Author Topic: Keysight B2980A series electrometer/picoammeter review/teardown  (Read 25988 times)

0 Members and 1 Guest are viewing this topic.

Offline TiNTopic starter

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Yes, it's time to catch some femtoamps  :-DMM.

In this thread we will teardown Keysight Technologies B2980A series instrument which include four models, from basic high-performance picoammeters (B2981A and B2983A) to fully featured electrometer equipped with integrated high-voltage source and high-impledance voltage input (B2985A and B2987A).



In this article we will take a look on top SKU - Model B2987A Electrometer/High-resistance meter.



To understand why high impedance of the voltmeter can be important, we can try to compare portable DMM and high-end DMM with electrometer, in role of the voltage measurement device. As a device under test we pick some high-impedance output source, such as light sensor or humidity sensor. For sake of this mind experiment we picked Zsource = 100 MOhm.



Values above are NOT complete accuracy specifications and not include all error contributors of the measurement setup, but give the basic idea how DUT loading with meters own input impedance can affect the measurement results.

Manuals references

To our misfortune there are no service or calibration manuals available for any of the B2980A series instruments.

B2980A series datasheet
B2980A User’s Guide, Edition 4, March 2016
B2980A Programming Guide, Edition 2, March 2016
B2980A SCPI Command Reference, Edition 3, March 2016
B2980A Configuration guide
B2980A Product Fact Sheet
AN: Reliable High-Resistance Measurements Using the B2985A/87A
AN: Low Current Semiconductor Measurements Using the B2980A Series Ammeter
AN: Measuring Insulating Material Resistivity Using the B2985A/87A and N1424
N1414A High Resistance Measurement Universal Adapter brief
N1420A Setup Integrity Checker Function brief
Ion Beam Current Measurement - Product Fact Sheet
AN: Wide Range of Resistance Measurement Solutions from μOhm to POhm
AN: Capacitance Leakage Current Measurement Techniques Using the B2985A/87A
AN: Real-time Noise Monitor Function in B2980A
N1424, N1425, N1427, N1428 Accessories for Keysight B2985A/B2987A

Exterior

Instrument designed around common half-rack format, with large display and programmable front panel buttons. All inputs and connection ports are located on the instrument rear side. Optional battery is installed in the special bay located on the top side. Front and rear plastic bezel frame are protected from damage by rather massive rubber boots. Boots and adjustable handle can be easily removed for rackmount installation.   



This electrometer is 100% silent, as it does not contain any fans. Fans generate vibration which can cause big problem for sensitive instrument like this. Vibration would translate into electrical noise by the triboelectric effects in cables and piezoelectric effects in ceramic-based components like surface-mount capacitors. The triboelectric charge is generated when friction stress applied to both cable conductor and insulator, resulting in charge stored on the insulator surface. This effect can be seen when vibration coupled into cable, and as a result, it is likely that there are many risks to suffer from this noise. Any cable on test setup and interconnect can act as the noise source due to triboelectrics.

   

Since this meter runs some sort of RTOS(Real-time Operation System), it takes about 53 seconds to boot up before it's ready for operation. It's not important for typical workflow, as you need few hours warm-up time to reach specified accuracy but some engineers do like faster boot-up time for some reason. Front panel sports full-color TFT panel without touchscreen, falling for modern trend with large LCD on everything, rather than expensive and fragile high-contrast VFD displays. Of course both display types have strong and weak spots, but contrast and visibility angles of the VFDs are not possible to beat with TFT panel. OLED provides great contrast too, but many of them fall by low life period. It is not uncommon for expensive test equipment to be in use for decades without powering off, so display life is one of the parameters to consider.

Also fonts used on display... Our apologies to Keysight engineers, but meter have at least five different fonts used in UI as they were not sure which one looks better. Why, Keysight?

Courier New font in my opinion not the best font for relatively small LCD display either. Modern Keithley LCD GUI design in that respect much more pleasing, even though overall colorful TFT LCD displays in professional instruments more of a gimmick, than necessary functionality. Also LCD do always come out ugly on photographs, just a reviewer note.

Overall UI performance is rather underwhelming, it does not take lot of time to notice lags and lot of interface quirks during standard operation from front panel. Yes, yes, this instrument supposed to be put in the rack, connected over GPIB/LXI and never seen again, but it's fair to expect wee bit more effort put into $11000 instrument..



Rear panel have lot of various interfaces and connectors:

* Voltage source HI/LO shrouded 4mm banana ports
* Analog output HI/LO shrouded 4mm banana ports
* Ammeter gold-plated 3-lug triaxial input (2 pA to 20 mA ranges)
* Voltmeter gold-plated 3-lug triaxial input (2 and 20V ranges)
* GPIB remote control port
* USB Device remote control port
* LAN remote control port with LXI support
* Trigger In and Out BNC ports for precise timing
* Digital I/O to interface DUT fixtures
* Interlock input for DUT fixture, to safely limit voltage source below 21V during DUT replacements
* Humidity digital interface port, designed to support E+E EE07 probe
* Type-K miniature thermocouple input with CJC(Cold Junction Compensation) sense
* Common point for chassis connection 



Those triaxial inputs look so pretty with surface gold plating. We also see trigger BNC ports to allow syncronisation of the meter with another experiment equipment for precise timing. Also glad to see GPIB is still present on this modern instruments to allow easy integration within older instrumentation.



Use for triaxial inputs is essential for sensitive instrument like this to prevent leakages of the current between input signal terminal and outer shield/common return. We will cover more on this topic further in the article during the circuit design review. Operation of both current meter and voltage source is decoupled by own ON/OFF buttons to allow flexible operation and easy zero offset corrections.



Design and internal design review


-1200 and +1200VDC voltages present inside of this electrometer when operation are dangerous and even lethal. Pay attention of what you are doing, and keep all connections nice and tidy. Use hands-free probing technique to ensure safety. Use tools rated for CAT III operation while servicing high-voltage live circuits in this instrument.

After removing rear bezel and metal cover with help of typical HP size Torx screwdriver internal boards are available for access. Top side reveals high-voltage output assembly, mains AC-DC switching power supply brick, battery charger PCBA and various interconnects. This is modern instrument, so majority of the components are mounted by SMT, as expected. Everything inside is organized and tidy, as you would expect from $11K professional instrument.



Cable harness carefully fixed to the chassis parts, so no loose wires running around. Top board in the back of the unit is high-voltage output source. B2981A and B2983A units will not have this board present.



Bottom side reveals back side of main acquisition board, small digital brain SOC module and multiple interconnect flat ribbon cables.



Analog board have fully shielded areas, enclused from both sides, but Image above have shield removed for clarity. Electrometer input circuitry is extremely sensitive and must be kept clean to maintain instrument's performance, so use of proper gloves to prevent board contamination is mandatory here.

Mains AC-DC power supply

Input power supply is high-quality switching module is 50W 24VDC output, made by TDK-Lambda, model ZWS50BAF-24. It retails for $80 USD, and provides output current up to 2.1A from the universal input 85-264 VAC mains. Output noise and ripple maximum specification is 150 mV pk-pk so it is not low noise supply. It's not a problem as this PSU output will be regulated and isolated from any sensitive circuitry by onboard POL(point-of-load) regulators on main boards.



Also front panel board is visible behind the power supply. We will not be taking that part off, as there is nothing interesting on the front panel. Just Xilinx Spartan 6 FPGA with DRAM chip to handle color TFT interfacing and GUI refresh, keyboard readout and related DC-DC power for digital domain of the instrument. Most likely this front panel board reuse design from other instruments in series as well.



Back to power supply module, mains input arrive to the instrument from the right side connector, pass onboard line fuse, go thru passive CMC and protection components, rectified to high voltage DC by a bridge and storage capacitors and then AC/AC controller and power FET provide drive to a switching pulse transformer. AC/DC module have top tier Chemicon KXJ +105 °C rated capacitor on primary side and pair of Chemicon LXZ low-ESR +105 °C capacitors on the output side. No cheap capacitors of unknown original, nothing to complain about here.



There is little adjustable trim-pot resistor near the output connector. Perhaps it's available to adjust output voltage, if needed.



Power supply module was manufactured in Malaysia around week 49, year 2015. Bottom side is packed with SMT components, there is room even for one tiny SOT-363 device. Most of surface mount passives are 1206 and 0603 size. Two optocouplers provide feedback signal to the high side for output regulation and protection.

Interesting detail visible on the mains cable to the power supply. Line wire before entering PSU have small current transformer, with secondary connected to front panel PCBA. Perhaps this is used to detect the mains grid frequency for automatic PLC timing setting? I see no obvious reason why meter would possible need measurement of it's own input power otherwise.

Charger board B2981-26603 for optional battery pack (only for B2983A and B2987A)

Looking at the charger PCB it become also obvious why main AC/DC supply is selected to provide +24VDC output. As we know, battery nominal voltage is 14.4V, so charger circuit should receive higher voltage for good regulation margin. Based on Rev.003 label Keysight engineers had few revisions to get this little charger module working properly before entering production in this unit. 



Board still have Agilent logo, and date code on Linear (today part of Analog Devices, sadly) LTC4100EG charger controller reveals manufacturing period around week 16, year 2014. LTC4100EG is the Smart Battery Charger Controller with digital SMBus telemetry support, wide 3.5 to 26V output voltage range and up to 4A of charging current capability. It is rather complex all in one chip with lot of functionality and protection available for safe charging and battery management.



Rear side of the board also have components populated. These include pair of ON Semi FDS5672 60V N-channel MOSFETs, three ON Semi FDD4685 -40V P-channel MOSFETs, Coilcraft inductor, some 1210-sized MLCC(metal-layer ceramic capacitor)s and two shunts for current monitoring. There is tiny DS201 green LED for charge indication and couple of test points, marked SDA and SCL. These are obvious related to SMBus interface and available for easy probing.

Keysight does offer battery as an option N1418A, however priced at $871 USD, that is almost five times over the original RRC2024 pack. Benchtop charger available from Keysight, option N1419A goes at $1081 USD, clearly targeted for enterprise/business market customers.

Original battery from OEM manufacturer that Keysight use for B2987A is Lithium-ion pack RRC2024. This battery have 6600 mAh capacity, rated for 14.4V and supports up to 4.6A charging current. It have internal cell 4-series 3-parallel (4s3p) topology. We have already ordered 3rd party battery to try in unit in standalone operation.



Battery and charger interface using blade-type connector, typical for power packs. There are five contacts to provide power and telemetry connection between the unit and the battery pack. According to the user's manual operation time on battery power for fully featured B2987A listed as mere 5 hours. Picoammeter B2983A without high-voltage source specified for 7 hours of battery life. This gives us some interesting math to guess. Meters power consumption = 95.0 WAh / 5 hours = 19 W. Two extra hours between the units means that high-voltage output option board takes about 5.4W of power alone. Would be interesting to know power savings possible if one would ditch power-hungry TFT front panel interface and its related FPGA logic? Hint to Keysight for NFP-version of the instrument for production test rigs :)



Interesting design note - charger board does not have rigid mount to the chassis and have small movement freedom. Two plastic holders keep the board in place, but allow flex to prevent contact damage and excessive stress during battery pack replacement/installation.

Main processor board / module P0500-63001

Heart of the digital sub-system is STMicro SPEAr eMPU (ARM Cortex-A9) with support of Xilinx Spartan-3E FPGA. Xilinx FPGA provides acceleration and additional I/O for the SPEAr SOC to help interfacing all those peripherals onboard of B2987A.

Processor board is not unique design only for this instrument, but shared module design. It is also used in other Keysight instruments like B2900A series SMU, Agilent/Keysight 532xx series frequency counters, arbitrary signal generators. Some Keysight 1000X/2000X/3000 X-series oscilloscopes also share similar CPU topology.

Small investigation work revealed that Keysight B2987A running Windows Embedded CE 6.0. One hardware hacker even discovered how to run legendary DOOM game on his Keysight DSOX1102G oscilloscope. Maybe next step is to try Quake on B2987A electrometer? ;)



This board use mechanical SO-DIMM type slot to interface with front panel PCBA. This allows easy development and reuse of the existing design in various instruments and not uncommon concept in world of mixed-signal instrumentation. Board have metal U-shape plate glued to the processor (main BGA chip in center) to provide additional surface area for better thermal performance.



We can also find 63-VFBGA (9 x 11 mm) package with marking NQ432 is Micron MT29F1G08ABADAH4-ITX:D NAND Memory with 1Gb (128M x 8) capacity and parallel interface. It stores OS and firmware code of the instrument. This NAND Flash is already EOL'd in February 2018 by the manufacturer, but reference price for it was about $3.5 USD a piece.



TI DS90C385AMT in TSSOP package is programmable LVDS Transmitter, designed to convert parallel 24-bit RGB bus from SPEAr processor into serial Flat Panel Display Link. This interface is quite similar to other modern display interfaces, but adapted for flat-panel interconnect.

Another Texas Instruments part - TPS73725 is 1 Amp LDO take care of memory bus termination voltage and related power delivery. Vitesse VSC8641XKO is low-power 1GbE MAC+PHY interface chip that handle LAN/LXI connectivity of the instrument.



84-FBGA (8 x 12.5 mm) D9LHR hiding under the heatsink plate is Micron MT47H64M16HR-3:H DDR2 SDRAM chip, also with 1Gbit capacity organized as 64M x 16 bit parallel bus. This lowest-grade 3ns RAM rated to run at 333 MHz (DDR-667 effective).

We can also find various support components, 32768 Hz XO for RTC timekeeping function and main power regulator, based around Texas Instruments TPS65023 PMIC. This regulator have three separate switching DC/DC regulators for SoC/FPGA power and three LDO regulators for auxilary power. There is even serial I2C interface for management and telemetry.

High-voltage output board B2981-66602 (only for B2985A and B2987A)

This assembly generate programmable voltage output in either polarity with two available ranges. With only 1 W of available power it is not high power, but suitable for many biasing applications and test conditions.







Interconnect for HV source is done via two 100 mil pitch 10 pin headers located on sides and going to main analog board. Digital control and programming performed thru FPC ribbon cable, close to digital isolators.



All active power devices have heatsinks mounted on them to help with thermal management. Unit does not have fan, so board location on top, right under the chassis vents definately helping here.



This PCBA is using 6 layers and have components populated on both sides. There are no special requirements for high leakage isolation so we don't see much guarding or slots here.

High voltage reed relays are manufactured by TPM Sanyu Electric Model UPM-16205YHL designed for 1000V switching voltage and insulation resistance minimum 100 GOhm. Relay contacts rated to switch currents up to 1 A, way over the specification of the B2987A HV output. Relay activation coil voltage is 5V, and this type is non-latching.



Galvanic isolation between high-voltage assembly and rest of the instrument provided by pair of Texas Instruments ISO7641FM 6kV[~pk~] low-power digital isolators. Each of these have four independent channels and rated for isolation at least 4200 VRMS. This is quite fast isolator with 7 ns typical propagation delay.



Pretty looking magnetic device is custom planar transformer. They got name from planar design made of metal windings etched in multilayer PCB and coupled magnetic core thru machined slots. This construction type of magnetics and conductors assembly have some clear advantages (and drawbacks too) over traditional wire-wound transformers. Advantages are compact footprint with low-height, low leakage inductance and excellent thermal characteristics. Drawbacks are capacitive coupling and more expensive manufacturing and design evaluation, however all these factors can vary a lot depending on application requirements and design.

This is still actively developed field as modern electronics demands even higher power density and better efficiency. There is interesting PhD thesis paper by Ziwei Ouyang on these devices type. Worthy read if you into DC/DC converters and power conversion design.



This board also have FPGA, Xilinx Spartan-6 LX, XC6SLX9. Used version have 9k logic elements, over 600 Kb of integrated memory, 16 DSP blocks and feature up to 102 configurable I/O pins in used TQFP-144 package.

Tiny QFN-14 chip near DACOUT as you may have guessed already is Texas Instruments DAC8832 voltage programming DAC. This model is low-noise 16-bit voltage output single-channel DAC, operating from external reference. It have good INL (1 LSB) and fast settling time about 1 μs, which is important for fast response. It is connected directly to FPGA I/O for fast interfacing. Texas Instruments TPS54040 next to L304 33nH inductor is 42V 0.5A step-down buck regulator with adjustable switching frequency. Larger TSSOP package is another TI part - TPS54295. Another synchronous step-down buck switcher with intergrated FET and two output channels. Inductors next to this regulator are tiny, almost same size as 0805 passive component package, but wider. U303 catch the eye too, it looks like tiny 8-pin DFN package but with 0805-size capacitor on top. It has to be some local LDO regulator.
« Last Edit: March 04, 2019, 07:04:35 am by TiN »
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 

Offline TiNTopic starter

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: Keysight B2980A series electrometer/picoammeter teardown
« Reply #1 on: March 04, 2019, 04:55:56 am »
Analog front-end assembly B2985-61101

Well, you didn't expect me stopping teardown just at the high-voltage source board and digital controller, right? Acknowledge neccessary warning, and move on to the more nerve-wrecking depths.



Time to look at the heart of the electrometer - main analog front-end and acquisition board, located on bottom side of the meter. After some careful fiddling with assemblies I was able to get this PCBA out for detailed inspection.



Rear side view already reveals electrically floating isolated areas, because this same PCB also have digital interfaces like GPIB, Network and USB. All sensitive stuff is protected by dual-sided metal shields. Lucky for us, these shields are not soldered directly to PCB, but held with spring latches. 



Electrometer triaxial inputs and thermocouple sensor input are located on "second shelf". Shielded low-noise high-insulation cables from triax inputs go directly inside shielded cans. No intermediate connectors or exposed contacts to avoid even tiny chance of the leakage.



Yellow label tells us assembly name, which is B2985-61001, Made in Malaysia. Probably lower end B2981A and B2983A have slightly different analog PCBA as those picoammeters do not have voltage input channel/high-voltage source assembly installed.



Here are visible interlock and digital humidity sensor inputs, using Phoenix terminal block 2.5mm pitch connectors, DSUB-type digital I/O connector and Omega miniature Thermocouple Type-K input. Maxim MAX31855 provide cold-junction compensated digital readout for the remote probe temperature.



This top board is using standard materials and have OSP surface treatment.



Can also spot another planar transformer on analog PCBA..



Input power to the DC/DC converter provided by pair of FETs from top area and output is rectified and filtered by circuits around CR1601-1604 bridge + L1602,L1603 inductors. 



There are bunch of Texas Instruments regulators sitting at the output and even 79M15BG LDO in DPAK package nearby. According to silkscreen labels these regulators provide +26, -26, +18, -18, +5, +3.3 and +1.8 VDC rails for onboard devices.


Whole regulator area located next to yet another Xilinx FPGA, this time Spartan-6 XC6SLX25 in BGA package. FPGA is paired with the Micron MT48LC16M16A2B4-6A 256Mbit SDRAM memory.



Pretty planar transformer have Agilent marking B2981-26610, and carry Rev.001 revision number. Wonder how many metal layers does it have?



Now time to caaaarefully remove shields and reveal the secrets.



Bottom side (facing outside of the instruments) give up easily, but we don't see much here. Except beautifully guarded and slotted areas around sensitive nodes.



Footprint in the center looks like SMA connector?



Top side in it's glory. That was not an SMA connector, but special metal-packaged capacitor! RefDes C1171 and 3pF label gives it away. Current input cable connection reveals graphite coating around center conductor and guard shield connected directly to PCBA guard plane. Switching for input current ranges done by rather usual Panasonic AGN210S4H relays in SMT-package. I expected some special reed relays like COTO, however I'm sure Keysight know what they are doing and extensive guarding (all that exposed gold-plated plane) helps to keep stray leakage in check. 

Image below shows front-end area of direct competitor for B2987A unit, which is Keithley 6517A electrometer.



At xDevs.com we have fixed few 6517A's before, so there is photograph of it's front-end for comparison. Keithley opted for LMC6001 as input amplifier. Largest feedback resistors used in Keithley meter are thin-film 4 TOhm and 1 TOhm.



Main voltage reference is typical LM399, selected by Keysight for their specifications. Chip is relabelled with custom Keysight partnumber 1826-1249-5. Reference manufactured around week 19 in 2015.



PCBA surface contamination control could be better, but I can't blaim Keysight on this, since instrument was acquired already used at fraction of the price. Macro photograph reveals some dust particles on random parts around. Areas under shield covers were good and clean. Would be a mistake to do any unnecessary cleaning with liquids in non-sensitive areas of the assembly here, risking further contamination.



Little TSSOP chip marked U604 in the bottom left corner of the photo is Analog Devices AD7767BRUZ. This is 24 bit high-performance SAR A/D converter, capable for speeds up to 128 kSPS. DC performance is not bad, sporting maximum +/-7.6 ppm INL and gain drift 0.4 ppm/°C. Key parameter of this ADC would be large dynamic range, about 110 dB, depends on conditions.



It is not too often that we see SAR ADC used as main detector in instrument designed to provide 6.5 digits of resolution.

But it's not a surprise, because of B2987A's ability to digitize inputs with 20 us intervals to internal memory. It can provide up to 12500 readings/s to the GPIB interface.



Guard shielding metal is all that exposed gold-plated area you see under the covered area. Surface leakage would be cought by exposed metal and prevent the sensitive signal loading and coupling. Guarding also reduce parasitic capacitance, because there is also no charge to accumulate between equal potentials.

Common and effective way to reduce the leakage current is to use the guarding technique. Enclosing the entire electrical signal path connected to the ammeter by the conductive material, but biased at the identical voltage level by additional low-impedance source, equal in potential to ammeter input terminal called guarding. Guarding also implemented in special cables between instrument input and device under test, by using triaxial low-noise cables, which have not one but two inner shields, separated by different dielectric layers. Inner shield in triaxial cable biased by guard voltage, removing leakage to center signal conductor, while outer shield is connected to common or ammeter Lo.



This concept may sound easy, but actual hardware design and component selection for electrometer front-end requires very careful design. Working with femtoamp level signals from the input triax port requires full guarding and isolation from any signaling or power nodes. Any stray currents from the instrument's circuitry, digital control signals or power regulators that may leak into the input node will cause errors the measurement. Running input signals across PCBA make it even more difficult as leakage in board material is not easy to characterize. Would not be surprising if Keysight carefully tested and selected special material for the outer dielectric layers of this analog board to keep PCB leakage low.

Other technique to mitigate PCB leakage is to use PTFE standoffs, however it's only viable for thru-hole components and makes circuit board much larger physically (also larger area for guarding against leakage). This method was often used back in the days of thru-hole assembly decades ago, but as this special case demonstrate old tricks can be very useful even today.



Older HP/Agilent/Keysight instruments are also known to implement manually-assembled air wiring design, for example in Agilent 41421B Source-Measure Unit module designed for Agilent 4142B semiconductor parametric analyzer.



Photograph of exposed 41421B SMU front-end reveals isolated assembly with PTFE standoffs, pressed into the standard PCB board.

The instrument's exposure to changing environment in both temperature, humidity and pressure affect performance as well. Humidity and temperature both have major impact on dielectric surface resistance. To provide confidence in results it is very useful to track and sync environment monitoring data at same time with measured currents/voltages. That is why Keysight B2980A series, well as Keithley 6517 series have thermocouple and humidity sensor inputs, so related data can be captured by same instrument in the lab with timestamps, correlated to measurements.



Vishay TR10X high-voltage thick-film resistors are used in input opamp feedback path, selected according to used current range. Highest value is 500 GOhm with 5% tolerance, rated working voltage 10 kV. Temperature coefficient of these high-value resistors rarely good, confirmed by specification 500 ppm/°C. Usually that is not a problem as temperature correction can be applied or calibrated out, however voltage coefficient (change of resistance under applied voltage) is more important specification in this case. And sure enough, these TR10X resistors are specified for better than 1 ppm/V VCR.



Pretty looking hermetic gold-plated reference 3.3pF capacitor used in integrator. Together with feedback resistor this network set bandwidth limit of the current input front-end. 500 GOhm feedback resistor at most sensitive 2 picoamp range together with 3.3pF capacitor have bandwidth limited at 1.7 seconds (DC to 0.588 Hz). In typical configuration with oversampling meter produce useable data each 16 seconds. Next higher 20 picoamp range use 10 times smaller resistor that translate bandwidth into magnitude faster response (DC to 5.88 Hz). This idea follows for higher ranges, with fastest response of 100 us on milliamp ranges.



Here we can also see that this board marked as Hi-Z suggesting special treatment requirements during assembly/factory testing process.

Voltage meter input have separate guarded can on top PCBA surface. It is obvious because voltage and current inputs are separate in B2987A and can be driven to different guard levels, according to measured signal potentials.



Voltage input front end have Texas Instruments LMC662 dual opamp with ultra-low input bias and offset current (maximum 2 pA, typical 2 fA!)



Core of the voltage measurement function in B2987A is Analog Devices AD7767BRUZ 24-bit 32/64/128 kSPS SAR ADC. It has +/-7.6 ppm maximum INL specification and able to provide over 100 dB of dynamic range, when used with 5 VDC voltage reference. And that is exactly what is present onboard next to ADC chip, in shape of Analog Devices XFET ADR445. This reference designed to provide 5.000 VDC with initial accuracy +/-2 mV and temperature stability 3 ppm/°C. It's low noise reference, with 2.25 uVpk-pk typical over 0.1-10 Hz bandwidth. Pair of ADG413BRUZ used to for inputs multiplexing and range switching. Voltage measurement function in electrometer have just two ranges, +/-2 V and +/- 20 V. Based on brief circuit analysis base range is 2 V, while higher range use x10 attenuation by front-end before ADC.

There are also three Analog Devices OP1177ARMZ amplifiers

Datasheets for all these components available for detailed study:

* AD7767 24-bit SAR ADC Data Sheet, Rev.C
* Analog Devices XFET ADR440-ADR445 series voltage reference datasheet
* ADG411/412/413 series Precision Quad SPST Switches datasheet
* OP1177/2177/4177 Precision Low Noise, Low Ib Opamp Datasheet





There you have it, Keysight B2987A Electrometer/High-resistance meter in it's beautiful construction and design.

Battery installation and initial checks

Before diving into performance and experiments, let's install some compatible battery. This enables field operation of the B2987A, hopefully also allow to evaluate noise performance with and without mains power, if there is any visible difference.



As result got two battery packs, one from Inspired Energy, model NL2024HD and second is RESPIRONICS REF900-102. Both packs have same configuration, form-factor and can be installed in B2987A.

 

After installation unit can charge battery normally and also report the battery charge cycle correctly:



Firmware and software

Firmware revision as received : 2.2.1745.7190
CPU FPGA version: 03.29
Module Revision : FPGA, (13.88, 23.29)
License: None
System uptime: 3 day 3 hour 39 min 15 sec
Battery Cycle count: 0 (no battery installed)



Firmware blobs are unified and available for all B2980A series instruments for user upgrade.

Firmware revisionReleased binary
B2980A series firmware 2.2.1834.5220 (latest) ZIP-archive
B2980A series firmware 2.2.1745.7190 ZIP-archive
B2980A series firmware 2.1.1645.5880 ZIP-archive
B2980A series firmware 2.0.1608.8800 ZIP-archive
B2980A series firmware 1.1.1527.1533 ZIP-archive
B2980A series firmware 1.1.1508.3400 ZIP-archive
B2980A series firmware 1.0.1430.6131 ZIP-archive

Upgrade process from version 2.2.1745.7190 to latest 2.2.1834.5220 was straight forward with random USB 2.0 flash drive, and took about 10 minutes.







Keysight also provides software to remotely control and interface with the instrument:

Released binary Description     
QuickIV ver.4.1.1821.3680 software suite Binary setup file

Web-interface

Nearly useless, thanks to obsoleted Java. None of my usual browsers (Opera, Firefox, Chrome) had Java support to run webapp. :)

Few more photos of teardown also available on my site.
« Last Edit: March 05, 2019, 12:14:49 pm by TiN »
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 
The following users thanked this post: Echo88, e61_phil, doktor pyta, serg-el

Offline TiNTopic starter

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: Keysight B2980A series electrometer/picoammeter teardown
« Reply #2 on: March 04, 2019, 04:58:33 am »
Part 3 with test results and measurements data  >:D
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 

Offline doktor pyta

  • Frequent Contributor
  • **
  • Posts: 488
  • Country: pl
Re: Keysight B2980A series electrometer/picoammeter review/teardown
« Reply #3 on: March 04, 2019, 08:14:53 am »
You made my day  :-+
TiN, thanks for Your effort and beautiful photos!

Offline HighVoltage

  • Super Contributor
  • ***
  • Posts: 5453
  • Country: de
Re: Keysight B2980A series electrometer/picoammeter review/teardown
« Reply #4 on: March 04, 2019, 08:39:42 am »
Nice teardown, Thank you.

Looking forward to your measurements.
The interface is a little awkward to get used to.

There are 3 kinds of people in this world, those who can count and those who can not.
 

Offline HighVoltage

  • Super Contributor
  • ***
  • Posts: 5453
  • Country: de
Re: Keysight B2980A series electrometer/picoammeter review/teardown
« Reply #5 on: March 04, 2019, 11:47:40 am »
My B2987A was a little older and came with Firmware 1.0
Updating to the latest FW v2.1.1645.5880 did not work and resulted in many errors.
I had to re-install FW v1 and update in sequence:
v1.1.1508
v1.1.1527
v2.0.1608
then finally v2.1.1645

After these steps, all was perfect.


There are 3 kinds of people in this world, those who can count and those who can not.
 
The following users thanked this post: TiN

Online Berni

  • Super Contributor
  • ***
  • Posts: 4922
  • Country: si
Re: Keysight B2980A series electrometer/picoammeter review/teardown
« Reply #6 on: March 04, 2019, 11:50:48 am »
Nice one, looking forward to the rest.

Im also curious about that PCB planar transformer.
 

Offline HighVoltage

  • Super Contributor
  • ***
  • Posts: 5453
  • Country: de
Re: Keysight B2980A series electrometer/picoammeter review/teardown
« Reply #7 on: March 04, 2019, 11:52:11 am »
Also, you need an interlock connector, that was not easy to get.
I got one from Keysight directly.

Keysight Part Numbers:
0360-3085
0360-3086


« Last Edit: March 04, 2019, 12:06:38 pm by HighVoltage »
There are 3 kinds of people in this world, those who can count and those who can not.
 
The following users thanked this post: TiN

Offline HighVoltage

  • Super Contributor
  • ***
  • Posts: 5453
  • Country: de
Re: Keysight B2980A series electrometer/picoammeter review/teardown
« Reply #8 on: March 04, 2019, 11:59:36 am »
And if you are looking for a humidity sensor I found the company E+E in Austria, that make this fitting sensor, called Model: EE07-PFT

When you order it, make sure to also order a fiting cable with connector.
In my first attempt I only go the sensor.
 
https://sensorpros.com/products/ee-elektronik-ee07-humidity-and-temperature-probe-with-digital-output?variant=7250484199458
« Last Edit: March 04, 2019, 12:06:21 pm by HighVoltage »
There are 3 kinds of people in this world, those who can count and those who can not.
 
The following users thanked this post: TiN

Offline TiNTopic starter

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: Keysight B2980A series electrometer/picoammeter review/teardown
« Reply #9 on: March 04, 2019, 12:18:01 pm »
interlock connector, that was not easy to get.

It's standard Phoenix block, got few from Digikey, P/N 277-1432-ND. $6.20 a pop. I was thinking about getting humidity sensor, but opted against it for time being, as I use BME280 on my GPIB-logging FPGA board already.
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 
The following users thanked this post: HighVoltage

Offline MadTux

  • Frequent Contributor
  • **
  • Posts: 785
Re: Keysight B2980A series electrometer/picoammeter review/teardown
« Reply #10 on: March 04, 2019, 12:36:28 pm »
Is that a loaner or company owned instrument or did you really pay $11k from private funds for this teardown?
 

Offline TiNTopic starter

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: Keysight B2980A series electrometer/picoammeter review/teardown
« Reply #11 on: March 04, 2019, 12:41:40 pm »
I only teardown stuff that I own. You don't want to teardown somebody else's device, break something (accidents happen even if you careful  :-BROKE) and get yourself into trouble.  :)
Got it from keysight ebay second-hand store. They still have few listed, but doubled in price since.  :-DD
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 
The following users thanked this post: MadTux

Offline e61_phil

  • Frequent Contributor
  • **
  • Posts: 962
  • Country: de
Re: Keysight B2980A series electrometer/picoammeter review/teardown
« Reply #12 on: March 04, 2019, 01:50:02 pm »
It would be interesting which value the highest feedback resistor for current measurement has. Many Keithley instruments I had a look into use quite low resistance values.
 

Offline TiNTopic starter

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: Keysight B2980A series electrometer/picoammeter review/teardown
« Reply #13 on: March 04, 2019, 02:02:26 pm »
Quote
Many Keithley instruments I had a look into use quite low resistance values.
You sure about that? 6517A I've repaired before have 1 Teraohm and 4 Teraohm resistors in front end  ;D.
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 

Offline e61_phil

  • Frequent Contributor
  • **
  • Posts: 962
  • Country: de
Re: Keysight B2980A series electrometer/picoammeter review/teardown
« Reply #14 on: March 04, 2019, 02:18:31 pm »
Quote
Many Keithley instruments I had a look into use quite low resistance values.
You sure about that? 6517A I've repaired before have 1 Teraohm and 4 Teraohm resistors in front end  ;D.

Yes, I'm sure about the ones I had a look into (old ones).

1T and 4T sounds good
 

Offline TiNTopic starter

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: Keysight B2980A series electrometer/picoammeter review/teardown
« Reply #15 on: March 05, 2019, 11:59:08 am »
Updated second post with analog board photos. Enjoy  ;).
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 

Offline HighVoltage

  • Super Contributor
  • ***
  • Posts: 5453
  • Country: de
Re: Keysight B2980A series electrometer/picoammeter review/teardown
« Reply #16 on: March 05, 2019, 12:59:10 pm »
Very impressive, your teardown pictures.

And more impressive, that you take such instrument so much apart.
I was happy, that my B2987A was working!
There are 3 kinds of people in this world, those who can count and those who can not.
 

Offline Echo88

  • Frequent Contributor
  • **
  • Posts: 821
  • Country: de
Re: Keysight B2980A series electrometer/picoammeter review/teardown
« Reply #17 on: March 05, 2019, 01:14:25 pm »
Impressive article, as always!  :-+
Amazing how they achieve that resolution with smd-components instead of air-wiring.
 

Offline TiNTopic starter

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: Keysight B2980A series electrometer/picoammeter review/teardown
« Reply #18 on: March 11, 2019, 08:08:03 am »
Thats my work lab where I do non-metrology EE  projects. But some gear like logic analyzer or upside down DPO is mine.

Vince is doing overclocking and pouring liquid nitrogen, while I do design work.
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 
The following users thanked this post: 3roomlab

Offline TiNTopic starter

  • Super Contributor
  • ***
  • Posts: 4543
  • Country: ua
    • xDevs.com
Re: Keysight B2980A series electrometer/picoammeter review/teardown
« Reply #19 on: March 11, 2019, 10:32:59 am »
Quote
you should tear down that 288 channel logic analyzer !
done while ago
YouTube | Metrology IRC Chat room | Let's share T&M documentation? Upload! No upload limits for firmwares, photos, files.
 
The following users thanked this post: 3roomlab

Offline denny_cheng

  • Newbie
  • Posts: 2
  • Country: cn
Re: Keysight B2980A series electrometer/picoammeter review/teardown
« Reply #20 on: March 31, 2020, 01:43:26 am »
what  interface does it use between the diffrent board?SPI?
 

Offline Kirill V.

  • Regular Contributor
  • *
  • Posts: 118
  • Country: ru
Re: Keysight B2980A series electrometer/picoammeter review/teardown
« Reply #21 on: April 01, 2020, 01:17:44 pm »
Strange device. They proudly claim that this electrometer can measure 10 aA, but the offset specified in the accuracy specification is 3 fA, which completely clogs all three lower counts.
Also, there is no external amplifier and the source is connected only with a cable, which makes it almost pointless to measure attoampere currents.
And maybe I personally do not understand something, but VNIIFTRI (all-Russian research Institute of physical, technical and radio engineering measurements) has certified this meter and indicated that the range of MEASURED currents begins only from 1 fA. In other words, only in this range does the tool provide the accuracy stated in the specification. The three lower counts don't matter. Of course, the range of DISPLAYED currents starts at 1 aA, but this cannot be used for measurements because accuracy is not guaranteed
 
The following users thanked this post: bsw_m

Offline guenthert

  • Frequent Contributor
  • **
  • Posts: 706
  • Country: de
Re: Keysight B2980A series electrometer/picoammeter review/teardown
« Reply #22 on: April 01, 2020, 05:44:02 pm »
[..] Of course, the range of DISPLAYED currents starts at 1 aA, but this cannot be used for measurements because accuracy is not guaranteed
  What?  There is value in resolution even without accuracy.  It would only be meaningless, if it wouldn't be stable during the time of interest.
 

Offline Kirill V.

  • Regular Contributor
  • *
  • Posts: 118
  • Country: ru
Re: Keysight B2980A series electrometer/picoammeter review/teardown
« Reply #23 on: April 01, 2020, 07:20:42 pm »
How is it useful to use the resolution of this electrometer? I'm interested in a specific example where you really need a resolution of 10 aA without accuracy
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 14073
  • Country: de
Re: Keysight B2980A series electrometer/picoammeter review/teardown
« Reply #24 on: April 01, 2020, 08:19:26 pm »
The main problem with the low currents is some background  / leakage current that can slowly change with time / temperature. This limits the accuracy for the very small currents.

For some experiments one can disable / enable the current one is interested, so use the electrometer for low frequency AC, not true DC. In these cases an offset is not so bad. The tricky part could however be to measure such tiny currents fast.

In times of computers reading the numbers anyway it even helps to have a little more numerical resolution than accuracy and  noise limit. This way one can avoid the quantization errors and the PC usually does not mind storing 1 more digit. Things have kind of changes from do not write down the non significant digits to hey I want some 1-2 digits more than significant. With multiple ranges this may lead to a digit that is not really needed / useful.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf