Are you still referring to the original circuit you posted in post #1? Here's an article which ends up with a very similar circuit arrangement and provides some analysis on using negative feedback from a class B output stage back to an op amp:
http://www.allaboutcircuits.com/vol_6/chpt_6/10.html

As far as your "middle stage," since your op amp is directly driving the output transistors (currently) and their gain is low, doing a darlington configuration for each output transistor would be one technique to reduce the op amp's current load.

 

Edit: I can use a tl072/82 for stereo operation. Any other design explanations or tips to this circuit?

You can not use a dual op-amp with this configuration because the power supply pins of the ic drive the power transistors.

Use a single op-amp for each channel.

Timescope

 

Are you still referring to the original circuit you posted in post #1? Here's an article which ends up with a very similar circuit arrangement and provides some analysis on using negative feedback from a class B output stage back to an op amp:
http://www.allaboutcircuits.com/vol_6/chpt_6/10.html

As far as your "middle stage," since your op amp is directly driving the output transistors (currently) and their gain is low, doing a darlington configuration for each output transistor would be one technique to reduce the op amp's current load.

That page is where I started from but the reccomended output transistors cannot be driven to produce 15w like the page suggest.

Yes, something like a darlington configuration is what I am trying to learn about. I have all the transistors for it already. I am just wanting to know if there is anything else I would need to do other then just adding to-92 transistors for the darlington configuration? Any other components I may need?

You can not use a dual op-amp with this configuration because the power supply pins of the ic drive the power transistors.

Use a single op-amp for each channel.

Timescope

Oh okay...using seperate packages wouldn't be a problem.

 

That page is where I started from but the reccomended output transistors cannot be driven to produce 15w like the page suggest.

Ah! Well that explains the similarities

On the darlington outputs, yes, that should solve your current (i) problem. Op amps usually like to work in small currents (usually <40mA) and, even then, the maximum output swing will decrease as output current increases. The downside to using darlington outputs is that your maximum output peak now becomes the maximum op amp peak minus 2 Vbe drops.

Also, you should probably add four diodes or a "Vbe multiplier" to get the amp into Class AB territory. The op amp's feedback will take care of most of the crossover distortion without this addition but it will be forced to swing its voltage upward or downward very quickly over 4 Vbe drops each time the voltage crosses the zero line. You'd have to just listen to it both ways to determine if it's worth the extra parts or not.

And in the CMoy schematic you posted, I'd recommend putting a small value cap in parallel to the 8.2k resistor, perhaps 100pF. The purpose of this is to roll off ultra-high frequencies beyond the audio spectrum and avoid high-frequency oscillations.

 

Ah! Well that explains the similarities

On the darlington outputs, yes, that should solve your current (i) problem. Op amps usually like to work in small currents (usually <40mA) and, even then, the maximum output swing will decrease as output current increases. The downside to using darlington outputs is that your maximum output peak now becomes the maximum op amp peak minus 2 Vbe drops.

Also, you should probably add four diodes or a "Vbe multiplier" to get the amp into Class AB territory. The op amp's feedback will take care of most of the crossover distortion without this addition but it will be forced to swing its voltage upward or downward very quickly over 4 Vbe drops each time the voltage crosses the zero line. You'd have to just listen to it both ways to determine if it's worth the extra parts or not.

And in the CMoy schematic you posted, I'd recommend putting a small value cap in parallel to the 8.2k resistor, perhaps 100pF. The purpose of this is to roll off ultra-high frequencies beyond the audio spectrum and avoid high-frequency oscillations.

That all helps alot...

I have a question pertaining to the pre-amp(cmoy). When I play with mine and measure voltage output in my oscope I can't get more then about 200mV of output. If my rails are +/- 9v and my op-amp swings even within 3v from the rails, shouldn't I be able to get like +/-6V of swing? I am thinking that it is maybe the current limit of the op-amp that starts causing the clipping and nasty distortion? I remember when I used a push pull pair of transistors before and I recall seeing as much as like 2V on my scope driving a 6ohm speaker.

 

Hmm, sorry, I don't quite understand the question or the schematic you're currently referring to. The CMoy amp as depicted in your first post is clipping at 200mV??? That doesn't sound right. What if you attach a 1k resistor as your load? Does the output look correct on a 'scope? Perhaps you can clarify or post the schematic under test.

 

Hmm, sorry, I don't quite understand the question or the schematic you're currently referring to. The CMoy amp as depicted in your first post is clipping at 200mV??? That doesn't sound right. What if you attach a 1k resistor as your load? Does the output look correct on a 'scope? Perhaps you can clarify or post the schematic under test.

I should Re-test to confirm, maybe i'm off. Currently I do not have a load resistor and neg feedback network is 6.8k from output to neg in, and 2.2k from neg in to GND. So it is not the same schematic as above...but the same style as in a non-inverting amp.

 

Boosting weak devices

Bill's Blog

 

Guys I have a major issue with the circuit posted in teh OP. The class B amp that is on in the chapter section. We all know that there needs to be added transistors to create darlington outputs to the original circuit so....asuuming that was added...I have constructed the circuit and i tested it on two 9 volts dual 9v supply...It worked on loads as low as 4ohms nominal...I was excited but I didnt get more then 1.7v out of it before clipping but I figured that was due to the 9vs not having enough capacity....So..I bought 4 6v cells and wired them for dual 12v supply and I get one of my transistors overheating and smelling. Once before I tried using dual 12v from an ac transformer supply when i built the circuit months ago and it blewup my transistor and i figured I just wired it wrong or something or my power supply was wrong. So for some reason it has isssues with dual 12v supply....The transistor that over heats is the to-220 tip41c the other transistors seem fine and im sure the tip41 would literally fry or blow if i hadnt removed the power right away.

What can it be? Why is dual 9v fine but not dual 12v? To me it seems as if the 9v's internal resistance is a safety net for the current and maybe that's why it dosnt blow with the 9v....Well I should mention that I did not add any resistors or anything anywhere in the output stage I did a simple darlington without resistors and the emitters of the to-220s is the positive the output to the speaker. Actually the circuit I build does not have any of the caps, or biasing diodes in the OP circuit. All it has is the to-92s bases common to eachother, connected to the opamp output, then the to-92s emitters to the to-220 bases(bases on to-220s arent common i dont think) then the emitters of the 220s to the speaker as said before. And the negative feedback loop is there at the speaker positive ofcoarse. So yes, I realize that I could have added diode for crossover distortion elimination etc etc but I was experimenting to see it for myself and go from there.

Is it possible I am having thermal run away and need resistor(s) on the to-220 output?

 

You described several differences between your circuit and the one in the OP. Post your exact schematic. Written descruptions get confusing really quickly.
Include how your source and load are connected to the amplifier.

 

You described several differences between your circuit and the one in the OP. Post your exact schematic. Written descruptions get confusing really quickly.
Include how your source and load are connected to the amplifier.

I understand, sorry comming right up..

 

Okay, let me say a few things before you view the schematic...you can view it first or after but please read all my comment below, I wrote them clear this time:

It shows a "power supply" to the bottom left, this is 4 6v batts wired for a dual 12V supply. Also, the single power supply is the power source for everything including two single op amps used and the transistor rails. Oh and the 2200uF caps on the rails are obviously to try and keep things balanced or good in all honesty I dont know if it will do anything or what exactly it will do, especially whenI already have DC? I just figured it will help somehow, is it required?

Also, keep in mind I omitted the orignal biasing diode, and the few caps etc for a reason, I was experimenting and wanted to add them along the way of my testing to see the results of changing the circuit...a learning experience.

Last and most importantly, as I have mentioned in the past, I constructed this circuit before months ago except I built it with everything in the original circuit(no darlington) but I used a center tap transformer, dual 12V supply and made a bridge rectifier, and used two capacitors for the ripple. When I built the circuit before, It literally BLEW one of the two TIP transistors every time I applied voltage, it was the same one overheating now I assume so it's the same issue. The transistor literally poped violently. So...Now when I use a dual 12V with batts I have the same issue basically but I assume the transistor dosn't pop because internal battery resistance but who cares the point is that there is an issue when using 12V. What is also important is that when I used a dual 9v battery supply, there was no overheating issue and I was able to run a 4 ohm speaker. Keep in mind when that even when the transistors exploded on the circuit constructed long ago, there was no music input signal even applied at the time.

Can it be that I didn't use any current limiting resistors somewhere in the circuit causing thermal run away? And if it is true then why the heck didn't the original circuit posted not include resistors itself and thus causing my original circuit I built months ago to explode?

Oh and btw I think on any power like dual 9v for instance, the speaker have a constant DC push...Basically the cone moves out and stays there idle with no signal. I think this is refered to as a DC offset? idk..

 

Can you disconnect the power transistors and put the feedback/output wire directly to the second op amp's output? You should get exactly 0 volts with no signal present. Do you get 0 volts at the 1st op amp's output? Sounds like you're definitely getting a DC offset which will quickly ruin the speaker. The other concern is the variable pot... you should perhaps put a 1k resistor between the pot and the 2nd op amp so there's a minimum load of 1k... if you turn it too far one way, the load on the left side is too small of a load for the op amp and if you turn it past the center point you are putting the op amp into an attenuation rather than amplification mode which will cause stability issues. AFAIK, pure class B won't ever cause thermal runaway but hopefully someone will correct me if I'm wrong.

 

Your schematic looks OK, except you should have 0.1uF ceramic bypass caps to ground from all the supply pins of the op amps. Short leads, as close to the packages as possible. Oscillation might damage your transistors, if you don't have big heat sinks on them.
I found datasheets for BC547 with different pinouts. What brand of BC547 and BC557 are you using?
Have you double-checked the pinouts of all 4 transistors against the datasheets?
Also - have you tested the circuit without the speaker? Does it still overheat your transistor? What is the output voltage in this case?

 

Can you disconnect the power transistors and put the feedback/output wire directly to the second op amp's output? You should get exactly 0 volts with no signal present. Do you get 0 volts at the 1st op amp's output? Sounds like you're definitely getting a DC offset which will quickly ruin the speaker. The other concern is the variable pot... you should perhaps put a 1k resistor between the pot and the 2nd op amp so there's a minimum load of 1k... if you turn it too far one way, the load on the left side is too small of a load for the op amp and if you turn it past the center point you are putting the op amp into an attenuation rather than amplification mode which will cause stability issues. AFAIK, pure class B won't ever cause thermal runaway but hopefully someone will correct me if I'm wrong.

Good point on the pot. I would use a fixed 10k input resistor, and 10k in series with a 100k pot connected as a rheostat for the feedback resistor. This will give a gain range of -1 to -11.
At and near unity gain, the Darlington may add enough phase shift into the loop to make it oscillate.

 

Oh, and one more stability tip: put a ~100pF ceramic cap directly across op amp 2's (-) terminal and its output. This is in addition to the mandatory decoupling as described by Ron H. This cap will route ultra-high frequencies directly back to the op amp so they are not amplified.

 

Okay guys I have added all caps...caps to the opamps power pins...ceramic .1uF on each(4 total) and I added the 100pF ceramic cap to the neg feedback loop however I noticed a killed output after soldering the 100pF but it may be due to my batteries dying. Keep in mind this testing was with 9vs again since the dual 12v scarred me and I was testing to make sure the BJTs didnt fully blow. It worked just as it did before.

Well, right now I am going to test using the dual 12v supply again hopping it will work, but I want to add the input resistance like you guys suggested...how exactly do you mean to add them? a 10K resistor in series with the 2nd op amps input, a 10k POT in rheostat mode, and another 10k after the rheostat? I ran out of 10k resistors and have 6.8k so I will use those and keep the rheostat from the end setting. Btw, does the 10k rheostat mean I have a variance between 0-5k ohms across the rheostat?

 

Your schematic looks OK, except you should have 0.1uF ceramic bypass caps to ground from all the supply pins of the op amps. Short leads, as close to the packages as possible. Oscillation might damage your transistors, if you don't have big heat sinks on them.
I found datasheets for BC547 with different pinouts. What brand of BC547 and BC557 are you using?
Have you double-checked the pinouts of all 4 transistors against the datasheets?
Also - have you tested the circuit without the speaker? Does it still overheat your transistor? What is the output voltage in this case?

oh btw, idk the brand but I believe I am correct on the CBE because it has already worked with my dual 9v battery supply...I am just limited to 1.7v output supplying a 4 ohm load.

 

oh btw, idk the brand but I believe I am correct on the CBE because it has already worked with my dual 9v battery supply...I am just limited to 1.7v output supplying a 4 ohm load.

I wonder if your power trannies are working with ±9V, you should get more than 1.7V (is that peak?) from fresh 9V alkaline cells for a few minutes.

You didn't answer this question:

Also - have you tested the circuit without the speaker? Does it still overheat your transistor? What is the output voltage in this case?

Do you have heat sinks on your power transistors? If not, why?

 

I wonder if your power trannies are working with ±9V, you should get more than 1.7V (is that peak?) from fresh 9V alkaline cells for a few minutes.

You didn't answer this question:

Do you have heat sinks on your power transistors? If not, why?

the 1.7V is measured form my oscope and because I am a newb I do not know if a scope measure p2p, peak, or RMS? I was going to ask that...I assume they measure p2p? idk..

havent tried without the speaker load. And yes I have heatsinks. This project is driving me nuts because I was using it on my 4ohm speaker on 9vs watching a movie on an ipod and the sound was getting weaker and clipping at the same pot setting it was at when it wasnt clipping before so I figure the 9vs are simply dying or getting low. Well, when I measure voltage from each one is lower then the other so they are un balanced. And now every since the sound was cutting out after attempting to listen to it, it will NOT work, and the led satay on regaurdless of if the power switch is on or not. I have the led in paralell on a 100ohm resistor to the +/- rails, not ground. So..I figure, one of the batteries is fighting the other, it sound very logical. The led used to turn off with the switch open but I have had this issue randomly before it started cutting out the this last time what i did to fix it was nothing really I just looked under the board for any wrong connections, moved some wires and it worked and I figured some wires were touching or something even though I didn't see any, but now I cannot get it to work no matter what on the 9vs.

So, going off the hunch that the batteries were fighting eachother, I wired two of my 6vs together for dual 6v configuration which I figured would be safe and wouldn't cause one of the to-220s to overheat but guess what happens, one of my to-92s was getting really hot(no output on speaker ofcoarse).

Another cariable is that when I measured the opamp supply pins I measured about 5v on the negative sides, but 0 volts on the positive sides with the 9vs. I know the 9vs are low obviously, but what is weird is how with LESS supply voltage(dual 6 vs dual 9) the to-92 transistor over-heats.

So what it looks like: The 9vs are un balanced and fighting eachother and one batt is forcing its polarity over the other one that is much "deader", and when using the 6v cells in any fashion, the internal resistance is a bit lower allowing them to allow to much current to flow...the 9s had their limitation and when moving the pot a little past center point the amp clipped. So as far as the to-92 overheating, do you thin it's because I didn't yet change the gain or potentiometer setup? I just left the 100k there like I have had it. I guess a test could be to lower the pot from center point after replacing the to-92 and re test with the dual 6v power supply? I have more 9vsaround here somewhere but I havent found them...