hi anatech !
I was interested in the DF, because i want to understand why one
of my amps have extreme bass, the other not. And why another
one has audible low DF, or whatever it is. (unprecise whobbely bass)
I never intended to achieve highest DF, but at least in sims the one
amp with the bass has DF of 4300 into 4ohms. I wanted to know if
this might be the reason. I don't know why it is that high, it has only
OL-gain of 1:56000 (95db as i know)
Mike
I was interested in the DF, because i want to understand why one
of my amps have extreme bass, the other not. And why another
one has audible low DF, or whatever it is. (unprecise whobbely bass)
I never intended to achieve highest DF, but at least in sims the one
amp with the bass has DF of 4300 into 4ohms. I wanted to know if
this might be the reason. I don't know why it is that high, it has only
OL-gain of 1:56000 (95db as i know)
Mike
anatech said:
Too many ?
The question that drives me, is it wrong what my amp does ?
At least, now the speakers sounds like what the speakersoftware
said, too much deep bass. (+4db @ 60hz) Hehe, i like it, as long as
i don't put in Peter Gabriel... (Bass can hurt !)
So you mean, when the speaker fights back, there's a chance that
one stage saturates, but shouldn't this sound ugly ?
Mike
Hi Mike,
Barring any other factors, the response of your speaker in software assumes a voltage drive, unlimited current. This is what an amplifier in the normal sense is designed to approximate. The speaker can only get worse as the output impedance gets higher. The output starts to resemble a current drive, so the response begins to follow the impedance curve of the speaker, plus other humps and dips. I use this principle to test speakers using a 1K (or higher) resistor in series.
The speaker load "fights back" by generating back emf. The amplifier is not getting it's way so it has to correct. Some amplifiers become a little unstable depending on how much back emf is generated and the circuit design.
When a stage saturates, it does sound ugly for the instant the stage is saturated. Normally seen at high output levels. This is caused when corrections force that stage past's linear operating range. Back emf does not cause this, it is the nature of too much feedback. If one stage is too slow, the preceeding stage(s) may saturate trying to correct. Back emf just throws another wrinkle into the feedback waveform.
-Chris
Barring any other factors, the response of your speaker in software assumes a voltage drive, unlimited current. This is what an amplifier in the normal sense is designed to approximate. The speaker can only get worse as the output impedance gets higher. The output starts to resemble a current drive, so the response begins to follow the impedance curve of the speaker, plus other humps and dips. I use this principle to test speakers using a 1K (or higher) resistor in series.
The speaker load "fights back" by generating back emf. The amplifier is not getting it's way so it has to correct. Some amplifiers become a little unstable depending on how much back emf is generated and the circuit design.
When a stage saturates, it does sound ugly for the instant the stage is saturated. Normally seen at high output levels. This is caused when corrections force that stage past's linear operating range. Back emf does not cause this, it is the nature of too much feedback. If one stage is too slow, the preceeding stage(s) may saturate trying to correct. Back emf just throws another wrinkle into the feedback waveform.
-Chris
Hi anatech,
Does this mean, there's a possibility that weak bass results from
unstable circuit, as it starts oscillating due to fightback from speaker ?
I ask that, because my latest experiment had very weak bass, and
it is not really stable. I thought bass shouldn't be a problem with
a not perfect stable amp... So i putted back the amp into the box...
Mike
Does this mean, there's a possibility that weak bass results from
unstable circuit, as it starts oscillating due to fightback from speaker ?
I ask that, because my latest experiment had very weak bass, and
it is not really stable. I thought bass shouldn't be a problem with
a not perfect stable amp... So i putted back the amp into the box...
Mike
Hi Mike,
Weak bass can result from a high output resistance, poor power supply or roll off in the circuitry. This is assuming another amp in the same setup does not have that problem. Rooms and positioning can absolutely kill the bass.
Back EMF from the speaker will generally not cause instability in an amplifier. Capacitance will for sure. Test your amp on dummy loads, then bypass the dummy loads with a 0.05uF cap and test again. If your amp breaks into oscillation, it will not reproduce anything well. Bass included.
-Chris
Weak bass can result from a high output resistance, poor power supply or roll off in the circuitry. This is assuming another amp in the same setup does not have that problem. Rooms and positioning can absolutely kill the bass.
Back EMF from the speaker will generally not cause instability in an amplifier. Capacitance will for sure. Test your amp on dummy loads, then bypass the dummy loads with a 0.05uF cap and test again. If your amp breaks into oscillation, it will not reproduce anything well. Bass included.
-Chris
"This is assuming another amp in the same setup does not have that problem."
Yep, that's the point, it's the same room, the same speaker, the same
cable, the same cd-player and the same PSU, only other amp...
You're right, this amp is not stable into capacitive loads, already in sims
it oscillated with 100nF load. I thought, just test the topology for it's
basical behaviour, didn't believe bassperformance should be changed
from beeing sligthly unstable.
I made some experience with an older amp, at this time i didn't realize
it was oscillating, this one sounded ultraboring. It played, but had
no details/dynamics/trebles, sounded just like my Yamaha.
Just another question, protecting the amp with a small coil (~5uH)
against capacitive loads avoids oscillating, is the amp still stable
enough, even if it would oscillate without the coil ?
I ask this, because my latest amp is designed this way.
(but it sounds really fine)
Mike
Yep, that's the point, it's the same room, the same speaker, the same
cable, the same cd-player and the same PSU, only other amp...
You're right, this amp is not stable into capacitive loads, already in sims
it oscillated with 100nF load. I thought, just test the topology for it's
basical behaviour, didn't believe bassperformance should be changed
from beeing sligthly unstable.
I made some experience with an older amp, at this time i didn't realize
it was oscillating, this one sounded ultraboring. It played, but had
no details/dynamics/trebles, sounded just like my Yamaha.
Just another question, protecting the amp with a small coil (~5uH)
against capacitive loads avoids oscillating, is the amp still stable
enough, even if it would oscillate without the coil ?
I ask this, because my latest amp is designed this way.
(but it sounds really fine)
Mike
Hi Mike,
The coil is normally bypassed with a resistor 5 - 10 ohms. Yes, this helps to stabilise an amplifier against capasitive loads. A zobel network is also normally included (R-C network from output to ground).
You may have to adjust the dominate HF pole downward to get more loop stability. Simulate a square wave input and look for ringing. Then do the same thing on the real amp with a 1 KHz square wave and watch the output with a 'scope.
This is just one problem. There may be other issues with operating points in the design. But that's the fun of it.
-Chris
The coil is normally bypassed with a resistor 5 - 10 ohms. Yes, this helps to stabilise an amplifier against capasitive loads. A zobel network is also normally included (R-C network from output to ground).
You may have to adjust the dominate HF pole downward to get more loop stability. Simulate a square wave input and look for ringing. Then do the same thing on the real amp with a 1 KHz square wave and watch the output with a 'scope.
This is just one problem. There may be other issues with operating points in the design. But that's the fun of it.
-Chris
Hi anatech !
Yes, i sim'd with a 20khz squarewave and adjusted until the corners
were nice and round (No overshooting and no ringing). Also i adjusted
with acsweep the falloff of bandwidth to have no spikes and moved
polefreq far below unitygain.
The adjusting to avoid any overshoot had a big audible effect, the
sound got much more pleasant, seperation and dynamics improved
and trebles got much clearer.
I'd like to test the squarewave in real world, but have no scope...
Yes, it's fun ! Any optimizing in proper behaviour improves the sound,
makes it more pleasant.
One more question: Is it worth to avoid any overshooting inside
the amp ? As the diffamp is much faster than following stages, it
always overreact a lot, driving into clipping with squarewaves fed.
Mike
Yes, i sim'd with a 20khz squarewave and adjusted until the corners
were nice and round (No overshooting and no ringing). Also i adjusted
with acsweep the falloff of bandwidth to have no spikes and moved
polefreq far below unitygain.
The adjusting to avoid any overshoot had a big audible effect, the
sound got much more pleasant, seperation and dynamics improved
and trebles got much clearer.
I'd like to test the squarewave in real world, but have no scope...
Yes, it's fun ! Any optimizing in proper behaviour improves the sound,
makes it more pleasant.
One more question: Is it worth to avoid any overshooting inside
the amp ? As the diffamp is much faster than following stages, it
always overreact a lot, driving into clipping with squarewaves fed.
Mike
Hi Mike,
Experiment with an R-C network between the collectors of the diif pair. Just to slow it down a little. You would like to adjust for little or no overshoot. When you do this, look at the entire amplifier design so you don't end up optimizing where you shouldn't. In other words, try to fix problems at the source, not compensate for them in the wrong spot.
I don't have a simulator, or know how to use one. So you are up on me there. I have to work on things real time, smoke and all.
-Chris
Experiment with an R-C network between the collectors of the diif pair. Just to slow it down a little. You would like to adjust for little or no overshoot. When you do this, look at the entire amplifier design so you don't end up optimizing where you shouldn't. In other words, try to fix problems at the source, not compensate for them in the wrong spot.
I don't have a simulator, or know how to use one. So you are up on me there. I have to work on things real time, smoke and all.
-Chris
Hi anatech !
The RC-Net between collectors of diffamp was the last thing i did
to "round" the edges on squarewaves. This did really improve things,
but the diffamp itself still overshoots, only the output doesn't.
I am trying to get rid off this, but seems to be very hard to get both,
not overshooting diffamp and stability...
I used 1nF+150ohm for the RC, quite big i thought, and i feared that
trebles would go, but the opposite happened, they got clearer...
One problem, i don't know how good an amp can get, so when should i stop ?
It already outperforms my commercial amps, but i have the feeling, that
there's still something with trebles that shouldn't be there.
Mike
The RC-Net between collectors of diffamp was the last thing i did
to "round" the edges on squarewaves. This did really improve things,
but the diffamp itself still overshoots, only the output doesn't.
I am trying to get rid off this, but seems to be very hard to get both,
not overshooting diffamp and stability...
I used 1nF+150ohm for the RC, quite big i thought, and i feared that
trebles would go, but the opposite happened, they got clearer...
One problem, i don't know how good an amp can get, so when should i stop ?
It already outperforms my commercial amps, but i have the feeling, that
there's still something with trebles that shouldn't be there.
Mike
Hi Mike,
A little overshoot won't hurt at all. Look elsewhere in your circuit. Sometimes you will get as far as a design or layout will let you. What I like to do is build two of a circuit. I improve one and use the other as a control. Very useful to verify that an improvement actually is.
Don't stop trying to improve your circuit. It's a great way to learn.
-Chris
A little overshoot won't hurt at all. Look elsewhere in your circuit. Sometimes you will get as far as a design or layout will let you. What I like to do is build two of a circuit. I improve one and use the other as a control. Very useful to verify that an improvement actually is.
Don't stop trying to improve your circuit. It's a great way to learn.
-Chris
Hi anatech,
Yes you're right, i've learned a lot while tuning my circuits. And seemed
that the overshooting from the diffamp was simply the amount of
energy the vas needed to charge/discharge capacities. So i simply
speeded up the vas by adding some resistors, i will do listeningtest
this evening...
Mike
Yes you're right, i've learned a lot while tuning my circuits. And seemed
that the overshooting from the diffamp was simply the amount of
energy the vas needed to charge/discharge capacities. So i simply
speeded up the vas by adding some resistors, i will do listeningtest
this evening...
Mike
Maybe it doesn't actually overshoot in the way overshoot is usually undersatood.
Since the diff-amp does actually do what its name tells: Amplify a difference it will show the difference between the input signal and the amps output signal. If the output rsponds slowly, what happens to the difference....... ?
Maybe it is this effect that you see ?
Regards
Charles
Hi charles !
Yes that's right, but i wanted to ensure that this overshooting does
not drive the diffamp into clipping.
And the real improving was avoiding any overshoot at output, or
already at vas. So now overshooting ends at input of vas.
This gave a much clearer sound and better details / separation,
even with much lower closedloopbandwidth.
Mike
Yes that's right, but i wanted to ensure that this overshooting does
not drive the diffamp into clipping.
And the real improving was avoiding any overshoot at output, or
already at vas. So now overshooting ends at input of vas.
This gave a much clearer sound and better details / separation,
even with much lower closedloopbandwidth.
Mike
Yep, hunting a ghost is not fun...
I had 2 ghost the last days:
1, I had some odd "oscillations" with Ib on small transistors.
But this was/is spice problem, it's not accurate enough. Changing
accuracy in options removed this, but then sometimes spice stops with
these odd "convergenceproblems"...
Spice uses some tolerance of ~1uA, with Ib having 3uA ...
It's some wonder that spice works at all ?
2, I hunted a spike in Ib of driverbjt, forgot to see that it's crossover-related.
But i will further investigate this.
It's fascinating how many designflaws a squarewave can reveal !
Even more fascinating: removing these flaws really improves sound.
Mike
I had 2 ghost the last days:
1, I had some odd "oscillations" with Ib on small transistors.
But this was/is spice problem, it's not accurate enough. Changing
accuracy in options removed this, but then sometimes spice stops with
these odd "convergenceproblems"...
Spice uses some tolerance of ~1uA, with Ib having 3uA ...
It's some wonder that spice works at all ?
2, I hunted a spike in Ib of driverbjt, forgot to see that it's crossover-related.
But i will further investigate this.
It's fascinating how many designflaws a squarewave can reveal !
Even more fascinating: removing these flaws really improves sound.
Mike
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.