The ECG263/4 are similar to TO-220 versions of the TIP142/7. I’d keep the voltage at 60 or less if for no other reason than keeping the heat down. Got a 36 or 40 volt transformer laying around? It would be perfect. Don’t even need a center tap. The input transistor can be just about anything - just scale for the required voltage.
I've got a load of ECG263/4's,, 260/261's, 270/271's and transformers out the ying yang to play with.
Oh what fun!
I've gathered from my stock all the components for this fun amp.
I'm gonna use the NTE 264/263's for testing, and have the NTE 270/271's if I decide to do a beefier version.
I came across a 33 ohm wirewound trimmer for the bias adj.
And a 2N5831 for the input driver (180v c/e)
I'm gonna use the NTE 264/263's for testing, and have the NTE 270/271's if I decide to do a beefier version.
I came across a 33 ohm wirewound trimmer for the bias adj.
And a 2N5831 for the input driver (180v c/e)
oh cool, you decided to make the bias adjustable with trimmer
Im sure my jibber jabber in post #18 makes sense to most.
but figured show visual correction in red for adding the trimmer.
just for reference if anyone else build the amp
with No adjustable bias the 47 ohm resistor
is bypassed like post #17 so if adding the 4th
diode bias is not to high
Well, after breadboarding this amp design, I decided it's not worth my time.
In fact, it's plain crap.
In fact, it's plain crap.
Until you get the bias up high enough to turn on the actual output transistors, it’s going to sound pretty nasty. It might require some 20 to 40 mA. Depends on what the Rbe value is on the second transistor in the darlington. The lower that is, the more current flows in the driver before the output actually starts conducting. When you finally get at least a few mA in the outputs themselves it will be 1000% better.
If you were to use an actual vbe multiplier, it would be easy(er) to see what’s going on. It would start out at zero, then come up quickly and eventually level off. That’s just the drivers turning on. If you kept going would start to rise more quickly again, and that’s when you actually turn on the outputs. In a regular EF2 output stage, the Rbe on the outputs bypasses the emitter resistor, so when you measure the output stage current (across the emitter resistor) the driver current is ignored. If you use darlingtons with the built in resistors, you need to get the driver and output currents up to the same level in order to reduce gross crossover distortion. It’s harder to do because it’s not clear, especially with a diode bias circuit that has a very narrow range of adjustment. You almost have to deliberately over bias it to be sure. Then you absolutely need the emitter resistors AND good thermal contact between the transistors and sense diodes. A tall order for a beginner most of the time. The stock circuit is deliberately underbiased, so that typical newbie mistakes won’t result in a thermal runaway. If you have the time and patience, try raising the output stage bias current to the levels I describe and you’ll be surprised how much better it will sound.
There ARE ways to get a fully class B output stage to work without a critical bias adjustment but they require the rest of the amplifier to be more sophisticated to do it well. One stage with schade feedback won’t cut it. It has hardly any distortion correction, so you have to get the output stage linear to begin with.
If you were to use an actual vbe multiplier, it would be easy(er) to see what’s going on. It would start out at zero, then come up quickly and eventually level off. That’s just the drivers turning on. If you kept going would start to rise more quickly again, and that’s when you actually turn on the outputs. In a regular EF2 output stage, the Rbe on the outputs bypasses the emitter resistor, so when you measure the output stage current (across the emitter resistor) the driver current is ignored. If you use darlingtons with the built in resistors, you need to get the driver and output currents up to the same level in order to reduce gross crossover distortion. It’s harder to do because it’s not clear, especially with a diode bias circuit that has a very narrow range of adjustment. You almost have to deliberately over bias it to be sure. Then you absolutely need the emitter resistors AND good thermal contact between the transistors and sense diodes. A tall order for a beginner most of the time. The stock circuit is deliberately underbiased, so that typical newbie mistakes won’t result in a thermal runaway. If you have the time and patience, try raising the output stage bias current to the levels I describe and you’ll be surprised how much better it will sound.
There ARE ways to get a fully class B output stage to work without a critical bias adjustment but they require the rest of the amplifier to be more sophisticated to do it well. One stage with schade feedback won’t cut it. It has hardly any distortion correction, so you have to get the output stage linear to begin with.
I never got that far, running it up slow on a variac, the outputs started to reach 180ma at 20v B+, got hot heatsinks.
Stability was crap.
No need for further discussion, I've UN-breadboarded the mess.
The End.
I too was experimenting with some circuit of the internet, not with darlingtons, regular c5200/a1943, because I have many original transistors, so I thought why not. Many times circuits on the web were crap, downright wrong. But I never give up. I identified the problem, and that gave me workable output stage. I optimized the bias and made a series of measurements bias vs distortion. Offcourse, more classA I run it, lower the distortion. It had plenty of power, but I wanted nice classA buffer like F4. I put 3 pairs of outputs, and had to use more powerful driver, but with some modifications it can work in any power voltage from +/_15 to +/-50, running cool as classAB with minimum bias, even nasty classB if I want to, or pure classA with ~0.4A per transistor. I am finishing the box for it. It sounds wonderful. I got plenty of nice tube pre, so I am looking forward to test it on the main system.
So do not give up just yet, find out what's wrong first. Fix it, make it work.
So do not give up just yet, find out what's wrong first. Fix it, make it work.
If you're having problems with something like this oscillating, wait till you try it with something with two gain stages. Or three. Improper grounding and/or decoupling is the likely culprit here - and the more complicated you make things the harder it will get. Doesn’t get much simpler than one gain stage + an emitter follower, as far as stability goes.
I've been working on and making amps for decades.
At 16, I was building tube amps - sucessfully.
Today I think I'm educated sufficiently, 40+ years in the professional service business.
Thousands of satisfied/delighted customers.
But once again, that BS design is crap.
I know crap when I come across it.
Drop the first stage, work on the output first, use just resistors and trimmer for bias. Drop the diodes. It will work.
adason, I'm done with it.
I stated earlier that I unpopulated the breadboard and am now on something else.
Until you figure out what went wrong on that simple circuit, the “something else” will probably give you even more grief.
I guess you just drive that with a 5532 op amp?
From your bias resistor values, I assume you mean to run it full class A. How much bias drift after an hour or so without ANY thermal feedback besides the emitter resistors?
From your bias resistor values, I assume you mean to run it full class A. How much bias drift after an hour or so without ANY thermal feedback besides the emitter resistors?
Currently driving it with headphone output directly from sony cd player.
First version with just one driver and one output pair had small heatsink, so I biased it accordingly not to overheat.
Version with three output pairs and one driver is on 4U standard modushop heatsink, just like AlephJ or F4 and will run ~0.4A per transistor. With helper fans. Just like I always do. With driver, total of ~1.5A. No thermal feedback, just fans. Have not finished the box yet, so no idea about long term drift.
First version with just one driver and one output pair had small heatsink, so I biased it accordingly not to overheat.
Version with three output pairs and one driver is on 4U standard modushop heatsink, just like AlephJ or F4 and will run ~0.4A per transistor. With helper fans. Just like I always do. With driver, total of ~1.5A. No thermal feedback, just fans. Have not finished the box yet, so no idea about long term drift.