Keithley 181 Nanovoltmeter
I got my Keithley 181 cheaply from a garage sale after it had collected dust for about a decade on a shelf. The first time I powered it up it became evident that the readings would occasionally jump violently around. It turned out that this behavior was merely caused by unreliable connections between the analog board and the main board. I only had to remove the analog board and put it back into place again to get rid of the problem! For a long term solution the pins of connectors J1009 and J1010 were cleaned up, however. Then, even when the violent jumps were gone, the readings for the most sensitive ranges were in general very noisy and drifty. By taking a closer look at the pre-amplifier, I found that the copper lugs that are crimped to the copper pins of R436 located at the very input of the 181 were not tightly secured to the board, and even looked slightly oxidized. I therefore gently cleaned the lugs and the pads on the board with a glass fibre brush. While doing this, I discovered that one of the lugs was not firmly connected to R436. I crimped the lug, and then screwed the lugs firmly into the board. After this simple operation, the 181 became quite another beast and could now be put into actual use.
The graph shows the measured residual noise with the 181 connected to a Keithley 1406 cable with its copper crocodile clips shorted. Note that this measurement was carried out during a night where the 181 was left alone with no operators close by to cause turbulence in the air and temperature gradients.
The voltage was measured on the 1000x analog output by a Fluke 8842A connected to GPIB. The measured 28.2 nVpp noise over the observation period of some 6½ hours are within the specified 30 nV. Note, however, that the noise check in the 181 service manual is done over a period of 2 minutes only and with the model 1488 low-thermal shorting plug, not a cable. |
The pitfalls of thermal effects apply to any low voltage measurement setup, and should be taken seriously, underlined by this odd example. The graph shows the residual noise with the 181 connected to a shorted Keithley 1406 cable. So, what happens in the middle? During the first half the operator (yours truly) sits close to the setup. Then, the operator leaves the room where the setup is located, causing the readings to stabilize, most likely caused by the reduction of air turbulence around the setup.
In addition to the essential recommendations of using pure (and clean) copper interconnections, or gold-plated copper-tellurium interconnections, with low thermal EMF, I would recommend to insulate external connections by wrapping them in a sheet of foam to reduce thermal gradients caused by air turbulence. |
Keep in mind that the 181 has a plastic casing with no shielding of the electronics, and offers no real protection against today's wireless technology. Inside the 181 there's a shielding box covering almost all of the analog board, and there's a small shielding box covering the most sensitive parts of the pre-amplifier. However, large parts of the design are left unshielded, so the 181 does not take lightly to a mobile phone nearby. The figure shows a measurement session started when a mobile phone, located 30 cm away and in front of the 181, is powered on. A transmitter burst after the phone has booted clearly inteferes with the 181.
|
The fastest acquisition time is obtained by running the Keithley 181 in Talk Only mode, which allows you to gather the measurements as soon as they become available. The manual claims "four readings per second", but the average interval between the measurements turns out to be about 280 ms.
The LabVIEW code shown here is an example of how to get data from the Keithley 181. I have used this code successfully with a GPIB-USB-HS interface between the 181 and a PC. For each round in the while loop the code gets a measurement, with the speed determined by the 181. The command string sets up the 181 for the 2 mV range, it shuts off filtering completely and enables the Damping (thereby removing the non-linear speed-up feature). |
The graph on the left show the frequency response of the 181 with the digital filter set to '1' (normal filtering) and to '0' (filter 'off'). In both cases the Damping function is enabled. The responses are not specified by Keithley, but the -3 dB cut-off with the normal filter is about 35 mHz, and about the half with the slow filter, both having a third order roll-off.
If you want to use the 181 as a low-level digitizer, you would likely prefer to disable the filters entirely. Setting the filter to 'off' equals filter P1, so you need to issue GPIB command P0 to remove filtering alltogether. By doing so, the response stays essentially flat to the Nyquist limit. If you want to use the filters while using the 181 as a low-level digitizer, you should enable the Damping function. With this enabled, you do not actually enable anything, rather you disable the non-linear speed-up feature, which is used to lower the settling time for signals that change above a certain threshold. However, such a feature is not appropriate if you are trying to digitize a signal faithfully. If you need to speed up the response of the 181, this can only be done through GPIB by disabling its digital filter entirely. |