Thanks, I will look to that. But after it nulls again (earth problem fixed and put the solartron on the same group as the calibrators the problem is less. So I am going to make more work from the shielding and guarding. A replaced the opamp in the 731A. I did not believe it was the problem but it was not much work so worth to try. But it did not matter. No difference. I simululated the circuit in spice and when I short out the compensating transistor I get the same results as in real live. I think my Vref is now just the voltage divider made from R5 ( paralleled by R8A) and R8B so that is why the other dividers have influence. So first I will test the behavour of Vref (Most times I try to figure it out on paper and simulation after first global measurement results, and if I find the problem like that, I start to measure the specific things I need to know if I'm right. Maybe a bit an unusual methode but it works for me ;-) ) I bought a BBC Servogor writer. i connected my 845 between the 332 and my homebuild reference. Put it on the 30uV scale, the writer on the output. In the 10mV/cm position. But I am terrible at math. So 30uV is 100 cm, 1 cm is 300nV. I placed a 10uF cap over the writer output to suppress spikes and short time effects, ect and now it is writing at 3cm/hour. I turned it on when still cool but it got hot this afternoon In about an hour the line moved a few mm to the right (also still a bit warming up of the instruments I think) but after that the writer writes a straight line. I leave it on this night. Nice experiment. After that I want to look at short time noise but then I have to limmit the bandwidth. What is the best way to get a good measurement ? Fred Will schreef op za 01-10-2011 om 11:43 [+0300]: > > If i do a calibration on 10 V. It showes 10.000,000 on the display, but if I > > switch the 720 to 1.000,000 Volt the display shows 0.999,980 Volt. If I > > switch the solartron to the 1 V range it reads 1.000,000.x Volt. So the 10 V > > range is not lineair. The stupid thing is, I can get it right. If I use the > > filter while meauring the 1 V in th 10 V range it is correct but then it > > will not read the 10 V correct. So it looks like there is some noise of > > common mode problem. I have not checked the 1 V range the same way. I will > > do that too. > > > > Does that ring a bell to someone ? > > > > Fred PA4TIM > > > The 7061 input resistance loads the 720A output. 10 Gohm should result > approximately 4ppm error. > > IF the 720A is working perfectly, the input resistance of your 7061 > (including connection cables) is lower than specified. Could be dust, > fingerprints... > > Will > > _______________________________________________ > 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.