Keithley 2000 TF-245 version / year question

[Overspeed]

Hello

Following a doc locate on this forum ( see linked pic ) there is several version of the TF-245

My question : does there is equivalent , different design but pin to pin alternative ?

I plan to make a patch base on 0603 component

Regards
OS

[Kleinstein]

I don't think the diagram of the A version makes much sense. it could be more like showing the measured resistance, not counting indirect paths.
The verion B should be the right one.

The resistor array should be rather stable. So a circuit from separate 0603 size resistors could be tricky. A final solution may want at least some resistor arrays for parts of the job.

One may not have to replace the whole array. There are different sections:
1) ref. amplification can get the auxiliary 13.3 V for the ohm part.
2) an auxiliary voltage for pin 14  for the fine slope of the ADC (not that critical)
3) the 3 resistors ~ 30 K for the integrator input   (could likely be 3 equal values)
4) the gain part (1K+9K+90K)

[Overspeed]

Hello

Dear Kleinstein

Thanks

Thanks I work on a patch base on a modified SAMTEC header and a PCB on the top to allow 0603 resistor , I can solder easily 0603  perhaps 0402

Just make a mock-up right now , TF-245 is  real source of problem and as 2000 / 2700 and other Keithley are now quite old that avaibility problem

Regards
OS

[coromonadalix]

other thread who talk about it

https://www.eevblog.com/forum/repair/keithley-tf-245-replacement-options-(k2000-models)/

[marcumr]

HELLO.
Some time ago one could order from Keithley-Tektronix (U.S.A.)the famous and sought-after TF-245.  I never managed it.

[picburner]

Some time ago I wrote directly to Tektronix to have some spare parts including some TF-245.
They told me they didn't sell to private individuals and directed me here where they got everything for me.
This at least for buyers in Italy.

[Overspeed]

Hello

Keithley have been purchased by Tektronix , my last call 4 years ago regarding my 2182 was more than disappointing ..... but Tektronic France have a long story of  :-

Another concern is the cost regarding the value I try to keep them alive at affordable cost , only parts located in the net are at 113 USD per unit without warranty

I have mostly complete the mock-up with the 0603 resistor I will try the week to check mecahnical fitting if that fit and select the best 0603 resistor I can locate 

Regards
OS

[Kleinstein]

It would be tricky to get good performance from individual resistors.  Resistor arrays (e.g. MORN series) can get quite good matching, especially with intermediate resistor values. If one would not mind a slightly higher supply current, one could separate the ohms part (the 13.3 V - should be less critical) and the reference amplification. This way 4 x 10 K could be used instead of the (1K+9K) + 10 K +20 K. 
The 30 K are not standard values, so maybe as 10K+20K or an array with 10 x 10 K.
The amplifier gain (1K+9K+90K) could use a resistor array (VTF330).

At the very lest one could use ACAS resistor arrays, which are the same size as 4 x 0603.

[Overspeed]

It would be tricky to get good performance from individual resistors.  Resistor arrays (e.g. MORN series) can get quite good matching, especially with intermediate resistor values. If one would not mind a slightly higher supply current, one could separate the ohms part (the 13.3 V - should be less critical) and the reference amplification. This way 4 x 10 K could be used instead of the (1K+9K) + 10 K +20 K. 
The 30 K are not standard values, so maybe as 10K+20K or an array with 10 x 10 K.
The amplifier gain (1K+9K+90K) could use a resistor array (VTF330).

At the very lest one could use ACAS resistor arrays, which are the same size as 4 x 0603.

Hello

Thanks for the advice regarding resistor net I check this solution this week

That very sad to not see a robust design in the Keithley serie 2000  with a small SOIC resistor net rather to use through the hole type as the Caddock ceramic , and worth as there is Caddock resistor on the PCB for other purpose

I can stack two pcb on the SAMTEC header pin I link a first CAD design with 0603 , done to check the mechanical fitting only

Regards
OS

[Kleinstein]

The Cadock ceramic networks that are used for high voltage dividers are usually thick film and thus not super stable. They are special to withstand high voltages, but for low voltage more normal thin film types would be preferable.

Chances are there would be more space available, possibly a vertical mounted PCB.

[Overspeed]

The Cadock ceramic networks that are used for high voltage dividers are usually thick film and thus not super stable. They are special to withstand high voltages, but for low voltage more normal thin film types would be preferable.

Chances are there would be more space available, possibly a vertical mounted PCB.

Hello

I agree , my remark was more on the size room on the board compare to the soic 16 broad

I check the vertical solution PCB , my first concern is to limit as possible force on the SOIC solder tab

Question : As the Keithley board have a stiif aluminium hood with a 4 mm pillar on the front , Do you think that could be possible to extend this hood as a PCB support and link the SOIC foot print with wires ? The aim will be to avoid mechanical stress on the small SOIC solder tab and also allow to install real resistors net without room limitation.

Regards
OS

[Kleinstein]

The mechanical stress could be indeed limiting. It should still be OK for normal use, but shipping it (e.g. for calibration) could become an issue.

Using wires and an edditional external PCB could work, though it adds a little copper resistance. The ref. amplification for some reason uses a rather fast amplifier but should be OK with the relatively large capacitors.
Still the extra capacitance is not great - so the cable should not be long.

[Overspeed]

The mechanical stress could be indeed limiting. It should still be OK for normal use, but shipping it (e.g. for calibration) could become an issue.

Using wires and an edditional external PCB could work, though it adds a little copper resistance. The ref. amplification for some reason uses a rather fast amplifier but should be OK with the relatively large capacitors.
Still the extra capacitance is not great - so the cable should not be long.

Hello

Could be possible to install a plate like the white page on the picture
Wiring will be less that 40 mm long as the resistor PCB will be on the top of the 245 foot print

If other people are interested I can laser cut some and propose kits after final design testing

Regards
OS

[Kleinstein]

The screws close to the rear side are likely not suitable, as they would be at or near the ground / outputside potential. The hood screw position does not look that bad. 60-100 mm of cable would likely be acceptable from the capacitance. For holding the PCB could likely extend to the screw area. The first point is seeing if the meter is working at all with modified resistors.

[Overspeed]

The screws close to the rear side are likely not suitable, as they would be at or near the ground / outputside potential. The hood screw position does not look that bad. 60-100 mm of cable would likely be acceptable from the capacitance. For holding the PCB could likely extend to the screw area. The first point is seeing if the meter is working at all with modified resistors.

Hello

I agree
As my 2000 ( this one ) is dead right now and I have also a 2700 with problems , I can built a prototype and test various circuits / components to see to result . accurate resistor are easy to locate when they are bigger too , Vishay foil as example

I will post progress

Mouser have 0603 and 0805 in 0.05% 10ppm SMD that quite cheap as under 1 Euro per unit and enough to make first test

Regards
OS

[Kleinstein]

There is no need to go for low tolerances - there is essentially no gain from this. The meter should tolerate quite some deviation from nominal. Even with 0.05% resistors one would still need recalibration as the gain with change a little.  Some compensation is possible by changing R300 if needed.

The point is more with matching between the three ~ 30 K  resistors towards the integrator and the chain with 20K 10K 9K1K from pin 1 to 3. Here it could make sense to use 4 x 10 K as a set with 100 K in parallel to 10 K to get close to 9 K.   ORN / MORN resistor arrays are not that expensive (~ 4 EUR) and usually good matching (e.g. 1 ppm/K range for 10 K).  ACAS would be small and cheap.
Even with seprate resistors it can make sense to have at least the 20 K as 2x10K.

[coromonadalix]

would a tf-245 equivalent feasible, is there some companies who can do custom value / part / like this one

will we scrap meters because of this %@ of part ?   that's why i never bought theses series of meters because of that single part ....

i do  hope some substitution will be feasible

[Overspeed]

There is no need to go for low tolerances - there is essentially no gain from this. The meter should tolerate quite some deviation from nominal. Even with 0.05% resistors one would still need recalibration as the gain with change a little.  Some compensation is possible by changing R300 if needed.

The point is more with matching between the three ~ 30 K  resistors towards the integrator and the chain with 20K 10K 9K1K from pin 1 to 3. Here it could make sense to use 4 x 10 K as a set with 100 K in parallel to 10 K to get close to 9 K.   ORN / MORN resistor arrays are not that expensive (~ 4 EUR) and usually good matching (e.g. 1 ppm/K range for 10 K).  ACAS would be small and cheap.
Even with separate resistors it can make sense to have at least the 20 K as 2x10K.

Hello

Thanks for advises , I will check and sort the resistor I have a kelvin setup for SMD and I can try to adapt a jig for null meter in the aim to control SMD

Question : ppm/C for resistor from the same batch are normally the same OR I need to to make measurement at several temperatures ? my kelvin jig is in controlled temperature enclosure

Setup of resistor : as on the ''top plate roof'' there is room to place a 32 mm x 80 mm PCB  I can use several setup without real problem I can even set a small thermal insulation box and why not an oven if necessary

Regards
OS

[Kleinstein]

There is no real need to go over the top with the resistors. After all it is only a 6 digit meter.

With resistor from the same batch one often gets some matching, but of cause not perfect and not guaranteed. Things can also be different between types.
There should be no need for super accurate resistor value matching - if at all it would be TC matching and getting comparable long time drift / PCB stress effects.

For the self heating effect (especially the 30 K at the ADC input) the tight thermal coupling in an array would be an additional advantage over separate resistors.
Still 30 K as an array are a bit tricky (3x10 K or 10 K +20 K).
It is a bit clear if the resistors for the references should be exactly equal or a slight difference (e.g. on the order of 1/128) wanted - whicht way fits depends on how the result is calculated. So an extra series resistor (e.g. ~ 234 ohm) for pin 11 may make sense to have the option to adjust.

[Overspeed]

There is no real need to go over the top with the resistors. After all it is only a 6 digit meter.

With resistor from the same batch one often gets some matching, but of cause not perfect and not guaranteed. Things can also be different between types.
There should be no need for super accurate resistor value matching - if at all it would be TC matching and getting comparable long time drift / PCB stress effects.

For the self heating effect (especially the 30 K at the ADC input) the tight thermal coupling in an array would be an additional advantage over separate resistors.
Still 30 K as an array are a bit tricky (3x10 K or 10 K +20 K).
It is a bit clear if the resistors for the references should be exactly equal or a slight difference (e.g. on the order of 1/128) wanted - whicht way fits depends on how the result is calculated. So an extra series resistor (e.g. ~ 234 ohm) for pin 11 may make sense to have the option to adjust.

Hello

Ok I make a trial and post the results step by step

Regards
OS

[Overspeed]

would a tf-245 equivalent feasible, is there some companies who can do custom value / part / like this one

will we scrap meters because of this %@ of part ?   that's why i never bought theses series of meters because of that single part ....

i do  hope some substitution will be feasible

Hello

I am in process to try

Even as used a 2000 and some other Keithley cost $$$$ and use the 245 component and that sad or stupid to be force to scrap 600 USD or more measuring instruments IF there is a robust solution to repair

Regards
OS

[voltsandjolts]

that's why i never bought theses series of meters because of that single part

You have the option to buy a newer version that doesn't have the TF245, uses off-the-shelf arrays instead.

[Overspeed]

that's why i never bought theses series of meters because of that single part

You have the option to buy a newer version that doesn't have the TF245, uses off-the-shelf arrays instead.

Hello

Yes but there is a lot of old models still in use and or need repair so rather to put them in the trask bin , I try to repair / upgrade

Take in consideration that other Keithley use also the 245 component not only the K2000 as the 2306 have the two TF-245 and 2010 also
In the 2010

the fact to use a bespoke connector with the 245 SMD foot print allow to install the patch on mostly all 2000 serie Keithley

Regards
OS

[coromonadalix]

that's what i want to see



OP  have you seen theses ??  Dip to SOIC adapters

https://www.epboard.com/eproducts/protoadapter3.htm#DIP-16%20to%20SOIC-16%20WIDE%20IC%20Adapter

[Overspeed]

that's what i want to see



OP  have you seen theses ??  Dip to SOIC adapters

https://www.epboard.com/eproducts/protoadapter3.htm#DIP-16%20to%20SOIC-16%20WIDE%20IC%20Adapter

Hello

Thanks , interesting
I have located that a SOIC header on which I can fit a socket to link my patch PCB by wire as I search / prefer to avoid stress on the main PCB but also be able to manage temperature / pcb stress of my in process patch

I link the ''quick and dirty first aluminium plate base on the CAD I have cut this day , patch pcb will not fit directly on this plate but on a SS304 plate with O ring to limit stress and dilatation effect

7 mm Holes net have been added to avoid to trap the heat of the microcontroller

Regards
OS

[Overspeed]

Hello

Update , this Harwin connector ( header ) can be soldered in place of the 245 component

Regards
OS

[Kleinstein]

With the resistor values I don't know if one needs rather accurate ratios at some places. It depends on the software inside the meter, if the use some kind of norminal values for the slow ration or do an extra measurement (e.g. special factory calibration or in between normal measurements). Ideally one wants an option to trim 2 of the resistors, to adjust the ratio of the positive , negetive and small slope.

[Overspeed]

With the resistor values I don't know if one needs rather accurate ratios at some places. It depends on the software inside the meter, if the use some kind of norminal values for the slow ration or do an extra measurement (e.g. special factory calibration or in between normal measurements). Ideally one wants an option to trim 2 of the resistors, to adjust the ratio of the positive , negetive and small slope.

Hello

I have draw two PCB base on 0603 resistor ( 1K and 10 K in serie for most of them ) and header i Have checked the stack height I continue to work on the header connector BOM

I can sort / measure the 0603 on my Kelvin 4 wires test jig

I link some screen prints

As I have some room on the PCB I can add additional 0603 for trimming , my first goal is to test a solution asap , after I can re design PCB with resistor net if necessary , 0603 are easily available from Mouser

Regards
OS

[Overspeed]

Hello

a question for Kleinstein

On Xdev website there is a table value of the 245 resistances

27.5 K looks to be closer than 27.4 than 27.5

Note : thanks to Xdev for his amazing work

Regards
OS