A3 ADC has no way to measure meter's reference drift, as all of the measurements are relative to A9 output and/or 40k STDR. In this respect ref output is taken as constant. As result drifty reference will cause all readings to drift as well after self-calibration, because DC constant is changed only after external DCV 10V cal. So you have either zero A9 drift assumed from ACAL DCV or zero drift ADC A3 assumed from CAL 10V. By playing with time you can narrow the more drifty component. Good stable 3458A stable to <0.2ppm over week. On November 24, 2017 9:32:30 AM GMT+08:00, Randy Evans <randyevans2688@gmail.com> wrote: >Illya, > >Can you explain why you say " If ACAL DCV does not remove drift then A3 >is >probably fine". I don't really follow the argument. > >Randy > >On Thu, Nov 23, 2017 at 5:18 PM, Illya Tsemenko <illya@xdevs.com> >wrote: > >> Well, one thing you can know for sure , that reference is indeed >broken. >> Drift over a day should be way below the noise floor (<0.1ppm). With >1.7x >> gain of ADC scale that drift rate gives you around 0.8ppm +/- 0.3ppm >due to >> zener noise. So it is in line of 1.1ppm/day. If ACAL DCV does not >remove >> drift then A3 is probably fine. >> >> Since reference is much easier to troubleshoot and fix I would go >with >> replacing LTZ chip and testing if drift go away, if that have any >help on >> your desire to keep meter. >> >> Also serial number range in SN doesn't mean much for you, as meter >history >> is unknown and it still may have been serviced at some time. >> >> On November 24, 2017 8:40:00 AM GMT+08:00, Randy Evans < >> randyevans2688@gmail.com> wrote: >>> >>> Illya, >>> >>> I ran the test you suggested over 4 days and got 7.19114068 VDC on >the >>> start of the test and 7.19113736 VDC at the end of the fourth day >(96 hours >>> later). I calculate a total of 0.46 ppm drift, which seems >excessive but >>> does not account for the 1.1 ppm/day I measured overall. I suspect >the A3 >>> card is the primary source of drifting. Since the unit is a late >model >>> Agilent unit, that is well beyond the expected range of units >described in >>> Service Note 18. What do you think? >>> >>> Thanks, >>> >>> Randy Evans >>> >>> On Sat, Nov 18, 2017 at 4:47 AM, Illya Tsemenko <illya@xdevs.com> >wrote: >>> >>>> Since you have 732A, testing should be easy enough. Calibrate >faulty >>>> meter for zero and DCV 10V to 732A, record CAL? 2,1 value. This is >your LTZ >>>> output. Then leave it running for few days to drift away and >calibrate >>>> again to same 732A. Check CAL? 2,1 again. Calculate the difference >and if >>>> it matches output drift (that 1.1ppm/day you mention) - you can be >80% sure >>>> that A9 is a problem. Other 19% go to A1 and A3 circuits, as 7V is >not used >>>> directly in the meter, and there are still gain parts to get +12 >and >>>> -12VREF on A3 and 10Vish bipolar levels on A1. If your CAL? 2,1 >stays same >>>> (within 0.3ppm) then A9 is fine. >>>> >>>> >>>> On November 18, 2017 12:59:53 PM GMT+08:00, Randy Evans < >>>> randyevans2688@gmail.com> wrote: >>>>> >>>>> I just received an Agilent 3458A that has a problem with noise and >a >>>>> drifting voltage measurements. I am using two Fluke 732As to >compare >>>>> absolute voltage measurements over time against the Agilent and an >HP >>>>> 3458A. The HP unit has a new A3 ADC card and seems to be very >stable and >>>>> low noise, so is being used for comparison. I have been doing >simultaneous >>>>> absolute voltage measurements and DC Cal Constant measurements >several >>>>> times a day and then calculating the drift rates of the two units >using the >>>>> HP Service Note 18 procedure. >>>>> >>>>> >>>>> >>>>> The results indicate the Cal Constant drift rate of both units is >very >>>>> similar and within spec per Service Note 18. However, the >absolute value >>>>> measurements show the Agilent unit changing 1.1 ppm over a day >whereas the >>>>> HP unit is within a tenth of a ppm over a day. In my way of >thinking the >>>>> Cal Constant procedure assumes the voltage reference board in the >3458A is >>>>> stable, hence the absolute value reading should remain essentially >constant >>>>> after each ACAL DCV, which is the case with the HP unit. Since >the Agilent >>>>> unit shows a steady drift in the absolute reading, this would >indicate to >>>>> me that the voltage reference board is likely the cause of the >problem, and >>>>> is also likely the cause of the noisy readings. If so, this is a >>>>> “relatively” easy fix (I have several 3458A voltage reference >boards, one >>>>> of which has been continuously powered up for several years). >>>>> >>>>> >>>>> >>>>> The issue is that I have to make a decision to keep or return the >Agilent. >>>>> It has a cal seal on it and if I open the unit up to change the >voltage >>>>> reference board, I own it and can’t return it. I would appreciate >an >>>>> opinion from the members of the group as to what they think the >odds are >>>>> that the voltage reference board is the source of the problems >with the >>>>> Agilent 3458A. >>>>> >>>>> >>>>> >>>>> Thanks, >>>>> >>>>> >>>>> >>>>> Randy Evans >>>>> >>>>> >>>