My experience is that boxes makes the time constants for humidity longer but does not take it away. Heating in boxes takes the change due to humidity down a little but it doesn´t go away. I have had boxes with LM399 and amps for 7-10v with wirewounds for more than 20 years and they have definitely had seasonal variations. Also if you have a completely sealed box with varying temperatures the relative humidity in the box changes if the humidity is not close to zero. A 10V 2ppm ref on Ebay was said to have no humidity effect but my unit definitely varied about 20ppm due to humidity during a long humidity test. One guess I have is that some units have much longer time constants due to better sealing but I am quite sure it doesn´t go away complete. Another problem with humidity I have seen is that you may be cheated by humidity changing over the season and cancelling the aging. This happened during my first tests of AD587KRZ many years ago. During the first four months two samples both had less than 2ppm drift and the humidity in the room changed from 45 to 25%RH so they looked very promising. But during the next four months when the humidity went up they drifted 15ppm. Probably the 5 and 2ppm/C versions is manufactured in the same way. The humidity graph under severe humidity and temperature testing shows a 500pm difference for your chosen resistors. So doesn´t say so much. Hopefully they are below 1ppm/%RH with low loads. Lars Från: Ian Johnston<mailto:ian@ianjohnston.com> Skickat: den 3 juni 2016 20:42 Till: volt-nuts@febo.com<mailto:volt-nuts@febo.com> Ämne: Re: [volt-nuts] LTZ1000 project build Hi, The product I designed to use the 2ppm/degC resistors is sealed in an enclosure where the LM399AH etc live, so I don't expect much problem with humidity. I use 2K2, 3.3K, 10K & 22K. It's the RG2012L series from Susumu:- http://datasheet.octopart.com/RG2012L-103-L-T05-Susumu-datasheet-13139546.pdf http://www.susumu.co.jp/common/pdf/n_catalog_partition01_en.pdf The 2nd PDF above is mainly for the 5ppm versions but does contains various graphs including humidity. Ian. On 03/06/2016 17:01, Lars Walenius wrote: > Hi, > > Do you have any idea of the aging and humidity sensitivity of these 2ppm/C resistors? > > I have tested many resistors during the years and my opinion is that tempco is not the biggest problem for normal nuts use but of course may be if you sell a product. For a nut it is quite easy to check the tempco but not aging and humidity sensitivity. > > Of the resistors I have tested it is only hermetical sealed resistors that I don´t see a humidity sensitivity on. Even if they are bulk metal foil, wire wounds or metal films with low tempco´s they seem to have very varying humidity sensitivity if not sealed. Lower values of resistance (100-1k) normally seems to be better than 10-100kohm values that in all families may have up to 1-2ppm/%RH. > Last year I bought wire wounds, to be used in a LTZ1000 based design, that were supposed to be insensitive to humidity but they were not. They were even worse than other brands of WW and had several months of time constant. After a while the manufacturer admitted that it was a problem in the manufacturing and they were humidity sensitive. So far I haven´t received a replacement. > > Of course if you have a resistor with 2ppm/%RH and a seasonal variation of 50%RH it is only maximum 1ppm seasonal variation on the output on the LTZ. That is not easy to measure for most of us nuts. > > Lars > > Från: ian@ianjohnston.com<mailto:ian@ianjohnston.com> > Skickat: den 26 maj 2016 13:35 > Till: illya@xdevs.com<mailto:illya@xdevs.com>; volt-nuts@febo.com<mailto:volt-nuts@febo.com> > Ämne: Re: [volt-nuts] LTZ1000 project build > > Hi all, > > TiN, out of interest I spent ages fishing for low ppm/degC resistors for my own reference......ended up with 2ppm units from DigiKey. Yes, they are expensive and send the BOM into the clouds!..............so I know how you feel! > > Ian. > > > ----- Original Message ----- > From: Illya Tsemenko [mailto:illya@xdevs.com] > To: volt-nuts@febo.com > Sent: Thu, 26 May 2016 18:42:54 +0800 > Subject: [volt-nuts] LTZ1000 project build > > I'm glad my little project got so much attention. Worth to mention few > points regarding my (or any other LTZ1000) unit: > > A. One should consider cost of whole BOM, not only LTZ chip. In my case > resistors from VPG exceed cost of LTZ1000 by far. > B. PCB I designed was targeted best to my own application need, where > compact size and direct output (whatever is coming out of LTZ) were key > importance. Since modules are part of bigger system with DAC synth, > specific value, like 7.1500000 was not important for me. > C. If there is large enough group to have this project going, I can > actually modify original design (e.g. for easier hand soldering, or > added buffer or ratio circuits, onboard LDO, you name it) and get PCBs > made for interested nuts, at small cost. > > You can also find lot of temco/stability data with > temperature/environment logs with mentioned LTZ modules on my site here: > https://xdevs.com/services/ in LTZ1000 voltage reference tests section. > RAW DSV data are available. I still have 3 modules at hand for any > testing/checks (one was sold while ago, one was installed in HP 3245A > source to replace LM399). > > Illya "TiN" Tsemenko > > _______________________________________________ > volt-nuts mailing list -- volt-nuts@febo.com > To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts > and follow the instructions there. > _______________________________________________ > volt-nuts mailing list -- volt-nuts@febo.com > To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts > and follow the instructions there. > _______________________________________________ > volt-nuts mailing list -- volt-nuts@febo.com > To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts > and follow the instructions there. -- Ian Johnston www.ianjohnston.com https://www.youtube.com/user/IanScottJohnston /"Because it feels good to make stuff!"/ _______________________________________________ volt-nuts mailing list -- volt-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts and follow the instructions there.

Could this be caused by ionic contamination of the printed circuit board? The quick test I have used for this is to blow on the board with a straw. In extreme cases solving it usually involved dropping the board into an ultrasonic cleaner with water and a bit of dish soap or TSP, then a rinse in pure (deionized, reverse osmosis, distilled, whatever) water, and then another cycle in the ultrasonic cleaner with pure water. Dry with an air compressor and then bake out in an oven. In production, we had to run boards through a dishwasher using detergent and pure water before loading them with parts. On Fri, 3 Jun 2016 23:47:05 +0200, you wrote: > >My experience is that boxes makes the time constants for humidity longer but does not take it away. Heating in boxes takes the change due to humidity down a little but it doesn´t go away. I have had boxes with LM399 and amps for 7-10v with wirewounds for more than 20 years and they have definitely had seasonal variations. Also if you have a completely sealed box with varying temperatures the relative humidity in the box changes if the humidity is not close to zero. > >A 10V 2ppm ref on Ebay was said to have no humidity effect but my unit definitely varied about 20ppm due to humidity during a long humidity test. One guess I have is that some units have much longer time constants due to better sealing but I am quite sure it doesn´t go away complete. Another problem with humidity I have seen is that you may be cheated by humidity changing over the season and cancelling the aging. This happened during my first tests of AD587KRZ many years ago. During the first four months two samples both had less than 2ppm drift and the humidity in the room changed from 45 to 25%RH so they looked very promising. But during the next four months when the humidity went up they drifted 15ppm. > >Probably the 5 and 2ppm/C versions is manufactured in the same way. The humidity graph under severe humidity and temperature testing shows a 500pm difference for your chosen resistors. So doesn´t say so much. Hopefully they are below 1ppm/%RH with low loads. > >Lars

the issue of resistor drifts is mainly related to the amplifier (7 to 10V) gain resistors, the other resistor drifts do have a very limited impact on the output (see data sheet). > Gesendet: Freitag, 03. Juni 2016 um 23:47 Uhr > Von: "Lars Walenius" <lars.walenius@hotmail.com> > An: "volt-nuts@febo.com" <volt-nuts@febo.com> > Betreff: Re: [volt-nuts] LTZ1000 project build > > > My experience is that boxes makes the time constants for humidity longer but does not take it away. Heating in boxes takes the change due to humidity down a little but it doesn´t go away. I have had boxes with LM399 and amps for 7-10v with wirewounds for more than 20 years and they have definitely had seasonal variations. Also if you have a completely sealed box with varying temperatures the relative humidity in the box changes if the humidity is not close to zero. > > A 10V 2ppm ref on Ebay was said to have no humidity effect but my unit definitely varied about 20ppm due to humidity during a long humidity test. One guess I have is that some units have much longer time constants due to better sealing but I am quite sure it doesn´t go away complete. Another problem with humidity I have seen is that you may be cheated by humidity changing over the season and cancelling the aging. This happened during my first tests of AD587KRZ many years ago. During the first four months two samples both had less than 2ppm drift and the humidity in the room changed from 45 to 25%RH so they looked very promising. But during the next four months when the humidity went up they drifted 15ppm. > > Probably the 5 and 2ppm/C versions is manufactured in the same way. The humidity graph under severe humidity and temperature testing shows a 500pm difference for your chosen resistors. So doesn´t say so much. Hopefully they are below 1ppm/%RH with low loads. > > Lars > > Från: Ian Johnston<mailto:ian@ianjohnston.com> > Skickat: den 3 juni 2016 20:42 > Till: volt-nuts@febo.com<mailto:volt-nuts@febo.com> > Ämne: Re: [volt-nuts] LTZ1000 project build > > Hi, > > The product I designed to use the 2ppm/degC resistors is sealed in an > enclosure where the LM399AH etc live, so I don't expect much problem > with humidity. > > I use 2K2, 3.3K, 10K & 22K. > > It's the RG2012L series from Susumu:- > > http://datasheet.octopart.com/RG2012L-103-L-T05-Susumu-datasheet-13139546.pdf > http://www.susumu.co.jp/common/pdf/n_catalog_partition01_en.pdf > > The 2nd PDF above is mainly for the 5ppm versions but does contains > various graphs including humidity. > > Ian. > > > > On 03/06/2016 17:01, Lars Walenius wrote: > > Hi, > > > > Do you have any idea of the aging and humidity sensitivity of these 2ppm/C resistors? > > > > I have tested many resistors during the years and my opinion is that tempco is not the biggest problem for normal nuts use but of course may be if you sell a product. For a nut it is quite easy to check the tempco but not aging and humidity sensitivity. > > > > Of the resistors I have tested it is only hermetical sealed resistors that I don´t see a humidity sensitivity on. Even if they are bulk metal foil, wire wounds or metal films with low tempco´s they seem to have very varying humidity sensitivity if not sealed. Lower values of resistance (100-1k) normally seems to be better than 10-100kohm values that in all families may have up to 1-2ppm/%RH. > > Last year I bought wire wounds, to be used in a LTZ1000 based design, that were supposed to be insensitive to humidity but they were not. They were even worse than other brands of WW and had several months of time constant. After a while the manufacturer admitted that it was a problem in the manufacturing and they were humidity sensitive. So far I haven´t received a replacement. > > > > Of course if you have a resistor with 2ppm/%RH and a seasonal variation of 50%RH it is only maximum 1ppm seasonal variation on the output on the LTZ. That is not easy to measure for most of us nuts. > > > > Lars > > > > Från: ian@ianjohnston.com<mailto:ian@ianjohnston.com> > > Skickat: den 26 maj 2016 13:35 > > Till: illya@xdevs.com<mailto:illya@xdevs.com>; volt-nuts@febo.com<mailto:volt-nuts@febo.com> > > Ämne: Re: [volt-nuts] LTZ1000 project build > > > > Hi all, > > > > TiN, out of interest I spent ages fishing for low ppm/degC resistors for my own reference......ended up with 2ppm units from DigiKey. Yes, they are expensive and send the BOM into the clouds!..............so I know how you feel! > > > > Ian. > > > > > > ----- Original Message ----- > > From: Illya Tsemenko [mailto:illya@xdevs.com] > > To: volt-nuts@febo.com > > Sent: Thu, 26 May 2016 18:42:54 +0800 > > Subject: [volt-nuts] LTZ1000 project build > > > > I'm glad my little project got so much attention. Worth to mention few > > points regarding my (or any other LTZ1000) unit: > > > > A. One should consider cost of whole BOM, not only LTZ chip. In my case > > resistors from VPG exceed cost of LTZ1000 by far. > > B. PCB I designed was targeted best to my own application need, where > > compact size and direct output (whatever is coming out of LTZ) were key > > importance. Since modules are part of bigger system with DAC synth, > > specific value, like 7.1500000 was not important for me. > > C. If there is large enough group to have this project going, I can > > actually modify original design (e.g. for easier hand soldering, or > > added buffer or ratio circuits, onboard LDO, you name it) and get PCBs > > made for interested nuts, at small cost. > > > > You can also find lot of temco/stability data with > > temperature/environment logs with mentioned LTZ modules on my site here: > > https://xdevs.com/services/ in LTZ1000 voltage reference tests section. > > RAW DSV data are available. I still have 3 modules at hand for any > > testing/checks (one was sold while ago, one was installed in HP 3245A > > source to replace LM399). > > > > Illya "TiN" Tsemenko > > > > _______________________________________________ > > volt-nuts mailing list -- volt-nuts@febo.com > > To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts > > and follow the instructions there. > > _______________________________________________ > > volt-nuts mailing list -- volt-nuts@febo.com > > To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts > > and follow the instructions there. > > _______________________________________________ > > volt-nuts mailing list -- volt-nuts@febo.com > > To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts > > and follow the instructions there. > > -- > > Ian Johnston > www.ianjohnston.com > https://www.youtube.com/user/IanScottJohnston > /"Because it feels good to make stuff!"/ > > _______________________________________________ > volt-nuts mailing list -- volt-nuts@febo.com > To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts > and follow the instructions there. > _______________________________________________ > volt-nuts mailing list -- volt-nuts@febo.com > To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts > and follow the instructions there. >

Lars wrote: > if you have a completely sealed box with varying temperatures the > relative humidity in the box changes if the humidity is not close to zero. As I have mentioned before, for many years I have put precision circuits that may be sensitive to humidity into gasketed metal boxes with fresh silica gel packets. I have opened some of these after 2 and 3 decades, and even if there are some small leaks, or a vent hole to equalize pressure, the silica gel has still been fully dry. This is an easy solution, and works extremely well. I second the suggestion that the main drift problem in a DIY voltage reference (apart from the reference itself) is the resistors that set the scaling amplifier gain (or the DAC or PWM drift, if you go that way). If you can live without scaling, by all means do so. Best regards, Charles

-------- In message <57527F8A.6070108@yandex.com>, Charles Steinmetz writes: >As I have mentioned before, for many years I have put precision circuits >that may be sensitive to humidity into gasketed metal boxes with fresh >silica gel packets. I have opened some of these after 2 and 3 decades, >and even if there are some small leaks, or a vent hole to equalize >pressure, the silica gel has still been fully dry. Only thing to remember is that it also reduces air circulation in the box and therefore increases thermal gradients. -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 phk@FreeBSD.ORG | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence.

On 4 Jun 2016 20:44, "Poul-Henning Kamp" <phk@phk.freebsd.dk> wrote: > > -------- > In message <57527F8A.6070108@yandex.com>, Charles Steinmetz writes: > > >As I have mentioned before, for many years I have put precision circuits > >that may be sensitive to humidity into gasketed metal boxes with fresh > >silica gel packets. > Only thing to remember is that it also reduces air circulation in the > box and therefore increases thermal gradients. I don't know how practical they are, but I have see humidity calibration standards (e.g 50%). These use a chemical reaction to keep a volume at a constant humidity. That might be a way of keeping the humidity constant, without introducing increasing thermal gradients. They might introduce a whole host of other problems, so such a method would need researching. If a new PCB was designed, it would be nice if there were pads to take the expensive Vishay foil resistors, in addition to a cheaper choice. Dave

Poul-Henning wrote: > Only thing to remember is that it also reduces air circulation in the > box and therefore increases thermal gradients. The starting point is a sensitive circuit, which I would put in a close-fitting, fully shielded box in any case. So even if it weren't quasi-hermetically sealed, there would still be essentially no exchange of air between the inside and outside of the box. In that respect, providing the moisture seal changes nothing. (I assume that by "air circulation in the box" you mean the exchange of air between the inside and outside of the box. Quasi-hermetically sealing an already tight shielded box would not change the air circulation *inside* the box.) Further, a relatively close-fitting sealed box tends to minimize convection currents inside the box and to create an isothermal environment, avoiding the effects of drafts and circulation patterns that often plague sensitive circuits sharing a larger volume with other circuitry. So, yes, as always, you have to think about heat flow, but quasi-hermetically sealing an already tight shielded box poses no additional complications. In fact, using a close-fitting sealed box tends to minimize thermal gradients inside the box. Charles