Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Small power amplifier

Status
Not open for further replies.

Dr.EM

New Member
Although I had bought a TDA2030 chip, I realised quickly it wasn't going to fit in breadboards, or stripboard for that matter. So, I decided to make this circuit because I had the parts around:

**broken link removed**

Well, it turned out to work quite well. I got reasonable volume and a decent frequency response, but I noticed it got exceedingly hot. I checked my test speaker and found it was actually 4 ohms and figured that was why. I might like to use this amplifier in an amplified test speaker design, but would like it to work with the 4 ohm speakers. Would this shown modification work? I don't have another set of BD139, BD140s to test anything with unfortunately.
 

Attachments

  • 2 watt power amp 4OHM.gif
    2 watt power amp 4OHM.gif
    9.4 KB · Views: 368
As far as I can remember, the TDA2030 is suitable for 4 ohms?, as suggested you should use a good sized heatsink - you should also check with a scope that it's not oscillating, layout is pretty critical on these chips - or they oscillate at high frequency.

I hope you're not trying to use a breadboard?, and if using veroboard make the layout as tight as you can, and remove all unused tracks and parts of tracks.
 
Yeah, I think the 2030 is suitable for 4 ohms, but its 5 pin to220 case won't fit into stripboards spacing :(

I thought that 2 watt circuit might be quite suitable as it has the advantage its current draw is low enough for wall wart powering. True, I probably could use a very large heatsink on it, but it would actually be cheaper to user another set of transistors on the output, so long as that would work?
 
The extremely simple transistor amplifier has only a 1.8k resistor turning on its PNP output transistors. So its negative-going output swing will have a reduced amplitude causing very high distortion and reduced power to the speaker. Bootstrapping the resistor would help but then the amplifier would probably oscillate.
 
I noticed that the negative swing clipped before the posistive one when I tested it with a sine wave generator. The output level was reasonable up to this point though, but are you saying that if it had to drive 2 PNPs, it would worsen the effect?
 
Dr.EM said:
I noticed that the negative swing clipped before the posistive one when I tested it with a sine wave generator. The output level was reasonable up to this point though, but are you saying that if it had to drive 2 PNPs, it would worsen the effect?

Just bootstrap the output stage, I see no reason for that to make it oscillate (it's a standard technique used on most amplifers?). However, that is a pretty naff circuit, I would suggest improving it a little by adding the usual components that aren't fitted - such as a miller capacitor, and some HF negative feedback, and even a zobel network.
 
Ok, i've heard of all those things, but am not sure on how any apply to this design. So far I concluded that the bootstrap would be a capacitor in parallel with the 470ohm trimpot? A zobel network is a resistor and capacitor in series to ground at the output? The miller capacitors are placed between C and B of the output transistors? For values, the bootstrap is in the order of 47uF? The zobel uses a 4 ohm resistor and a small value capacitor. Not sure on miller capacitors, i'm assuming they are quite small? The HF feedback i'm not sure how to implement.
 
Dr.EM said:
Ok, i've heard of all those things, but am not sure on how any apply to this design. So far I concluded that the bootstrap would be a capacitor in parallel with the 470ohm trimpot?

No, you split the bottom 1.8K into two series resistors (820 ohm should be OK), and connect a capacitor from their centre to the junction of the four 1 ohms (the output of the amplifier before the blocking capacitor). As it's a capacitor it's frequency dependent, between 47uF and 470uF is pretty common.

A zobel network is a resistor and capacitor in series to ground at the output?

Yes, 3.9 ohm and 0.1uF in series should be fine.

The miller capacitors are placed between C and B of the output transistors?

No, between C and B of the driver, the 2N3906, 15pF-47pF should be fine.

The HF feedback i'm not sure how to implement.

Small capacitor in parallel with the 22K feedback resistor, the larger the value the greater the HF roll off - try a couple of hundred pF for a start.

I also never like the two diode bias arrangement, I'd use a Vbe multiplier instead.
 
Thanks a lot, i'll try all of those out. I don't like the look of the diode arrangement either. In this case it seemed to work quite well though, I couldn't see crossover distortion even with the smallest amplitude sine waves. Mabye I just chose some very well matched diodes at random :D
 
Dr.EM said:
Thanks a lot, i'll try all of those out. I don't like the look of the diode arrangement either. In this case it seemed to work quite well though

Oh yes, it works - I just don't like it :D

Much prefer a Vbe multiplier!.
 
Hi Dr. EM,
The diodes in your circuit don't have to be matched since they are in series and are in series with the pot that changes the idle current in the transistors and therefore how much crossover distortion.
The temperature change of the diodes match the temperature change of the output transistors so they should be clamped to the heatsink together.
 
5 pin TO 220 can fit nicely in a breadboard. Just bend the pins out a bit. Use a breadboard as reference. I have some mosfet drivers in such a packege.This may be a bit harder if the pins are aranged in two rows.
 
I'll have to try the chip amp out again, problem is it'll need a big heatsink, and that gets in the way with breadboard layouts.

I still haven't tried those modifications, i'll get and do that. If I ever make this on stripboard, i'll be sure to fix the diodes to the heatsink. I guessed the pot affected crossover distortion, but I couldn't see much anywhere except fully to one end when it suddenly became very distorted.
 
Dr.EM said:
I guessed the pot affected crossover distortion, but I couldn't see much anywhere except fully to one end when it suddenly became very distorted.

That's the whole point of it, you adjust the bias current until the distortion just vanishes - increasing it past that point won't have much effect on the sound, but will increase the chances of thermal runaway.
 
Ok, well i've added all the things you've suggested. It seems to behave much the same, but I assume the intention of most the parts is to improve stability? I think the transistors run slightly cooler, mabye the zobel network helps control the lowest impedance of the speaker. One nice change is the turn on pop seems much smaller, it used to push the cone right out slowly, but seems to not now. It still seems to clip on the negative half much sooner than the positive half of the wave, but the sine wave is unconfortably loud and annoying by that point anyway, lol. I'll adjust that trim pot more accurately I think.
 
Hi Dr. EM,
The zobel network applies a load to the amplifier at high frequencies to help keep it from oscillating. The speaker is an inductance with a high impedance at high frequencies.

To have symmetrical clipping, adjust the value of one of the two resistors that are a voltage divider for the 1st transistor's base.

Adjust the trimpot for about 20mA to 30mA in each output transistor without a signal and after it has warmed up. You can measure 20mVDC to 30MVDC across each 1 ohm emitter resistor.
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top