V7-49 electrometer produced in Minsk, Belarus.

[exe]

How big is the percentage of bad compared to the total. Is it 100%? 10%? 1%.

I think even one bad cap can be enough to ruin measurements. Caps are there for a reason...

[2N3055]

That's dead alright..
4R ESR is only good for 10uF/63V elko...
2200uF elko must be in tens of milliohms...

[MegaVolt]

I think even one bad cap can be enough to ruin measurements. Caps are there for a reason...

Here is more a question of the search method. Or we change all capacitors. Or we turn on the device and see where the ripple is more than necessary and change only there. It all depends on the percentage of marriage. If 1 capacitor per device, then the second way is easier. If 90%, then by itself the first.
I think the storage and operating conditions of the device before this are still strongly affected.

[MegaVolt]

Storage conditions of my devices looks good. All packages come in very good conditions, no oxidize into device or accessories. There is no sign of damp.
When i check 5-7 caps, all of it is died, then i make decision - do not waste time for check and replace it all.

I understood

[asis]

Hi
I read this exciting story with great interest.
I remember with pride our instrumentation.
Keep pouring balm on your soul, please.

Good luck.

[Vgkid]

Thanks for the link, keep it up.

[exe]

Don't they have different proximity to the speakers? Also the sound was too quiet, if one wants to listen to Lordi, crank up the volume to 140dBa or more (so your lovely neighborhood can enjoy it too).

[bsw_m]

So what about noise of this instrument?
I'm was collect  data  from instrument in 1*10^-12А range and settling time 10s. Time for data collection is ~1Hour (Instrument in stock condition, no modifications, only a bug fixed with resistors in TIA)
For graph I'm use the Stable32 program.

So now time for modding this instrument. 

RAW data from instrument attached. Sampling time 0.1s

Program for working with this instrument trough GPIB written in C and use linux-gpib library. Later I will post the code.

[bsw_m]

Some new data and new graph

[Kleinstein]

There are regular jumps, like some periodic adjustment steps to null the amplifier. those steps seem to be quite large in the beginning. Maybe the meter was not yet fully warmed up. Otherwise jump of up to 5E-16 A are quite large compared to the noise level shown before (5 E-17 range for a few minutes).

[bsw_m]

You absolutely right!

[bsw_m]

Second attempt to plot noise floor for this instrument with capped input.
Environmental temperature is quite stable: not more that 1Celsius change. Instrument is properly warmed-up.
Sampling rate in file is 10sps. Data collection time about 11hours.

[bsw_m]

My idea with REF200 has fail. i'm found some interesting things - that JFET's need a programmable current source to set extremely low leakage mode, REF200 can't be programmed.

You right. For low gate current (less than 1*10^-14) input jfet in this instrument work on special combination with drain voltage/source current.
This interesting mode is described in more detail in this patent

[bsw_m]

10 fA... hmmm... that is too big leakage current for me. If i can't drop down current to less 1fA, i will try ADA4530-1 or some similar.
But now i need stabilize zero at first, because zero stability is affect to leakage current directly.

Less that 10fA, in real world less that 1fA ;-)

[bsw_m]

Right!

[MegaVolt]

My TC test of Vbe change is accomplished, i'm select 3 NPN's (2 as 50-150uA driver, 1 as temp.co. stabilizer) to create dual channel JFET current driver, calculations shows possibility to create programmable  50-150uA current source with up to 100ppm/C temp.co.

Can I sketch a diagram ?? At least on a piece of paper to understand what is at stake?

[MegaVolt]

And what parameter of the device is planned to be increased by such revision?

[bsw_m]

Irrc 1aA is around 63 electrons/sec, so some form of sub-atomic catapult might be in order. 

1aA is around 6.3 electrons/sec.
And I'm got the "sub-atomic" catapult  This is NK4-1. Soon i will open the topic for this tool

[bsw_m]

I'm got the "sub-atomic" catapult

Yup, that NK4-1 catapult use for calibration all Keysight/Keithley electrometers equipment in our country.
I go to your's lab with my devices to calibrate my 2985.

So, this is not problem, Andrey! I will be glad to talk live.   

[Kleinstein]

To adjust the drift of the FETs it may actually help to independently adjust the currents. Via the current ratio one should be able to adjust the differential temperature drift of the offset. Just adjusting the current would alow to adjust the bias current - though likely no big gain of reduces current noise by actually getting it down to zero, as there will likely be some compensation of a positive and negative current from different parts of the fet.

For calibrating very small currents, there would be single electron turn styles, so counting electrons, one at a time. I don't remember calling them catapult. This is still quite new and may require low temperature.

[MegaVolt]

I have to buy the most sensitive electrometer on planet Earth to test these OpAmp's and select one OpAmp with lowest leakage.

To measure small currents, you can watch the charging time of a small capacitor connected to the input.

[bsw_m]

So, I performed a rapid test on the lower ranges.
Used current source: NK4-1 calibrator.
For 100fA using differentiator №1, for lower current using differentiator №2
IMHO I see very good results for this simple electrometer.

[Kleinstein]

The low bias OPs have a relatively high offset drift and at higher temperature there can be drift in the bias current.
Still those OPs are often better than the single FET solutions, unless the TC is really well trimmed.

The circuit shown looks odd: As a follower for a low bias TIA input the ADA4530 should get away with a even lower supply. It would be only for higher voltage readings that the higher supply is needed.  In combination with the ADA4530 there is no real need to trim the offset of the OP177. Adjusting the offset would also change the TC. So with trim the OP177 has very limited advantage compared to a 741 in the given circuit. As a low temperature would help, I would consider more like a low power OP.

[Kleinstein]

There is nothing wrong with the 2 OP configuration. If only use as a TIA it would be more like a lower supply to the ADA4530, more like +-5 V or +-3 V.  The +- 7 V mA be needed for use in voltmeter mode however. You may have to check if this is enough - some electrometers use am extra bootstrapped supply or driving the opposite side, so it depends on the rest of the circuit if the +-7 V are sufficient and needed.

Both the ADA4530 and OP177 are supposed to have a rather low offset - so I doubt a trim would be needed. There is still the option to use the lower power LT1097 if wanted. Noise would be dominated by the ADA4530 anyway.

[Kleinstein]

Is the V7-45 OK with working horizontally ? The mechanical part may be effected from working in a different orientation.

I would guess the V7-49 is working in voltage reading mode. In current mode it would not make much sense to connect 2 TIA like instruments.

What type of offset compensation is used, that it effects the input current ? I would not expect the voltage offset adjustment to effect the current so much.