Output DC offset.

[nigelwright7557]

I noticed on my amplifier designs that I was getting a volt DC output with no signal applied.

I put a pot on the long tail of the long tailed pair to adjust this.

I also found a couple of diodes in series with one leg of the LTP also nulled out the DC offset.

I found in the end the problem was one collector of the LTP was to B+ while the other was through a 12k resistor to B+.

Reducing the 12K resistor to 1K got rid of the DC output offset completely.

The 12K compared to the connection to B+ was unbalancing the LTP.

[EchoWars]

There are much better ways to go about this, such as injecting tiny currents into the base of one of the transistors to compensate, or even a servo. But I have to say that if you have a full volt of offset, there's something broke...either a component, or the design.

[wg_ski]

It sounds to me like the 12k collector resistor was too big to begin with. If that resistor is also the base-emitter resistor of the VAS, that means it will force that leg of the LTP to develop about 0.6V (plus whatever's across the degeneration resistor) across 12k. That's about 50 uA. If your LTP current is 100 uA, then you'll end up with low offset. If it's more (typically 500uA-5mA) it will be severely out of balance and you'll have an offset and high THD in the input stage! Reducing the collector resistor or adjusting the LTP current may very well have put it back in balance. You can't just arbitrarily assign values - you need to think about what you're doing.

[AndrewT]

Hi,
your changes to the tail voltage or to the collector load resistor were achieving the same end, but for different reasons.

Changing the collector load allows the VAS to set the collector resistor current. That current flows down one side of the LTP and into the tail load. A tiny portion is tapped off to flow out of the base lead.

The tail load tries to flow a constant current. The other half of the LTP gets the tail current less what the VAS forced down it's partner's side.

The NFB (Negative FeedBack) tries to make the output voltage = Gain times the difference in voltage between the two LTP bases. By changing the collector load you have made the base voltages more equal, resulting in less output offset.

The alternative is to change the tail current. Adding diodes to the tail reduced the voltage across the tail resistor. That in turn reduced the tail current. Now we know that the two LTP halves pass a current virtually equal to the tail current. So reducing the tail current has reduced the current in the -IN transistor. This has led to the difference between voltages at the differential input being more equal and the output offset went down.

These two changes could just as easily have made the output offset go up, it's just luck of the draw.

There are other methods of balancing the -IN & +IN base voltages. I like the VAS current adjustment method whereby the voltage across a degenerated VAS is changed by changing the VAS CCS load.

In all of these methods, the LTP works best if the two transistors are at the same temperature and amplify by the same factor. This requires that the power in each device is the same and that they are chosen to have the same Vbe at the quiescent current they are required to operate at. They also require to have very similar hFE so that as the operational currents change with signal then the two halves behave in a similar fashion.
To get all that to work, both sides of the LTP should pass the same quiescent current and have the same Vce and be thermally coupled and shielded from the heat emitted by any hot devices.