Thank you all for your time and your responses, they have been very precious and gave me a lot to think about and a lot of ideas to try out. It's my first proper analog project and my inexperience is really showing :^)
I think I'll temporarily drop the power output stage keeping only Q7 and Q6 and I'll up the load to 10k-100k in order to try to get at least a working kind of "discrete opamp". You all mention potential problems with this CFP topology, and I want to tackle one problem at a time. I'll reattach the power output stage in the future, for now I want to keep it simple and get at least something to work properly.
I'll also decrease the closed loop gain to around 10x.
I'm sorry to make you wait as your responses were very quick, but I haven't been able to test the health of the transistors yet because I was out of the house all day. I think I'll hopefully have some time to resume work on this project during the weekend.
Thank you again.
I think I'll temporarily drop the power output stage keeping only Q7 and Q6 and I'll up the load to 10k-100k in order to try to get at least a working kind of "discrete opamp". You all mention potential problems with this CFP topology, and I want to tackle one problem at a time. I'll reattach the power output stage in the future, for now I want to keep it simple and get at least something to work properly.
I'll also decrease the closed loop gain to around 10x.
I'm sorry to make you wait as your responses were very quick, but I haven't been able to test the health of the transistors yet because I was out of the house all day. I think I'll hopefully have some time to resume work on this project during the weekend.
Thank you again.
Be careful with that! Unless you have pretty big resistors between the emitters, or you'll instantaneously fry Q7 and Q6 even if you have a very high impedance load. Remember that the bias current has nothing to do with the load impedance. It passes between the transistors.
I built it up on a breadboard and it works just fine!
However, the bias current was about 200mA, rather than the 90mA I got in SPICE. That's probably due to variations in the forward voltage of the diodes and/or hFE in the transistors. At that current (and even at 90mA), the poor little BD139/140 get very hot. I don't have mine on a heatsink yet, so I could only run it for a few seconds at at time. As you can see, the DC offset if just a few mV.
So, in conclusion, this circuit CAN work, but you have to make sure everything from the diodes and onwards are mounted on a heatsink, close to each other. You should probably add a couple of 0.33ohm resistors on the emitters of the BDs. I also strongly recommend you replace the dropper diodes in the bias circuit with a VBE multiplier as I showed above.
I've attached the exact circuit I built. You'll notice some 2N3904/6. That's because I ran out of BC546/556.
May I ask why you opted to clone the STK? There are some much better designs if all you want is a simple Class AB amplifier. Not to engage in shameless self-promotion, but my "Bog Standard" isn't half bad. 😉
I hope you found this helpful!
However, the bias current was about 200mA, rather than the 90mA I got in SPICE. That's probably due to variations in the forward voltage of the diodes and/or hFE in the transistors. At that current (and even at 90mA), the poor little BD139/140 get very hot. I don't have mine on a heatsink yet, so I could only run it for a few seconds at at time. As you can see, the DC offset if just a few mV.
So, in conclusion, this circuit CAN work, but you have to make sure everything from the diodes and onwards are mounted on a heatsink, close to each other. You should probably add a couple of 0.33ohm resistors on the emitters of the BDs. I also strongly recommend you replace the dropper diodes in the bias circuit with a VBE multiplier as I showed above.
I've attached the exact circuit I built. You'll notice some 2N3904/6. That's because I ran out of BC546/556.
May I ask why you opted to clone the STK? There are some much better designs if all you want is a simple Class AB amplifier. Not to engage in shameless self-promotion, but my "Bog Standard" isn't half bad. 😉
I hope you found this helpful!
When you first test it, I recommend shorting the bias-generating diodes so you get zero bias. This will make it run in pure Class B and it'll have more distortion, but you should be able to get it running without the risk of blowing transistors due to the high bias current.
If you remove C1 and the current sourse is gone, and the neg distortion is very present.
Duke
Duke
Thank you very very very much for your time and for your patience, the fact that you got it working is reassuring.
This means that the problem is on me, I'm doing some idiotic systematic error in my build which I'm not seeing. I'll try again starting back from ground zero as soon as I can.
By the way, I tested all transistors tonight with a multimeter in diode check mode and they appear to be fine; if I'll have the opportunity I'll check them on the curve tracer in the university lab (which, unfortunately, I can't access freely). Still, I'll order some new transistors so I can safely restart from scratch.
Long story short, for mainly two reasons:
1- I bought a Nikko NA-440 on ebay sold as "nonworking" for very cheap (20 euros) as a repair project. While I was waiting for it to arrive, I began researching STK amplifying modules as I thought that the STK inside the amp was blown. I looked on the internet for replacements, and I discovered that the market is flooded with STK465 clones made with discrete components. I wanted to try to make my own clone of a STK465 to replace the broken one inside my amp (mainly as a learning experience, I wasn't hoping to beat the performance of the original as I know Sanyo engineers did things properly). When the amp arrived, I discovered it just had a blown fuse on one of the channels while everything else worked just fine. So I had a working amp and new stimulating challenge to tackle.
2- Italian universities, for some reason, love theory and despise labs (we have an electronics lab of course, but us students don't have free access to it; we must always ask permission to professors, which is a very slow process). I wanted to see what I'm studying in practice, not just on paper or on a simulator; if my uni won't provide this experience to me, I'll seek my own. I thought a simple AB amp would be an interesting build as I studied them in an analog electronics class but I never built one myself.
I'll get back to this project as soon as I'll have a bit of free time (unfortunately I got to study for exams).
Again, as always, many thanks!
The hardest part of cloning one of those STK modules is making a mechanical assembly that fits the available space, using packaged output transistors that are Up To The Task. You can SMD most of it and that does help. But don’t ask TO-126’s to run at full voltage. Even TO-220’s will be thermally challenged given enough supply voltage. There are some 40, 50, 60W modules floating around using NJW0281/0302, using aluminum-cored PCBs.
As far as building things outside of the lab/class, those that do generally do better on their exams than those who don’t.
As far as building things outside of the lab/class, those that do generally do better on their exams than those who don’t.
FINALLY! It works properly!!! It sounds better in real life, my phone is total crap and makes everything sound tinny.
I think I had a bad transistor which somehow tested fine when checked with my multimeter, if I'll have the opportunity I'll check them all with the curve tracer in my uni lab. I changed R1 to 10k to lower the input stage bias current a bit, R4 to 10k to lower the gain, and added a VBE multiplier with a 10k potentiometer (in lieu of R12) in order to regulate the bias current of the output stage.
Now that I have a working basis I can begin experimenting, starting from the suggestion you posted above. I'll also have to conduct a criminal investigation to find the culprit of my (many) previous failed attempts.
Many thanks once again for your invaluable help!
I think I had a bad transistor which somehow tested fine when checked with my multimeter, if I'll have the opportunity I'll check them all with the curve tracer in my uni lab. I changed R1 to 10k to lower the input stage bias current a bit, R4 to 10k to lower the gain, and added a VBE multiplier with a 10k potentiometer (in lieu of R12) in order to regulate the bias current of the output stage.
Now that I have a working basis I can begin experimenting, starting from the suggestion you posted above. I'll also have to conduct a criminal investigation to find the culprit of my (many) previous failed attempts.
Many thanks once again for your invaluable help!
Food for though:
For the best DC offset R9 & R4 should be equal. R8 needs to be set for gain.
R4 may needed to be passed by RC to control the HF BW.
Q10 CE should have a large cap (1-100uf) to insure a stable DC BIAS.
R13 might need to be split and a small 1k pot to set the output bias current.
The bias current IR dropping resistor(s) may need to be added, like 2 ea. 0.1R in each collector circuit of Q8 & Q9.
Use a clampon clip to join Q1 & Q2 thermally stable to each other.
Duke
For the best DC offset R9 & R4 should be equal. R8 needs to be set for gain.
R4 may needed to be passed by RC to control the HF BW.
Q10 CE should have a large cap (1-100uf) to insure a stable DC BIAS.
R13 might need to be split and a small 1k pot to set the output bias current.
The bias current IR dropping resistor(s) may need to be added, like 2 ea. 0.1R in each collector circuit of Q8 & Q9.
Use a clampon clip to join Q1 & Q2 thermally stable to each other.
Duke
Excellent, thanks. I'll take them in consideration while I design an improved version of this amp.
Interesting.
Never really looked at any STK module in detail.
Noticed the 365 was originally Quasi Output.
Glad your quick build worked.
To have actual working reliable amplifier for recommended original +/_ 28 volts
Would of course need actual power transistors.
And many of the noted changes.
With CFP needs the noted resistors so they can actually turn off.
Plus degen resistors for the mirror to function correctly.
Likely for a first build, would stick to Darlington output.
Never really looked at any STK module in detail.
Noticed the 365 was originally Quasi Output.
Glad your quick build worked.
To have actual working reliable amplifier for recommended original +/_ 28 volts
Would of course need actual power transistors.
And many of the noted changes.
With CFP needs the noted resistors so they can actually turn off.
Plus degen resistors for the mirror to function correctly.
Likely for a first build, would stick to Darlington output.