Manufacturing custom high-precision low-TC resistors

[TiN]

And failure is the best teacher, the more you fail the better you learn. Especially when each attempt tied to significant cost.
That will optimize your efforts a great deal. 

[MisterDiodes]

Exactly...You learn about a ten million ways to do it wrong, and then you begin to appreciate how hard it is to get right.

The really difficult part is testing a resistor on week one and it looks pretty good, maybe you got 8 or 10ppm TC if you're lucky - and then come back 6 months or 6 years later to find it absolute crap for stability and high drift rate.  You might be able to claw your way to lower PPM but after a while just purchasing quality resistors is a much more practical solution if you want to get your circuit idea completed sometime this century.

In the end what you might learn is: you're grateful to be able to buy a lot of custom precision and stability for $5~$10 each in a quality PWW (and spend your time designing areas of a circuit that you CAN solve)...but you had fun learning why they are very hard to build along the way. 

Have fun!

Again:  If you haven't done so and think making a PWW resistor means wrapping up some wire and making a couple crimped connections - Make sure you read Edwin's excellent article that appeared in EDN and learn what's been done long ago, and why various techniques that seem obvious solutions won't work.:

https://www.edn.com/design/analog/4427151/The-last-half-century--Wirewound-resistors-Part-one

https://www.edn.com/design/analog/4427940/The-last-half-century--Wirewound-resistors-Part-two

[ramon]

Back to the basics--

Why could not the wire be terminated by some mechanical means? Some sort of screw terminal?

(No one would want this in production, but should be ok for hobbyist use, surely).

Well, the position of the wire can move, changing the value of the resistor and/or the joint can corrode, changing the value of the resistor-- and also, there might be more thermal EMF than a properly welded joint.  Just my $0.02 worth...

What if we apply a silver coating at the ends? Won't this solve all this three issues at once?

Too easy, right? I guess someone tried that before and failed.

[zhtoor]

Back to the basics--

Why could not the wire be terminated by some mechanical means? Some sort of screw terminal?

(No one would want this in production, but should be ok for hobbyist use, surely).

Well, the position of the wire can move, changing the value of the resistor and/or the joint can corrode, changing the value of the resistor-- and also, there might be more thermal EMF than a properly welded joint.  Just my $0.02 worth...

What if we apply a silver coating at the ends? Won't this solve all this three issues at once?

Too easy, right? I guess someone tried that before and failed.

interesting... high conductivity and solderability, but coatings could result in semiconductor action resulting
in maybe a diode in series with a resistor.

regards.

-zia

[David Hess]

What if we apply a silver coating at the ends? Won't this solve all this three issues at once?

Too easy, right? I guess someone tried that before and failed.

interesting... high conductivity and solderability, but coatings could result in semiconductor action resulting
in maybe a diode in series with a resistor.

I have seen this done with copper but not silver.  One problem with silver is that the tin in the solder would very quickly dissolve it unless silver bearing solder was used.

[branadic]

In therory there is a possibility to use homebrew galvanic kits to coat some nickel + gold or silver on the end of the wire, to make it solderable. You don't want to mess around with chemical deposition. But don't forget to remove the enamel.

-branadic-

[Edwin G. Pettis]

Ramon:  Besides Goodfellow, this source has various wire wlloys available:

https://www.pelicanwire.com/product-category/resistanceheating/insulated-resistance-wire/?filter_alloy=advance-294&filter_product-availability=in-stock

As I've mentioned before, resistance wire is not cheap, particularly good wire and there is a lot of questionable wire floating around on the internet.  Sometimes you will find some 'surplus' wire available on eBay for a good price but I haven't seen much showing up there.

All alloys have a standard ±xxxPPM/°C range, a given spool will be somewhere within that range.  The only way around getting a random TCR on a spool is to order wire with a specific TCR such as I do, and yes that does cost a pretty penny.  For the finest wire gauges, I can expect to pay upwards of $20,000/lb., that isn't a typo.  Of course nobody orders a full pound at those prices.  The more common wire sizes that I use for LTZ resistors are still mostly above $1,500/lb.  That is one of the reasons why PWW resistors aren't bargain basement cheap, every manufacturer pays roughly the same high costs for wire, mine is a bit higher because of my specifications.  You simply can't make a low TCR precision resistor with cheap wire.  Wire costs have ballooned since about 2000, they have stabilized in the last few years but are much more expensive than they used to be, I've got wire quotes from some years ago and current pricing is just about 300% higher.  If you're wondering, film/foil uses the same/very similar alloys as I do, just in a different form.

[zhtoor]

Hello,

question for Mr. Edwin:

your comments on silver plated evanohm wire soldered to terminals, what would you expect in terms of TCR?

regards.

-zia

[Conrad Hoffman]

Many years ago I walked into H.P. Reid in Palm Coast Florida and asked if they had any partial spools I could buy. They fixed me up with a nice quantity of manganin and Rediohm 800 for under a couple hundred bucks. Times have changed and I doubt they'd (now Kanthal) do it today, at least not for what I paid back then.

[Edwin G. Pettis]

Yes, H. P. Reid was the best resistance wire supplier in the country, sadly (in a manner of speaking), they were bought by Kanthal about ten years ago, while the quality of wire is still top notch, they are expensive.  I don't know if they still would sell to 'walk-ins' these days or not.  I do know that they have maintained a very small inventory of unsold wire from customer runs that didn't require the minimum line item purchase the last time I checked in with them.  I have no doubt that you won't get wire for anywhere near what you paid years ago.  The commodity metals market went crazy after 2000 and while some metals have come down in cost since those crazy days, there has been no such drop in wire pricing that I can see.

[Edwin G. Pettis]

Silver has a TCR of +3800PPM/°C, close to copper and while silver is 'friendly' with nickle, it will add to the overall TCR, how much depends on what percentage its resistance is to the Evanohm resistance, this will include the 'interface' resistance between the silver plating and the Evanohm.  In other words, you are adding two small silver resistors in series with the Evanohm and just like the copper leads will add a small resistance in series.  EMF wise, these three materials, Evanohm, Silver and Copper all have close compatibility so thermal generation isn't going to be a big issue.  Of course, low resistance values will show a larger effect than on higher values.

[zhtoor]

Silver has a TCR of +3800PPM/°C, close to copper and while silver is 'friendly' with nickle, it will add to the overall TCR, how much depends on what percentage its resistance is to the Evanohm resistance, this will include the 'interface' resistance between the silver plating and the Evanohm.  In other words, you are adding two small silver resistors in series with the Evanohm and just like the copper leads will add a small resistance in series.  EMF wise, these three materials, Evanohm, Silver and Copper all have close compatibility so thermal generation isn't going to be a big issue.  Of course, low resistance values will show a larger effect than on higher values.

thanks.

it seems that this is a good solution for DIY PWW's at least for medium ohmage 100 Ohm -> 100K Ohm,
assuming EVANOHM is available with a reasonable (sub 5ppm) TCR.

regards.

-zia

[Echo88]

So we now need someone who can galvanize stuff, ramon ships him some of his wire, branadic prints the needed bobbins and now we can get in competition with Vishay and Edwin.   

[zhtoor]

So we now need someone who can galvanize stuff, ramon ships him some of his wire, branadic prints the needed bobbins and now we can get in competition with Vishay and Edwin.   

Andreas/TiN : TCR Testing
Ramon: Wire Source
Branadic: Bobbin (Maybe collapsable for stress relief)
Xxxxxxx: Silver Plating (preferrably electroless)

if silver plating can be found in Germany, most of the work can be done in Germany 

regards.

-zia

[Conrad Hoffman]

One of the old precision resistor companies was Julie research. They made KVDs and various other things. I have some of their old literature and they were quite proud of their resistors. Loebe Julie was a bit of a character. By modern standards, does anybody know how good their resistors were? I know my KVD (serial #1) has remained within a couple ppm.

[IconicPCB]

I do immersion silver plating.

[zhtoor]

Loebe Julie was a bit of a character.

one hell of a chap.
definitely a man on par with (or exceeding) the jim williams and pease, its a pity he was not as "visible"
as others, and as many other greats, was not treated in the best of ways in the US.
such a man should have been revered by the electronics engineering community.

regards.

-zia

[zhtoor]

I do immersion silver plating.

maybe you could post the method to immersion silver plate the wire by ramon?
without changing the TCR of the wire.

regards.

-zia

[ramon]

Edwin:

Thank you so much. Actually I already saw that pelicanwire had evanohm and several months ago asked them about their spools. What they sell on the website is wire by length (50, 100 or 250 ft) without any other information. I didn't wanted that. As you know, the manufacturer sell spools by weight, and they specify the TCR in the spool label, and put some serial number or code to track any quality issue. However pelicanwire doesn't have any single picture of the actual spool. So far I have no response from them (don't know if that was because I clearly said in my first contact that I was in Taiwan, so maybe they don't sell overseas).

The only manufacturer that didn't completely ignored my quote request was Jelliff. They asked me what I wanted, how much quantity ... and that's it. So I am very happy with them that the took the time to write a few emails.

So, as last resource, I bought the wire on a famous auction site. According to the seller it was from military surplus. It has at least a serial number, the actual weight (gross/tare/net), the measured resistance per length, and the TCR specified. I don't expect less than that.

I completely understand that some business don't want to deal with one-off customers, or low quantity orders. At work, sometimes a customer wants to buy some product we don't have stock, and production is costly, and our providers ask for a minimum order quantity, and we ask the customer too for a minimum order quantity or we offer them an alternative product in stock. But not a single alloy distributor or manufacturer have ever replied to any quote I asked, they haven't either told me the required MOQ.

And this reminds me that we are really lucky to have your presence here in this forum. It is amazing that we have one of the most respected manufacturers willing to sell a kit of 5 precison resistors to individuals all over the world (and even discussing how to make them !). With all the troubles, and loss of time, that that will carry. Exactly the reason that no single manufacturer or distributor has ever replied to me. So, I just want you express grattitude for what you are doing.

I wonder If you ever had that problem too. Don't know If you have ever needed some spool gauge at hand for some customer request but you couldn't meet the MOQ from your suppliers, then you was forced to look for alternatives or even deny the job.

Now I ask you (because you already offered to sell resistors) if you could sell a high quality spool of wire too?

I suppose that it will not be that much different to sell the bare spool than the finished resistor. It can be even less work for you, and you maybe can have the added benefit of any volume discount (well, maybe not, as not sure how many here will be interesed too). For me (or us), the benefits are that we can get a good spool from a reliable source. Also some people want to do some special resistors (like a PWW resistor network) that no manufacturer is willing to make just only because it doesnt fit the production line, then they will have a way to make it. Obviously, we are not going to get the performance of your resistors, so this will not decrease any of your sales. And those who want a high quality resistor will buy it directly from you anyway, as they will not want to loose their time and mess with winding wire.
What do you think?

[Edwin G. Pettis]

Hello Ramon,

First, thank you for the compliments.  You are quite right about sellers like Pelican and Goodfellow, they buy wire and resell it in small quantities for a significant markup, they fill a niche for buyers who don’t need a lot of wire or can’t meet minimum purchase.  They really should supply the wire data for the wire they sell since all wire manufacturers provide that information on the spools, most resistor manufacturers would not accept such unlabeled spools.  MWS is another source which does not provide wire to specific TCR as I have talked with them and they won’t provide anything beyond the stated ‘normal’ range of TCR, such as ±20 PPM/°C for Evanohm.  They may have changed that policy since I last talked to them but I kind of doubt it.  When I asked them for a given TCR, all the salesmen will say it’s within ±20PPM/°C.  There aren’t that many ‘original’ wire sources around and they know it so they can be selective in their customers and have no interest in small orders.  It may be that is it simply too expensive for them to handle small orders but that doesn’t mean they shouldn’t at least reply to inquiries.  More than likely, if you’re using an individual’s e-mail, they won’t reply, a business e-mail might at least get a burp out of them.

In general, a PWW resistor house can make any value of resistor if they have the proper wire/bobbin size available, some will charge a premium for nonstandard values but that is more marketing policy than real cost involved.  I will make any resistor value if I have the wire and bobbin size available in inventory and there is no premium for oddball values.  The main limitation I have is limited inventory of wire and bobbin sizes available in stock, replacement costs are high so it is difficult to increase the number of wire sizes I have.  Yes I have had to no bid some resistors due to inventory limitations.  Unless it is a rather large order that can support the purchase of inventory to do it, I have to turn it down.  Another limitation is very high values, since the huge metal price increases since 2000, wire cost has nearly tripled and that makes resistors up around >500K more expensive.  I’ve generally only made high values to repair ESI decades or instruments that need PWWs because factory replacements are even more expensive.

On the one hand, I don’t have any problems purchasing wire, that said, I’m still subject to their minimum order and I’ll also note that because they make the wire to my specifications, it raises the price from the standard stuff.  The price of the wire depends on the wire size, larger sizes are lower cost, finer wire more expensive.  Each purchase is quoted at the time and sometimes, they’ll change the quote more than once before all is said and done.  How much wire is on a given spool also varies by the size (you’ll normally see the weight given on the spool), most of the time a given spool is in the hundreds of dollars each or higher.

I know this is going to sound self-serving but….while I encourage experimentation, if you really need <5PPM/°C TCRs for a project, you’re much better off in the long run to just buy reputable resistors.  As proven in actual industry, it took decades to get TCRs down below 10PPM/°C with reasonable yields.  It isn’t a case of somebody just tripping over a solution, it requires considerable knowledge of all of the factors that goes into a low TCR and stable resistor and most resistor houses are still struggling with many of those factors, the customer just never is privy to all that goes on behind closed doors and manufacturers don’t like their faults being put on display very much for some unknown reason (tongue in cheek).

One of things I see here is that a given parameter such as TCR is latched onto and is pursued, unfortunately unless all of the given parameters of construction are correct, the wire’s TCR is the least of your worries.  I can guarantee that even if you have wire with 0PPM TCR, the end result is not going to be even close to that.  There is a laundry list of parameters that will affect the finished TCR of a resistor and stability.

Experimenting will help you learn how things work and what causes this effect or that effect or even more than one effect but you don’t need very low TCR wire to do that.

Best regards,

Edwin

[Conrad Hoffman]

I notice that when I search for Rediohm 800, it's used for biological work (cell recording?), and not resistors. It's also about $2 per foot! That's just crazy.

Does anybody know how Julie terminated the wire in their resistors, especially the NBS resistors?

Because the thermal conductivity of resistance wire is low, thermal effects should be local to the joint. How much of an effect might be expected if a very short portion of the wire was overheated by brazing it into a small copper socket or similar? It seems like all forms of termination involve some degree of compromise.

[Edwin G. Pettis]

Rediohm800 is what H. P. Reid called their recipe for Evanohm, there are a number of slight (and I do mean slight) variations on the metallurgy of the original Evanohm developed back in the 1940s but basically there is little difference.  The mixture is tweaked to get a specific characteristic of the end result.  Every batch (ingot) of an alloy is slightly different, there are limits as to how precisely the metals mix can be controlled.  Each batch has to be processed slightly different in order to get a specific characteristic, such as TCR, depending on how good things go, they might hit the target on the first round, if not, they have to start over and redo the heat treatment process from the start.  All of this has a bearing on final cost.

If you're thinking that Loeb had some special termination that was better than others, not so much, he had yield problems like everybody else and some of his claims were disproved in the end.  Loeb was a very good engineer for that matter, but when it came to resistors he had manufacturing problems like everybody else.  Like most manufacturers, they don't like having their flaws put out in the public and it often leads to a silly argument merely to 'protect' inside knowledge.  In the resistor industry, there are very few real secrets if truth be told, most industry knowledge has been spread around over the years so the any given manufacturer is using whatever techniques they think works for them and if you read my white paper on the history of resistors, you'll find a whole lot of 'borrowing' (to be kind) that has gone on for years.

The problem here when shooting for very low TCR is that even localized effects can have significant effect on the overall end product.  You are quite right, all forms of termination involve some form of compromise, a good match of materials and processes can produce a very low TCR consistently but that is not easy to do even with the right materials and processes.  Years ago when we did a long term study of resistor terminations with the help of Northrup Metallurgy Labs, brazing did not produce a particularly good termination, it did produce a better termination than many commonly used methods in the industry for that matter but that isn't saying much, a little better isn't really good enough for very low TCRs and long term stability.  Resistor termination has been a thorn in the industry for many years.

I don't recall specific numbers for brazed terminations but they were still significant in the 0-10 PPM/°C range.  Short of a correctly welded joint, a proper brazed joint was about the middle of the range of termination effects as measured by Northrup.

[branadic]

I found an interesting paper by NIST, that could be of interest for you and give you some inspiration of what is possible with the resistor wire:

Calculable Coaxial Resistor for Precision Measurment

-branadic-

[Conrad Hoffman]

Edwin, that's great info- thanks! I may still try a braze. I've got a Smith "Little Torch" that's capable of very tiny work and I should be able to keep the area exposed to high heat reasonably short. Ultra low TC is nice, but I can live with a bit higher, so long as the resistance is stable over the years. I note you worked at Shallcross. Though I have many GR resistance boxes, the Shallcross is the one I use the most. Not terribly sexy, but a great product.

[cellularmitosis]

So, I just want you express grattitude for what you are doing.

Me too!