hi all, back to trying to build my amp, other assignments took a while longer than i would have hoped.
trying to get my head round sorting the positive/negative feedback, i understand what your saying but have seen a few schematics for three stage amplifiers that are similar to the sort of thing im trying to achieve.
Infact most 10 watt amp schematics i've seen are 3 stage and feature a long tail pair for feedback.
For example:

From what i can see this would also mean that the feedback loop would be inphase with q1's emitter at the emitter of q2?
yet this is a tested schematic so i assume it works, am i missing something to do with the sziklai pair?

 

Yes, the feedback path must be in phase with the input signal to accomplish diminishing the voltage gain of the differential pair.

For a moment, in your mind, ignore Q8 and Q5. They are slaves to Q7 and Q4 and do nothing except enhance the current flow.

If the input voltage moves more negative, more current flows through R1.
More current through R1 means more positive on the base of Q3.
More current through Q3 means more current through Q4 and the output voltage moves toward less positive (more negative).

Feedback more negative voltage to base Q2 and you have accomplished negative feedback.

 

From what i can see this would also mean that the feedback loop would be inphase with q1's emitter at the emitter of q2?
yet this is a tested schematic so i assume it works, am i missing something to do with the sziklai pair?

What is the question here?

R8/R4 forms the feedback resistor divider and is fed into the base of Q2.

The output to the next stage of the longtail pair is taken from the collector of Q1 which is in antiphase to the output of Q2 collector.

The output of Q2 collector is in antiphase to the output of Q1 collector and to the input at the base of Q2.

So the output of Q1 collector is in phase with the feedback input at the base of Q2. (see edit)

Edit
I should have said the output of Q1 collector is in phase with the % output fed back to the other input of the long tail pair.
Of course the input to the amp is to the base of Q1 and therefore in antiphase to the collector of Q1 and therefore in antiphase to the feedback.
So the feedback is negative, as required.

Do not compare the emitters of a long tail pair as they operate in antiphase to maintain the constant current in the tail. Both are doing their (opposite) thing at the same time into the same load. The collectors, however are separate and therefore can be considered separately.

 

i think i get what you mean, i was thinking along the lines of if there in phase at the emitters then positive voltage + positive voltage = greater positive voltage, silly me ^^ it makes sense when explained like this.

 

I love the different voices here! I accomplished the shared emitter concept so long ago that I neglected to mention it in my analysis. Then I am reminded that some of the beginners haven't absorbed that part yet. Thanks studiot.

 

ahhhhhhh, ok, fully understand now, looking at the input pair entirely from the wrong perspective, thanks guys.

 

hi guys,
updates, so heres what i've been toying with so far...

added an output pair, am i right in thinking i need some arbitrary small value resistors in r9 and 10 to stop damage with continual output? any other thoughts/suggestions? still not sure what to do about balancing the positive and negative output cycles with regards to R6, is it likely to affect my output massively? id ideally like to avoid using paired transistors for bias currents and keep that as kind of an optional upgrade for if i have time/things are going well.

i have also been looking at output transistors that would be suitable and are generally in stock with farnell.
http://www.farnell.com/datasheets/1818698.pdf MJE2955T/MJE3055T
http://www.farnell.com/datasheets/696013.pdf D44C8


or there are some all in one darlington pairs such as
http://www.farnell.com/datasheets/1685024.pdf BD677
http://www.onsemi.com/pub_link/Collateral/BD675-D.PDF BD675

am i looking at the right sort of thing? i've generally been aiming for something that gives reasonable gains at 2 Amps, can handle up to 3 or 4 A to deal with a 2.5A max (Vcc @ 22V, output peak @10v) and has a lower Vceo - around 60V rather than larger 100V, since these seemed to give better current gains at lower voltages compared to larger transistors.
any downsides to the all in one darlington pairs? obviously the circuit would need changing up a bit but it does mean a few less things to worry about.....

 

Emitter resistors R9 and R 10 are usually less than 1 ohm, but will need to be power types as they carry the output current.

Here is the classic Texas Instrument amp of this type from their manual.

They use this configuration to create amps from 10W to 30W.
The circuit description and the whys and wherefores of the design decisions are quite illuminating.


Note the amplified diode transistor VT4 instead of bias diodes D1 and D2 that you have. This allows setting the bias to optimise low level distortion.

 

Here are some missing values. PM me an email address for scans of the full article if you want. There are 20 more pages including design variations and their implications.

 

makes sense, not entirely sure i understand why its wired up between c7 and r17, come to think of it R17 seems a bit of an enigma altogether, i shall have to see if i can dig up a copy of the TI manual. I have been thinking that there is a distinct lack of trim in my current circuit....

 

The parallel RV2 & VT4 is interesting.
I'd like to play with just that and see if it can be a general variable current source.

 

RV2 and VT4 are an adjustable spacer. They compensate for the Vbe cost of the driver transistors VT5 and VT6. You use them to set the idle current through R15 and R16. Then attach VT4 to the hottest place you can find on the heat sink for VT5 and VT6.

 

So it can not be used as a current regulator?

 

Of course it's used for current regulation. It regulates the idle current through VT5,6,7, and 8.

 

Hello,

Here is a piece of theory behind the "rubber diode" as the RV2 and VT4:
http://www.st-andrews.ac.uk/~www_pa/Scots_Guide/audio/part2/page3.html

Bertus

 

I guess I did not express myself well.
I know, in general, what it is used for in an amp.
I was interested how it might be used in other applications to regulate current.
Say, through a string of LEDs for example.

Good page, bertus. I'll have to read it again and think (work!) to better understand it.

(Reminds me of Maynard G. Krebs)

 

I can see there is demand for more of this amp and it really is a good study design.

 

Again

 

And again : There is another 5 page section about protection circuit additions.

 

I feel a need to correct myself. A rubber zener does not make a current regulator all by itself. It's a voltage regulator that is very convenient to apply to a push-pull stage because you can force it to track temperature changes.