[time-nuts] Low Cost Voltage Transfer References

[Mike Monett]

  > I thought  I might make a couple of comments about the  Geller SVR
  > board (I am a long time Time-Nuts reader, although  sometimes only
  > an occasional reader of the archives).

  > The SVR  board is nothing more than a high quality platform  for a
  > monolithic reference chip. The goal of this project was to provide
  > a relatively  low   cost   way   to  transfer  voltage calibration
  > (primarily with hobbyists and amateur scientists in mind). The SVR
  > board is  not  intended  to be a  voltage  standard.  The absolute
  > calibration of  our set point is directly traceable  to  our Fluke
  > 732B, which  is  in  the Fluke calibration  program.  We  rate the
  > transfer accuracy at +/- 5 ppm which I believe to be conservative.

  > With some care, generally a 2 to 3 ppm absolute  transfer accuracy
  > is realizable. In recent years, most SVR boards are  burned-in for
  > 200 hours before final calibration.

  > Although there  has  been some industry support  for  the project,
  > including some  support  from ADI, this project  has  largely been
  > accomplished at  a financial loss. Our costs include  the hp3458A,
  > the Fluke 732B, (including overnight shipping, calibration cost is
  > over $500 / year) not to mention component parts,  shipping boxes,
  > and eBay and PayPal fees.

  > Since the relatively low cost AD587LN grade is now discontinued, I
  > have been debating whether to continue testing and  design efforts
  > and to continue the low cost transfer project, or to  just abandon
  > the effort  altogether.  There  have  been  several  promising new
  > reference chips  in recent years, yet none so far  that  match the
  > performance of  the   AD587LN   considering  stability  with time,
  > stability with  temperature, output noise spectrum (both  below 10
  > Hz and  higher  frequency  noise),  stability  with  power supply,
  > stability with load (output drive), and ease of assembly and use.

  > I agree that the LTZ1000 appears to be the best ovenized reference
  > available, especially with burn-in temperature  cycling techniques
  > and/or part selection. However, at >$35 per part in  quantity, and
  > the need for an amplifier, and ultra low tempco scaling resistors,
  > it does not appear to be practical for use in a low cost product.

  > Regards,

  > Joe Geller

  Hi Joe,

  I want  to  congratulate  you for your excellent  work.  You  have a
  well-deserved reputation for quality and fair dealing.

  It is  unfortunate the AD587LN is no longer available,  but  would a
  part like the AD780BN be suitable? It lists for only $6.39, which is
  considerably considerably better than the LTZ1000:

 
http://www.analog.com/en/other/militaryaerospace/ad780/products/product.html

  The AD780BN spec is 3ppm/C part vs the 5ppm/C of the AD587, so it is
  a bit  more stable with temperature. Both references  have  the same
  noise spec  of 100nV/sqrt(Hz). The initial accuracy is a  bit better
  at 1mV,  but I don't see anything on long-term stability  for either
  part.

  If it  is  suitable, it seems your customers may  not  mind  a small
  increase in  price,   especially   considering   the  improvement in
  performance over  the AD587LN. Also, I'm sure if you  put  the above
  explanation on your web site, your customers would certainly find it
  understandable if  you  increased your price to cover  the  costs of
  maintaining the  product. In fact, many products do better  when the
  price is raised to match the customer's perception of the  value. So
  keeping a  low  price may give the wrong  impression  of  the actual
  caliber of the service.

  It would be a shame to see such a valuable concept disappear. I'd go
  for it.

  Regards,

  Mike Monett

  P.S. I  liked  your page on the HP 53310A. I have the  file  you are
  missing, so if you would like, I can email it to you along with some
  others that might be useful.