On 23/02/2015 10:09, Jim Lesurf wrote:
In article , Eiron
wrote:
Hi experts,

How much DC offset would you consider acceptable in a power amp? What
can you do on a rainy Sunday afternoon? I was poking around inside an
old Rotel RA-970BX amp. It works OK except that the DC offset is about
100mv and 50mv on the two channels, which seems rather high to me. There
is no adjustment, except for bias current, so any tweaking would
probably have to be adjusting the current in one of the long-tailed
pairs by swapping resistors.

FWIW I'd tend to go along with aiming at around 10 - 20mV or less. And if
its a long-tail-pair (ltp) started design you'd probably need to examine
that end if you want to improve it.

If it starts with an ltp then does it use two devices in one pack? If not,
then you may be able to tweak by selecting a pair. If it is a packed pair
you could find a better matched example.

I don't know the design, but if the ltp is bipolars, consider the
resistances though which each device gets its bias current. If they don't
present similar resistancesfor the bias paths then that may be offsetting
the result. That kind of problem made me prefer matched pairs of fets in a
single pack for this kind of job.

Two complementary long-tailed pairs feeding the + and - sides of the
output stage. And the transistors are bonded together after installation
presumably to equalize the temperature.
It's the same basic design as all Rotel amps from the 90s.

I guess the components are matched before installation but as there is
no offset adjustment it either passes or fails on testing.
I wonder what the factory limit is?

An interesting point is that the feedback is taken from the speaker side
of the fuse so if you remove the fuse there is no feedback
and the output voltage drops to -38v.

--
Eiron.