RoadTest: Keithley 2450 SourceMeter SMU Instrument
Evaluation Type: Independent Products
Did you receive all parts the manufacturer stated would be included in the package?: True
What other parts do you consider comparable to this product?: We have not tested any other SourceMeters, but Agilent has the B2900 series units which are a little more expensive, but have the capability of more channels. It is possible to recreate the functionality of the SMU with a separate power supply and volt meter, but the integration of the two functions is really nice.
What were the biggest problems encountered?: While learning the unit, I was able to put it in a state where I thought it should have been outputting a voltage, but it would not. I had to reset the unit several times. Also, the autoranging function interfered with an attempt to use the unit as a calibration standard.
Below is our review of the Keithley 2450 SMU. We decided to do the review differently then most reviews: we put together a panel of 3 people, and all three had hands on time with the 2450.
We thank Element14 and Keithley for the opportunity to evaluate a piece of equipment that we could not have afforded, but that turned out to be very useful in testing components and hardware designs such as analog to digital converters.
Keithley 2450 Road Test
Tracy Hooker, esalazar Michael Harris
The Keithley 2450 SMU unit is basically a high resolution multimeter (volt, ohm, amp, watts) coupled with a high precision voltage and current source. Keithley has combined these two units into one box, along with abundant connectivity options (LAN, USB, IEEE488). It is easy to interface the unit with a PC, allowing the PC to write instructions and read data from the unit.
If you needed a highly integrated single box solution, this would be it. If you simply wanted a high resolution DMM and a high precision power supply, it would be less expensive to purchase separate units.
Originally arrived in factory packaging, which was adequate. There were two custom-cut pieces of foam supporting the ends and an anti-static bag containing the instrument. The power cord and instruction manuals were loose inside the box. There was also a rattling sound coming from inside the 2450 after it was first un-boxed. The repaired unit came in a generic box, which was not sized for the instrument. It was "foamed in" with some expanding packing foam, though, which was sufficient for the trip.
Overall, the packing was not as nice as some gear that we've gotten from other manufacturers, but it got the job done. It would be nice to see a little more effort go into the box, especially since the power cord and manuals were not in their own interior box.
The calibration certificate was poor. The unit was supplied with a single page and just said that the unit was within tolerance. Even our Agilent 1272A handheld multimeter came with an extensive multi-page document detailing the calibration procedure and lists of test measurements. For such a precise instrument and expensive, we would have expected more. The unit was re-calibrate on repair and came with a new certificate, but it was still only a single page and listed little more than the instruments used as standards and that the the 2450 was within tolerance.
Customer service for the instrument was friendly and prompt. I talked to them both via email and phone regarding the loose component mentioned before. We had to contact them due to a loose component inside of the unit when it arrived. They offered a quick repair and re-calibration and paid for shipping both ways.
This is an odd addition. It’s easy enough to build the connector, but we can’t see it being anything more than an annoyance in practice. An interlock only makes sense if it actually disabled the output altogether. Instead, the output is just limited to 38V.
The touchscreen is a new innovation in test instrumentation, and it seems to work well. It is not too sensitive, but still seems to respond well to finger presses. We found it easy to make parameter changes, and the screen was informative. There is a strange-looking rainbow pattern at certain angles, but it's not a distraction. The viewing angle is OK. For someone like myself who is spoiled by recent smartphones and whatnot, the resolution is underwhelming, but more than adequate for the functionality. One very nice touch is the fact that the screen is flush with the the rest of the front panel. Overall, we like the addition of the touchscreen, especially on an instrument with so many features.
There was some confusion over what test leads should have been included with the 2450 data sheet shows 8606 Test Leads. Quick start guide shows 8608 Safety clip-lead set
The website at http://www.keithley.com/products/dcac/currentvoltage/2450smu/?path=8010/ACC_8010 shows something entirely different.
The test leads that were actually included (8608) have very nice, flexible insulation, but the plastic on the connectors and probe ends feels somewhat cheap. They work great, though.
The front panel has banana jacks for + Source, - Source, + Sense, - Sense, Ground. There is also a USB jack for upgrades, storing data, etc. The front panel connectors are placed intuitively and seem to very high quality connectors.
The rear connectors consist of computer connectivity (USB, LAN, RS232, and IEEE488 (HPIB/GPIB), an Interlock connector, a pair of TSPLink connectors, the standard AC Power connector, and 4 Triaxial connectors. These are mainly used when sourcing and measuring very low currents. Triax connectors may be difficult to find, and I believe there are two different types (with different angles between the bayonet lugs). It would have been nice if Keithley had supplied a set of them. The power connector has a built in fuse, make sure you change the fuse if you operate the 2450 in a location that has a voltage other than 115 VAC.
Since the Keithley 2450 is very precise, I began my review by using the 2450 to calibrate an older HP bench DMM, the HP3478A. This DMM was out of cal, and was inaccurate in its DC VOLTS readings. The 2450 unit should easily be able to provide the calibration signals, since it can produce high accuracy voltage and current.
After placing the older HP DMM in calibration mode, I began the calibration process using DC Volts. This, however, became a frustrating hit and miss process, because the HP DMM was changing its input impedance during calibration mode, which caused the Keithley 2450, which was in autorange, to change ranges. This in turn created a change in DC voltage, and caused the DMM to reject the DC voltage being created by the 2450. It took several times, but I eventually got through the entire DC VOLTS cal procedure on the 3478A.
My second task using the Keithley 2450 was to characterize an Analog to Digital converter. We developed a battery monitor which consists of a voltage monitor and a precision current monitor. This circuit provides a 12 bit digital serial stream in I2C format to an Arduino board, and finally to a PC.
I manually set the 2450 to various voltages and monitored the digital data stream with both a logic analyzer and the PC serial window. The 2450 proved to be easy to set, and with both voltage and current limits available, was also safe for the circuit and the operator. I enjoyed the ease of setting the voltage to the required value, and I knew it was accurate. I did not have monitor it with another DVM. I tested the circuit from 0 to +80VDC. It was very nice not having to change power supplies or worrying if the voltage was accurate.
I performed this test manually, however, future similar tests will be conducted under the control of a PC. That would totally automate the process and allow us finer detail in setting the voltage. This would allow us to evaluate the ADC more thoroughly, such as finding missing or stuck bits. This is the power of the 2450, with both instruments under one ‘roof’ so to speak.
I wanted to check the absolute accuracy of the 2450. The most accurate meters we have access to are 1) an older HP3455A and 2) a brand new 34461A. Both are 6 ½ digit meters, and both read within microvolts of each other, quite a testament to the HP engineers of olde.
I set the Keithley 2450 into “Power Supply” mode, which basically shows both voltage and current on the front panel, and allows constant voltage (CV) or Constant current (CI) operation. I ran both meters against the 2450 from 1uV up to 200 VDC, and as mentioned before, both meters read within microvolts of each other. At +200 volts, the 2450 was off about 30mV, still within specifications of .015% + 24mV.
I then set up the meters to read current, and I tested the output current setting accuracy of the 2450 from 1uA up to 1 Amp, the maximum the 2450 can produce. In all cases, the 2450 output setting and readback functions were spot on, and highly repeatable.
I have every confidence in the accuracy of the voltage and current outputs (and readback) of the Keithley 2450.
My fourth and final evaluation of the unit was to measure the input impedance of several of our DMMs. Since the Keithley 2450 has superb low level performance, and displays high resolution, even below 1nA, I thought that measuring the input impedance would be a good test. What I did not realize at the time was the fact that Keithley 2450 can source voltage (or current) and measure resistance! That made this test trivial.
First I set the 2450 to source voltage and measure current. After looking at the input impedance of the Agilent 34461A, and doing some math, it came out to be 10M ohm. No surprise there. At that point I remembered about the ability to read resistance, and let it do the math for me. A couple of presses on the touchscreen later, and it is sourcing voltage and reading resistance. I discovered that the Agilent 34461A has to input impedance settings: 10M and Hi-Z. 10M is just that, about 10M ohm, whereis high Z is more like 10G ohm!
The input impedance of the Agilent U1272a was also measured to be ~10M ohm. The older HP3455A, on the other hand, changed its input impedance as the range was manually selected. I think at 100K and above its input impedance went to 10G ohm.
The Keithley 2450 is an accurate volt/amp meter (which also computes resistance and power), combined with a sophisticated computer controlled power supply. The unit contains sophisticated display software which can show trends of a parameter. It also allows on-screen diagramming of trigger conditions, almost like building a flowchart in MS Visio. The 2450 can be set up to perform voltage/current sweeps AND take data at the same time, all from the front panel GUI. There is a wealth of sophistication (i.e. a small computer) packed into the box of the 2450.
The touchscreen makes it easy to set up the unit, however, in several instances I was able to somehow place the unit in a mode where it should have been supplying a voltage, but was not. I simply reset the unit to put it back into a known state. As I got more familiar with the menu structure and how to navigate it, I did not have that problem, so I will chalk it up to operator error.
I used the graphing capability several times during my reviews. It is nice to be able to see trends, not for long term monitoring of a signal. The x axis (usually time) is scaled automatically, and as time continues, the x axis becomes compressed. I tried to manually manipulate the x axis, but every time I did so, the data on the graph would stop updating. I had to exit, and reset to get the graphing function working again. I think it would be better to have a selection to allow the user to look at some time period, say 1 minute, 10 minutes, 10 seconds, and it would just give you a rolling window with that duration.
The real power of the 2450 is when it is connected to a PC, and software written to test devices like in my Review #2 above. Sadly, I was not able to control it in this manner, but other reviewers were, so please read their review of the unit.
Overall, I like the unit, and I think Keithley has a nice instrument here. Their innovative use of a touchscreen is, almost, revolutionary in test equipment arena. One irritating area that needs improvement is that their brochure, website, and data sheet do not match as to what accessories are supplied with the unit, which caused us to delay our review (see our evaluation of the packaging and shipments at the beginning of this review).
The overall DC source accuracy is approximately .020%, and the current readback resolution is phenomenal, with 10fA (thats Femtoamps, as in 1x10-15 amps, as in .000 000 000 000 010 amps) resolution in the lowest range! I had to look Femto up to make sure I had the right number of zeros in there. However, with a list price of ~$5500 USD, it is expensive. You are paying for Keithley’s unique experience with generating and measuring low currents. The rear triaxial connectors are unique to low current applications, and I did not have a change (or the parts) to test them. If you need it, you gotta have it!
Very high precision, high accuracy DMM and Power Supply in one integrated package, LAN, IEEE488 and USB connectivity, great touchscreen GUI, wide voltage range (up to +200 volts source AND sink), ability to create graphics and statistics on screen.
Price (~$5500 USD), documentation does not match, sometimes confusing menu structure (multiple places to adjust the same parameter), interlock (should have a user front panel defeat). Additionally, I was able to lock up the unit several times, probably due to too much 'button pushing'. The graphing feature is nice, but it does not allow enough customization: the x axis does its own thing, and the user does not have any control over it. I suspect anybody who needed graphing would logically just hook a PC to it and bring the data in to view in either open source or proprietary graphing software.
As a programmer, I was most interested in in writing programs to do automation with the Keithley 2450. The 2450 has a wide variety of options when it comes to programming. It can connect with either Ethernet, USB, or GPIB and communicate using with the Standard Commands for Programmable Instruments (SCPI), or the Test Script Processor (TSP). After looking at the manual, I found that the default protocol is SCPI and decided to run my experiment using that.
For connection computer communication I decided on standard ethernet even though we also had the instrument connected to our GPIB network. By default, the 2450 will obtain an IP address with DHCP. To find the IP address you can either navigate the menus on the device, obtain from your router or DHCP server by knowing the MAC address, or by using the Keithley LXI discovery tool (free download). I chose the last option.
Once you have the IP address, you can enter it into a web browser to verify connectivity and view the device status.
In order to communicate with the 2450 using SCPI you can simply use Telnet or a similar program to connect to port 23 of the device’s IP address. To test, I used the free terminal client putty (link)
After telling in there is no real feedback of saying whether the connection was made successfully. Also, typing in generic commands like `help` provide no real feedback. Apparently, SCIP is not a very verbose protocol. However, there are some commands you can type that will give you feedback such as:
In the manual, I came across an example that uses two 2450 to do MOSFET curve tracing. Although I don’t have a second Keithley, I figured I could do a similar experiment on a NPN BJT transistor using a power supply and a current meter.
For review a BJT will take a current across the base and emitter and amplify it by a certain factor across the collector and emitter. This is easily illustrated using the classic “transistor man” illustration.
Although the gain factor is not consistent at every voltage and in fact the relation is non-linear. To charicerarize how a particular transistor will respond. manufactures will provide IV curve traces in the datasheet.
Most of these charts will have a fixed current across the base and emitter and then sweep a voltage across the collector and emitter measuring the voltage. For my experiment I will use a power supply, 10k resistor and current meter to set that Base-Emitter current and then write a python script which will control the 2450 that is capable sweeping a voltage and measuring its current. I will also repeate this processes 4 time for different base voltages and combine the data into one graph. For software I am using Python2.7 on windows along with PySide and Matplotlib. PySide is providing the socket layer in which I use to connect to the 2450 over tcp port 22 and I use Matplotlib to render the graphs. Connecting to the Keithely is relatively easy as illustrated in the code below.
if __name__ == "__main__":
s = QtNetwork.QTcpSocket()
#Ready any junk
For the initial setup I want to source a voltage and measure current:
Here I am going to sweep across 51 data points between 0 and 5v:
s.write("SOUR:SWE:VOLT:LIN 0, 5, 51, 0.01\n")
After the the sweep is complete, I will read in both the voltage reading and the current readings:
s.write("TRAC:DATA? 1,51, \"defbuffer1\",SOUR\n")
source_str = s.readAll()
s.write("TRAC:DATA? 1,51, \"defbuffer1\",READ\n")
read_str = s.readAll()
Last, I take the comma separated values and plot the results, I also run through this entire process 4 time for different base currents:
volts = 
for n in source_str.split(','):
current = 
for n in read_str.split(','):
The complete code can by found at my GitHub repository at: https://github.com/visgence/keithley2450
Below are the results from the voltage sweep with 4 different base currents. The results look typical for a 2N2222 transistor.
Overall, I was very pleased with the versatile programming capabilities of 2450. There are still a lot more functions and modes for me to explore. I can see how this device is valuable for engineers designing automated test equipment. My only concern was the limited current rage. If I wanted to test large power transistors or MOSFETs I would need to be able to source at least 10amps. As a tool without computer automation the 2450 provides a nice touch color display that is easy to setup a wide range of various test.
Tracy, I am the Product Manager for the 2450 at Keithley Instruments. It was a pleasure to read your review and thank you for taking the time to review the product. I want to make contact with you directly to share some thoughts with you and get additional feedback. Would you kindly send me an email at email@example.com with your email address so I can correspond with you.
Thanks and enjoy the 2450.
Nice review but the current accuracy of the 2450 far surpasses the 34661A below about 100mA so you were really checking the 34661A current accuracy in that range. The 2450 only sources 1A as far as I know