« Close

Datasheets and User Guides

App Notes

Software & Driver

 

14.0 AIN

T7 AIN Overview

Analog Inputs: 14

Voltage Ranges: ±10V, ±1V, ±0.1V, and ±0.01V (Appendix A-3 and Appendix A-3-2)

ADC Resolution: T7 = 16-bit
T7-Pro = 16-bit and 24-bit


Effective Resolution:
11-bit to 22-bit (Appendix A-3-1)

Max Data Rate: 100 ksamples/second in stream mode (Appendix A-1)

 

Available Inputs

The LabJack T7/T7-Pro has 16 total built-in analog inputs. Two of these are connected internally (AIN14=Temperature & AIN15=GND), leaving 14 user-accessible analog inputs (AIN0-AIN13). The first 4 analog inputs, AIN0-AIN3, appear both on the screw terminals and on the DB37 connector.  The remaining 10 analog inputs appear on the DB37 only.

 

Table 14.0-1. Positive Channel Numbers

Positive Channel #
0-13 Single-Ended
0,2,4,6,8,10,12 Differential
14 Temp Sensor (deg K)
15 GND
16-127 Extended Channels

 

Table 14.0-2. Negative Channel Numbers

Negative Channel #
1,3,5,7,9,11,13 Differential
0,15,199 Single-Ended (GND)
24-127 Extended Channels

 

Extended channels 16 through 127 are created with external multiplexers or the Mux80 accessory.  See Section 14.2 for more information.

Resolution Index

The analog inputs have variable resolution.  Sample time increases as the resolution index increases.  Resolution indices range:

  • from 1 to 8 for the T7
  • from 1 to 12 for the T7-Pro

Setting resolution index to 0 sets the default resolution index, which is:

  • 1 for stream mode
  • 8 for command-response mode on a T7
  • 9 for command-response mode on a T7-Pro

See Appendix A-3-1 more details about resolution index.

Impedance, Gain, Single-ended vs. Differential, and Floating Inputs

The analog inputs are connected to a high-impedance instrumentation amplifier, as shown in Figure 4.2-2. This in-amp buffers the signal for the internal ADCs, allows for single-ended or differential conversions, and provides gains of x1, x10, x100, and x1000 (corresponding to ranges of ±10V, ±1V, ±0.1V, and ±0.01V).

Differential channels are adjacent even/odd pairs only, such as AIN2-AIN3. Thus the positive channel must be even and the negative channel must be +1.

The analog inputs are not artificially pulled to 0.0 volts, as that would reduce the input impedance, so readings obtained from floating channels will generally not be 0.0 volts. The readings from floating channels depend on adjacent channels and sample rate and have little meaning. See related floating input application note.

Extended Features

Various extended features are available for analog inputs to allow them to do thermocouple math, RMS calculations, and more.  See Section 14.1.

Analog Input App Note

For information regarding typical analog input connections, please see the Analog Input App Note.

  Register Listing

Analog Input Registers
Name Start Address Type Access
AIN#(0:13) 0 FLOAT32 R
AIN#(0:13)_RANGE 40000 FLOAT32 R/W
AIN#(0:13)_NEGATIVE_CH 41000 UINT16 R/W
AIN#(0:13)_RESOLUTION_INDEX 41500 UINT16 R/W
AIN#(0:13)_SETTLING_US 42000 FLOAT32 R/W
AIN_ALL_RANGE 43900 FLOAT32 R/W
AIN_ALL_NEGATIVE_CH 43902 UINT16 R/W
AIN_ALL_RESOLUTION_INDEX 43903 UINT16 R/W
AIN_ALL_SETTLING_US 43904 FLOAT32 R/W
AIN#(0:13)
- Starting Address: 0
Returns the voltage of the specified analog input.
  • Data type: FLOAT32  (type index = 3)
  • Read-only
  • This register may be streamed
Expanded Names Addresses
AIN0, AIN1, AIN2, AIN3, AIN4, AIN5, AIN6, AIN7, AIN8, AIN9, AIN10, AIN11, AIN12, AIN13 Show All 0, 2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 22, 24, 26 Show All
AIN#(0:13)_RANGE
- Starting Address: 40000
The range/span of each analog input. Write the highest expected input voltage.
  • Data type: FLOAT32  (type index = 3)
  • Readable and writable
  • Default value: 0
  • T7-specific details: Valid values/ranges: 0.0=Default=> +/-10V, 10.0=> +/-10V, 1.0=> +/-1V, 0.1=> +/-0.1V, or 0.01=> +/-0.01V.
Expanded Names Addresses
AIN0_RANGE, AIN1_RANGE, AIN2_RANGE, AIN3_RANGE, AIN4_RANGE, AIN5_RANGE, AIN6_RANGE, AIN7_RANGE, AIN8_RANGE, AIN9_RANGE, AIN10_RANGE, AIN11_RANGE, AIN12_RANGE, AIN13_RANGE Show All 40000, 40002, 40004, 40006, 40008, 40010, 40012, 40014, 40016, 40018, 40020, 40022, 40024, 40026 Show All
AIN#(0:13)_NEGATIVE_CH
- Starting Address: 41000
Specifies the negative channel to be used for each positive channel. 199=Default=> Single-Ended.
  • Data type: UINT16  (type index = 0)
  • Readable and writable
  • Default value: 199
  • T7-specific details: For base differential channels, positive must be even channel from 0-12 and negative must be positive+1. For extended channels 16-127, see Mux80 datasheet.
Expanded Names Addresses
AIN0_NEGATIVE_CH, AIN1_NEGATIVE_CH, AIN2_NEGATIVE_CH, AIN3_NEGATIVE_CH, AIN4_NEGATIVE_CH, AIN5_NEGATIVE_CH, AIN6_NEGATIVE_CH, AIN7_NEGATIVE_CH, AIN8_NEGATIVE_CH, AIN9_NEGATIVE_CH, AIN10_NEGATIVE_CH, AIN11_NEGATIVE_CH, AIN12_NEGATIVE_CH, AIN13_NEGATIVE_CH Show All 41000, 41001, 41002, 41003, 41004, 41005, 41006, 41007, 41008, 41009, 41010, 41011, 41012, 41013 Show All
AIN#(0:13)_RESOLUTION_INDEX
- Starting Address: 41500
The resolution index for command-response and AIN-EF readings. A larger resolution index generally results in lower noise and longer sample times.
  • Data type: UINT16  (type index = 0)
  • Readable and writable
  • Default value: 0
  • T7-specific details: Valid values: 0-8 for T7, 0-12 for T7-Pro. Default value of 0 corresponds to an index of 8 (T7) or 9 (T7-Pro).
Expanded Names Addresses
AIN0_RESOLUTION_INDEX, AIN1_RESOLUTION_INDEX, AIN2_RESOLUTION_INDEX, AIN3_RESOLUTION_INDEX, AIN4_RESOLUTION_INDEX, AIN5_RESOLUTION_INDEX, AIN6_RESOLUTION_INDEX, AIN7_RESOLUTION_INDEX, AIN8_RESOLUTION_INDEX, AIN9_RESOLUTION_INDEX, AIN10_RESOLUTION_INDEX, AIN11_RESOLUTION_INDEX, AIN12_RESOLUTION_INDEX, AIN13_RESOLUTION_INDEX Show All 41500, 41501, 41502, 41503, 41504, 41505, 41506, 41507, 41508, 41509, 41510, 41511, 41512, 41513 Show All
AIN#(0:13)_SETTLING_US
- Starting Address: 42000
Settling time for command-response and AIN-EF readings.
  • Data type: FLOAT32  (type index = 3)
  • Readable and writable
  • Default value: 0
  • Minimum firmware version: 0.9328
  • T7-specific details: 0 = Auto. Max is 50000 (microseconds).
Expanded Names Addresses
AIN0_SETTLING_US, AIN1_SETTLING_US, AIN2_SETTLING_US, AIN3_SETTLING_US, AIN4_SETTLING_US, AIN5_SETTLING_US, AIN6_SETTLING_US, AIN7_SETTLING_US, AIN8_SETTLING_US, AIN9_SETTLING_US, AIN10_SETTLING_US, AIN11_SETTLING_US, AIN12_SETTLING_US, AIN13_SETTLING_US Show All 42000, 42002, 42004, 42006, 42008, 42010, 42012, 42014, 42016, 42018, 42020, 42022, 42024, 42026 Show All
AIN_ALL_RANGE
- Address: 43900
A write to this global parameter affects all AIN. A read will return the correct setting if all channels are set the same, but otherwise will return -9999.
  • Data type: FLOAT32  (type index = 3)
  • Readable and writable
  • Default value: 0
  • Minimum firmware version: 0.9328
AIN_ALL_NEGATIVE_CH
- Address: 43902
A write to this global parameter affects all AIN. Writing 1 will set all AINs to differential. Writing 199 will set all AINs to single-ended. A read will return 1 if all AINs are set to differential and 199 if all AINs are set to single-ended. If AIN configurations are not consistent 0xFFFF will be returned.
  • Data type: UINT16  (type index = 0)
  • Readable and writable
  • Default value: 199
  • Minimum firmware version: 0.9328
AIN_ALL_RESOLUTION_INDEX
- Address: 43903
The resolution index for command-response and AIN-EF readings. A larger resolution index generally results in lower noise and longer sample times. A write to this global parameter affects all AIN. A read will return the correct setting if all channels are set the same, but otherwise will return 0xFFFF.
  • Data type: UINT16  (type index = 0)
  • Readable and writable
  • Default value: 0
  • Minimum firmware version: 0.9328
  • T7-specific details: Valid values: 0-8 for T7, 0-12 for T7-Pro. Default value of 0 corresponds to an index of 8 (T7) or 9 (T7-Pro).
AIN_ALL_SETTLING_US
- Address: 43904
Settling time for command-response and AIN-EF readings. A write to this global parameter affects all AIN. A read will return the correct setting if all channels are set the same, but otherwise will return -9999. Max is 50,000 us.
  • Data type: FLOAT32  (type index = 3)
  • Readable and writable
  • Default value: 0
  • Minimum firmware version: 0.9328
  • T7-specific details: 0 = Auto. Max is 50000 (microseconds).

Some Examples

Analog Input Example: To read a voltage connected to AIN2, perform a read of AIN2 (address 4).  The result will be in the form of a floating point number, like 8.82332V.

Range Example: If the voltage source connected to AIN1 has a known range of 0 to 0.7V, the appropriate range for AIN1 is the ±1V range. Configure this by writing 1.0 to AIN1_RANGE (address 40002).

Differential Analog Input Example:  To take a differential reading on AIN2, set AIN3 as its negative channel so that the measurement is AIN2-AIN3.  Configure this by writing a value of 3 to AIN2_NEGATIVE_CH (41002).  To set it back to single-ended (AIN2-GND), write 199 to AIN2_NEGATIVE_CH (address 41002).

Resolution Index Example: Change the AIN1 resolution index to 5 by writing a value of 5 to AIN1_RESOLUTION_INDEX (address 41501).

Settling Example: Change the settling time of AIN3 to 500 µs by writing 500 to AIN3_SETTLING_US (address 42006). The value 0 sets automatic settling, which is recommended for most applications.

Address Increment Size

Addresses of the FLOAT32 type increment in steps of 2 because FLOAT32 uses two sets of 16-bits. FLOAT32 registers include AIN#, AIN#_RANGE, and AIN#_SETTLING_US.

Addresses of the UINT16 type increment in steps of 1 because UNIT16 uses only one set of 16-bits. UINT16 registers include AIN#_NEGATIVE_CH and AIN#_RESOLUTION_INDEX.

Extra Details

The AIN#(0:13)_RANGE and AIN_ALL_RANGE parameters are controlling the gain of the internal instrumentation amplifier. The in-amp supports gains of x1, x10, x100, and x1000. If you set range=10, you get gain=x1, and the analog input range is ±10 volts. If you set range=1, you get gain=x10, and the analog input range is ±1 volts. Note that the device knows what the internal gain is set to and adjusts the return values to give the voltage at the input terminals, so if you connect a 0.8 volt signal to the input terminals, it will be amplified to 8.0 volts before being digitized, but the reading you get back will be 0.8 volts. Write range=10 to get a range of ±10V (default), range=1 to get a range of ±1V, range=0.1 to get a range of ±0.1V, or range=0.01 to get a range of ±0.01V. If you write a value in between the valid ranges, the larger range will be used.

The AIN#(0:13)_NEGATIVE_CH and AIN_ALL_NEGATIVE_CH parameters pertain to differential versus single-ended readings (not to be confused with bipolar and unipolar ... the T7 is always bipolar).  On the T7, differential channels are adjacent even/odd pairs only, such as AIN2-AIN3. Thus the positive channel must be even and the negative channel must be +1.  Only an even channel can have an associated negative channel, so you will never write to odd channels of this register (e.g. never write to AIN3_NEGATIVE_CH/41003).  Channel numbers in the extended range (above AIN15), are connected to AIN0-AIN13 and those dictate the even/odd rule, not the extended channel numbers (see the Mux80 datasheet).

The AIN#(0:13)_RESOLUTION_INDEX and AIN_ALL_RESOLUTION_INDEX parameters affect command-response readings from the ADC.  A higher Resolution_Index results in lower noise and thus higher effective & noise-free resolution, with the tradeoff of longer sample times.  See the Noise and Resolution Appendix.  The value passed for Resolution_Index is from 0-8, where 0 corresponds to default, 1 is roughly 16-bit resolution (RMS or effective), and 8 is roughly 19-bit resolution. The T7-Pro has additional Resolution_Index settings 9-12 that use the alternate high-resolution converter (24-bit sigma-delta) and correspond to roughly 19-bit to 22-bit resolution.  For command-response readings, the default value of 0 corresponds to Resolution_Index=8 on a T7 and Resolution_Index=9 on a T7-Pro.  For typical noise levels and sample times at different combinations of resolution index and gain see Appendix A-3-1 Noise and Resolution.  For general discussion on the meaning of resolution see the Noise and Resolution App Note.  This register is for command-response and AIN-EF only.  Stream mode has its own resolution index register documented in the Stream Section.

The AIN#(0:13)_SETTLING_US and AIN_ALL_SETTLING_US parameters are setting the time from a step change in the input signal to when the signal is sampled by the ADC, measured in microseconds. A step change in this case is caused when the internal multiplexers change from one channel to another. In general, more settling time is required as gain and resolution are increased. The default “auto” settling time ensures that the device meets specifications at any gain and resolution for source impedance up to at least 1000 ohms. This parameter applies to command/response mode and AIN EF.  Stream mode has its own settling parameter. The timings in Appendix A are measured with “auto” settling.  This register is for command-response and AIN-EF only.  Stream mode has its own settling register documented in the Stream Section.

 

Duplicated Terminals on the T7 (AIN0-AIN3)

AIN0-AIN3 appear on the built-in screw-terminals and also on the DB37 connector. Users should only connect to one or the other, not both at the same time.

To prevent damage due to accidental short circuit, both connection paths have their own series resistor. All AIN lines have a 2.2k series resistor, and in the case of AIN0-AIN3 the duplicated connections each have their own series resistor, so if you measure the resistance between the duplicate terminals you will see about 4.4k.

 

21 comments

What is the channel-to-channel sample delay for the T7?

For the UE9 it's mentioned in http://labjack.com/support/ue9/users-guide/appendix-a

The interchannel delay (or channel-to-channel delay) in stream mode has not been measured yet.  In fact I see that the U6, which has pretty much the same AIN system, just specifies the max possible times (1/MaxSampleRate) rather than actual measurements.  See Section 3.2 of the U6 User's Guide.  The T7 will stream at least as fast as the U6 (we expect a limit of 50-100k), so we know the max interchannel delay will be < 10-20us.

Hello, I am using the Labjack T7 with my own Modbus TCP programing and I am having troubles to set the AIN Range. Upon startup, it reads 10V and I am able to read the voltage in less than 3ms. If I set a range value, any value, it is accepted but read back as 0.01 and then my voltage reading are wrong and take 12ms... Here are the Modbus TCP frames exchanges after T7 is powered on : 1) read AIN#0_RANGE sent: 00 00 00 00 00 06 00 04 9c 40 00 02 received: 00 00 00 00 00 07 00 04 04 41 20 00 00 AIN#0_RANGE = 10 (0x41200000) 2) write 10 (0x41200000) in AIN#0_RANGE sent: 00 01 00 00 00 0c 00 10 9c 40 00 02 00 04 41 20 00 00 received: 00 01 00 00 00 06 00 10 9c 40 00 02 No error sent back 3) read again AIN#0_RANGE sent: 00 02 00 00 00 06 00 04 9c 40 00 02 received: 00 02 00 00 00 07 00 04 04 3c 23 d7 0a AIN#0_RANGE = 0.009999999776482582 (0x3c23d70a) The problem is probably with the write but I do not see what. Please help !

It looks like your write command packet is incorrect. Try something like this instead:

sent: 00 01 00 00 00 0b 00 10 9c 40 00 02 04 41 20 00 00

The problem in your packet starts at byte 12. Byte 12 is the Byte Count (2 * Num Reg) byte and is only one byte. Byte 13 is where the write data starts. It looks like you were treating Byte Count as a 2 byte value.

For Modbus protocol details on our site take a look here:

http://labjack.com/support/modbus/protocol-details

Thanks a lot !

You saved my life ;-)

Hello guys,

I have a question about the negative channels! You wrote

Differential Analog Input Example: To perform differential readings between AIN2(pos) and AIN3(neg), change the negative channel of AIN2 to AIN3 by writing a value of 3 to address 41004.

Could it be possible, that you meant that if I want the negative channel of AIN2 to AIN3 that I have to write the 3 to the address 41002 (negative_ch_2)?

With friendly regards

MT

Dear Labjack Support-Team,

I have a little problem again. I am writing a software in combination with the T7.
The User has the choice to change the configuration of the analog and digital input channels.
Could it be possible that some channels or some configurations need time to take effect, before you can
reconfigure the channel? Because in my software the configuration is changed as the following:

first resolution
second range
third negative channel

But sometimes I just get two answers instead of three!
Do you think it is the fault of my software or is it because it takes time before I can reconfigure these channels again?

I am looking forward to hearing from you!

With friendly regards
Mr. T

If you send 3 commands you should get 3 answers.  You can set resolution/range/negative in any order and do not need any delays.  To troubleshoot the details of this I suggest starting a forum topic or emailing [email protected].  Let us know what you are using for software, what comm interface (USB, Eth, or WiFi), LJM or direct Modbus, and any other details about the code.  Make sure you are using the latest of everything (see "Updates" on the T7 Support Homepage).

What is the input impedance on the analog inputs? Are they essentially floating op amp inputs, with some protection beyond the normal input range?

I don't see that spec either, so we need to add it.  Look at Appendix A of the U6 User's Guide (note #9) as it has the same analog input system.  Yes, each analog input is connected to an instrumentation-amplifier input through multiplexers.

Specs added in A-3.  The + and - inputs are routed to in-amp inputs as shown in the hardware diagram in the Hardware Overview section.

Note that as is the case with most instrumentation amps, the input bias is usually of more concern than the input impedance.  Also, if scanning multiple channels at higher speeds you have to think about how high source impedance might slow down settling.

Dear LabJack Team,

I am using a LabJack 7 Pro to read a differential input coming from a load cell. In order to read the differential input, I have connected the positive signal line of load cell to AIN6 in T7 and negative signal line from load cell to AIN7. Then to read the differential input (AIN6-AIN7) from T7 I have configured the AIN6 as follows:

# Setup and call eWriteNames to configure the AIN on the LabJack.

numFrames = 3

names = ["AIN6_NEGATIVE_CH", "AIN6_RANGE", "AIN6_RESOLUTION_INDEX"]

aValues = [3, 10, 0]

ljm.eWriteNames(handle, numFrames, names, aValues)

 

When I try to run the program to read AIN6 it shows an error, which is

raise LJMError(error, errAddr)

labjack.ljm.ljm.LJMError: Address 41006, LJM library error code 2373 AIN_NEGATIVE_CHANNEL_INVALID

Can you please explain me the reason for the error.
Thanx

Dear LabJack group,

The T7's analog inputs appear to have a range of +/-10V (Gain =1), even in single-ended mode. 

Is there anyway to configure the T7 to utilize full-resolution with a uni-polar input signal (0-5V, 0-10, 0..20mA, ...)?

I see that even the analog 'tick' accessories output only unipolar signals, so the resolution is probably halved if those were used.

I am not opposed to minor, warranty-breaking, pcb modifications (if that's an option).

Best regards,

Chris

You could remove a resistor and change 1 other resistor to make the max range unipolar 0-10V (or anything smaller).  Contact [email protected] if you want more detail.

The default resolution index is 0.  You only state that resolution index 1 ~ 16bit resolution.  What does 0 correspond to then?

For command-response readings, the default value of 0 corresponds to Resolution_Index=8 on a T7 and Resolution_Index=9 on a T7-Pro. For stream readings the default of 0 corresponds to Resolution_Index=1. For typical noise levels and sample times at different combinations of resolution index and gain see Appendix A-3-1 Noise and Resolution.  For general discussion on the meaning of resolution see the Noise and Resolution App Note.

Hi guys,

I have a problem with reading differential analog inputs of T7-PRO over Modbus/TCP (Ethernet wired connection) which I assume is related to extra initialization available over USB.

I described the test I made with Kipling to confirm the statement above in my comment I posted here:

labjack.com/support/kipling

The readings I receive differ from the rather precise source I applied. My software working via Modbus over TCP/IP returns results that are on 0.01V range about 50% of what I expect and see on the source display. That issue shows the most on the -0.01..0.01V range which I haven't tested until I got the sensors hooked up.

I am testing single readings in C-R mode as I would like to use highest index to get best precision, especially on the -0.01..0.01V range.

The discrepancy I am seeing is different for different ranges, say

* 0.01V shows ~0.00498V

* 0.1V shows 0.042V on the -0.1..0.1V range

* 0.09V shows 0.072 on the -0.1..0.1V range; 0.082 on -1..1V range and 0.091 on -10..10V range

I probably can calibrate the device but that is outside of my current scientific or hobbyist's  interests unless of course this is the only way ;-)

If my assumption about extra initialization available over USB is not correct, I'd like to try to find out what is that I'm doing wrong with modbus initialization.

Here is the list of values I initialize (some of them are irrelevant in the case of single readings, others are irrelevant in the streaming case):

STREAM_SCANRATE_HZ         4002     FLOAT32 = 1*100.0
STREAM_NUM_ADDRESSES         4004     UINT32    =  1
STREAM_SAMPLES_PER_PACKET     4006     UINT32    = 512
STREAM_SETTLING_US         4008     FLOAT32    = 100.0
STREAM_RESOLUTION_INDEX     4010     UINT32    = 7
STREAM_BUFFER_SIZE_BYTES     4012     UINT32    = 32768
STREAM_SCANLIST_ADDRESS#(0:127)    4100     UINT32    = (0 4 8 .. 24)
AIN_ALL_RANGE             43900     FLOAT32 = 10.0
AIN_ALL_NEGATIVE_CH        43902    UINT16 = 199
AIN_ALL_RESOLUTION_INDEX     43903     UINT16 = 12
AIN#(0:13)_RANGE             40000     FLOAT32 = different (0.1 10. 10.0 0.01 10.0 0.01 10.0 twice in a row for differential mode so I write each value twice 0.1 0.1 10.0 10.0 10.0 10.0 0.01 0.01...)
AIN#(0:13)_NEGATIVE_CH        41000    UINT16 = (1 3 5 7 9 11 13) into even slots only, leaving default 199 in the odd ones as is
AIN#(0:13)_RESOLUTION_INDEX    41500    UINT16    = 12 (7 times)
AIN#(0:13)_SETTLING_US        42000    FLOAT32 = 100.0 (7 times)

Thank you in advance!

There is no extra initialization required for USB compared to Ethernet, or vice versa.  The communication interface should be unrelated to analog input stability/quality.

One possibility is your power supply voltage is out-of-spec.  Measure VS with a DMM, and if it is too low ( <4.15 volts at room temp or <4.75 volts at 70 deg C), you are at risk for bad readings.  The supply we ship with the T7 is chosen because it holds a solid output voltage at high loads.

Thank you, for the answer.

As I wrote on the Kipling page we are using adapter that came with T7 device. The voltage on VS is 5.02V and never drops lower at any point of operation.

We don't have another power supply with more than 1A output (I tried 1A and it has the same problems as expected), so we'll buy one and try next week.

If it will not work that'll probably mean faulty T7-PRO device, although the fact that it is working from notebook USB port seems encouraging.

What are you measuring?  Since it's not a power supply issue, the next likely cause is a a grounding issue.  Your notebook is providing an isolated power source, but the wall power is connected to earth ground.  The difference is that in the case of the Laptop, the negative analog input channel is actually becoming ground, but when you have the LabJack connected to the wall power, ground becomes earth ground, and then the negative analog input isn't referenced properly anymore.

Try connecting a large resistor (100k - 1M) between GND and your differential signal's (-) analog input.

http://labjack.com/support/app-notes/differential-analog-inputs  (See the section about common mistakes for differential signals).

You seem to be right again :-)

With a resistor it works as expected with 2 different wall power supplies and with the notebook. Furthermore, it now doesn't work with notebook without ground - not sure why it worked on Friday as both notebook and wall power supply should not provide any connection to the earth.

We are measuring 0..10kOhm potentiometers (I use your 200uA and 10uA for them) and outputs of battery-powered amplifiers that are in tern connected to magnetic field sensors. The purpose of using differential channels is #3 on the page you posted a link to (to reduce any noise from wiring).

I talked to our hardware engineer and it totally makes sense to connect all the grounds to device reference ground via individual resistors. We'll do it in the connection box right by LabJack device.