Small, Low Inductance, High Precision Resistors

[RogerEvans]

I am trying to find suitable replacements for 2k, 1% 1/8 watt carbon composition resistors in a Tektronix 7S14 sampling plugin. They are located in the 1GHz sampling circuitry, hence the need for 'low inductance' but I don't have a spec for the allowed inductance. The originals look like ordinary wire ended carbon resistors around 7mm long, the wire ends are not particularly short so there was no worry about the lead inductance, they are normal through hole mounting.

Precision carbon composition resistors seem to be a thing of the past so any pointers would be very welcome. I am also wondering about suitable modern surface mount parts which could be properly attached at one end and a short jumper wire at the other end. I know nothing about their construction techniques and specifications so I am trying to tap into the knowledge here rather than endlessly search the manufacturers spec sheets.

I don't believe the precise value is important, they could be 1.8k or 2.2k but need to be matched and stable to keep the DC balance OK over time.

Many thanks for any help

Roger

[Craig Sawyers]

Tek used Allen Bradley carbon composition resistors, because being a rod of carbon and binder the inherent inductance was very small (not counting lead inductance).

Of course AB carbon composition resistors were real devils for moving higher in resistance as time went by.

I like the idea of surface mount resistors. If you get physically quite large parts, handling and soldering is fairly straightforward. You can get either metal film or thick film construction, and 1% tolerance is easily obtained.

The physically large SM resistors might seem to have high power handling (several watts), so should be overkill in that department.

Craig

[The Philpott]

Have you got a picture of them Roger?
Dave

[Radio Wrangler]

SMT resistors are trimmed with an L-shaped laser cut. MELFs and carbon-film resistors get a spiral cut to trim them and to make higher values, so they are quite inductive.

The usual ruthenium oxide or tantalum nitride thick-film SMT resistors are noisier than metal film types, but then carbon comps are noisier still.

Precision carbon comps weren't better made and more stable than usual, they were just the fruits of selection by a tighter binning process.

I agree. Solder some legs on some 1% SMT parts.

David

[Craig Sawyers]

As David said - avoid MELFs (Metal Electrode Leadless Face) - they are surface mount, but tubular construction, and inductive.

Craig

[factory]

They weren't precision (only 5%) but I did manage to order some new Arcol carbon composition resistors only three years ago from RS, now of course discontinued.
https://uk.rs-online.com/web/p/throu...stors/0386032/

David

[GrimJosef]

Hifi Collective still stock the Arcols, I believe https://www.hificollective.co.uk/com...resistors.html. The nominal tolerance is 5%, but the last time I bought any they measured comfortably within that. They're small, but not tiny though.

Cheers,

GJ

[John_BS]

I can let you have a couple of 2k05 1% resistors if they're suitable. May be AB? They measure 2k05 and appear to be, if anything, slightly capacitive at 200MHz.

John

[G0HZU_JMR]

I had a quick skim of the sampler manual and the 2k resistors in the sampler seem to be these ones from Allen Bradley as the BOM from the sampler manual describes them as BB2025 and the manufacturer cross reference code is 01121 Allen-Bradley Company, Milwaukee USA.

The UK company below claim to have extensive stocks of these resistors although they are going to be old stock assuming they really are genuine Allen-Bradley BB2025 parts.

https://www.hificollective.co.uk/cat...2j-p-6836.html

BB2025: 2K 0.125W Allen Bradley 5%


Datasheet here:
https://www.hificollective.co.uk/sit...data_sheet.pdf

[G0HZU_JMR]

I think it is worth trying to obtain the original parts because a 2k resistor will misbehave quite a bit by 1GHz. There will be a certain amount of lead inductance, a certain amount of capacitance in the resistor connections and there will also be distributed capacitance to ground/free space along the length of the resistor body.

This means that each resistor technology will produce a different impedance and response up at UHF. A lot depends on the size of the part and the internal construction and any proximity effects from nearby components and other hardware.

[Craig Sawyers]

You have to try very hard indeed, even with physically tiny RF surface mount resistors to get a flat response to >1GHz

https://www.mouser.co.uk/datasheet/2/427/ch-1764162.pdf

The original AB carbon comp resistors will be far worse than that.

To the OP - what makes you think you have a problem with R11 and R13?

Craig

[G0HZU_JMR]

The point I'm trying to make is that the original circuit will presumably be designed around the frequency response limitations of the original resistor. The datasheet shows the 1k version of the AB part is already misbehaving by 100MHz. I'm suggesting that it would be wise to fit a part with the same profile of misbehaviour. A close to perfect chip resistor might not be the correct choice because of this.

[G0HZU_JMR]

I had some free time this evening and I had a go at measuring a couple of resistors on a VNA.

I compared a basic 2k2 MF25 1/4W leaded resistor from Farnell against a 2k2 SMD resistor in an 0603 package.

https://uk.farnell.com/multicomp/mf2...oBdApoq1Oxydve

I tested both up to 2GHz and the result is below for parallel capacitance and parallel resistance. It shows that the 2k2 0603 cap has about 0.04pF self capacitance and the 2k2 MF25 has about 0.3pF all the way up to 2GHz although this is climbing upwards slightly as the frequency increases for the MF25 resistor.

The 2k2 parallel resistance Rp is seen to droop in both cases as the test frequency increases up past 1GHz and this will be due to the impact of the package parasitics of both resistors. The 0603 resistor performs better here as expected.

The MF25 resistor does quite well here although I did crop the leads to be just under 2mm long, just long enough to solder neatly and easily. Sadly, my resistor kit from Farnell is branded Multicomp so I'm not sure who actually made the MF25 2k2 resistor. I don't know who made the 0603 SMD resistor either but the results for the 0603 resistor are what I would expect to see.

Note again, the plot below is for Rp and Cp and not series Cs and Rs. I used an 18GHz rated test fixture for the measurement and a decent lab VNA.

[John_BS]

It occured to me some time ago, (!!) that many components in real-world circuits possess (parasitic) transmission-line characteristics.

Normally a component will be only a couple of mm from an earth plane, or a few mm from the walls of the enclosure.

For a small resistor in the immediate vacinity of an earth plane, you might expect a Zo of 50 to 100 ohms: it's difficult to find pcb-based transmission lines with Zo much higher than 200 ohms (I recall my old boss announcing that you can construct a 600 ohm transmission line by stringing a 40swg wire up the centre of a tube tunnel).

So, coming back to our 7mm-long resistor, let's say it forms a 70 ohm transmision line within its enclosure, which yeilds c. 47pF per metre, = 0.3pF shunt C. We're in the right ball-park for both Jeremy's and my measurements (in post 8): it's not necessarily a parameter of the resistor itself.

[RogerEvans]

Thanks for all the replies and offers of help. The reason for believing there is a problem with the 2k resistors (actually R1 and R2 between the bias batteries and the sampling diodes) is that Ch1 used to work perfectly and now cannot centre the trace using the DC offset control, Ch2 was never completely up to spec but I didn't at the time try to identify the problem - this was probably about 3-4 years ago and I never found a problem for which the 7S14 was the solution. Such are the perils of collecting interesting 7000 series items!

Oddly enough now both Ch1 and Ch2 show R1 is around 3k and R2 is still close to 2k, I suspect some electrolyte leakage from the alkaline button cells that I used originally as a quick fix to replace the original Hg cells and then left in place for too long! The only sign of corrosion is a slight dulling of the solder at one end of R1, same in both channels. If I tack in place a 5k6 resistor in parallel with R1 then I can get approximate DC balance and see a signal which has substantial transient errors, if memory is correct the falling edge is much worse than the rising edge. The feedback signal to Q40B shows a damped oscillation which shouldn't be there.

I think my best course of action in the short term is to move the 'good' R2 from Ch2 to replace R1 in Ch1, I will make a more careful check of any change in value before I do this. Then I will follow up the suggestion of Jeremy and see if I can get a pair of reasonably well matched resistors to put in to Ch2. John_BS, thank you for the offer of a pair of 1% resistors, I may well be glad to take that offer up later. I suspect Ch2 has a 'leaky' sampling diode which could explain the slightly odd waveforms I saw in the past but at the moment there is an odd problem where I see no 'signal' in response to an input square wave (1MHz or 60MHz) but the Ch2 trace will move vertically in response to changes in the DC value of the Ch2 input. Identifying just where the problem lies is made difficult since a lot of the sampling circuitry is under an RF shield and the power and signal feeds are provided by traces on the 'shield'. With the shield removed nothing works!

Recent threads on TekScopes have suggested that 7S14s are extremely rare so I am looking to pass this one on to someone who will enjoy having it rather than lying forgotten in my cupboard.

Thanks again,

Roger

[Craig Sawyers]

Another cause for a DC offset is a sampling diode gone leaky in reverse bias

Craig