Siglent Bench DMMs SDM3065X, SDM3055 and SDM3045X

[Kleinstein]

With the high input impedance it is absolutely normal that the meter is not reading zero with open inputs. It is more like a good sign that the reading is still within range. 1.4 mV is actually more like suspicious small: either a low input impedance or a lucky very low bias.

It would be an issue if the 1.4 mV are measured in the 10 M input impedance mode. This would be 140 pA of input current and this is on the high side.

One could do a simple test of the input resistance and input bias current with a capacitor (e.g. 10 nF low leakage type like PS or PP).
The capacitor is initially charges (e.g. +20 or -20 V for 2 separate runs) and the drift / decay of the voltage recorded in manual 20 V range.  The rage of the voltage change gives the input current as I = C*dU/dt. Often one would get something like a shifted exponential decay, like the classical RC discharge and from the time constant can extimate the input impedance.

[skander36]

SDM3065X open input 10M, AUTO, NO REL.

[LastManStanding]

@KungFuJosh
I recommend the following video by Dave. Here, the phenomenon is explained in detail, and some of your questions are addressed.

[KungFuJosh]

@KungFuJosh
I recommend the following video by Dave. Here, the phenomenon is explained in detail, and some of your questions are addressed.

Thanks! That was a helpful video. Inserting the leads and twisting them is certainly effective.
Leads inserted but not twisted:


Leads inserted & twisted:

SDM3065X open input 10M, AUTO, NO REL.

I guess maybe I have a little more stray AC than you do? It's a known issue here, that's why I always have filtering on the other DMMs. It's just odd to me how it works in this model. PLC < 1 shows around 4mV, but PLC ≥ 1 shows 1.4mV.


Thanks,
Josh

[KungFuJosh]

With the high input impedance it is absolutely normal that the meter is not reading zero with open inputs. It is more like a good sign that the reading is still within range. 1.4 mV is actually more like suspicious small: either a low input impedance or a lucky very low bias.

It would be an issue if the 1.4 mV are measured in the 10 M input impedance mode. This would be 140 pA of input current and this is on the high side.

It is 10M input. It's also 3 times higher than skander36's example at 10M input. Which of us should be more concerned? 10G input imp just shows OVERLOAD on the screen.

[skander36]

10G input imp just shows OVERLOAD on the screen.

This is normal.

[KungFuJosh]

10G input imp just shows OVERLOAD on the screen.

This is normal.

Yeah, I was just pointing that out since he asked if it was 10M input imp.

[KungFuJosh]

As good as my other bench DMMs are, I'm tempted to sell them both and buy a second SDM3065X. I reeeally like the stability. Am I crazy?

[Kleinstein]

Getting an a reading of 1.4 mV with the 10 M input would suggest an input bias of 140 pA. That is nearly 3 x higher than the specs, that call for < 50 pA for the bias (called offset). The test should be done at not too high a tempererature (e.g. max ~ 75 F) and with not high humidity (e.g. RH < 70% maybe) and the meter well warmed up ( e.g. > 1 hour on). High temperature or high humidity could increase the input bias. I am not so sure of the specs are for the AZ on or off mode and if this makes a difference.  I don't kown what the AZ setting means for this type of DMM.

It could make sense to record the reading with different conditions, like 1 PLC, 10 PLC, maybe 100 PLC and both AZ on and off.

A 2.5 V reading is a rather odd test point for stability. The usual stable reference are 10 V or 7 V and maybe 1.018 V. 2.5 V is also more on the low end of the 20 V range. Chances are the 2.5 V ref for the test can be the limiting factor more than the DMM.

[Martin72]

As good as my other bench DMMs are, I'm tempted to sell them both and buy a second SDM3065X. I reeeally like the stability. Am I crazy?

Then I'm crazy too.
I've twice had the opportunity to buy a used SDM3065X that was still in mint condition and twice I hesitated too long.
That won't happen to me a third time.

[KungFuJosh]

Getting an a reading of 1.4 mV with the 10 M input would suggest an input bias of 140 pA. That is nearly 3 x higher than the specs, that call for < 50 pA for the bias (called offset). The test should be done at not too high a tempererature (e.g. max ~ 75 F) and with not high humidity (e.g. RH < 70% maybe) and the meter well warmed up ( e.g. > 1 hour on). High temperature or high humidity could increase the input bias. I am not so sure of the specs are for the AZ on or off mode and if this makes a difference.  I don't kown what the AZ setting means for this type of DMM.

It could make sense to record the reading with different conditions, like 1 PLC, 10 PLC, maybe 100 PLC and both AZ on and off.

The AZ setting seems to do nothing. 🤷 My room conditions are good for your test, but I'm not sure what you're asking me to do. I assume a WIMA MPK4 10nF cap is fine? Beyond that, I'm also assuming you want me to do more than connect it directly to the DMM in DCV mode?

I also asked a question, I don't know if you saw it. Skander36's post above shows that my voltage reading with no leads attached is about 3 times higher than his. Should he be testing this too?

A 2.5 V reading is a rather odd test point for stability. The usual stable reference are 10 V or 7 V and maybe 1.018 V. 2.5 V is also more on the low end of the 20 V range. Chances are the 2.5 V ref for the test can be the limiting factor more than the DMM.

I was lazy with the ref, switching the jumper is a lot of work. 😉 The point was to compare stability at a specific range between my 3 bench DMMs, which it did. But here's a 10V check after no warmup time. Still nice:

[Kleinstein]

The simple input bias test is with just the 10 M resistance from the divider. This needs not external capacitor.
The reported 1.4 mV (converts to 140 pA) would be a problem - possibly a reason to return the meter as out of spec.

To really test in the high Z mode one could use an external 10 M resistor.  To also test at different voltages one could use the 10 nF MKP4 (should be OK). The test needs a way to record the changing voltage over some time. For just a points the internal memory may be good enough (e.g. 2 readings 10 seconds appart give 1 mV of change per 1 pA of input current). One can charge the capacitors to different starting points as needed: e.g. some  +20 V , -20 V and maybe 0 V. It may need a few seconds to maybe a minute to stabilize the DA part.  The drift rate than gives the bias current at the choosen voltage. The difference between different voltages gives the input impedance (actually the conductance, but that is what should matter).  The cables to connect the capacitors should also be low leakage or handled with care (hold on insulators). The insulation resistance of bad cables may be comparable to a good meters input impedance.

If the AZ mode and different PLC rate do not change things for the test with the resistor (or 10 M mode) one could get away with just one case for the test with the capacitor.

[Martin72]

I also asked a question, I don't know if you saw it. Skander36's post above shows that my voltage reading with no leads attached is about 3 times higher than his. Should he be testing this too?

I briefly switched on my SDM3065X, it also shows something between 1.1 and 1.3mV DC with open inputs.
I don't feel this is anything to be concerned about.

[KungFuJosh]

I also asked a question, I don't know if you saw it. Skander36's post above shows that my voltage reading with no leads attached is about 3 times higher than his. Should he be testing this too?

I briefly switched on my SDM3065X, it also shows something between 1.1 and 1.3mV DC with open inputs.
I don't feel this is anything to be concerned about.

No no, it's definitely messed up. You should trade for my two 5.5 digit DMMs. 😉

[Performa01]

I’ve once checked the input current with regard to the input voltage for several bench DMM. I thought I would show the results here, just to give you an idea what to expect from an average bench DMM in practice.

Notice that most manufacturers do not even specify input current, only input resistance, which is just a limit value that can (and should) be exceeded by far. Furthermore, the input of a DMM is not resistive anyway, but can have a rather non-linear voltage-current characteristic.

As a consequence, I would never rely on the input characteristic of a modern DMM for any critical task, just because even the best will still have a substantial input current that is not negligible for high precision measurements.

Caution: it should be kept in mind that the input impedance / input current of a DMM depends on the environment, like ambient temperature and relative humidity and it is subject to component tolerances. Furthermore, all of these instruments (except Picotest M3500A and Rigol DM3068) have been acquired from the 2nd hand market and their history is largely unknown. For the given reasons, the following test results can only be a guideline, only valid for that particular unit under normal laboratory conditions.

Here come the test results in alphabetical order:

1. Fluke 8842A
This venerable 5.5-digit instrument is famous for its exceptional high long-term stability, which puts many 6.5-digit meters to shame. Yet at higher sensitivities the Fluke 8842A is very noisy and requires heavy filtering, which makes especially the 20 mV range very slow – and short-term stability aka noise is still nothing to write home about.

The Fuke 8842A provides high input impedance up to 20 V; it draws a constant input current of 50 pA from 1 - 19.9 volts. The 8840A is largely identical in this regard.


Fluke_8842A_input


2. Keithley 2001
This 7.5-digit instrument certainly is not an “average” bench DMM. It is easily one of the best and combines excellent long-term stability with low noise. The Keithley 2001 provides high input impedance up to 22V and its input current is almost linear rising from 12 to 38 pA over the range of 0 – 22 volts. Its input characteristic can be approximated with 909 GΩ input impedance and 12 pA input bias current.


Keithley_2001_input


3. Keithley 2015 THD
This 6.5-digit instrument is certainly a good average lab tool. It provides high input impedance up to 12V and its input current is almost linear rising from 450 pA to 890 pA over the range of 0 – 11 volts. The input impedance remains high up to 12 volts, yet there is a steep increase of the input current beyond 11 V. The input characteristic of the Keithley 2015 THD can be approximated with 24 GΩ input impedance and 447 pA input bias current.


Keithley_2015THD_input


4. Picotest M3500A
This 6.5-digit instrument is the same as the Keithley 2100 and certainly needs not hide behind a Keithley 2000/2015. The Picotest M3500A provides high input impedance up to 12V and its input current is rather non-linear, rising from 134 pA to 370 pA over the range of 0 – 11 volts. The input impedance remains high up to 12 volts, yet there is a steep increase of the input current beyond 11 V. The input characteristic of the Picotest M3500A can be approximated with 56 GΩ input impedance and 134 pA input bias current.


Picotest_M3500A_input


5. Rigol DM3068
This 6.5-digit instrument is the one I don’t have a lot of experience with. It provides high input impedance up to 22V and its negative(!) input current is pretty linear, decreasing from -115 pA to -87 pA over the range of 0 – 22 volts. The input characteristic of the Rigol DM3068 can be approximated with 793 GΩ input impedance and -115 pA input bias current.


Rigol_DM3068_input


6. Solartron/Schlumberger 7150plus
This 6.5-digit instrument is actually only 5.5 digits where one additional digit is obtained by averaging. This is a very old design and it certainly cannot compete with the more modern DMMs. The Schlumberger 7150plus provides high input impedance up to 2V and its negative input current is all over the place, yet very low and never exceeds -10 pA over the range of 0 – 2 volts.


Schlumberger_7150plus_input


If you miss the Siglent SDS3065X here – well, I just don’t have one. Evaluating their oscilloscopes and waveform generators is sufficient to keep me from getting bored 😉


As can be seen, only the very old Schlumberger 7150plus has a really low input current (and this is not an isolated case, since I have four of them and they all behave similar). Yet this instrument is not very useful for precision tasks, as it lacks stability (both long- and short-term) and high input impedance up to only 2 volts isn’t overly useful either.

Because of all this, I’ve once built an electrometer amplifier (with <1 pA input current, voltage gain of 1 and <1 ppm/°C stability) that I used for sensitive measurements, whenever accuracy would have been ruined by the input current of a DMM.


DCA_R02_Inside_Front_01

[KungFuJosh]

I played with EasyDMM some more, connected to the SDM3065X. It's nicer than I thought, but a bit clunky. They could make this really nice with a little more development.

[alm]

I’ve once checked the input current with regard to the input voltage for several bench DMM. I thought I would show the results here, just to give you an idea what to expect from an average bench DMM in practice.

Thanks for sharing these measurements!

As a consequence, I would never rely on the input characteristic of a modern DMM for any critical task, just because even the best will still have a substantial input current that is not negligible for high precision measurements.

I'd expect a good bench DMM to have a reasonably low bias current in the order of 100 pA or less. Though it certainly doesn't hurt to verify. Particularly if the bias current is very critical, like for high source impedance (like a high-impedance voltage divider). Fluke has a document how to account for bias current when using DMMs with a Fluke 752A voltage divider which is discussed here.

2. Keithley 2001
This 7.5-digit instrument certainly is not an “average” bench DMM. It is easily one of the best and combines excellent long-term stability with low noise.

I find its noise performance disappointing. It may well be the highest noise 7.5 digit DMM from a reputable brand. Even the 6.5 digit Keithley 2000 has lower noise than the 2001. I don't think the stability is particularly impressive for a 7.5 digit meter using an LM399 as reference, while most competitors (contemporary Datron and Solartron, modern Keysight and Keithley) used better references. The K2001's 1 year DCV accuracy is barely better than the Fluke 8842A. I think it's a nice and versatile bench meter (with features like measuring current in-circuit), though.

3. Keithley 2015 THD
This 6.5-digit instrument is certainly a good average lab tool. It provides high input impedance up to 12V and its input current is almost linear rising from 450 pA to 890 pA over the range of 0 – 11 volts. The input impedance remains high up to 12 volts, yet there is a steep increase of the input current beyond 11 V. The input characteristic of the Keithley 2015 THD can be approximated with 24 GΩ input impedance and 447 pA input bias current.


Keithley_2015THD_input

This is about an order of magnitude higher than I'd expect. See for example this discussion about the Keithley 2000 bias current. The DMM part of the Keithley 2000 and 2015 should be identical.

[tautech]

If you miss the Siglent SDS3065X here – well, I just don’t have one. Evaluating their oscilloscopes and waveform generators is sufficient to keep me from getting bored 😉

Hoping you get the opportunity to work on SDM4065A development. 

[Martin72]

I briefly switched on my SDM3065X, it also shows something between 1.1 and 1.3mV DC with open inputs.
I don't feel this is anything to be concerned about.

Today at work, with a 5.5 digit Keysight 34450A:
2 mVdc with open inputs.

[KungFuJosh]

Hoping you get the opportunity to work on SDM4065A development. 

I would not be sad if my second DMM was the 4065A instead of the 3065X. Any idea how long until they release them?