EEVblog® Electronics Community Forum
Electronics => Metrology => Topic started by: Overspeed on September 20, 2025, 06:25:15 pm
-
Hello
I my quest to locate a DIY solution for DC nanovoltamplifier the Keithley 103 looks interesting not regarding the age of this amplifier from the 70's but on the schematic as 103 use a stack of x 10 amplfier
But in 2025 there is 2000 gain very low noise amplifier as the AD8428 are available or LT1028 x 3 as diff amp
So does there is still a gain or an possible improvement to stack x10 amplifier or just go with last generation high gain low noise amplifier IC in // or even use a SSM2220 pair as MAT03 are obsolete with a ultralow noise opamp ??
Regards
OS
-
It looks like this is about an AC amplifier. DC nV range voltage is a quite different story.
Unless one has a super low impedance source one also has to look at current noise and the product of voltage noise and current noise is good performance parameter. The ready made instrumentation amplifiers can simplify the internal circuit a little and this way may have better noise performance.
One can stack multiple stages for more gain. There is no need to have all the gain in one stage. This is especially the case for higher bandwidth.
Some of the INAs get lower noise with high gain. So more than 10 x gain may be a good idea.
One needs to be a bit careful with decompensated OP-amps like the LT1028 in the 3 OP-amp INA circuit. The common mode gain is less and the amplifier may tend to oscillate in common mode mode.
-
It looks like this is about an AC amplifier. DC nV range voltage is a quite different story.
Unless one has a super low impedance source one also has to look at current noise and the product of voltage noise and current noise is good performance parameter. The ready made instrumentation amplifiers can simplify the internal circuit a little and this way may have better noise performance.
One can stack multiple stages for more gain. There is no need to have all the gain in one stage. This is especially the case for higher bandwidth.
Some of the INAs get lower noise with high gain. So more than 10 x gain may be a good idea.
One needs to be a bit careful with decompensated OP-amps like the LT1028 in the 3 OP-amp INA circuit. The common mode gain is less and the amplifier may tend to oscillate in common mode mode.
Hello
Kleinstein , thanks
The case of the Keithley 103 was to illustrate the architecture x10 + x10 ... not to use it as a base design , Keithley 1801 is more close to my need or better the A10 already built and proof tested at a more reasonable price
Xdev have made impressive works on 1801 and EM10 https://xdevs.com/review/ema10/
My need is DC only , that to amplify very low voltage level in a 4 wire resistor setup
As INA do you have a proposal ? INA103 is on the low side regarding performances but perhaps an AD8428 in //
That strange there is a lot of schematic and article on AC amplifier but mostly nothing on DC ,
Agilent 34420 schematic show a pair of JFET one op amp nothing which look as a top of the range op amp
I link an interesting setup by using two diff amp
Regards
OS
-
For nV level DC an instrumentation amplifier would not directly help. It may be useful further down after some gain, but not for the initial part.
For the choice it helps to know the source resistance and requited noise / voltage level.
A DC amplifier is different as it needs some kind of chopping / AZ switching to avoid DC offsets and drift. Another point is that AC can use a transformer to match the impedance.
-
For nV level DC an instrumentation amplifier would not directly help. It may be useful further down after some gain, but not for the initial part.
For the choice it helps to know the source resistance and requited noise / voltage level.
A DC amplifier is different as it needs some kind of chopping / AZ switching to avoid DC offsets and drift. Another point is that AC can use a transformer to match the impedance.
Hello
The amplifier will use for resitivity measurement , the metal part ( 4 probes setup) have a very low resitance , voltage is in the range of 250 nV to 2 uV perhaps more if I succeed to increase the current .
Do you think this setup as linked is possible ?
Regards
OS
-
Hi OS,
can the stimulus made AC?
You could use a bandpassed- or a lock-in-amplifier.
Good luck
-
For a resistance measurement one could actually change the excitation to low frequency AC and than use an AC amplifier. This would simplify things quite a bit and get better sensitivity. Instead of a constant current source one may also measure the voltage on a ref. resistor in series. Even a ratgher low frequency in the 5 Hz range may be easier than DC.
AC excitation would also avoid possible issues with thermal EMF at the DUT.
-
Offset compensated DMM ohms measurement effectively uses a single AC cycle (either positive and zero current or positive/negative) for this reason - you can null out thermal errors.
-
Offset compensated DMM ohms measurement effectively uses a single AC cycle (either positive and zero current or positive/negative) for this reason - you can null out thermal errors.
Hello
I use pulsed current or polarity inverting , I have spend months to reach ppm accuracy DC and I still work on ... I have make some measurement with my INA circuit , I need now a portable dedicated circuit linked with my Ioki 7.5 digit or even a 24 bits DAC .
I agree that AC current can solve problem but not solve others
Regards
OS
-
I built an nV amplifier using the LT1028 but its performance was poor, likely due to excessive 1/f noise.
Later, l tested the RM3545 (which is based on the ADA4528-1) as an nV meter, and its performance was much better than the previous one.
So, l put together a simple x1000 amplifier with 4 parallel ADA4528-2(1 in use), and the measurement results were satisfactory.
Therefore, l plan to continue with this approach and build a 16-parallel-channel nV amplifier, mainly for low-impedancesignal sources.
-
I built an nV amplifier using the LT1028 but its performance was poor, likely due to excessive 1/f noise.
The LT1028 already has low 1/f noise.
My differential low noise amplifier used a pair of LT1028s in the common 3 resistor x1000 differential amplifier configuration, with an LTC1151 dual high voltage chopper stabilized amplifier suppressing flicker noise and drift below about 1 Hz. This design was an extension of Jim William's single ended design to fully differential operation. The frequency breakpoint between the LT1028 and LTC1151 was tuned by monitoring output noise with a high resolution DC multimeter. When I did it, the frequency breakpoint exactly matched the crossover point between the LT1028 and LTC1151 noise curves, like it should.
A design for higher input impedance would use low noise JFET operational amplifiers in place of the LT1028s, or low noise JFETs from a company like Linear Systems. I have heard of designs using small vertical power MOSFETs, but never seen any test data for them.
My design relies on the offset null capability of the operational amplifiers, so many modern low noise parts like the OPA140 are not suitable.
-
For an offset correct ohms measurement or with true AC excitation there are ready made low noise INAs like the AD8428/AD8429.
These are lower noise than the classic 3 OP-amp amp solution with LT1028 that has the additional stability issues (marginally stable for CM mode mode).
There is little need to go really super low in the frequency, as with low resistance ( << 1K) the settling is usually fast. So the 1/f noise should not be that critical. How low one needs to go depends a bit on the ADC: some SD ADCs need extra settling time and thus don't work that well for less than some 100 ms. ADCs with low latency (SINC1 filter) should also work with a little more than 1 power line cycle per phase (e.g. 22 ms).
-
Hello
Thanks for your answer , I will try the solution proposed by Zylmex by using two ADA4528
composite ampfier looks more complex for my electronic knowledge
Regards
OS
-
Analog Devices application note AN-245 discusses how instrumentation amplifiers can be used in fully differential applications to improve signal-to-noise over a single ended design.
-
Analog Devices application note AN-245 discusses how instrumentation amplifiers can be used in fully differential applications to improve signal-to-noise over a single ended design.
Hello
Thanks , I have located the AN21 1986 Jim Williams , with schematic that more easier for me to understand .
Regards
OS
-
Analog Devices application note AN-245 discusses how instrumentation amplifiers can be used in fully differential applications to improve signal-to-noise over a single ended design.
Thanks , I have located the AN21 1986 Jim Williams , with schematic that more easier for me to understand.
Which schematic in Linear Technology application note AN21?
My differential low noise amplifier was based on figure 14, but extended to differential operation. Jim Williams published other examples of that type of composite amplifier.
-
Analog Devices application note AN-245 discusses how instrumentation amplifiers can be used in fully differential applications to improve signal-to-noise over a single ended design.
Thanks , I have located the AN21 1986 Jim Williams , with schematic that more easier for me to understand.
Which schematic in Linear Technology application note AN21?
My differential low noise amplifier was based on figure 14, but extended to differential operation. Jim Williams published other examples of that type of composite amplifier.
Hello
Yes the schematic 14 , even as single ended that can a nice experimentent without to break the bank account , other advantage as that a 10 V Vs no worrie regarding common voltage ( I think )
Regards
OS
-
Thanks , I have located the AN21 1986 Jim Williams , with schematic that more easier for me to understand.
Which schematic in Linear Technology application note AN21?
My differential low noise amplifier was based on figure 14, but extended to differential operation. Jim Williams published other examples of that type of composite amplifier.
Yes the schematic 14 , even as single ended that can a nice experimentent without to break the bank account , other advantage as that a 10 V Vs no worrie regarding common voltage ( I think )
Also check out Figure 4 on page 3 of Linear Technology application note 45 which shows the same idea, but with a low noise differential JFET pair. It could be modified to use small power MOSFETs for lower noise. It could be the next step up in performance over LT1028s. This configuration was used in application note 124 for measuring low noise reference noise.
-
David , thanks
Do you think the old and now obsolete 2N6485 can be replaced by a TI JFE2140 ?? as TI propose a schematic
or any JFET in SOIC if possible as replacement.
Regards
OS
-
Do you think the old and now obsolete 2N6485 can be replaced by a TI JFE2140 ?? as TI propose a schematic
or any JFET in SOIC if possible as replacement.
Sure, that will work fine.
-
One can of cause substiture the FETs. The old 2N6485 are anyway a bit noisy and if at all would be a thing for more higher impedance.
For a low impedance source one could also use BJTs instead. For lowest noise even slightly larger transistors than the normal small signal ones.
With a lower noise AZ amplifier one would consider less the 100 K resistors for the filters. The 100 K add noise than can be relevant compared to the OP-amp.
Depending one the FET and AZ amplifier noise one would likely shift the cross over frequency between the fet and OP-amps part more towards a lower frequency. The ADC behind the amplifier and measurement cyle (turn the current on / off) would determine which frequencies really matter - the JFETs may not be that relevant afterall.
-
One can of cause substiture the FETs. The old 2N6485 are anyway a bit noisy and if at all would be a thing for more higher impedance.
For a low impedance source one could also use BJTs instead. For lowest noise even slightly larger transistors than the normal small signal ones.
With a lower noise AZ amplifier one would consider less the 100 K resistors for the filters. The 100 K add noise than can be relevant compared to the OP-amp.
Depending one the FET and AZ amplifier noise one would likely shift the cross over frequency between the fet and OP-amps part more towards a lower frequency. The ADC behind the amplifier and measurement cyle (turn the current on / off) would determine which frequencies really matter - the JFETs may not be that relevant afterall.
Hello
Thanks
On resistor it s seems that lower value generate less noise
On BJT I agree but question is to select one ..very loise transistor as LM394 but also obsolete , MAT12 looks in production but price is very hot .
I have located the circuit linked in picture base on LM394 , slow as that a 50 KHz amplifier
LM108 can be replaced by TLE2061MD still available or if someone have a better solution ???
SSM2212 is available and cost is reasonable , 8 Euro on Mouser
Regards
OS
-
PMI published some more recent implementations of the same instrumentation amplifier configuration, shown below.
The linearized cross-quad in the second example is a whole subject by itself.
-
The MAT-04 in Figure 6 is configured as the little known Caprio Cross-Coupled Cell, don't see that very often :-+
Long ago we utilized that along with a differential Cherry-Hooper Amp done in SiGe BiCMOS for some MW/MMW very high DR amps and mixers.
Best
-
The SSM2212 is a reasonable replacement for the LM394.
The cascode position transistors are less critical and could get away with simpler transistors (like HN4C51J). This especially the case if the amplifier is for low noise and super low distortion / super high linearity.
If the measurement cycle is relatively fast the offset and offset drift and thus transistor matching is not that critical. So one could even get away with single transistors. For the OP-amp the LTE2061 may not be good enough. Depending on the details the gain of the transistor stage may not be that high. If only for small signals, one could have relatively large resistors on the upper end and thus no need to have CM range close to the positive supply.
-
The thing which these integrated matched monolithic transistors provide, besides matching and good hfe over a wide current range, is low base spreading resistance which allows for lower noise. Some discrete transistors also have low base spreading resistance, including transistors with "perforated" and "ring" emitter construction. The old Zetex "super e-line" transistors would be good, and I am guessing modern high hfe and low Vce saturation parts from other manufacturers. I think Microchip makes the Zetex parts now.
THAT Corporation has matched quads, but I do not know what kind of construction they use.
Linear Systems has duals, but I do not know if they have low base spreading resistance.
-
The MAT-04 in Figure 6 is configured as the little known Caprio Cross-Coupled Cell, don't see that very often :-+
Long ago we utilized that along with a differential Cherry-Hooper Amp done in SiGe BiCMOS for some MW/MMW very high DR amps and mixers.
Best
Hello
I agree little known Caprio Cross-Coupled Cell , another electronic niche domain ....but I enjoy to learn
Regarding the Amp Op , on the LM394 schematic they use a old LM108 , on your PMUI schematic with MAT04 that a OP-17
I am not an expert in Op Amp but I have seen other schematic ( Mat03 ) with OP-07.
Perhaps SSM2210 can be wired a this schematic , they cost 8 Euro per piece a possible budget and easy to purchase.
Do you think a OP-27 can works as that available , that upgrade OP-07 and that available at acceptable price and quantity but that not a FET.
Regards
OS
-
The OP27 or a similar modern one is a reasonable choice for the OP-amp with a discrete amplifier. It depends a bit on the frequency range of interest and gain wanted. With high gain one may want a reasonable high GBW op-amp.
-
Regarding the Amp Op , on the LM394 schematic they use a old LM108 , on your PMUI schematic with MAT04 that a OP-17
I am not an expert in Op Amp but I have seen other schematic ( Mat03 ) with OP-07.
The LM108 (usually an LM308) was the best part at the time, at least from National. The uA725, predecessor to the OP-07, would have been suitable but is a Fairchild part. In general a precision operational amplifier is called for and we have lots of good parts now.
Perhaps SSM2210 can be wired a this schematic , they cost 8 Euro per piece a possible budget and easy to purchase.
The SSM2210 will work fine, but parts like the MAT02E have lower temperature drift which may be called for in more precise applications.
Do you think a OP-27 can works as that available , that upgrade OP-07 and that available at acceptable price and quantity but that not a FET.
The OP-27 and similar parts have all of the advantages of the OP-07 at higher speed and lower noise so will be excellent in a higher bandwidth design.
-
The OP27 or a similar modern one is a reasonable choice for the OP-amp with a discrete amplifier. It depends a bit on the frequency range of interest and gain wanted. With high gain one may want a reasonable high GBW op-amp.
Hello
I search to reach 2000 as gain so 1000 for the fet and x 2 for the Op Amp ... perhaps that better to set a lower gain on the FET stage and increase the final gain by using the OP27
My DC pulsed current is low in the 100 Hz range maximum .
Regards
OS
-
Regarding the Amp Op , on the LM394 schematic they use a old LM108 , on your PMUI schematic with MAT04 that a OP-17
I am not an expert in Op Amp but I have seen other schematic ( Mat03 ) with OP-07.
The LM108 (usually an LM308) was the best part at the time, at least from National. The uA725, predecessor to the OP-07, would have been suitable but is a Fairchild part. In general a precision operational amplifier is called for and we have lots of good parts now.
Perhaps SSM2210 can be wired a this schematic , they cost 8 Euro per piece a possible budget and easy to purchase.
The SSM2210 will work fine, but parts like the MAT02E have lower temperature drift which may be called for in more precise applications.
Do you think a OP-27 can works as that available , that upgrade OP-07 and that available at acceptable price and quantity but that not a FET.
The OP-27 and similar parts have all of the advantages of the OP-07 at higher speed and lower noise so will be excellent in a higher bandwidth design.
Hello
I try to locate and secure components with a good life time not to focus on obsolete one as that looks most of the time as saint graal quest .
SSM2210 and OP27 are easy to purchase and the market prices are reasonable and available in SOIC so easy to solder
I start an updated schematic and a BOM and I start the PCB with SOIC foot print
Question : does SSM2210 need a special pcb track / guard ?
Question : rather to use a DC /DC converter for the +15/-15 V can I use a set of 94 x3.7 V)x 2 to genrate +/- with lithium battery and generate 14.8V with lowest noise ? this circuit electrical consumption shall be very low so a set of battery can do the job
Regards
OS
-
The SSM2212 does not need special guards. If at all one would consider cut outs to reduce stress. However this would mainly be for long time precision, not for a relative short measurement cycle with current on/off.
For a low voltge amplifier one should not use a +-15 V supply, but less. So the question is of one can get away with a +-7.2 V or even less. Direct supply from the Li cells could work. It depends a bit on the details how good the PSRR is and how close one gets to the positive rail for the OP27.
SO8 OP-amps use a standard layout - so easy the change if needed.
The discrete amplifier stage will usually not reach a gain of 1000. It would be more like 20 x for JFETs and maybe 100 x for BJTs. With a cascode stage the gain can be a little higher with the JFETs. So the OP-amps noise and drift may matter and thus look for something better than the TLE2061.
So the OP-amp will naturally provide more of the gain, but this would not be an issue at the relatively low frequency.
-
In the pic of AN45, I can see the 2 op-amps and 2 JFET's, but are those 2 baby bottles by Figure 4 ?? They don't really look like pens or pencils, :-//
-
These are Jim Willliam's time measurement units :-)
From the first page of AN45:
The circuits in this collection represent our efforts, which
stopped when he (more or less) began sleeping through
the night. Most of the breadboarding occurred between
feedings, with design reviews and discussions during
feedings. As such, the circuits are annotated with the
number of feedings required for their completion; e.g., a
“3-bottle circuit” took three feedings. The circuit’s degree
of difficulty, and Michael’s degree of cooperation, combined
to determine the bottle rating, which is duly recorded in
each figure.
-
As was mentioned earlier AC is a better solution for measuring low resistance values for several reasons- No thermals, no issues with correcting DC offsets etc. The HP 4328A is an example of how to do it: https://www.hpmemoryproject.org/technics/bench/milliohm/bench_milliohm.htm (https://www.hpmemoryproject.org/technics/bench/milliohm/bench_milliohm.htm)
They have got expensive recently, $800, but the principles should be helpful for your project.
-
As was mentioned earlier AC is a better solution for measuring low resistance values for several reasons- No thermals, no issues with correcting DC offsets etc. The HP 4328A is an example of how to do it: https://www.hpmemoryproject.org/technics/bench/milliohm/bench_milliohm.htm (https://www.hpmemoryproject.org/technics/bench/milliohm/bench_milliohm.htm)
They have got expensive recently, $800, but the principles should be helpful for your project.
Hello
Thanks
AC measurement genrate skin current that not a problem on most electronic component , that could be a problem for mettalurgy purpose , I use DC reverse current
Regards
OS
-
Reversal of current is also a form of AC. A point that could make a difference is a pause between the polarities. This gives time for the inductive effects to settle.
-
Maybe one of the Vahalla microohmmeters would be better suited? https://www.ebay.com/sch/i.html?_nkw=vahalla+ohmmeter&_sacat=0&_from=R40&_trksid=p4624852.m570.l1313 (https://www.ebay.com/sch/i.html?_nkw=vahalla+ohmmeter&_sacat=0&_from=R40&_trksid=p4624852.m570.l1313) I belive they use a reversal technique to remove thermal errors. They also make igniter testers which are specal low current ohmmeters. Vahalla : https://valhallascientific.com/product-category/micro-ohmmeters/ (https://valhallascientific.com/product-category/micro-ohmmeters/)
-
Maybe one of the Vahalla microohmmeters would be better suited? https://www.ebay.com/sch/i.html?_nkw=vahalla+ohmmeter&_sacat=0&_from=R40&_trksid=p4624852.m570.l1313 (https://www.ebay.com/sch/i.html?_nkw=vahalla+ohmmeter&_sacat=0&_from=R40&_trksid=p4624852.m570.l1313) I belive they use a reversal technique to remove thermal errors. They also make igniter testers which are specal low current ohmmeters. Vahalla : https://valhallascientific.com/product-category/micro-ohmmeters/ (https://valhallascientific.com/product-category/micro-ohmmeters/)
Hello
Thanks , Valhalla 4300 is an interesting system , on uOhm they use 10 Amp and 20 mV voltage , a very low value . Schematic will be interesting to watch
Regards
OS
-
The igniter testers are also interesting. Failure there could be catastrophic. https://valhallascientific.com/shop/igniter-testers/4314-ki-wide-range-ultra-safe-digital-igniter-tester/ (https://valhallascientific.com/shop/igniter-testers/4314-ki-wide-range-ultra-safe-digital-igniter-tester/) One on eBay for $150: https://www.ebay.com/itm/197197314308?chn=ps&norover=1&mkevt=1&mkrid=711-166974-028196-7&mkcid=2&mkscid=101&itemid=197197314308&targetid=2304343365564&device=c&mktype=pla&googleloc=9032039&poi=&campaignid=22556502643&mkgroupid=179708627419&rlsatarget=pla-2304343365564&abcId=10314741&merchantid=8536794&geoid=9032039&gad_source=1&gad_campaignid=22556502643&gbraid=0AAAAAD_QDh9lTNOWETSqcDx6v_PH-8M56&gclid=Cj0KCQjw6bfHBhDNARIsAIGsqLhJg4POnkLEaSvVkMmGPx2EUrYWICVpyZm3zdbCUcu4HOu0jXaZXaoaAsOREALw_wcB (https://www.ebay.com/itm/197197314308?chn=ps&norover=1&mkevt=1&mkrid=711-166974-028196-7&mkcid=2&mkscid=101&itemid=197197314308&targetid=2304343365564&device=c&mktype=pla&googleloc=9032039&poi=&campaignid=22556502643&mkgroupid=179708627419&rlsatarget=pla-2304343365564&abcId=10314741&merchantid=8536794&geoid=9032039&gad_source=1&gad_campaignid=22556502643&gbraid=0AAAAAD_QDh9lTNOWETSqcDx6v_PH-8M56&gclid=Cj0KCQjw6bfHBhDNARIsAIGsqLhJg4POnkLEaSvVkMmGPx2EUrYWICVpyZm3zdbCUcu4HOu0jXaZXaoaAsOREALw_wcB)