Author Topic: Manufacturing custom high-precision low-TC resistors  (Read 6411 times)

0 Members and 1 Guest are viewing this topic.

Offline Edwin G. Pettis

  • Frequent Contributor
  • **
  • Posts: 375
  • Country: us
  • The plural of anecdote is not data.
Re: Manufacturing custom high-precision low-TC resistors
« Reply #75 on: February 01, 2018, 12:24:48 pm »
David:

You're welcome, no reason why you can't try it and see what happens.  Actually it was my colleague Michael Chesselet that was at Shellcross, he was the front man so to speak and I was consulting from here.  He was able to move around more than I could.  We had a presence at quite a few of the resistor houses around the country over the years.  At one point, Vishay tried to hire me as an applications engineer and Mike as consultant but that's another story.
 

Offline ramon

  • Contributor
  • Posts: 43
  • Country: tw
Re: Manufacturing custom high-precision low-TC resistors
« Reply #76 on: February 16, 2018, 12:36:27 am »
Edwin,

When I asked you for help to find wire I didn't mean that I want the same wire provider and specifications that you use on your resistors. Of course, I don't want you to disclose (to any competitor) any information about your wire manufacturer and specifications.

I currently have a 0.03mm (.0012") spool. It has a high resistivity, 1920 ohms/m, and that's a good thing. But when working with such a thin wire diameter I really need to be careful handling it, and I also need to slow down everything. I cannot see the wire if it's not over a some white background. It bends so easily ... and I always lose the ends.

So I think I need a thicker wire: between 0.0024" and 0.0040" (0.06mm or 0.1mm). I also want it to be enamelled: to speed up winding and for compactness.  And I also want it to be of Carpenter's Evanohm S alloy. I've just found in their website/datasheet that Evanohm S has much better TCR coefficient than Evanohm R: 1ppm instead of 3ppm (Due to its different composition: Ni72, Cr20, Al3, Si1, and Mn4).

This wire is almost impossible to find for anyone that wants to experiment on his own.

I have been lucky lately. I have just found (and ordered) a enamelled spool of Evanohm 0.03mm with 0 ppm TCR between 25C and 125C. Yes, I don't need zero ppm wire to find out which parameter (substrate, winding, epoxy, leads ...) is changing the TRC, but it doesn't hurts either.

I haven't received the spool yet, so I don't know if this enamelled wire (extra thickness and stiff?) will be easier and faster to work with than my current bare 0.03mm wire. So still wondering if you could help us to order enamelled Evanohm S spools with .0024" - .0040" diameter and low TCR (1ppm or less).

Best regards,
Ramon
 

Offline Edwin G. Pettis

  • Frequent Contributor
  • **
  • Posts: 375
  • Country: us
  • The plural of anecdote is not data.
Re: Manufacturing custom high-precision low-TC resistors
« Reply #77 on: February 16, 2018, 06:54:03 am »
Ramon,

I know you weren't asking explicitly about my wire specs or source, that information would not directly help a competitor anyway since those specs are specific to my processes.

Welcome to the world of fine wire handling, what you have is at the bottom range of the mid-size diameters, Evanohm can be drawn down to a mere ,0004" diameter and if you think .0012" is tough to see and handle, try some of that <grinning>.  Some of the Evanohm variants are quite stiff and do not like to be wound around a small form, they will spring like a clock's main spring if let loose and it is a mess to straighten out.  Handling is difficult to say the least with the finer wire sizes.

The standard specification of TCR for Evanohm 'S' is 0±5PPM/°C, meaning that a given spool of wire can be anywhere within ±5 PPM and considered in spec.  Indeed a given batch of wire can be any TCR within that range, if it happens to be 0 TCR, they just happened to hit the mark during heat treating, it is not a real precise process and depends a great deal on the knowledge of the heat treat process and the alloy.  You can certainly 'order' a spool of 0 TCR but it isn't going to be cheap, base pricing is calculated upon the standard ±5 PPM spec, when you ask for a specific TCR, the marketing cash register starts dinging big time, the price is near exponential as the TCR approaches zero TCR.

Carpenter is quite possibly the most difficult alloy source to deal with, they absolutely do not like small potato buyers, I've had quotes from them and you will likely end up with something on the order of 10 lbs minimum unless they've changed lately, that is a lot of wire.  They will provide your spec willingly if you buy enough wire.  I generally do not see very much Carpenter alloy on the 'surplus' markets, why I don't know for sure, they are one of the largest alloy suppliers around along with Kanthal.  You were very lucky to find a spool with a zero TCR on it, those are relatively rare as the original owners usually won't part with that TCR wire.  Be advised, like most alloys, cold working the alloy will change the apparent TCR, it is a mechanical stress and adds to all of the other mechanical stresses that accumulate during resistor manufacturing.  As I've noted before, the finished wire TCR is inherent to the wire and cannot be changed by normal resistor manufacturing, it is the mechanical stresses introduced during manufacture that make the overall apparent TCR different and depending on the resistor manufacturer, they will have come up with various ways of trying to reduce that stress with varying success.

The wire enamel (and there are several classes of enamel) won't have too much effect on the 'ease' of handling it.

The short answer to your wire buy request is no, as I noted above, Carpenter's minimum is large and 0±1 PPM TCR would definitely be a special order from any wire source, including mine.  What TCR you end up with on a 'finished' resistor will be different than the ±1 PPM/°C you started with.  I am not trying to dissuade you from trying, I'm just trying to let everyone know how difficult it is to achieve very low TCR, there are so many variables involved, you just might get a resistor here and there with low TCR but it will likely be by accident.  The wire is only the beginning.

If I could persuade my supplier to produce 0±1 PPM wire, their minimum purchase will run pretty close to $600 my cost and depending on the wire size, how many feet that will get us won't be known until the quote is made.  If you can persuade enough people to purchase enough of the spool, I can absorb some of the wire myself if it is a size I need.
 

Offline ramon

  • Contributor
  • Posts: 43
  • Country: tw
Re: Manufacturing custom high-precision low-TC resistors
« Reply #78 on: February 18, 2018, 07:28:17 pm »
Yes, it's like a clock spring. If I touch or cut them wrong they will permanently curl. For me it's impossible to straighten out. Also If I put two or more wires together they will suddenly make a sailor knot. Impossible to untangle.

The first spool I bought was 1 oz. (29g) and around 3 miles length (5 Km). I think that is already a life-time supply of wire. The second and third spools I've recently ordered weights 4 oz (112 gr) and 5.5 oz (155 gr). That is 11 and 15 miles (18 and 24 Km). That is a lot of wire, I certainly don't know what I will do with 10 lbs.

The second spool is actually 0/3 ppm. 0 TCR between 25-150C, and 3 ppm between -55-25C. And the third one is -1/+1 ppm (-1 ppm between 25-150C; and 1ppm between -55-25C).

Coating says 'TE 40' and I have not found any information on Kanthal website. Also I don't know how I will remove it without damaging the wire.

As no one has expressed their interest I think no one other than me has interest in buying wire. I actually don't need to buy more spools (47 Km are already enough), It is just only for the reason of having another diameter much more 'easy' ... well, less difficult to handle. But 10 lbs is huge, If such an order is made I guess it will contain at least 20 spools. I don't think we will meet that number. (I gave some wire for free, and only three persons expressed interest.)

Don't know how often do you order wire. If yearly, quarterly, ... or if you already has a shelf full of spools and you do not need to order more wire in years to come. The idea was that it first benefits you. If for some reason you need some wire diameter (between .0024" - .0040") for some customer and you know that you will not use all the wire required in the MOQ. I thought it can benefit you (and us too) if we can share an order that otherwise will not be made (as you can choose to made the resistors with wire of different diameters from your stock). Ideally, everyone that joins the shared order will have it's own spool, so you don't need to re-spool it and everyone will have is own spool with low TCR labeled on it.
 

Offline TiN

  • Super Contributor
  • ***
  • Posts: 3610
  • Country: tw
  • xDevs.com
    • xDevs.com
Re: Manufacturing custom high-precision low-TC resistors
« Reply #79 on: February 18, 2018, 07:44:42 pm »
Quote
That is a lot of wire, I certainly don't know what I will do with 10 lbs.
Sell to next enthusiastic person, who get to thinking of making high-stability resistor is easy-peasy wiring some wire on bobin and job done.  :=\

On serious note, in production type designs and component production, I'd think most manufacturers prefer to overstock. Especially for high precision, this may result buying life-time supply for parts, as you never know when manufacturer will EOL that very specific component you need so much to ensure your design performance. Same reason why I try to buy things in as much multiple (resistors, connectors, LTZ chips, 8.5d DMMs....) as economically possible. Leadtime of such orders also not helping to "buy it when you need it" direction either.  ;) So I think Edwin likely to have many extra spools, proven spools and verified spools, that he will never sell at any cost, just to avoid the trouble of doing inbound QC and testing of each new batch of wire. It just makes no sense from business perspective. Once you figured magic formula that works, you'd do everything to keep it going. In the end, all that time spent on validating and testing worth way more than even 10lbs spool.

Same reason why high-performance resistors, ADC/DACs and opamps cost so much. Usually it's not because use of voodoo materials or expensive process, but because of expensive testing time. Making ATE production test system to test uA-offset and mV-offset opamp (and it's test time) is very different to making same principle but high-performance ATE system to test fA-level offsets and nV-level offsets.
« Last Edit: February 18, 2018, 07:48:44 pm by TiN »
xDevs.com YouTube | Have test gear documentation to share? Upload here! No size limits, firmware dumps and teardown photos welcome.
 

Offline ramon

  • Contributor
  • Posts: 43
  • Country: tw
Re: Manufacturing custom high-precision low-TC resistors
« Reply #80 on: February 19, 2018, 12:22:16 am »
I certainly never said it will be easy, and I cannot find that anyone else in this thread said that before.
Also, if you have read carefully I am not asking Edwin to provide one of his 'proven spools'.

DIY is exactly the opposite of production. Magnificient Bastard has already said before the advantages of that better than I can write.

Agreed.  But also, since Edwin is making resistors for profit, he has to keep the material and labor costs low so he can sell his resistors at a reasonable price.  There can be no doubt that there were compromises made in his resistor designs to hit a price target.  Conversely, a hobbyist can take all day long (if we want to) to make just one resistor, and since (we) are only making one (or a few), more expensive (even exotic) materials can be used, where that would be totally impractical for a resistor production line.  This means that (we) can make resistors even better than Edwin's (and even better than VPG foil resistors), if enough time and effort is put into it-- and this process will go much faster if we share information.  PWW resistors are an old technology-- so there is a lot of information available (for free) from the various patent filings and also from the many national metrology laboratories.  If you have access, there are also many peer-reviewed scientific papers available on this subject.

BTW, you don't need to wind wire on a bobin to get a high-stability resistor, you can just lay the wire straight. Have you read the NIST paper that describes the coaxial straight-wire evanohm resistor? "used as primary transfer standards for characterization of frequency dependence".

You will not find any manufacturer that does that for you even if you pay. 'Production issues'. The same reason you will never get a wirewound divider resistor from Edwin or any other manufacturer.
 

Offline ramon

  • Contributor
  • Posts: 43
  • Country: tw
Re: Manufacturing custom high-precision low-TC resistors
« Reply #81 on: February 19, 2018, 12:31:39 am »
Well, maybe I am wrong on that!  (So I will ask first)

Edwin, have you read the NIST paper about the straight coaxial resistor?

Is it possible that you can do that kind of primary coaxial resistor for us (the same way you provide LTZ1000 resistor kits) to allow us to have some kind of primary resistance for 8.5 dmm check?
 

Offline amspire

  • Super Contributor
  • ***
  • Posts: 3633
  • Country: au
Re: Manufacturing custom high-precision low-TC resistors
« Reply #82 on: February 19, 2018, 12:44:19 am »
Have you read the NIST paper that describes the coaxial straight-wire evanohm resistor? "used as primary transfer standards for characterization of frequency dependence".
It the paper you are talking about this one?

Calculable Coaxial Resistors for Precision Measurements
http://ws680.nist.gov/publication/get_pdf.cfm?pub_id=29665
 

Offline ramon

  • Contributor
  • Posts: 43
  • Country: tw
Re: Manufacturing custom high-precision low-TC resistors
« Reply #83 on: February 19, 2018, 03:20:57 am »
Yes, that same (four pages) document by Randolph E. Elmquist. Don't you think it's brilliant?

If you are going to get any trouble by winding the wire into a bobin, then use straight wire.
If you are going to get any trouble by having a some coating on evanohm wire, then use bare wire.
If you are going to get any trouble by applying any kind of epoxy to the resistor, then don't use any.
If you are going to get any trouble by welding evanohm to any dissimilar material, then weld it to evanohm itself.
If you are going to get any trouble by ...

I propose to extend that way of thinking further: and for proper attachment to 8.5 digits DMM we will also need Kelvin test leads made with evanohm wire welded into a 100% evanohm-S mounting post (or banana plug). Just kidding (... or not?)

Seriously. Going even further, If we take it to the extreme:

If we are going to get any trouble by doing a weld, then do not weld it.
If we are going to get any trouble with leads attachment, then don't use them.
 

Offline TiN

  • Super Contributor
  • ***
  • Posts: 3610
  • Country: tw
  • xDevs.com
    • xDevs.com
Re: Manufacturing custom high-precision low-TC resistors
« Reply #84 on: February 19, 2018, 04:48:51 am »
I'm just lost on what is the end goal of all this trouble?
Resistors and their specs are built to purpose (usually). What is it in here for DIY?
xDevs.com YouTube | Have test gear documentation to share? Upload here! No size limits, firmware dumps and teardown photos welcome.
 

Offline zhtoor

  • Frequent Contributor
  • **
  • Banned!
  • Posts: 337
  • Country: pk
Re: Manufacturing custom high-precision low-TC resistors
« Reply #85 on: February 19, 2018, 06:16:44 am »
and it also appears to be the best solution for very high and very low values.  There also needs to be a parallel effort in developing DIY hermetic packages.

hello,

i am not so sure about "very high" (100K +) and "very low" (100R -) to start with,
middle ground should be the more reasonable target.

and to do DIY PWW's a starter's approach may be use a manganin / zeranin (solderable)
enamelled wire selected for a -ve tempco (say around -5ppm / degC) and use a copper-wire
compensator to "tune" the total TCR.

as far as DIY hermetic packages go,

how about taking the top off a 2N3055, disconnecting / removing the transistor die
and mounting the resultant 2 1/2 terminal posts into a milled aluminum box and
making a metal-metal seal using some kind of loctite or "head gasket" sealing shellac.

regards, and btw i agree with you 100%

-zia
« Last Edit: February 19, 2018, 06:22:54 am by zhtoor »
 

Offline TiN

  • Super Contributor
  • ***
  • Posts: 3610
  • Country: tw
  • xDevs.com
    • xDevs.com
Re: Manufacturing custom high-precision low-TC resistors
« Reply #86 on: February 19, 2018, 06:18:15 am »
Magnificent Bastard,

I didn't say it is useless, quite the opposite actually :). I spent hundreds of hours and $$$$$ on gear, that I don't need for day-day work,
but if I go and post on forums, I usually try to specify what the goal is. This way other people can suggest some ideas towards the goal.
Right now it's a lot of guesses here, but not clear direction where we going. And boundaries in resistor technology for metrology today advanced far beyond ppm level ;).

There is not much learning in speculating about this and that, learning means the need in experimenting and documenting of observed results.

« Last Edit: February 19, 2018, 06:20:20 am by TiN »
xDevs.com YouTube | Have test gear documentation to share? Upload here! No size limits, firmware dumps and teardown photos welcome.
 
The following users thanked this post: SeanB, 2N3055

Offline GerryBags

  • Frequent Contributor
  • **
  • Posts: 334
  • Country: gb
Re: Manufacturing custom high-precision low-TC resistors
« Reply #87 on: February 19, 2018, 06:48:33 am »
Forgive me butting in with what may be an ignorant suggestion (especially if it has already been brought up and I missed it), but apparently silver solder is used to join evanohm wire to terminal hooks made from nickel plated oxygen-free copper.

 I know a few people who have been involved in the Hatton Garden jewellery trade for generations, who may know more reliable sources for SS than The Bay, I must remember to ask.
 

Offline branadic

  • Super Contributor
  • ***
  • Posts: 1130
  • Country: de
Re: Manufacturing custom high-precision low-TC resistors
« Reply #88 on: March 13, 2018, 06:55:26 am »
Ordered some hermetic feedthroughs from they bay:

Glass Hermetic 1 Pin Feedthrough Size 7.1mm x 4.5 mm
Glass Hermetic 1 Pin Feedthrough Size 10.4mm x 5 mm

that received today. Together with some brass or copper tube it's possible to create hermetic resistor cases. Here is a frist put together.

-branadic-
Prema 5000 | 2x Prema 5017 SC | GenRad 1434-G | Tek 2465A | VNWA2.x with TCXO upgrade and access to: Keysight 3458A, Keithley 2002, Prema 5017 SC, 34401A, 34410A, Keithley 2182A, HDO6054, Keysight 53230A and other goodies at work
 
The following users thanked this post: TiN, Andreas, cellularmitosis

Offline babysitter

  • Frequent Contributor
  • **
  • Posts: 751
  • Country: de
  • pushing silicon at work
Re: Manufacturing custom high-precision low-TC resistors
« Reply #89 on: March 13, 2018, 09:34:39 am »
Ordered some hermetic feedthroughs from they bay:

Glass Hermetic 1 Pin Feedthrough Size 7.1mm x 4.5 mm
Glass Hermetic 1 Pin Feedthrough Size 10.4mm x 5 mm

that received today. Together with some brass or copper tube it's possible to create hermetic resistor cases. Here is a frist put together.

-branadic-

I imagine a smaller brass tube sticking out, soldered shut to keep some dry gas inside. Also, a Pt100 glued to the housing. And resistance wire wrapped around everything.

I'm not a feature, I'm a bug! ARC DG3HDA
 

Offline Gyro

  • Super Contributor
  • ***
  • Posts: 3604
  • Country: gb
Re: Manufacturing custom high-precision low-TC resistors
« Reply #90 on: March 13, 2018, 09:55:55 am »
I have some of those feedthroughs myself and have similar ideas (possibly for a complete reference circuit).

Just looking at your arrangement, it strikes me that thermal expansion of the tube could put significant axial stresses on the resistor. I would suggest including strain relief loops in the leads... or maybe bring them out radially from a soldered copper foil box rather than a rigid tube.
Chris

"Victor Meldrew, the Crimson Avenger!"
 

Offline branadic

  • Super Contributor
  • ***
  • Posts: 1130
  • Country: de
Re: Manufacturing custom high-precision low-TC resistors
« Reply #91 on: March 13, 2018, 06:59:21 pm »
Quote
Just looking at your arrangement, it strikes me that thermal expansion of the tube could put significant axial stresses on the resistor.

Well, looking at the CTE of the different materials there is only a slight mismatch between the copper leads and the brass tube. However, I will use a tinned copper tube instead, but for a first put together I used what I had laying around.

CTE brass: 18.4 ppm/K
CTE copper: 16.5 ppm/K

-branadic-
Prema 5000 | 2x Prema 5017 SC | GenRad 1434-G | Tek 2465A | VNWA2.x with TCXO upgrade and access to: Keysight 3458A, Keithley 2002, Prema 5017 SC, 34401A, 34410A, Keithley 2182A, HDO6054, Keysight 53230A and other goodies at work
 

Offline Gyro

  • Super Contributor
  • ***
  • Posts: 3604
  • Country: gb
Re: Manufacturing custom high-precision low-TC resistors
« Reply #92 on: March 13, 2018, 08:48:12 pm »
I was thinking more of the length of resistor package that isn't copper wire - the bobbin etc. I'd still go for a slightly longer tube to leave space for strain loops.
Chris

"Victor Meldrew, the Crimson Avenger!"
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf