Boiling cycler nightmare

John Dixon jpcd100 at
Wed Aug 16 09:49:12 EST 2000

We have recently stumbled across serious problems with some of our
cyclers.  In one experiment with a (maximum) denaturing temp of
95degrees, some well to well transfer seemed to have occurred. By
repeating the program on a 96 well plate with "randomly" scattered
wells filled with 50ul orange gel dye (no oil - heated lid @ 110 degC -
sealed with manufacturers sealing lids), we found to our horror that
samples were getting from one well to another; and not just from OG
wells to empty ones but from adjacent OG to OG, some of which had gone
up to more than 50ul, indicating well to well variability around
100degC - as well as >5 degree overshoot of intended temperature.

We sent the machine off for repair and it came back "no apparent fault
found - recalibrated". 

We have since tested our cyclers using a thermocouple attached to an
old Intelligent Heating Block (Cambio, I think). A couple of our
cyclers are still not doing what they say they are (including the
"repaired" one), but we would like to check our thermocouple to ensure
that that is accurate - it does agree with another different
manufacturers cycler so we're fairly confident but this might get ugly
as the help we are receiving from the supplier isn't up to much as yet.

So far the only accurate temperatures we can think of to test our
thermocouple with are:

37C using waterbath with medical thermometer
100C boiling water bath

Anyone got any cunning ideas of any other defined temperatures to check
our thermocouples accuracy?


P.S. Full saga complete with manufacturers names to follow if support
isn't sorted soon. Off now to repeat a few months PCRs. 

P.P.S. Good luck finding a thermocouple and testing your own machines -
hope this doesn't spoil too many people's day (except the
manufacturers, of course, they've just spoiled months of our work).

John Dixon                    Lab 44 (1223) 334131
Wellcome/CRC Institute        Fax 44 (1223) 334089
Cambridge University
United Kingdom CB2 1QR       e-m: jpcd100 at

More information about the Methods mailing list