8
\$\begingroup\$

I'm having issues with deployed knob pots after they've been in use for a while. It's been an ongoing problem that we don't see at our local facility. I'm wondering if any of you may have experienced this problem, and more importantly, are aware of a solution.

Pot: Bourns 3590S-1-203L, 10-turn, panel-mounted, 20k, knob-pot.

https://www.bourns.com/pdfs/3590.pdf

Problem:

Durability. Common occurrence of glitchy output after some time (months/years) in use.

A picture showing traces from both a good (blue) and bad (black) pots, turned by hand:


enter image description here


  • Initially, there are no indication of problems. Glitches manifest after some time...

  • Once a pot develops glitches, they are permanent and always appear at the same location in the 10-turns.

  • Glitches may appear, seemingly, at any location in the ten-turn range.

  • Duty cycle is low. Lets say a full ten turns forward and back every day (which would still be an over-estimation).

  • The pot setting may remain (energized, or not) at the same location for days, weeks.

  • Although discouraged (and unusual), the system may be power cycled with the wiper set in any location. It should typically be set to full CCW.

  • We are very careful when hand-soldering leads, using a heat-sink on the pot tabs. High quality, temperature controlled irons. Experienced assemblers. SAC305/ No-Clean flux.

  • The pot wiper is placed at one extreme location prior to soldering (mainly as a way to determine if this problem could be a result of heat damage during soldering).

  • Working environments tend to be environmentally controlled. Even though high-humidity is unlikely, it's possible.

  • Knob pot dial and panel are well earthed, with a dedicated chassis conductor.

  • We've seen failures for pots with both plastic and metal shafts.

Here is a copy of the driver and buffer circuits for the 20k pot:


The problem pot is the "Panel_Cmd_Knob_Pot" connected to the top 3 yellow ports.

enter image description here

(Not shown above is U2A. This section is used as another low-speed buffer elsewhere.)


The +/-12VDC supplies come from +/-15 rails (switching supply), linearly post-regulated on-board (LM3937/LM2990 series).

All connections are 'permanent' (no hot-plugging).

What I plan on trying next:

Add a series resistor from the knob pot wiper to the U2B buffer input. Perhaps current-limiting power-up transients, that may or may not exist.

Any better ideas?

Updates:

  • 5/22: Sent suspect pot to Bourns for eval.
  • 6/5: Requested status update from Bourns. No response yet.
  • 7/22: Bourns has acknowledged that they have the pot in hand, and the fellow who would look into it is back from traveling. However ever since, they've gone dark and appear to be unwilling or unable to provide any insight. Suggestion: Don't do what I did and/or avoid these parts.
  • 9/10: Bourns has stayed quiet. Giving up, and closing this question.

-Chris

\$\endgroup\$
  • 2
    \$\begingroup\$ Have you presented this question to Bourns? What about changing to a different manufacturer? \$\endgroup\$ – AlmostDone May 15 '18 at 21:26
  • 1
    \$\begingroup\$ "Have you presented this question to Bourns?": I just did. Let's see if they have any suggestions... \$\endgroup\$ – Chris Knudsen May 15 '18 at 21:56
  • 2
    \$\begingroup\$ The rotational life is 1M cycles which you don't seem near reaching. Their QA definitely needs to know about this, and they'll probably request you send them some faulty units to investigate. If your company has a quality system in place where you can generate vendor non-conformance reports, that formality might step up their process to comply with their own qms. \$\endgroup\$ – AlmostDone May 15 '18 at 23:40
  • 1
    \$\begingroup\$ Are the positions randomly distributed, or do they correlate with something, e.g. each other, positions the thing is left in for a while. Also have you ruled out that you ever exceed the load life? Also you should take some apart and inspect the positions at which they fail. \$\endgroup\$ – PlasmaHH May 16 '18 at 20:37
  • 2
    \$\begingroup\$ I would be curious to learn what they find is the problem. Please post a followup when you find out. \$\endgroup\$ – AlmostDone May 18 '18 at 12:16
1
\$\begingroup\$

switch to a conductive plastic element or even a try the sealed bushing of the same pot. The problem is that it could have develop wear spots from the mechanical wiper, or crud/dust has got inside it. I'd probably try the 3590S-4-203L, if that didn't do the trick, then maybe just a seal shaft conductive plastic pot if it doesn't really need 10 turns.

If this design didn't have current flowing through the wiper you would have never knew of the developing wiper noise.

\$\endgroup\$
  • \$\begingroup\$ I need to keep the 10-turn spec... \$\endgroup\$ – Chris Knudsen May 15 '18 at 21:49
  • 1
    \$\begingroup\$ Will try a sealed version of the pots. Since I don't expect enough performance data to make any conclusions for months/years, I'm accepting drtechno's answer (@Sparky256 has enough points...). \$\endgroup\$ – Chris Knudsen Sep 10 '18 at 18:20
4
\$\begingroup\$

Adding a heavy duty RC filter (1 M * 100nF) will help reduce noise while pot is moving, but a bad spot is always going to be a bad spot.

I ran into the same problem with variacs under daily use. At some point the element had no conduction at the bad spot, but micro-arcing of current from using the same spots over and over again was the culprit, so every two years I had to replace them.

I did not have this problem with the pots you use. A 10K resistor in series with the wiper would reduce any shoot-through currents when the pot is at extremes of range, and maybe the op-amp conducts or clamps voltage close to their supply rails.

I did switch to 1K pots to reduce sensitivity to noise, but I never had one with bad spots even after ten years of heavy use, but I used conductive plastic potentiometers, not wire-wound.

\$\endgroup\$
  • \$\begingroup\$ Filtering is an option, but I'd rather fix whatever I'm doing wrong. \$\endgroup\$ – Chris Knudsen May 15 '18 at 21:48
  • \$\begingroup\$ @ChrisKnudsen. With all do respect, I think your mistake is in using wire-wound pots instead of long-lived conductive plastic, and inserting a basic RC filter to remove wiper noise and block any shoot-through currents. \$\endgroup\$ – Sparky256 May 16 '18 at 5:00
  • \$\begingroup\$ I think that you're probably on the right track. I'll let this post stew for a bit, and see if anything else comes up. \$\endgroup\$ – Chris Knudsen May 16 '18 at 12:07
1
\$\begingroup\$

You must draw some current from the wiper for it to be self-cleaning just like with switches. 1mA should be enough. If it's gold-plated, maybe 50uA is enough.

\$\endgroup\$
  • 1
    \$\begingroup\$ I've been looking for data on this, and have found none. Only in an app note for composition pots (non-metallic junctions): Bourns: "Here at Bourns Trimpot we’ve conducted extensive work to better define dry circuit conditions for non-metallic junctions in composition potentiometers." ... "...an absolute minimum wiper current of 25 microamps and preferably over 100 microamps. This should give you enough current through the wiper to help you avoid dry circuit problems over extended periods of time and temperature." No mention of this issue at all for wire-wound or hybrid pots. \$\endgroup\$ – Chris Knudsen Sep 11 '18 at 14:59
  • \$\begingroup\$ I'm not trying to say that this isn't an issue, just that I cannot find any published guidelines. \$\endgroup\$ – Chris Knudsen Sep 11 '18 at 15:01
  • \$\begingroup\$ @ChrisKnudsen I'm surprised it's an issue at all for non-metallic potentiometers since they don't experience any kind of oxidation. \$\endgroup\$ – τεκ Sep 11 '18 at 15:53
0
\$\begingroup\$

I've seen this type of failure before with potentiometers, where the track exhibits strange properties during the turn. Long story short, the tracks were disintegrating/failing due to vibration from the vehicle. We ended up switching to RVDTs.

\$\endgroup\$
0
\$\begingroup\$

(Wanted to COMMENT, but don't have enough points, so ANSWERing instead.)

Just a SWAG on my part, but could galvanic corrosion be the cause or a contributing factor? I remember being a new engineer working on phone systems and learning that the system voltage is ground and -48VDC, rather than ground and +48VDC, to counter this effect.

Referring to the schematic, the pot's wiper goes only to the non-inverting input of U2B, so the input bias current will flow though the wiper. The bias current spec is less than +/-1 nA (+/-1.5 nA over temp), so not a bunch, but still something. Maybe with the right set of conditions -- input bias current polarity and magnitude, shaft inactivity, humidity, temperature, (Elvis sightings??) -- a problem will occur. A clue would be if the bad spots are present only where the wiper has been parked for a while. If this does turn out to be an issue, one solution would be putting a resistor from the wiper to the appropriate end of the pot: the high end to source current into the wiper, or the low end to sink current out of the wiper. A large value resistor (e.g., 1 Meg) would probably do the trick and not affect the range of the pot.

Like I mentioned, it's just a SWAG, but wanted to pass it along. Maybe the engineer at Bourns could review (if he gets back), or maybe some folks on SE who are more familiar with the topic could comment.

Hope you get to the bottom of this and get it resolved!

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.