OWON XDM1041 the unknown multimeter...

[rsjsouza]

This is not uncommon. These DMMs usually have several OEMs and Amazon sellers sometimes test the waters with unknown brand names - I got a Sinerky UT136C (Uni-T UT136C), a no-name DM139C (Uni-T UT139C) and a single batch production Surpeer AV4 (HY-19 IIRC) - all at a significant price reduction at the time.

[Kleinstein]

That meter may show up under different names in the future. The OWON brand is nothing to be really proud off like a Fluke, HP or similar name.
It looks like the meter does not show the brand name and so the seller did not add it, which is OK. If the seller would add the brand one could also get suspicious.

The internals looks reasonable, with plenty of creapage distance as it should be. One point I don't like that much is that they use a SMPS and not a classic mains transformer. This may lead to extra common mode injected signal. The max. power at 15 W looks quite high. Even with just 15 VA and a relatively poor power factor it look high to me.

It is nice that they have the RS232 part separate (looks like a max232 or similar on the small board). So there may be a easy way to change to the more logical direct USB.
Chances are the RS232 version is some 20 cents cheaper.

[Doctorandus_P]

First:
Tnx for posting pictures of the internals.

I found this thread after first seeing it announced (from 2021-05-13) in a local shop:
 https://www.eleshop.nl/owon-xdm1041-multimeter.html

and then I started looking for more info, and found this thread.

I have been looking for a decent Benchtop DMM (Big display, fits on shelf, PC connectivity Push buttons instead of rotary selector etc.) for quite some time.
I do not have room for a 30cm deep meter though, and that is the reason I did not buy a Siglent 3045. The Siglent is also too slow for me in autoranging and connectivity test.

Somewhat slow autoranging is accepable for a low power battery operated instrument, but it's inexcusable for a benchtop instrument.

After I found out this uses the HY3131 I've been looking into that chipset a bit and it's quite popular for handheld meters:
EEVblog 121GW. (EUR200 meter)
Keysight 1280. ( EUR 600 meter)
HoldPeak HP770D ( EUR 50 meter)
Digilent DMM shield (EUR90) (No display, but Isolated interface)

Eleshop announces it (localy in EU) for EUR 130 and this is definitely in my price range, and I like the form factor, push buttons and big TFT.
I'm curious how far this HY3131 chipset can be pushed. Ideally it would be programmed to do some auto-sensing, and switches to a quick mode for auto-ranging, and to an accurate mode for more digits.

There is a EEVblog thread about reverse-engineering and alternate firmware for a Fnirsi scope. I do not understand that effort. After looking into it the frontend of the fnirsi scope is simply not good enough to make something decent out of it (Quite a shame, it could be a very decent thing for a BOM of EUR10 more...)

This XDM1041 looks extremely hackable though. The HY3131 complete with all measuring stuff and isolated interface is on a separate PCB from the "brains" part.
The whole microcontroller board can be easily replaced by your favourite, and then you can write any software you want for it, and if you stay with STM32 then once firmware development has surpassed what Owon does you can flash it into the on-board GD32F303.

I do not understand why Dave is complaining about why a benchtop DMM does not run on batteries. It seems a bit silly to me.
But if you must...
I find the Glue on SMPS module hilarious. It is also very easy to replace it with any power module you would like, including a battery pack, and I guess there is plenty of room left in the compact housing to add some decent batteries.

Another hackable upgrade for this scope would be to add a uSD or dataflash chip for datalogging. Should not be too difficult.

I do not give much for CAT ratings, I thing they're much overrated (for home or benchtop use). I'd rather have 50ct fuses then EUR10 fuses!
This XDM1041 has some decent sized fuses, widely spaced components and even some well placed isolation slots.

Overall, I'm not sure which path to follow.
Almost all decent benchtop meters are simply too big for me. I do not like the rotary buttons of the (cheaper) benchtop UNI-T meters, nor the simplistic LCD of the more expensive UNI-T, and others with fluorescent displays are also not for this generation anymore. Simply can't compete with a TFT.

I am looking forward to an in depth review of this meter. Especially auto ranging speed and the speed (and quality) of the continuity tester.
Another point that is a real deal breaker for me is a peep function that warns if the probe is in a current banana bus, while voltage measurement is selected. That is just a too simple mistake to make, and it either destroys your meter, or it destroys both a EUR10 fuse and your probe tips. Just happened too many times for me.

[SpecialK]

A couple days ago the Amazon link I posted went down to $148 CAD minus a $15 coupon.  I was very tempted even though I don't need it.

Now it's $218 minus $15.  To be fair, the Siglent 4.5 digit is like $500 and the 5.5 digit over $600.  There's very little to choose from in bench DMMs at hobbyist prices.  I think this should be a popular meter.

[coromonadalix]

Seeing other meters based on the hy3131  surely getting faster auto ranging should / could be possible ?

For a battery pack,  you have the battery bank for cell phones who could do a very good job, i have one based on 18650 cells  that could be called a 25000 ma  model loll

I think it could snug in there or behind it ??


Yeah the smps psu  was fun to see in there

[martinr33]

The original photos of the innards (on the brochure) show a power supply integrated with the PCB.

So we are looking at some kind of fix here. Esecially with the corner rounding and cutoff on the board - that's a reworked board from another design.


https://photos.app.goo.gl/oj4yj544yRkzSndb6

[rsjsouza]

Wow, that's an ugly fix! It seems rushed and desperate to put product out.

[theHWcave]

I just got mine thru Ebay, straight from China to the UK, took about 2 weeks although the seller said 4 weeks.
Interestingly, my power supply is exactly like in the brochure, no ugly fix
but they cheated on the fuses, instead of 2 decent 30 mm long high rupture ceramic as in the brochure, I got cheap 20mm long ones rated for 250V only, at least the 10A is ceramic.

Other than the slowness already mentioned, the meter is quite ok for what it is. After a weekend of intense testing my summary is:
the negatives:

• Capacitance above 500uF, auto range is not always working and one has to do manual ranging... capacitance is definitely the weakest mode anyway, the lowest range is just 50nF and 500pF or less shows only zero
• Ranging in volt mode is weird: it can auto range from the millivolt up to volts but not down from volts to millivolts. Same with manual ranging. To go from volts to millivolts or back you can't use the range up down, instead you have to select between volt and mV mode. You can get used to i, but interestingly this is just a problem of the user interface. The behaviour over SCPI is much more sensible, you can just go up and down using the RANGE command or use the CONF command without any issue.
• Rel mode is actually a maths function which means you can't have rel active and do averaging or dB/dBm at the same time since only one function can be active at a time 
• No smart data hold which would have been easy in software..., just a silly run/stop button which is like a dumb hold 

on the plus side:

• I really like the temperature mode. Very comprehensive and can do PT100 or K-type
• The SCPI interface (RS-232) has its quirks, syntax wise, but once figured out, it works well and can deliver more decimals than shown on the display. Using that the accuracy is pretty good.
• The display is easy on the eyes, especially if those eyes are not as young anymore...
• It fits exactly under the first tier of the shelf at the back of my bench. That space is very shallow and very low, which is why I got this meter for it. It does not take any extra space on the work area, its mains powered and can be on the whole day, and its display is exactly at the right angle .

And if you have not guessed it... the meter is always on as long as it's plugged into mains.The power switch on the front is only a soft-switch, but I measured the "standby" current to be only about 1mA at 230V, so its not too much of an issue.

[coromonadalix]

would be nice if Owon could read this thread and improve the meter in a fw update ??

[msquared]

It's finally official.

https://www.owon.com.hk/products_owon_4_1%7C2_digits_xdm1041_bench-type_digital_multimeter

Also looks like there's already a newer version with a USB data port instead of RS232.

So much for being an early adopter. Maybe time for a little internal USB mod...

[theHWcave]

I have this meter and I would strongly advise caution in trusting anything that's written on that Owon Hongkong page. Notice how they dither between USB or RS-232? They avoid making a clear statement. Of course nobody these days has RS-232 on a PC/laptop so we all use USB to RS-232 converters to talk to the meter using USB. And he presto, now your marketing people claim you "can"  talk to the meter over USB! Of course you do 

Could you easily add USB, yes and no. No, if you mean a fully capable USB interface that is supported by the processor to for example mount and use USB sticks that you may plug in. The reason is that there are only 3 signal wires (tx,rx,gnd) that go from the digital board to the interface board and these are already RS-232 on TTL level. All the interface board does is convert these to RS-232 levels. It would need a relative large revision to change the hardware and firmware to do proper USB. Maybe they will do it eventually but I would not hold my breath.   But if you want to replace the DB9 with a USB socket, nothing stops you from removing the level converter chip and replacing it with one of the standard USB-to-serial chips (usually some chip from Prolific or clone) that are normally embedded in the ubiquitous adaptor cables. That gives you a physical USB port for the marketing people but that "USB" would not give you any more capabilities than what you get now with a cable converter, i.e.  a simple COMx or ttyUSBx interface.

The greatest humbug IMHO is the record function. Absolutely useless! Yes the meter can record up to 1000 entries manual or automatically but there is no way to download them. I tried all kinds of methods. And don't be fooled by the Owon PC software. In desperation, I finally gave in and went through the pain of installing it. I also rigged up a complex setup with my scope to sniff and protocol-analyse the exchanges between the OWON PC software and the XDM1041 to finally learn the secret how they access and download the recorded data over the serial interface. Turns out they don't. What they do instead is make a new recording (reading new measurements via SCPI and storing them in a text file on the PC). I think there is simply no way accessing remotely whatever you may have already recorded on the meter. Even more expensive OWON meters have nothing of that sort in their SCPI implementation. I think in those meters you can dump the recording on a USB-stick but that would require a "proper" USB interface and that (see above...) 

[rdl]

What is the point of the big "Trigger" annunciation on the display? The little square next to it seems to blink in sync with display updates, but that seems pretty useless to me.

[theHWcave]

Yes, in the XDM1041 its only purpose is to show that the display is updated ... There is no setting or anything to influence trigger apart from the "run/stop" button but that is a simple hold (no auto-hold or anything fancy)

[coromonadalix]

Would it be better to call it  something like a "gate",  or "measurement in progress"   which could take too much space on the screen ...

[davebb]

Yes, in the XDM1041 its only purpose is to show that the display is updated ... There is no setting or anything to influence trigger apart from the "run/stop" button but that is a simple hold (no auto-hold or anything fancy)

Hi Great Youtube videos
it was me with the temp problem with my meter, as i said when the meter is cold the temperature Reading is very good
after about 10-20mins of the meter being on the reading reads 4-5d/c higher, what does your meter read with the k type
disconnected ?
can you please tell me is there a temp reference in the meter and what does it use to do this, or is the reference in side the meter HY3131 chip, i have re soldered the chip but this did not help,
also i took a photo of the ch 340 usb board that is fitted to mine, you can use it in your vids if you want to,
Thanks for your help ,
Dave 2E0DMB

[Kleinstein]

The type K sensor is a thermocouple and it thus makes no realy sense to look at the reading, when open circuit. This would be effekted by the input bias and the relatively high input impedance (likely 10 M or even higher).
It make sense to look at the reading with shorted inputs. This should read the internal temperature used for the cold junction. Dave has a relatively recent video on this.

[davebb]

Hi Thanks for the info, is the internal temperature used for the cold junction in the HY3131 chip or is it external to the chip,
Thanks i am trying to find out what is going on as i do not rely want to send it back to china as the meter is working well apart from this problem ,
Thanks
Dave

[Kleinstein]

Connecting a type K thermocouple just to the banana plugs is always tricky, as the banana plugs are usually copper / brass or similar. So the cold junction would be somewhere out at the connector. So it really depends on how the connections are made and it could be just a poor probe quality. The test with just a copper short would be more reliable than with a real probe. Some banana cable are also not that great and could cause a few µV of offset if one connector gets warm.

Chances are there is some sensor near the COM terminal (due to isolation requirements it would be hard to measure at the other connector).
One could test this by using a short with a copper cable and than warm up the COM connector (e.g. touch / warm with the hand). The closer the sensor to the terminal, the faster it can react.

For an accurate reading a PT1000 or NTC temperature probe would be better. Thermocouple with just 4 mm connectors is more like a crude relative estimate with some extra uncertainty.

P.s. : The linked teardown pictures don't show an extra sensor near the connector. So likely the HY3131 internal temperature only - ideally with some offset in software.

[davebb]

Hi Thanks for all that info, i have shorted out the input and yes temp reading is good but after about the same amount of time the temp goes up high about the same amount, just like it does with a probe, There is a relay in line in the circuit, so the next thing to try is to link it out to see if it is that,
Thanks Dave

[Kleinstein]

I don't expect the relay to be the problem.  Chances are the relay is a latching type and not getting really hot.

The more likely problem is that the whole meter gets warmer with time. With a copper short the meter should show the temperature of the connector(s). So some rise above the room temperature is correct. There is still a chance that the correction for the meter getting warmer is not correct (e.g. to much, as the chip likely heats up more than the connector).
With no sensor at the connector the meter can not reliably get that temperature - the cables can have quite some effect on the temperature at the connectors. So one can not expect a really good conpensation of the cold junction.

Besides the problem inside the meter, it is also the question where the actual cold junction is. Ideally it would be on the inside of the terminals, like the TC wire going all the way to the tip if the plugs or plugs made from thermocouple equivalent material. The more likely construction is having normal cheap plugs made from brass and thus a cold junction a bit outside. This could add some error, though likely less than 50% of the cold junction variations. So I don't hink the probe could explain a similator effect as with the short.

Is the thermocoule at least reading external temperature changes about right (e.g. some hot/cold water) ?

[davebb]

Hi yes it reads fine until about 10-20 mins and it follows another meter until it has been on for this time,
Thanks Dave

[theHWcave]

@Kleinstein:  I examined the board carefully and I don't think there is another sensor. I think they rely on the HY3131.

 I recorded the displayed temperature with shortened probes when the meter is turned on after being off overnight. Over 1 hour it climbed from 18.9 to 27 degC. But not smoothly. There were very distinct steps. This could be something else interfering as I was not paying attention, so I will repeat that again. Right now it shows 26.9 with probes shortened. With a K-type probe (borrowed from a BM235) it shows 27.7 The BM869S with a pretty good (=expensive) K-type probe shows 26.9 at the same location.

I will do some  more testing tomorrow

[davebb]

Hi Thanks
if you could run some more tests that would be good,
Thanks Dave

[wizard69]

Thermocouple could junction compensation needs to happen as close to the cold junctions as is possible.    Doing so on the PCB might result in a significant error.   

The first thing that should be done is to determine where the cold junction sensor is.   As has been explained that should be relatively close to the banana jacks, if not it is a design fail.

@Kleinstein:  I examined the board carefully and I don't think there is another sensor. I think they rely on the HY3131.

 I recorded the displayed temperature with shortened probes when the meter is turned on after being off overnight. Over 1 hour it climbed from 18.9 to 27 degC. But not smoothly. There were very distinct steps. This could be something else interfering as I was not paying attention, so I will repeat that again. Right now it shows 26.9 with probes shortened. With a K-type probe (borrowed from a BM235) it shows 27.7 The BM869S with a pretty good (=expensive) K-type probe shows 26.9 at the same location.

This shift is excessive and could indicate a part failure or simply the cold junction sensor being it the wrong place.   I'm thinking hard here but if the cold junction sensor was heating up then I'm thinking the displayed temperature should be going down.     Generally the offset associated with the cold junction is subtracted from the millivolt value at the measurement point.

A short, zero millivolts, represents 0ºC for a K-type thermcouple.   You would then subtract the reference junction voltage but here is the problem, you don't have another junction in an all copper connection scheme.   It would be interesting to see the behavior of the meter with a K type TC at the ice point.   This would create the second junction to be compensated at the meter terminals.

If you have a room temperature of 20 degrees a shorted input should result in an approximate negative 20 reading.    This if there is no second junction to offset the reading.  In any event my mind is a big foggy here and I worked extensively with type T TC's in the past where the copper conductor eliminated one joint requiring compensation.   My point is that reading room temperature might not make sense with a short.

I will do some  more testing tomorrow

One thing to consider is air flow.   IF this is indeed a reference junction compensation problem cooling the unit down after it heats up should result in fairly fast response on screen.

In any event I may need to think more on this.   I'm not at work at the moment so I can't set up anything to test the idea expressed above.

[wizard69]

As a follow up to the above post I remembered that my new EEVBlog meter does K type thermocouples and it indeed reads room temperature with the test lead shorted.    This is causing me a bit of confusion as it implies that they are adding the reference junction temperature to the reading.

It is getting late so I may end up sleeping on this.    As I've said it has been awhile since I was heavily involved in TC work and that was mostly with type T.  In this case though both TC are in very similar millivolt ranges at room temperatures.

in any event drift is the problem here.