Keithley 610C vs. Op Amp Current to Voltage Converters

[ocw]

Having just completed building a couple of op amp picoamp to voltage converters and then seeing a Keithley 610C Electrometer and its picoamp and below measurement capabilities available at a good price, I couldn't pass up the Keithley unit.  My first interest was in comparing the accuracy of them.

I originally built a current to voltage converter using an AD8067 op amp.  I chose that IC to add low RF current measurement capabilities in addition to DC.  I then built a similar circuit using an ADA4530-1 to improve the DC performance at lower currents.

The Keithley 610C is an old analog design which still a good circuit.  Besides it front panel meter, it has either a 3 VDC or 1 mA output which can be connected to a digital meter for greater resolution with the ability to graph and store the measurements.  Attached is a picture of the 610C with a test box which I have connected to its input.  Inside of the shielded box is a 1.5 volt battery which is connected to a 1T ohm (1,000G) series resistor which produces the displayed current of about 1.5 pA.

The 610C has two methods of current measurement.  The first is the usual way of measuring the voltage drop across a resistor.  I found that the series resistor used can significantly reduce the current and displayed reading due to its size (up to 100G ohm).  Currents which are dependent on the voltage level present were typically around 8% low.

Fortunately the 610C has a second "FAST" current measurement mode which uses feedback and a component level transimpedance amplifier which has a low shunt resistance having a burden voltage under 100 uV.  The FAST mode only works for currents under about 10 uA.  With the meter set for currents larger than that the meter can't be zeroed so a return to the normal mode is required.  On the attached spreadsheet I document the accuracy difference between the two modes.  For the normal mode I take the added meter shunt resistance into consideration and also show the accuracy when that resistance is taken into account.  Keep in mind that all currents are in the units shown in the light blue column.

My 610C tests point to its 10G shunt resistor as having marginal accuracy.  Most of its shunt resistors have 1% accuracy while the accuracy of the 1G, 10G and 100G shunt resistors isn't listed.  It may be 5% for that 10G resistor.  Those three giga-ohm resistors are the RX-1 model resistors which current have a price at Mouser for around $40 each.

My ADA4530-1 circuit had excellent accuracy with good ability to measure low currents.  I made the pictured test box to aid in low current measurement with limited influence by external noise.  Having just constructed that I have yet to determine its lower limit.  It has insignificant shunt resistance and burden voltage.  Its circuit is similar to the one shown on the ADA4530-1 evaluation board data sheets, see:
http://www.analog.com/media/en/technical-documentation/user-guides/ADA4530-1R-EBZ_UG-865.pdf
There is also good information on the IC's regular data sheets.  See:
http://www.analog.com/media/en/technical-documentation/data-sheets/ADA4530-1.pdf
I haven't yet seen significant lack of linearity which Analog Devices discusses with the resistors which I used.  My ADA4530-1 is also used in the transimpedance mode and has selectable 1k, 10k, 100k, 1M, 10M, 100M, 1G, 10G, 100G and 1T feedback resistors.  I have yet to figure out how best to label its 12 switch positions on its small enclosure.  The two extra switch positions include a feedback capacitor in parallel with two of the mentioned feedback resistors.  The feedback resistors in it have 0.01% accuracy for 1M and below resistance, falling to 1% worse case for all resistors except for a 10% 1T resistor.

The AD8067 circuit was more of a trial circuit.  Its range switch is hidden inside.  It has trim pot calibration for the 1k, 1M and 1G feedback resistor positions.  The 1G feedback resistor is the largest one in it.  I'm not sure that it would benefit with a larger one installed.  I did try using a 1,000 volt supply with a 1G resistor in series to the input of that circuit to extend its range.  I did that after trying the same thing using a LTC1150 circuit that I quickly put together.  I thought that I'd sacrifice a CMOS IC as a trial at that voltage.  However, both IC's survived!  I have yet to get up enough nerve to try that with my ADA4530-1 circuit.  But, I need to get some 1P ohm resistors to test that IC with 1,000 volts.  With the ADA4530-1's 1T feedback resistor switched on, 1P in series with 1,000 volts should have a 1 volt circuit output.  I've found the resistors to order:
http://www.guildline.com/media/k2/attachments/Guildline9336Datasheet.pdf
I'd go with the 10P resistors, but I couldn't handle their 30% accuracy.

There are pictures of the two IC circuits on the attached spreadsheet.  I still need to do complete testing of some of the other 610C functions.

[ocw]

Seeing a little higher inaccuracies on my Keithley 610C on the ranges using the highest two value resistors (10G & 100G) I thought that I'd replace them with 1% MOX-750231008FE and MOX1125231009FE resistors.  They were harder to get to than I expected.  Removing the function/range switch wasn't easy.  Replacing the resistors looked even worse!

I saw a film covering the glass sealed resistors that I'd be replacing.  That's not surprising given the equipment's age.  I thought that I'd settle with cleaning the surface of those resistors.  That may improve the meter's accuracy.  The largest resistor (100G) was made before gigaohms became more common.  It is labeled as 100KM and is a Dale M51-1009G 2% resistor.  The other two glass sealed resistors are 1G and 10G Victoreen resistors (labeled as 1000 and 10000 megaohms).  The largest resistor which was not glass sealed was a Dale M14H 100M 1% resistor.

Due to the switch mounting the glass sealed resistors are hard to see.  Due to that I thought that I should take some pictures of them.  Attached are four pictures of them.  You can see all of them in each, but the different shots have them moved for better individual viewing.