DIY cable for picoammeter

[Gyro]

How difficult is it for you to buy European / US commodity parts in the Russian Federation? An LMC662 costs $1 - $2 in single quantity. Probably much cheaper that trying to source low leakage FETs etc.

[David Hess]

I have not but Linear Systems also produces low leakage JFETs and JFET pairs as well as low leakage diodes.

These diodes are actually JFET connected as diodes to two external pins, right? Unfortunately, this is a rare and not cheap product. At least for prototypes or the first attempt, as in my case:)

They could be JFET dies connected externally as diodes but I do not know.  Planar diodes constructed like planar transistors do exist.

2N4117s and 2N4117As are inexpensive and readily available.  It only takes a couple minutes to test some very inexpensive 2N3904s or similar small signal transistors for low leakage.  So there are several inexpensive and readily available low leakage diodes available for those who need them.

How difficult is it for you to buy European / US commodity parts in the Russian Federation? An LMC662 costs $1 - $2 in single quantity. Probably much cheaper that trying to source low leakage FETs etc.

The LMC6081/LMC6082 is also suitable or the LMC6061/LMC6062 if lower supply current is required.

Testing an operational amplifier for input bias current is very easy; configure it as an integrator so that it integrates its own input bias current.

[Kirill V.]

LMC662 - it is available under the order
LMC6001 - around 15$
2N4117/PN4117 - not available in our stores
LMC6081 - not available in our stores

Probably much cheaper that trying to source low leakage FETs etc.

I had in view the use of their as diodes protection.
I can buy these things in Mouser, but they do not work with Russia directly, only through an intermediary in St. Petersburg. They charge a certain amount of money for their mediation. I've never done that before. I'll leave it to the future.
When you advise to buy something ready-made, you do not understand the specifics of it. I can't go to a store and leave with a package of parts. I can't even order it in Moscow or anywhere else, because these parts are not there

[Gyro]

When you advise to buy something ready-made, you do not understand the specifics of it. I can't go to a store and leave with a package of parts. I can't even order it in Moscow or anywhere else, because these parts are not there

I feel your pain.   I suspected that it might be difficult, I just wasn't sure how difficult.

Edit: Actually the LMC6001 cost is fairly similar to the UK, LMC662 is a good cheap bet if it is available to order, as you indicated - typical input current of 2fA (lower than the 6081, of course you have to be lucky to get the typical one!)

[Kirill V.]

Thanks
But don't try to talk me out of using the AD549 because it's already in my drawer waiting The strangest thing is that this item is available for purchase absolutely easily.
Products LIS or InterFet is absolutely not available from us even to order. The vast majority of people in our country have not even heard of such a thing.
H&S or Belden are available only under the order and the cost needs to request and the delivery time is not very fast. In fact, a Russian company will order it in America or Europe and sell it to me. They don't have it in stock.

[MegaVolt]

LMC662 - it is available under the order
...

https://efind.ru/offer/LMC662+?r=https%3A%2F%2Fefind.ru%2F&stock=

[DeltaSigmaD]

I would like to make some comments on this subject according to my experience.

Triboelectric currents and microphonic effects are a general problem for the cable connection between signal source and first amplifier, since the dielectric material of the cable cannot be freely selected. Teflon/PTFE has excellent DC isolation, but it has very high triboelectric generation and microphonics. This can be easily understood: PTFE has C-F-chemical bounds, where the difference of the electronegativity of C and F are very high. After mechanical treatment or soldering, PTFE can freeze electrical charges inside its bulk volume (which cannot be removed via surface) with relaxation times of hours to several weeks. The relaxation is so slow since the C-F-bound of PTFE is extremely rigid (bound energy far away from kT, only tunneling?). But a relaxing charge generates surface currents (displacement current, see Maxwell equations). Nearly perfect in this aspect are dielectrics like PE (polyethylene) or PP (polypropylene), where the difference of the electronegativity between C and H are weak. Even after machining only weak bulk charges are produced. The limiting factor with these materials seems to be chemical additives to the PE/PP polymer material. Afaik, no cable manufacturer mentions any concerning information in the data sheets. Therefore, it is also not clear which influence the degree of polymer cross-linking has. PVC (polyvinyl chloride) has much too bad isolation and microphonics, it is a very dirty material even due to the required softener(s). Glass has relatively high surface conductivity due to the Ca-, Na-, and K-ions on its surface (hygroscopic, good wetting with water!). With glass-insulated reed relay contacts, you can easily measure the wandering charges on the glass surface. Ceramics also have high difference of the electronegativity (e.g. Al2O3), and are not flexible.

One additional problem with all cables is that the noise is increased by the cable if non-steady-state (--> ac) signals must be measured. In most practical cases, the cable capacitance generates additional noise due to the amplification of the input voltage noise of the preamplifier, dominating factor is Ccable / Cfeedback.

Consequences:
1. Don't use PTFE, excepted you are measuring only DC signals and there is no movement or vibration.
2. When a PTFE-isolated cable (even triaxial cable) was moved, consider a relaxation time of hours to stabilise the displacement current. After that period, subtract the offset current to get reliable results.
3. Make the cable length as short as possible.
4. If the mechanical/practical constraints permit, use a transimpedance preamplifier directly at the source. The amplifier's box size will be only a few cm³. That's the preferable solution.

[Kirill V.]

Thank you
Almost everything you said is described in Soviet literature. I am once again convinced that this is a source of non-outdated knowledge.

[klimm]

.....
2) The BAV199 diode from a decent manufacturer (TI, NXP) will give you a much better performance than either BJTs or JFETs as diodes.

As with 2N3904s used as low leakage diodes, the BAV199 is only suitable if you grade them yourself; they are only tested down to like 2 nanoamps or something.

If you want a guarantied low leakage diode, 2N4117 or 2N4117A JFETs are the least expensive option.  But personally I just grade 2N3904s which only takes a couple minutes and can be done with a power supply and any 10 megohm input voltmeter.  Use the 0.2 or 2.0 volt scale as a picoammeter unless you have an electrometer handy.

According to the above described measurement procedure, can anyone estimate the  value of reverse current  one could  expect from B-C junction of 2N3904? If measured , even better. 
My measurement on a lot of 20 unknown bipolars(B-C, E not connected)  with about 5V, results in  410±10 pA. I get same result when measuring 2 pcs of 2N3904. Not much difference when measuring B-E. 
For comparison a 4148 diode resulted in 39 nA at 5V and 53nA at 10V. Temperature was about 25C

[klimm]

Well... I must admit,  I forgot to null the DMM after 1 hour on.  Therefore the intriguing current of 410pA, I got.
There was a certain offset that I was able to notice today in the same conditions: 4.1mV  which equates for 410pA at my 10M input resistance. 
Nulling  the offset I  now  read 000.00(4 1/2 digits)  when testing the  B-C junctions. It is fine to know that the junctions have a leakage<10pA but can't see any difference between the 20 transistors I tested. Was expecting to see some variation.   

[Kleinstein]

< 10 pA sounds reasonable. So it may need a more sensitive circuit or meter to see the differences.

[klimm]

What intrigues me is that mr Hess says "any 10 megohm input voltmeter"  and also said that  he grades the juctions which implies that is able to see differences among these. 

[Kleinstein]

Transistors from the same batch can be relatively similar. Grading in the case is more like check if they are actually <= 10-20 pA and not 1 nA.  Many simple DMMs only provide 0.1 mV resolution with 10 MOhms and thus 10 pA resolution.  In addition the measurement with such a high impedance may pic up mains hum and with a diode in the ciruit and not enough voltage this may see some rectification.

[bsw_m]

Hi!
Please note russian low noise cable AVK-6 (АВК-6) and cables in this series have a conductive layer on the insulation to suppress the triboelectric effect. But these cables is not a triaxial. But I not see any problem to make outer shield.
P.S. MNIPI used these cables for their electrometers.

Also I'm test B-E junction of 2T363B transitors at 1V reverse voltage.
On average, the leakage current was 50fA with extremes of 100fA and 10fA

[Kirill V.]

Hi all again,

But these cables is not a triaxial. But I not see any problem to make outer shield.
P.S. MNIPI used these cables for their electrometers.

I believe that the question of a triaxial cable can be closed in this particular case. This refers to homemade current meters for the home. As proof - FEMTO uses a coaxial cable
The results for Soviet transistors are interesting and valuable, thanks for the measurements