This post will be a bit long. Here it comes.
I have reported that we had "clipping" of the 1st stage.
This clipping was auible only at very high volumes of low frequency music (bass notes)...
Well, that was a wrong conclusion. (by the way, because that wrong conclusion I have spent hous studiyng the input stage).
Thomas an me started to measure with a scope further the amp, and the we have found very sharp peaks were alos present (high low frequency music) on the high voltage power supply.
So our attention went to the choke, and we have bypassed the choke with a resistor. The peaks were still there.
And at that moment our attention turned to the output stage.
Thomas went deeply into investigation, and he detected a very high frequency oscillation (during the high volume low frequency music), above 1 mHz. He has measured kind of envelope(s) with high frequency content.
At that moment the grid stopper was 330 ohm, and the ultralinear resistor was also 330 Ohm.
We have changed the grid stopper to 4k7, and the ultralinear also to 4k7. The nasty oscillation nearly disappeared.
I say nearly, bacause at the highest volumes, we still could detect a very small clicks in the music.
We need help. What to do, how to proceed further.
trif.
I have reported that we had "clipping" of the 1st stage.
This clipping was auible only at very high volumes of low frequency music (bass notes)...
Well, that was a wrong conclusion. (by the way, because that wrong conclusion I have spent hous studiyng the input stage).
Thomas an me started to measure with a scope further the amp, and the we have found very sharp peaks were alos present (high low frequency music) on the high voltage power supply.
So our attention went to the choke, and we have bypassed the choke with a resistor. The peaks were still there.
And at that moment our attention turned to the output stage.
Thomas went deeply into investigation, and he detected a very high frequency oscillation (during the high volume low frequency music), above 1 mHz. He has measured kind of envelope(s) with high frequency content.
At that moment the grid stopper was 330 ohm, and the ultralinear resistor was also 330 Ohm.
We have changed the grid stopper to 4k7, and the ultralinear also to 4k7. The nasty oscillation nearly disappeared.
I say nearly, bacause at the highest volumes, we still could detect a very small clicks in the music.
We need help. What to do, how to proceed further.
trif.
Just for fun, watch what happens to the output stage bias when driven at high volumes. I wonder if the clicking you hear is actually blocking.
The cathode is biased with a 270 Ohms resistor (and decoupled), there’s 10V DC on it.
But when the spike happens, this voltage suddenly drops to 8V and then come back to 10V with a kind of RC charge shape.
In fact by zooming with the scope on this effect, we see that it is a kind of bubble full of high frequencies (approx 1MHz).
But that happens only at very high levels…
..and we do not see a dip on the grid
But when the spike happens, this voltage suddenly drops to 8V and then come back to 10V with a kind of RC charge shape.
In fact by zooming with the scope on this effect, we see that it is a kind of bubble full of high frequencies (approx 1MHz).
But that happens only at very high levels…
..and we do not see a dip on the grid
Now we have on the EL84s 10k grid stoppers,
270 ohm catode rezistors with 220u and 0.82 u caps in paralel, the Ultra linear is via 100 ohm. The EL84 grid leak is 100k.
The cap between the splitter and the output stage is 330nF.
The concertina is with 10k anode, cathode, and grid stopper resistors.
The input stage is with 22 k grid stopper, 100 ohm catode rezistor with 220u and 0.82u caps , 15 k anode resistor, and 100k grid leak.
This means we have changed quite a lot of the resistor values.
And we STILL have that strange spike sound at higher volumes.
We have also used TRIANGLE signal to see/hear what happens.
With triangle input signal, at low volumes we have NO spiky sound, at mid potmeter position we have LOTS OF (repetitive) spiky sound, and surprisingly at top potmeter position NOW spiky sound, so it seams to us only in the mid position of the potmeter we have the spiky sound.
Gens, do you have ideas?
Please help.
Thomas and trif.
270 ohm catode rezistors with 220u and 0.82 u caps in paralel, the Ultra linear is via 100 ohm. The EL84 grid leak is 100k.
The cap between the splitter and the output stage is 330nF.
The concertina is with 10k anode, cathode, and grid stopper resistors.
The input stage is with 22 k grid stopper, 100 ohm catode rezistor with 220u and 0.82u caps , 15 k anode resistor, and 100k grid leak.
This means we have changed quite a lot of the resistor values.
And we STILL have that strange spike sound at higher volumes.
We have also used TRIANGLE signal to see/hear what happens.
With triangle input signal, at low volumes we have NO spiky sound, at mid potmeter position we have LOTS OF (repetitive) spiky sound, and surprisingly at top potmeter position NOW spiky sound, so it seams to us only in the mid position of the potmeter we have the spiky sound.
Gens, do you have ideas?
Please help.
Thomas and trif.
Further info: the HT rectifier is hybrid
the heater PSU for the tubes is DC
Removing one concertina does not affect the unwanted spiking/clipping in the other channel
Removing concertina + preamp tube from one channel does not help either - the effect is a bit smaller
the concertina consumes 9 mA
the EL84s consume just below 40 mA
the input stage needs 14 mA
the power trafo is rated 700mA for the HT
changing CD player does not help
disconnecting one input does not help
Now I am lost...
What is wrong?
Do I have to disassemble the amp and make another topology???
trif.
the heater PSU for the tubes is DC
Removing one concertina does not affect the unwanted spiking/clipping in the other channel
Removing concertina + preamp tube from one channel does not help either - the effect is a bit smaller
the concertina consumes 9 mA
the EL84s consume just below 40 mA
the input stage needs 14 mA
the power trafo is rated 700mA for the HT
changing CD player does not help
disconnecting one input does not help
Now I am lost...
What is wrong?
Do I have to disassemble the amp and make another topology???
trif.
Why it is necessary to connect the amp's 0 V wire to the amp's chassis?
Is there a leak?
Can you gens explain?
Note: if the chassis is connected or not to the saftey earth - what IS in normal case - does not make any difference.
Imagine that the chassis is made from wood/plastic.
Is there a leak?
Can you gens explain?
Note: if the chassis is connected or not to the saftey earth - what IS in normal case - does not make any difference.
Imagine that the chassis is made from wood/plastic.
I am a physicist.
Saying/writing that the amp's 0 v/ground has to be connected to the chassis to avoid oscillations, or saying because it has to be like that...DOES not give me an acceptable explanation. This sounds like kind of magic.
So, is here anybody who can explain in a rigorous way why it is necessary to connect he amp's 0 ground to the chassis?
I am not asking about the safety reason, that I know.
trif.
Saying/writing that the amp's 0 v/ground has to be connected to the chassis to avoid oscillations, or saying because it has to be like that...DOES not give me an acceptable explanation. This sounds like kind of magic.
So, is here anybody who can explain in a rigorous way why it is necessary to connect he amp's 0 ground to the chassis?
I am not asking about the safety reason, that I know.
trif.
Dave,
Thanks for your answer. Your answer seams to me also the most logic one=safety.
I just can not accept that if the amp oscillates, it is due to the missing connection between the amp's "OV" ground and the chassis.
Maybe in that case the amp will "brumm" or "buzz" but it should not oscillate.
In the meantime my amp had 4 variations.
Stefan
Thanks for your answer. Your answer seams to me also the most logic one=safety.
I just can not accept that if the amp oscillates, it is due to the missing connection between the amp's "OV" ground and the chassis.
Maybe in that case the amp will "brumm" or "buzz" but it should not oscillate.
In the meantime my amp had 4 variations.
Stefan
trifidmaster said:
Stray capacitance. There is inevitably capacitance between each and every part of the circuit and the chassis. Thus, there are two capacitors in series that connect the output of the (non-inverting) amplifier back to the input, potentially causing oscillation. Connect the chassis to 0V and you have shorted the junction of those two capacitors to 0V and broken the positive feedback loop.
And what is the value of the stray capacitance(s)?
The calculated value is a few pico.
What frequency would give this value (a few pico) with 8 ohm output, and 25 kohm input? That is far from audible frequency, isn't it?
Stefan.
The calculated value is a few pico.
What frequency would give this value (a few pico) with 8 ohm output, and 25 kohm input? That is far from audible frequency, isn't it?
Stefan.
Small update:
the "1st" version of my amp is ready - after lots of experimentation.
Now it is in "current injected" version - and it sounds VERY GOOD.
Soon I will check the amp with a network analyzer.
Note: the 6922s are so beautiful, they just need to be driven with the right plate current.
trif.
the "1st" version of my amp is ready - after lots of experimentation.
Now it is in "current injected" version - and it sounds VERY GOOD.
Soon I will check the amp with a network analyzer.
Note: the 6922s are so beautiful, they just need to be driven with the right plate current.
trif.
- Status
- Not open for further replies.