Need some help :D

Status
Not open for further replies.
I have some questions about the schematic bellow.
Since the "output stage " uses darlingtons and not drivers and power output transistors ( for thermall stability they must have been put on the same heatsink ). Just the TIP142-147 on the heatsink would be ok ? no thermall runaway ? Do i need to put the bias diodes on the Heatsink as well ? or replace them with a Transistor which i put on the heatsink ?.
Do i need to add a zobel for stability ?
50-100p caps between the base and collector of the TIP's for oscillations ?.
Thank you.
 

Attachments

  • TIP142-TIP147-NE5532.png
    TIP142-TIP147-NE5532.png
    15.6 KB · Views: 568
Last edited:
> 50-100p caps between the base and collector of the TIP's for oscillations ?

No. These are emitter followers. A cap there would just annoy the opamp.

> no thermall runaway ?

If the un-specified diodes are single diodes, it will never run-away.

Unless the un-specified bias trimmer value is set too high!

With some assumption about that un-specified bias resistor, the output stage will run stone-cold, with measurable crossover, greatly masked by opamp gain. I have seen/heard stages like this work "well". But it is not a "fine" design.

Build it. Simple, and doesn't cost much.
 
Add another regular diode in series with D1L, maybe a Schottky in series with D2L, mount all the diodes on the heatsink with the Darlingtons, and see what kind of idle current adjustment range you get. I don't think the pot needs to be much over 100 ohms.

BTW, in sim I have never seen much difference between the two electrolytics as shown and just one electrolytic between the two bases. Might save some space.

Since nominal supply is reasonably high as-is, I would also consider moving the opamp connection up to Q1L base while omitting R5L, slightly reduces maximum output but gives you some Class A bias for the opamp output stage along the way.

(The above two items are pretty much unavoidable when going for a Vbe multiplier bias instead.)

A 47-100 ohm series resistor at the opamp output may be advisable. Can't hurt to have a Zobel either.
 
The main reason for calling it "bad" - ESP 113 is a working circuit, your modification - is not really a well-working one 😉

As sgrossklass already mentioned - biasing circuit must be more "sophisticated" and thermally coupled with the heatsink. What is enough for a headphone amp is not enough for a power amp. As soon as you reach the right bias one way or the other, you will need a thermal feedback to keep it stable and avoid the thermal runaway.

Compensation is the other point of concern.
 
It helps.
You can test for thermal runaway with a current meter across emitter resistors, and a hair dryer on the heat sink. If it does you can 1. decrease the diode spreader stack resistor and or 2. increase the value of the output transistor emitter resistors.
Glad to see you trying something a little more conventional.
 
indianajo , I have built this circuit before , with bd139-bd140 and KD503 ( quasi out ) it worked fine when i had a fan on the heatsink , no thermal runaway. and without fan...things weren't so good. I read somewhere that you need to put the Drivers and Power outputs on the same heatsink, so i tought darlingtons might work well, no thermal runaway.. Do you recommend putting the diodes on the heatsink ? or using a vbe multiplier. Thanks 😀
 
" Circuits like this run on a knife edge. The voltage developed across the 680 ohms determines if the output transistors turn on or not, and that voltage depends on the current drawn by the IC.

It is not a good circuit for an audio amplifier imo. You would need to tweak the resistor values to suit your transistors and opamp and even then, it would mean running in essentially class B. " - Mooly
 
Yep.. well... pretty much any simple design like this is going to work like this. A headphone amplifier that only outputs a watt into 32 ohms is one thing... driving an 8 ohm load is another.

To be honest, if you want a power amp with a small amount of output power and a minimal parts count, build an LM1875 based circuit.
 
Status
Not open for further replies.