Siglent Bench DMMs SDM3065X, SDM3055 and SDM3045X

[oz2cpu]

remember i am not talking about what is visible on the display,
if that was the case, there would not be a reason to play with any of this :-)
it is important to repeat that statement..

see this : leak vs temperature ADG1408 datasheet snip inserted,
lets say the pcb near this part changes temperature from 30 to 32, just a very littly during the day,
and the source resistance is 100k
and in the 20V range, full scale is 200mV differential
now we see where the details i talk about, is comming from,
it is just under last digit visible :-) but i like it to be modified to be much better.. that is what i play with

[oz2cpu]

this night latest schematic update :-)

[Kleinstein]

R4 (100K) would not effect the 20 /200 V range, but only the 1000 V range. The 20 and 200 V range still see near 100 K of source impedance, but there is not much that can be done about this as this comes from the divider.

The ADG1408 is a odd choice in that circuit. It is rather low resistance (4 ohms) and thus higher leakage than the more logical choice ADG1208 or DG508.
I had not considered such a high leakage part - in this case the leakage would indeed be a possibly critical point.

[oz2cpu]

thanks a lot Kleinstein, you are right, i was wrong :-)  R4 is 1kv range

I updated the schematic, and also added a note about what lines are 20/200V range, and 1000V range,
now i will be less confused in the future :-)

Seems impossible to trace down the connection to U405 pin5 ??
i am trying to understand the origin of the 2.5V and 0.5V signal needed for the LO drive,
it must be via some sort of "selector" and it is not exactly 2.5000 as the ref is, but it does not matter what the LO drive it,
since it is also measured, in differential, see right side of U406,

U405 is the LO driver, both opamps are in parallel via two 49R9 resistors,

U400 ?? what is that doing ? no clue yet :-)

[Kleinstein]

The low half of U400 (oins 1,2,3) looks like a low pass filter for the input signal that can be switched on/off. This filter may also be part of the reason to have the 100 K for R4, as this resistor would be part of the filter. It is still odd to have a AZ OP for this purpose. A more normal JFET based OP would have been the more logical choice.

The upper half of U400 is still unclear.  If R32 goes to some signal source like a shunt, this could be some analog amplifiaction.

[oz2cpu]

all 2.5V LO noted nets, are confirmed to be hardwired to the LO terminal

but the lead on U400 as selecable filter might be a good clue,
this meter do have fast / slow setting, all i need to do it is change settings, while monitoring the A0 A1 A2 signals on ADG633
EDIT : A0 is HI, when Filter is on, and A0 is LO when filter is Off

I did not plan to write the entire schematic, 2W and 4W ohm, is not of big interest right now (for me at least)
if any one else like to join and draw that part, feel free to say so :-)
ALSO there is a TL072 performing the current resistor amplification,
just under that, a bridge rectifier, funny construction, i assume it is so the DC current always go the same way ?
or is it only for overload protection ? i did not trace that part, it would require the pcb to be removed,
that was also not my plan.

[oz2cpu]

my brand new replacement SDM3055 showed up a few days ago,
I did power it on for 2 days, before starting to test it,
but of course it is still in it very erly days of burn in,
each horizontal line in the chart = one last digit on the screen (more fair way to setup the chart)
so as expected, the noise is way under, what is visible on the screen readout,
so far i am super happy, let this test run a few days.

[Kleinstein]

The big bridge rectifier for the amps part should be for the protection (as 2 pairs of anti parallel diodes), to clamp the voltage over the shunts.
The TL072 would no be good enough for the main amplification of the DC signal, it may still be used for an TIA for the very low currents and for an auxiliary buffer to reduce the leakage from the rectifier (may no be used with a 5.5 digit meter, especially if the low currents would be with a TIA instead of a simple shunt). From the firmware theread I remember that there may very well be a TIA used for the lowest currents. The datasheet woth low burden for the lowest ranges also supports this.  I still dont' see an extra relay (or MOSFET) for the range switching.

The 2 V signal from AIN4  might very well to the ohms current source circuit (a + input at U501).

Looking at when A2 signal of the ADG633 changes would give a hint for what the upper half of U400 is used.

[oz2cpu]

correct A2 is DCI 200uA and 2mA range
i have updated the schematic this ADG633 selector bits

[oz2cpu]

bottom side of the board picture.
correct AC pins and DC pins of the bridge rectifier are connected together,
so it will protect the two sense resistors 1Ohm and 10mOhm
if your circuit draw high inrush

[oz2cpu]

new top pcb picture,
I will try to make a few more pictures, but this system will not accept large pictures

[oz2cpu]

new technical challenge : i took a lot of super good pictures of SDM3055 PCB,
using a digital microscope, with a lot of overlap,
now is the challenge to stitch them together into one total picture, ALL programs i found and tried, dont work,
anyone of you found a program that can work in flat mode ?
the total mosaik is about 200 pictures, each is 4032 x 3024,
solving this challenge will let us create amazing details of pcb pictures.

link to 3 x 3 test pictures, so you can test your stitch program is working
webx.dk/siglent

[Qw3rtzuiop]

anyone of you found a program that can work in flat mode ?

Lightroom works fairly well.

[oz2cpu]

great thanks, this looks amazing, all done automatic ?
ultra easy ?
can a noob do it ?
maybe you like to help ?
i could give you access to all files, 600MB :-)

[mawyatt]

PS choked on our larger stitching attempts. Ended up using PTGui for our chip imaging stitching, with Zerene to stack the images, and PTGui to stitch them together. One final chip image rendering was a massive ~30,000 by 20,000 pixels from 19 individual tiles, each tile composed from 200~400 individual 36MP images (total of ~200GP) from a complex custom semiautomated setup, custom lens assembly and Nikon D850.

This is professional level panorama stitching software and takes a little time to learn but very powerful. PTGui uses a unique method which employs selectable "points" which are blended and if done properly almost undetectable stitch, even zoomed at the pixel level.

Best,

[Qw3rtzuiop]

great thanks, this looks amazing, all done automatic ?
ultra easy ?
can a noob do it ?
maybe you like to help ?
i could give you access to all files, 600MB :-)

Yes very easy. Didnt change any settings. All automatic. I just reduced the jpg quality to get a file size that im allowed to upload here.
Send me a pm with a file link and i can try it with the 600mb

[oz2cpu]

files away, ha ha i cant wait to see the final result :-)
191 files = 725MBytes
ok halve of one picture is also the other halve on the two next pictures, and so on,
so in reality the final will be halve on each side = quater the size of 725MB is 181MB, in jpeg, darn it.. crasy..
lets see we can maybe size reduce ?
here is picture number 1002,
original 100% size is 4032 x 3024 pixels,
to show the size of material, and if size down is possible, see the 4 attached micro snips of the same picture,
showing 100%, 75% 50% and 30% scaled

[mawyatt]

Watch out, you are treading very close to the chip imaging rabbit hole

We fell in way back in ~2000, and were just able to get back out (truth is, we haven't designed a new chip since retiring)!!

Best,

[oz2cpu]

From the Spec sheet:
Continuity test specifications for the SDM3045X, SDM3055, and SDM3065X
Test Current Source: 1mA
Max Resolution: 0.01Ω
Input Protection: 1000V Input Protection
Open-circuit Voltage: <8V
Beep Threshold (short-circuit resistance): from 0 Ω to 2000 Ω

i like the feature beeper on/off and then 3 levels of beep,
ok so far so good, but what is the idea about the relay ??
every time you short the wires, a relay go click
this sounds a bit odd ?

[Kleinstein]

The DMM has the problem, that voltage higher than some 2.5 V can not be measured directly and need the relay to connect the divider. For the intermediate values resistos it can get more accurate when the divider is not connected. So it may be the relay disconnecting the divider.
I would prefer the divider always on in continuity mode / diode mode. Those modes don't really need linearity and high accuracy.

This leaves the highest ohms mode (e.g. > 20 M) case that may use the 10 M divider in parallel. This point could be common to many meters that use this trick to extrend the measurement range and use a relay to connect the divider. My preferred solution would be an SW option to exculde the highest range from auto-ranging.

[oz2cpu]

that makes no sense :-) why they made it this way then ?
if you select the Continuity test mode,
why would the user be interested in auto range behind it ?
Continuity test is only used for LOW values, right ?
the spec say from 0-2k
that means anything over 2k is always considered an open..

[ComradeXavier]

Is this a model difference? Or maybe a firmware version difference? On my SDM3065X with current (I think) firmware, there is no relay action in continuity mode. Open->short->open is silent (with the beep turned off).

[Kleinstein]

The input stage of teh SDM3065 is quite a bit different and also a bit unusual: it looks like the circuit uses a +-15 V supply and they still have 20 V range with high impedance.  Chances are high the move the low side relative to the supply, opposite to the high side. In addition the configuration is so that that there is buffer and than a divider before the ADC. With a high Z 20 V range there is no need for relay swiching in ohms mode or at 2 V.

Ideally the 3055 should also get away withput range / relay switching: either limit the voltage to 2 V or keep the 10 M in parallel and accept a slightly higher error - both would be better than relays switching.
One could be lucky and the auto range setting (maybe from ohms or voltage mode) could still effect the behavior.

[oz2cpu]

OMG.. now i turned on my SDM3055, and pressed the continuity test,
tried all possible things,
NO RELAY CLICKS !!
problem is gone :-) now a challenge : if you own an SDM3055, redo this and see if you get relay clicks when you short the wires,
together with the beep if enabled.
i was not drunk yesterday, you got to believe me..

[2N3055]

OMG.. now i turned on my SDM3055, and pressed the continuity test,
tried all possible things,
NO RELAY CLICKS !!
problem is gone :-) now a challenge : if you own an SDM3055, redo this and see if you get relay clicks when you short the wires,
together with the beep if enabled.
i was not drunk yesterday, you got to believe me..

It is late for that, we know the truth... You already posted the photo ....