PSN-L Email List Message
Subject: Re: Op amp front end noise - termal variation control
From: "Charles R. Patton" charles.r.patton@........
Date: Fri, 11 Mar 2005 09:51:33 -0800
Hi-stability oscillators use proportional oven control techniques such
as a copper container that has a heater coil and sensing. Then the ckt
is inside with some insulation. Very high quality variations of this do
two containers with separate heaters and sensing – one inside the other
with insulator between the containers. The idea is to have high thermal
conductivity containers separated by insulation to reduce gradients and
thermal transients. A cheaper variation would be to use aluminum pipe
rather than copper – not as good in the thermal mass dept., but still
pretty good thermal conductivity. There are lots of examples of these in
the amateur radio community – see issues of QEX for several examples.
A very simple way to reduce air current variation is to put your circuit
in a small, open-top box and pour in clean, dry, sand. I like #30 silica
sand (the fine, white sand you see in hotel lobby ash-tray cans), about
$4 for a 50 lb. sack at Home Depot. Also helps with microphonics – if
you have them. Just pour out the sand to work on the circuit.
>In a message dated 10/03/2005 16:11:50 GMT Standard Time, jpopelish@........
>Jack Ivey wrote:
>>Bret Nordgren wrote:
>>>Another factor that you may want to consider is thermal variation. At
>>>low frequencies, below 1Hz, the effects of micro-variations in the device
>>>temperature can add additional "noise".
>>I've seen this effect with thermocouple amplifiers, where moving your hand
>> near the circuit would move the air enough to create low-frequency noise.
>>It can be almost eliminated by pressing the circuit board between pieces
>>of foam rubber.
>It also helps a lot to keep the internal temperature rise of the front
>end opamp to a minimum. Reducing the opamp supply voltage as much as
>possible without degrading the performance of the amp helps keep the
>chip cool and reduce the thermal effect of changes in air currents.
>For this reason, if two amp choices have similar noise specs, but one
>may be operated at lower supply voltage or draws less supply current,
>its lower self heating may allow it to out perform its hotter
>competition in the low frequency realm.
> Assuming that you are using a 16 bit ADC with a range of +/-10V, one
>count is 305 micro volts. Normal amplifier gains can result in very significant
>count drifts with temperature unless great care is taken in the design and
> There are two different factors operating here. One is the
>temperature sensitivity of the opamp input circuit in micro V / C Deg. Remember that
>this relates to temperatures on the IC chip itself, so it is effected by the
> The CAZ type opamps have very greatly reduced thermal input drifts and
> The other is the signals derived from external thermo electric junctions
>and are rarely less than a few micro V / C Deg. These can be between the
>chip header and the socket or the wiring, or between cables and the input
>clamps, or even between different cables or connections. You will see differences
>across the circuit board, if there is a thermal gradient across it.
> Some resistors, like the metal oxide types, generate high EMFs if there
>is a temperature difference between the two ends. Don't even try to use
>carbon resistors, either composition or film.
> It can be an advantage to stick a strip of soft Al or Cu to the top, or
>even to both sides, of the input amplifier chip and bolt this onto the outer
>Al Screening Case. Another alternative is to use double sided circuit board.
>This greatly reduces temperature variations across the board. You can bolt a
>Cu chip cover strip onto the board. This is preferable to trying to reduce
>the dissipation by reducing the supply voltage. Having said this, it may be
>desirable to use separate IC regulators for the input opamp supply, to give low
>noise and drift and high AC supply rejection. The first amplifier does need
>very good supply noise decoupling.
> Seismometer amplifiers often have two distinct gain stages, with a high
>pass filter set to maybe 20 to 30 sec in between. This will greatly reduce
>thermal error signals and 1/f noise at the output. For geophone circuits, the
>filter maybe set to 1/10 the resonant frequency.
> The seismometer amplifier case is preferably made of metal and earthed.
>It should be kept dry, screened from drafts and any temperature variations
>should be minimised. It can be an advantage to fill the case with glass wool to
> You might include the LF412 for second amplifiers. They have quite low
> The INA118 is very useful as a low noise true differential input opamp.
> For information, noise calculation and selection of your photo diodes
> Chris Chapman
Public Seismic Network Mailing List (PSN-L)
[ Top ]
[ Back ]
[ Home Page ]