can use it as a switch unit only without the DMM. In fact, you can get it with Opt. 001, which means there is no internal DMM installed (it can be added later).
2. My understanding is that the internal DMM is based on the 34401A, for what it is worth.

Finally, some comments on efficiency measurements:
If your expected efficiencies are less than 80% or so, the accuracy requirements are not so critical. But, if you are measuring in the mid- to upper-90% range, you want to be very careful. Depending on relative measurements is a tricky business that can really come back to bite you. Current measurements are especially problematic. Shunts can change with temperature, and cheap shunts can drift a lot. You can not assume they will drift the same way, or be at the same temperature, etc., and shunt calibration is expensive. That being said, external shunts are often more accurate than internal shunts, especially as you get close to 5-10 A and above.

John

[Elmar]

I'm planning to measure efficiencies around 98.0 - 99.6% so things will defenitly get tricky.

I was planning to callibrate the shunts against each other as I'm interested in a relative measurement between in and output power of the converter.  That being said, a temperature difference between the two external shunts that is introduced after the callibration might still introduce an error. I can add a thermo couple to the shunts to measure the relative temperature difference between the two shunts, in combination with the temperature coefficient of the shunts this could give me some more accuracy there, or i can thermally connect them so temperature of both shunts will stay the same.

I'm trying to see if i prefer either a second hand "Agilent 34970A +34901A card with 20 channels + 2 1A channels" or a new "6.5dig Siglent SDM3065X with 12-channel multiplexer"  , similarly priced.  I'm starting too get less worries about the actual absolute accuracy of the meter as I'm interested in relative measurements. Some kind of multiplexer would really be a big plus as this saves me the development of my own multiplexder, this would be worth the extra $250 to me or so as it saves me a lot of work.

[nctnico]

You can use oversized current shunts and mount both on a large aluminium heatsink.
For example these in a SOT 227 Package. They are not accurate in an absolute sense (1%) but the tempco is 60ppm.

[Elmar]

You can use oversized current shunts and mount both on a large aluminium heatsink.
For example these in a SOT 227 Package. They are not accurate in an absolute sense (1%) but the tempco is 60ppm.

this is great advice, I will buy that exact model (TGHGCR0100DE   ,TGHGCR0200DE) , so the shunt issue should be solved now

still not sure what multiemter to get, really tempted to get something with an integrated multiplexer as is makes life easier, although i really don't mind making a little pcb with 4 relays either and multiplex myself

6.5dig Siglent SDM3065X with 12-channel multiplexer     ---> I'm worried about the max voltage ratings of the switch card and siglent is a new player on the market so no real trust in performanc/reliabilitye over time

Agilent 34970A +34901A card with 20 channels + 2 1A channels   ---> It's kinda dissapointing that it doesn't have front connection / a nice interface so i could also conviniently use it as a DMM and i would have to buy it second hand which is ok

[rsjsouza]

(...)
Agilent 34970A +34901A card with 20 channels + 2 1A channels   ---> It's kinda dissapointing that it doesn't have front connection / a nice interface so i could also conviniently use it as a DMM and i would have to buy it second hand which is ok

Did you look at the Keithley 2700?

[Elmar]

that unit looks great I'm gonna get a quote on that one


seems like 960EUR including VAT and multiplexer card for the 2700 second hand, comes pretty close to the other options

[tautech]

AFAIK you need to use the SC1016 or the DMM for measurements and can't use both together where you could instead use the DMM for the voltage measurement and SC1016 for the other measurements. 
I need to check this ^ in my demo unit to make sure.

Checked.
As above, one must either the DMM in normal mode and with a Dual measurement setup or the scanner card for which the ratings are not as good as the DMM inputs. In scanner mode the channel step rate is ~1Hz which is also the update rate of the scanned channel list and the measured value is displayed on the main display and added to the measurement table.
Very similar to this graphic when using the Hold measurement function.

[JanJansen]

I would not go for any siglent, on paper says nothing.

[tautech]

I would not go for any siglent, on paper says nothing.

Oh yeah, right ? 
So you haven't seen this thread:
https://www.eevblog.com/forum/testgear/siglent-new-bench-dmm-sdm3055/

Or don't know they come with a Cal cert guaranteeing Datasheet accuracy ?
Or know about their 3 year warranty ?

And can't find any recent videos about them ? 

Or are you just anti-Chinese equipment ? 

[nctnico]

If you buy a Keysight or a Fluke 6.5 digit DMM you know you buy from a manufacturer which has decades of experience making this kind of high accuracy equipment and you get good service on top of that. The price difference also isn't that big so why settle for a B-brand anyway?