Its a very fine line. Alan Parsons is a musician, producer, and mixing and mastering engineer. How do you suppose he selects the mastering compressors he will use for when he is wearing his mastering engineer hat? Do you think he doesn't listen to what the compressor does to the sound as he turns the knobs? As he is learning what the compressor does and how to use it, he is at some times more of an audiophile than a music consumer. After he has to listen to what the equipment is doing to the sound in order to do his job.
So, what about an amplifier designer? Is there part of the time he has to listen to an amplifier to see if its working the way he wants and expects, and if so, maybe at other times can't he sit back and just enjoy the music?
Anyway, I think these sorts of overly simplistic definitions are not necessarily helpful. If at one extreme you can't enjoy just listening to music then you have a problem. OTOH, if you design an amplifier and don't bother to give it a listening test before calling done, maybe you you will be embarrassed when you send it in for review. Never bothering to listening test something you designed is another way of being extreme. Most of us are not extreme one way or the other, rather we find some balance.
So, what about an amplifier designer? Is there part of the time he has to listen to an amplifier to see if its working the way he wants and expects, and if so, maybe at other times can't he sit back and just enjoy the music?
Anyway, I think these sorts of overly simplistic definitions are not necessarily helpful. If at one extreme you can't enjoy just listening to music then you have a problem. OTOH, if you design an amplifier and don't bother to give it a listening test before calling done, maybe you you will be embarrassed when you send it in for review. Never bothering to listening test something you designed is another way of being extreme. Most of us are not extreme one way or the other, rather we find some balance.
Well, I started to test with the resistors and capacitors I had in stock. I did all my tests on the right channel of the DIY amplifier and compared it with the right channel of the Rotel RB-1552 MKII. Tests were done at 3 kHz square wave and 3 Vpp at the output.
For the resistors; I made 2 Ohms with two 1 Ohms/8W in series, I made 4 Ohms with two 8 Ohms/2W in parallel and finally I use one 8 Ohms/2W for itself. For the capacitors; I use one 100 pF (that explain why the 100 pF captures are separated for the DIY and the Rotel), 10 nF and 100 nF.
I had to stop at these values because I made a mistake while testing 4 Ohms with the 100pF. I am pretty sure I made a short at the speakers posts because it happen while testing the 4 Ohms and 100 pF in parallel, But I had already did the test with 2 Ohms||100 pF and 8 Ohms||100 pF. I don't see why a middle resistor value could have made my output go crazy...
When I did the bad test (shorted, I think), the scope was showing a square wave without sync, and I saw smoke coming out from my DIY channel in test. I stopped everything. The two 8 Ohms resistors were so hot I couldn't keep them in my hand. The Zobel resistor R11, just before the L1 inductance make smoke and kept a brown color. When I compare it's value in circuit versus the other clean channel, I measure 3.3 Ohms for the bad one and 5.5 Ohms for the good one. Unfortunately, I don't have a replacement resistor in stock. Now I am wandering if I still can use the DIY amplifier knowing that this Zobel network is only active at high frequencies, am I right?
Anyway... Here are the tests results. The Yellow trace is the DIY output while the Cyan trace is the Rotell RB-1552 MKII output.
I started with 2 Ohms only:
4 Ohms only:
And 8 Ohms only:
I didn't know I had a 100 pF in stock. So the next tests has been made with the lowest capacitors I had.
This is 2 Ohms||10 nF:
Then 2 Ohms 100 nF:
Then I found one 100 pF. With only one cap, I had to do two separate captures; one for the DIY and one for the Rotel.
For the 8 Ohms||100 pF, I also made capture of the falling edges:
Unfortunately, while trying to test the 4 Ohms||100 pF, I made the short and had to stop.
And if it is really a short, it prove that the Short Protection circuit does not do it's protection :-/
I wanted to confirm that I did a short. So I redo the 4 Ohms|100 pF test, this time taking my time to assure a good connection. Everything is fine again...
4 Ohms|| 100 pF:
And I compared both channels open to see any difference resulting from my bad test. I don't see any difference. The Cyan channel here is not the Rotel, but the left channel of the DIY. I didn't take the time to replace the label.
4 Ohms|| 100 pF:
And I compared both channels open to see any difference resulting from my bad test. I don't see any difference. The Cyan channel here is not the Rotel, but the left channel of the DIY. I didn't take the time to replace the label.
I have some though about this. It doesn't make sense that a short was at the output seeing that the two resistors were very hot. It means a current was going into them. So what I suspect now is the output of my Rigol DG1022Z that had a glitch.
When I was doing some tests on this amplifier in the past, I noticed that sometime passing from 1 kHz to 2 kHz with the knob, the output voltage was doubling at the amplifier output. I never understood this side effect. Rebooting the generator solve the problem.
So I bet that when I activated the output yesterday, the voltage has suddenly raised at an unknown value, and that the DIY Amplifier output was to much high for the Load resistors to support it. I remember that the screen of the scope was out of scale vertically. That would make sense if the output was to high.
I will have to be careful in the future with this generator. It doesn't seem to be in health!
a toasting resistor in the Zobel network is often a telltale sign of oscillation -- your amp might be at borderline stability, and could be triggered into oscillation by an input signal. The frequency it oscillates at is often at least several hundred KHz where the capacitor in the Zobel network presents little impedance, hence the resistor seeing the oscillation in full amplitude.
Don't hook up an amp with questionable stability to a speaker that you mind it smoking, as the oscillation can be deadly to the tweeter and the midrange units.
Don't hook up an amp with questionable stability to a speaker that you mind it smoking, as the oscillation can be deadly to the tweeter and the midrange units.
But wouldn't that instability shouldn't have been also present in my latest test? With the same 4 Ohms||100 pF??
So you said that the action to toggle the generator Output from OFF to ON, at that specific time, could have hazardly initiate an oscillation?
It does seem there may be an issue here. You need to check and understand what caused your load resistors to get hot like that. Usually with severe oscillation, as nattawa says, the Zobel will also smoke. If you somehow had a large DC offset on the output however, the load would smoke, but not the Zobel. Just for completeness sake, have you done a loop gain plot of your amp in a sim? This will help you understand if you have enough gain/phase margin.
Yes. I have done a lot of simulations in LTSpice, Playing with C301 and C401 values. Here is one capture done with, If I remember well, the actual values.
Also, some scope outputs in real life. I am not sure if I should worry about the glitch. Sometime the Bode Wave of the scope have weird transitions like this one, and when you redo it the glitch is not present.
Can you confirm how did you simulate Phase Margin / Gain Margin ? I'm not sure what I see on that 1st plot (cursor selection)....
Edit: I see Bonsai's post now; yeah this plot, cursors, and numbers look weird...
Sevy, see this post for the guideline how to simulate Phase Margin:
https://www.diyaudio.com/community/...att-with-tmc-and-laterals.406282/post-7540040
Edit: I see Bonsai's post now; yeah this plot, cursors, and numbers look weird...
Sevy, see this post for the guideline how to simulate Phase Margin:
https://www.diyaudio.com/community/...att-with-tmc-and-laterals.406282/post-7540040
Last edited:
OMG!!
I am not sure my plot are acceptable and/or normal :-/
Please tell me yours impressions...
I followed your tutorial minek123. It is really well done and comprehensible. Thank you!
I did the tests with both your schematic and mine in parallel, steps by steps. Here are three screenshots. The first is the result of your tutorial. The second and third are the same simulation as yours but with my schematic. I don't see the Phase crossing the Gain more than once, but I admit I am a bit scare of the 1 kHz peak!
What do you think?