DC-offset in spite of perfect matched FETs!?

[Klaus]

I am a bit of irritated...

One of my Aleph 4 mono blocks had before a quite high DC-offset, 100mV – in spite of bought “good matched” input stage 9610-FETs. So I matched new FETs by myself – found perfect pairs with just some mV difference as well at 10 and 20mA. Expectantly I soldered them in and: starting cool with less than 10mV still 50mV offset after warming up! How is that possible? All the voltages in the amp fit to the schematic values, it works fine, both Fets are of almost equal temperature, 0,1% tolerance Resistors everywhere...

Has anyone a tip what the reason could be?

Klaus

[xsnailx]

try to match fets after heat-up
leave them for more time

[Klaus]

But how?

At matching they get 10 or 20mA and dissipate 4V = 0,04 or 0,08W. Even w/o cooler just a warm up of some degrees from ambient – never the real temp of 60°C like at later operation in the block. So how to warm up the FETs? Of cause I waited until they stabilized at matching.

[AudioFreak]

50-100mV is fine. In a design such as these, getting absolute offset much lower than that becomes rather difficult and it wont really be of any benefit anyway.

[xsnailx]

that is a real problem but for experiment you could heat them using real conditions
switch the voltage and circut and measure but it is way too complicated for me so i didnt tried that maybe nelson has some secret weapons in this topic
maybe use hairdryer to heat them?

[xsnailx]

AudioFreak, check this page - translate it with babelfish http://www.gmx.cz/ti_aleph.htm
and post what do you think

[AudioFreak]

xsnailx,

What language is that?

[Klaus]

My Aleph 4 begin to hum a little bit (out of the speakers) with increasing DC-offset. Maybe the AC-ripple can no longer be regulated out completely. That is my motivation for perfect matching.

[AudioFreak]

Your probably better to use a PI filter in the power supply than to try and get almost perfectly matched pairs ... the reality is that they will creep apart a little regardless of what you do. If you want to get rid of hum, you need to remove as much of the ripple from the DC rails as possible.

[uli]

Originally posted by Klaus
I am a bit of irritated...

One of my Aleph 4 mono blocks had before a quite high DC-offset, 100mV – in spite of bought “good matched” input stage 9610-FETs.
Klaus

Hi Klaus,
I would experiment with R3 (221R) as this resistor defines the
gatevoltage of the outputfet. If outputQ needs to have e.g.
higher Vgs than diff pair delivers then the diffpair gets out of balance thus creating outputoffset:
Replace R3 with a 200R in series with 100R and a 100R trimmer
parallel to that 100R thus creating a range 200 - 250R.
Trim for lowest possible outputoffset when warm(!) and replace
R3 with the adjusted value.

Uli