hi brianti
All parts of amp have an influence on the damping factor and it depends on the amp topology (voltage feedback, current feedback, no overall feedback, ...). In a voltage feedback amp for example , the ouput impedance of the amp in open loop is reduced by the amount of feedback applied. One way to get a high damping factor is to have the output stage equivalent impedance low along with high open loop gain of the amp (for a voltage feedback amp of course). All stages gain have their influence on the overall open loop gain value.
Fab
All parts of amp have an influence on the damping factor and it depends on the amp topology (voltage feedback, current feedback, no overall feedback, ...). In a voltage feedback amp for example , the ouput impedance of the amp in open loop is reduced by the amount of feedback applied. One way to get a high damping factor is to have the output stage equivalent impedance low along with high open loop gain of the amp (for a voltage feedback amp of course). All stages gain have their influence on the overall open loop gain value.
Fab
Also, you can use an output stage error correction circuit like the one discussed in this post
http://www.diyaudio.com/forums/showthread.php?postid=381999#post381999
or the whole thread on "Hawksford"
to increase the damping factor ( I suppose that it is the goal of every designers to have a high DF in an audio amp).
Fab
http://www.diyaudio.com/forums/showthread.php?postid=381999#post381999
or the whole thread on "Hawksford"
to increase the damping factor ( I suppose that it is the goal of every designers to have a high DF in an audio amp).
Fab
Hi Brianti!
Don't shoot me, ...BUT I do not have the goal to achieve a high damping factor.
As long as it is around 50 or higher, I tend to put my efforts to get
this damping factor constant all over the audiable frequency range.
Furtheron, I prefer stable frequency compensation for any load much more than tuned damping factor ....
What does damping factor mean? This value finally describes
the output impedance of your amp. This impedance is in series with the
speaker voice coil resistance and by this it has impact on the damping.
OK. But as long as the output impedance of the amp is much lower than the speakers impedance, the damping will be dominated by the speakers properties. Speaker damping will be the same at
an damping factor of 50 or an damping factor of 1000.
Especially if we look at the damping of the woofer many people are very concerned about the last mOhm output impedance of the amp, and the wires and put endless money to that..., while not noticing the 200mOhms of the choke in the passive cross over ...
Curiously waiting for fire from every side....
Markus
Don't shoot me, ...BUT I do not have the goal to achieve a high damping factor.
As long as it is around 50 or higher, I tend to put my efforts to get
this damping factor constant all over the audiable frequency range.
Furtheron, I prefer stable frequency compensation for any load much more than tuned damping factor ....
What does damping factor mean? This value finally describes
the output impedance of your amp. This impedance is in series with the
speaker voice coil resistance and by this it has impact on the damping.
OK. But as long as the output impedance of the amp is much lower than the speakers impedance, the damping will be dominated by the speakers properties. Speaker damping will be the same at
an damping factor of 50 or an damping factor of 1000.
Especially if we look at the damping of the woofer many people are very concerned about the last mOhm output impedance of the amp, and the wires and put endless money to that..., while not noticing the 200mOhms of the choke in the passive cross over ...
Curiously waiting for fire from every side....
Markus
Damping factor is an old term.
My rule of thumb is that an amp's output resistance (real part of it's output Z) should be <0.2 ohms over the audio band.
The more "real" it is the less likelyhood of undesirable reactive interaction with the speaker.
I believe damping factor is defined to be speaker resistance divided by amplifier output resistance. So an output R of 0.2 provides a damping factor of 40 for an 8-ohm resistance speaker.
You'll find that the resistive output R of an amp is not, in itself, one of the more critical factors to sound quality. For example, you can try adding half an ohm or so resistor in series with a speaker and listen to the difference. It will be very subtle unless the system is exceptionally good in other respects.
My rule of thumb is that an amp's output resistance (real part of it's output Z) should be <0.2 ohms over the audio band.
The more "real" it is the less likelyhood of undesirable reactive interaction with the speaker.
I believe damping factor is defined to be speaker resistance divided by amplifier output resistance. So an output R of 0.2 provides a damping factor of 40 for an 8-ohm resistance speaker.
You'll find that the resistive output R of an amp is not, in itself, one of the more critical factors to sound quality. For example, you can try adding half an ohm or so resistor in series with a speaker and listen to the difference. It will be very subtle unless the system is exceptionally good in other respects.
To stimulate thought on the question...
Just curious, how do you measure an amplifiers output resistance? You can't just look at circuit resistances, because they are electrically removed (at least for small to moderate power levels).
My curiosity comes from thinking of an amplifier running a closed feedback loop. Such an amp would appear to have a resistance of zero. However, zero is an impossible number to attain. In reality, output resistance may very well be some small value. And, is it actually a resistance (i.e. linear)?
I'm currently wondering if the output resistance is some frequency and output current dependent value, based loosely on the open loop gain of the amplifier or gain margin between the open loop gain and the closed loop gain combined with the load current being drawn.
Thoughts on this?
-Dan
traderbam said:
Just curious, how do you measure an amplifiers output resistance? You can't just look at circuit resistances, because they are electrically removed (at least for small to moderate power levels).
My curiosity comes from thinking of an amplifier running a closed feedback loop. Such an amp would appear to have a resistance of zero. However, zero is an impossible number to attain. In reality, output resistance may very well be some small value. And, is it actually a resistance (i.e. linear)?
I'm currently wondering if the output resistance is some frequency and output current dependent value, based loosely on the open loop gain of the amplifier or gain margin between the open loop gain and the closed loop gain combined with the load current being drawn.
Thoughts on this?
-Dan
Re: To stimulate thought on the question...
Given that the output impedance of the amplifier is not zero, such impedance and the load impedance forms a voltage divider. By measuring the voltage drop created by this voltage divider you can calculate the output impedance.
Here is a practical example. First, run your amp without a dummy load, and not into any speakers. Feed a sine wave of a certain frequency (say, 1KHz) into the input of the amp, and measure the output voltage of the amp, adjusting the input level as necessary to achieve a particular reference output level (for example, 1Vrms). Let's call this voltage V1. Don't set this voltage much higher than this to prevent overheating the amp or burning out the dummy load.
Then, connect an 8 ohm dummy load resistor (rated at least 5W or more, and use one that whose resistance is stable over a wide temperature range) to the output of the amp. Measure the voltage at the output again. Let's call this voltage V2.
V2 should be slightly less than V1 due to the voltage divider effect.
We can express the relationship as follows:
V1/V2 = (R1 + R2) / R2
where R1 is the output impedance of the amplifier and R2 is the dummy load resistance. Solving for R1 gives:
R1 = (R2 * (V1 / V2)) - R2
So, as an example, if your V1 was 1.000Vrms and V2 was 0.980Vrms, then the output impedance would be 0.16 ohms at the test frequency.
The damping factor, would then be 8 / 0.16 = 50
Note that to get accurate measurement you need to measure the exact resistance of the dummy load resistor rather than just use the nominal "8 ohm". Also, you should use a highly accurate AC voltmeter to do the measurement, especially if the output impedance of the amp under test is very low. You can repeat this test using sine waves at different frequencies to establish a output impedance vs. frequency curve. Most amps exhibit a slight rise in output impedance at higher frequencies. Your AC voltmeter should have good HF response for that.
You may find that using a 4 ohm dummy load would yield a voltage drop that is easier to measure.
-Ti
dkemppai said:
Given that the output impedance of the amplifier is not zero, such impedance and the load impedance forms a voltage divider. By measuring the voltage drop created by this voltage divider you can calculate the output impedance.
Here is a practical example. First, run your amp without a dummy load, and not into any speakers. Feed a sine wave of a certain frequency (say, 1KHz) into the input of the amp, and measure the output voltage of the amp, adjusting the input level as necessary to achieve a particular reference output level (for example, 1Vrms). Let's call this voltage V1. Don't set this voltage much higher than this to prevent overheating the amp or burning out the dummy load.
Then, connect an 8 ohm dummy load resistor (rated at least 5W or more, and use one that whose resistance is stable over a wide temperature range) to the output of the amp. Measure the voltage at the output again. Let's call this voltage V2.
V2 should be slightly less than V1 due to the voltage divider effect.
We can express the relationship as follows:
V1/V2 = (R1 + R2) / R2
where R1 is the output impedance of the amplifier and R2 is the dummy load resistance. Solving for R1 gives:
R1 = (R2 * (V1 / V2)) - R2
So, as an example, if your V1 was 1.000Vrms and V2 was 0.980Vrms, then the output impedance would be 0.16 ohms at the test frequency.
The damping factor, would then be 8 / 0.16 = 50
Note that to get accurate measurement you need to measure the exact resistance of the dummy load resistor rather than just use the nominal "8 ohm". Also, you should use a highly accurate AC voltmeter to do the measurement, especially if the output impedance of the amp under test is very low. You can repeat this test using sine waves at different frequencies to establish a output impedance vs. frequency curve. Most amps exhibit a slight rise in output impedance at higher frequencies. Your AC voltmeter should have good HF response for that.
You may find that using a 4 ohm dummy load would yield a voltage drop that is easier to measure.
-Ti
Ah,
I've reverse calculated things like that many many times, I don't know why I just didn't do it here! I guess I never thought of just hooking up my fluke to my amp. I always use the scope.
Anyway, I tried the amp that I've designed, and have found the output impedance to be about .026 Ohms. I did the measurements at 1Khz, repeated at three different power levels(1/10th watt, 1watt, and 10Watts), with results repeating. (Results repeated within the measurment cababilities of my Fluke189). Even though I suptracted lead resistance from the meter, and added lead resistance of the test load, the values are approaching the limits of instruments, and that makes me nervous about trusting, or "publishing" the results.
Anyway, I guess that makes the damping factor something over 300. When I get some more time, I'll test thing over the audio band.
I would like to test the amp over it's whole -3db bandwidth, but don't trust my fluke over the entire range of 1/2Hz to 600Khz.
Since this is more of an acidemic exercise, I really am interested in knowing the HF impedance. Like I said before, I suspect it has some response based on open loop gain, but am not sure how to test the whole thing.
Any ideas on how to get an accurate measurement at those frequency ranges above, say 20Khz??? At this point, I've run out of tools in my toolbox.
Thanks!
Dan
P.S. I would have posted again sooner, but my ISP started blocking my DIY audio mail as spam (aparently DIY audio has a bad reverse DNS lookup value), and did not recieve notification of the reply.
I've reverse calculated things like that many many times, I don't know why I just didn't do it here! I guess I never thought of just hooking up my fluke to my amp. I always use the scope.
Anyway, I tried the amp that I've designed, and have found the output impedance to be about .026 Ohms. I did the measurements at 1Khz, repeated at three different power levels(1/10th watt, 1watt, and 10Watts), with results repeating. (Results repeated within the measurment cababilities of my Fluke189). Even though I suptracted lead resistance from the meter, and added lead resistance of the test load, the values are approaching the limits of instruments, and that makes me nervous about trusting, or "publishing" the results.
Anyway, I guess that makes the damping factor something over 300. When I get some more time, I'll test thing over the audio band.
I would like to test the amp over it's whole -3db bandwidth, but don't trust my fluke over the entire range of 1/2Hz to 600Khz.
Since this is more of an acidemic exercise, I really am interested in knowing the HF impedance. Like I said before, I suspect it has some response based on open loop gain, but am not sure how to test the whole thing.
Any ideas on how to get an accurate measurement at those frequency ranges above, say 20Khz??? At this point, I've run out of tools in my toolbox.
Thanks!
Dan
P.S. I would have posted again sooner, but my ISP started blocking my DIY audio mail as spam (aparently DIY audio has a bad reverse DNS lookup value), and did not recieve notification of the reply.
Markus,
No flame. You are so right and it drives me nuts when people forget about the crossover and connection resistance. Tunnel vision.
Dan,
You will find that the HF response on your meter will change with pressure on the case. You can see this effect using a calibrator like a Fluke 5500A or better. It is very hard to adjust a meter correctly on the high frequency end. The newer Flukes use software calibration, but the effect is still there. A bench meter such as the HP 34401A is more accurate and doesn't suffer from this effect. You may be able to use a precision resistor in series and measure the voltage drop with your 'scope as a check. Still a small change.
-Chris
No flame. You are so right and it drives me nuts when people forget about the crossover and connection resistance. Tunnel vision.
Dan,
You will find that the HF response on your meter will change with pressure on the case. You can see this effect using a calibrator like a Fluke 5500A or better. It is very hard to adjust a meter correctly on the high frequency end. The newer Flukes use software calibration, but the effect is still there. A bench meter such as the HP 34401A is more accurate and doesn't suffer from this effect. You may be able to use a precision resistor in series and measure the voltage drop with your 'scope as a check. Still a small change.
-Chris
anatech said:
I've been thinking about this problem. I'm sure your're right about the meters. I've never tested meters at high frequencies. I'm a comfortable with results at around 1Khz, but not sure for frequencies much above that.
The measurement problem becomes difficult with such a low output impedance. The problem is that the voltage change with an 8 ohm load is very little (undetectable on a my scope). I would probably need to lower my load resistance. I'm guessing I'd need something in the order of 10 times the output resistance to mesaure changes reliably on the scope.
I'd need around .25 ohm load, which is a little lower than I'm confortable testing my amp with! Add to that the difficulty of gettng a calibration measurement of the resistor and leads, and compensating for temperature changes in the load resistor.
I may just try a .25 ohm load and run the test anyway, tho.
Overall, the problem is not an easy one...
-Dan
A method that gives good measuring accuracy (due to the fact that it is less demanding on the instruments accuracy) is the following: connect a load between the two channels of your amp. exactly as if you would use them in bridge mode. Now feed only one channel with a signal. The voltage that you see at the terminals of the undriven amp is proportional to the amps source impedance. The ratio of the voltage across the load resistor and the voltages accross the output terminal is the DF.
But take care with this method since it is demanding on the output devices' dissipation.
Regards
Charles
But take care with this method since it is demanding on the output devices' dissipation.
Regards
Charles
I'd have to agree that the absolute value of output resistance is not that critical in a general sense as long as it's over .. say 50 as a number. That is into an 8 ohm load (assumed normally). This excludes special circumstances, ie. weird loads. The rest depends on the quality of amplifier and the impedance characteristic of the load. I've only seen negative values on a couple DC power supplies.
Dan,
You can get lab quality four terminal (Kelvin connection) precision resistors and current shunts. The temperature co-efficient can be very low with these. Your Fluke has a four terminal shunt in it for the high current range. I have some 0.01 Ohm units I use for other things. I agree that you don't want to hang these across your outputs.
I normally measure the output voltage unloaded, then with a Dale 8 Ohm (corrected) 250W low inductance resistor. I use these figures to get a rough idea. I don't care to measure commercial amplifiers normally as it's pointless. I can't change the design.
-Chris
Dan,
You can get lab quality four terminal (Kelvin connection) precision resistors and current shunts. The temperature co-efficient can be very low with these. Your Fluke has a four terminal shunt in it for the high current range. I have some 0.01 Ohm units I use for other things. I agree that you don't want to hang these across your outputs.
I normally measure the output voltage unloaded, then with a Dale 8 Ohm (corrected) 250W low inductance resistor. I use these figures to get a rough idea. I don't care to measure commercial amplifiers normally as it's pointless. I can't change the design.
-Chris
Hmm, does this mean i can simply put a 0.1ohm resistor in series
with the speaker, and it shouldn't make an audible difference ?
With this 0.1ohm connected to an amp having high dampingfactor,
it should drop DF to 80. Hmm, i will try this...
BTW, the crossover in my box has 0.6ohms.
But this means, that DF is as useless as THD to describe the
quality of an amp ?
Obviously it is hard to build an amp with low DF, if a simple mosfetstage
without feedback already has DF of 100.
Are there ANY numbers that can describe the quality of an amp ?
How is it possible to hear immediately that an amp is not able
to damp/drive a woofer ? You can also see this, the membrane starts
to move uncontrolled with low volumes.
My experience is, that if the amp gives deep controlled precise bass,
the membranes show less movement. But what number can describe this ?
Mike, loosing his hair from all the
with the speaker, and it shouldn't make an audible difference ?
With this 0.1ohm connected to an amp having high dampingfactor,
it should drop DF to 80. Hmm, i will try this...
BTW, the crossover in my box has 0.6ohms.
But this means, that DF is as useless as THD to describe the
quality of an amp ?
Obviously it is hard to build an amp with low DF, if a simple mosfetstage
without feedback already has DF of 100.
Are there ANY numbers that can describe the quality of an amp ?
How is it possible to hear immediately that an amp is not able
to damp/drive a woofer ? You can also see this, the membrane starts
to move uncontrolled with low volumes.
My experience is, that if the amp gives deep controlled precise bass,
the membranes show less movement. But what number can describe this ?
Mike, loosing his hair from all the
Similiar to Charles' suggestion, there is a practical damping factor measurement in Citation.pdf by Nelson Pass.
In theory, damping factor is Zout/Zamp. If you follow this equation, Mike's thinking of putting R in speaker maybe increasing that damping value from that equation. But It maybe more burning output energy in the resistor rather than raising the capability of the amp to control the speaker.
I tought damping factor is more than that formula. it is the ability of one amp to control the speaker. If there is a signal, the amp should make the speaker move. If there is no signal, the speaker shouldn't move. Just how great the amp controlling this is the damping factor. Am I right here?
In theory, damping factor is Zout/Zamp. If you follow this equation, Mike's thinking of putting R in speaker maybe increasing that damping value from that equation. But It maybe more burning output energy in the resistor rather than raising the capability of the amp to control the speaker.
I tought damping factor is more than that formula. it is the ability of one amp to control the speaker. If there is a signal, the amp should make the speaker move. If there is no signal, the speaker shouldn't move. Just how great the amp controlling this is the damping factor. Am I right here?
With EC, you can make negative impedance, that is the sinusoidal trace becomes larger when you apply load. Is this a good thing?
lumanauw said:
Yes, you are right...
But... I played it through, no matter how much voltage the voicecoil
inducts, damping is theoretically limited by resistances in the crossover
and cable. The amplifier can only fight voltages he sees at his output.
So higher DF gives more stable voltage at output, but not at voicecoil.
Or ohms law does not apply here.
The formula exactly tells the ability of the amp to hold the voltage.
Means if DF is 100, the voicecoil inducts 1volt, this 1 volt is reduced
to 10mV.
Another question:
Is it possible due to the inducted voltage, that the resistance of the
speaker drops below it's DC-resistance ?
Mike
Hi Mike,
Now you are looking at phase angles. The induced current (being inductive) will not be in phase with the exciting current since the voice coil does not move instantly.
Looking from the amplifier terminals, the load will be the sum of all impedances (frequency dependant). Looking from the speaker driver, the damping it receives is dependant on the sum of all impedances. The energy is dissipated in these impedances and acts as a brake on the voice coil. This is why an active crossover driving individual amplifiers sounds so much better, provided the amplifier drives the speaker directly.
As far as numbers are concerned, they are one way of describing an amplifiers performance. Manufacturers can twist the testing method and manipulate the data to look great, falsely. So the numbers are a rough guide. You still have to look at the unit and listen to it since the specs do not describe what the amplifier will do dynamically. The history of a company or designer is worth more than the specs in my view. You can almost "shop by weight" for power, Carver is an example of an exception here.
As a final note, if you place 0.1 ohm in series with the speaker, you may or may not here a difference. Think about the ratio of the resistance of wire and the crossover network. It really depends on you, the room and the speaker (possibly the amp too). "Magic wire" guys take note.
-Chris
Now you are looking at phase angles. The induced current (being inductive) will not be in phase with the exciting current since the voice coil does not move instantly.
Looking from the amplifier terminals, the load will be the sum of all impedances (frequency dependant). Looking from the speaker driver, the damping it receives is dependant on the sum of all impedances. The energy is dissipated in these impedances and acts as a brake on the voice coil. This is why an active crossover driving individual amplifiers sounds so much better, provided the amplifier drives the speaker directly.
As far as numbers are concerned, they are one way of describing an amplifiers performance. Manufacturers can twist the testing method and manipulate the data to look great, falsely. So the numbers are a rough guide. You still have to look at the unit and listen to it since the specs do not describe what the amplifier will do dynamically. The history of a company or designer is worth more than the specs in my view. You can almost "shop by weight" for power, Carver is an example of an exception here.
As a final note, if you place 0.1 ohm in series with the speaker, you may or may not here a difference. Think about the ratio of the resistance of wire and the crossover network. It really depends on you, the room and the speaker (possibly the amp too). "Magic wire" guys take note.
-Chris
Hi anatech !
This throws a completely different light on DF. I forgot phasehift
with coils. In worst case phaseshift is 90°, means when voltage
hits 0, current is at max. So looking at DF with ohms law does not
work... This would mean, that DF needs to be very high, otherwise
the amp only sees low voltage and does not really compensate.
I mean, the errorvoltage might only be some mV, but the current
needed to compensate is very high. Or is this effect cancelled out
again by resistance from cable and crossover ?
Am i completely wrong ? I have some difficulties with these 90° phasehifts...
We shouldn't forget that the outputimpedance is a virtual resistance.
Mike, obviously not beeing an EE...
This throws a completely different light on DF. I forgot phasehift
with coils. In worst case phaseshift is 90°, means when voltage
hits 0, current is at max. So looking at DF with ohms law does not
work... This would mean, that DF needs to be very high, otherwise
the amp only sees low voltage and does not really compensate.
I mean, the errorvoltage might only be some mV, but the current
needed to compensate is very high. Or is this effect cancelled out
again by resistance from cable and crossover ?
Am i completely wrong ? I have some difficulties with these 90° phasehifts...
We shouldn't forget that the outputimpedance is a virtual resistance.
Mike, obviously not beeing an EE...
Hi Mike,
Okay. At high power levels, the induced back emf current from the speaker can be high. The output stage has to dissipate this energy in response to the error voltage on the error amp section (front end, or diff pair). This voltage has nothing to do with the input signal (it's out of phase) or distortion mechanisms in the circuit. Some of the generated current is lost in the series impedances. Therefore, the amplifier can never really control the motion of the speaker voice coil completely.
The more "stuff" you can eliminate between the amplifier output and the actual voice coil terminals will improve the situation. Speaker wire (within reason) is the least of the problems. So are the terminals (again within reason).
Remember, damping factor is an attempt to describe the ability of an amplifier to control the motion of a speaker cone. The only "constant" we have is the standard speaker impedance (8 ohms) and the *rated* amplifier output impedance. Everything else is a variable as far as the amplifier manufacturer is concerned. It's only a rough guide.
To answer your question. A few mV of back emf is what shows up across your output terminals. The magnitude depends on the amplifiers output impedance. It's ohm's law from here. What is lost in the crossover is not reasonable to calculate. It's too complicated to be worth while. Plus, you can't change it.
The dynamic character of your amplifier is worth more than a high damping factor. Many Nelson Pass designs do not have a high DF, but sound great. Other amplifiers may have a high DF but sound really bad. There are other examples the other way, such as the Marantz 300DC. I think the DF of the Marantz is around 60.
-Chris
Okay. At high power levels, the induced back emf current from the speaker can be high. The output stage has to dissipate this energy in response to the error voltage on the error amp section (front end, or diff pair). This voltage has nothing to do with the input signal (it's out of phase) or distortion mechanisms in the circuit. Some of the generated current is lost in the series impedances. Therefore, the amplifier can never really control the motion of the speaker voice coil completely.
The more "stuff" you can eliminate between the amplifier output and the actual voice coil terminals will improve the situation. Speaker wire (within reason) is the least of the problems. So are the terminals (again within reason).
Remember, damping factor is an attempt to describe the ability of an amplifier to control the motion of a speaker cone. The only "constant" we have is the standard speaker impedance (8 ohms) and the *rated* amplifier output impedance. Everything else is a variable as far as the amplifier manufacturer is concerned. It's only a rough guide.
To answer your question. A few mV of back emf is what shows up across your output terminals. The magnitude depends on the amplifiers output impedance. It's ohm's law from here. What is lost in the crossover is not reasonable to calculate. It's too complicated to be worth while. Plus, you can't change it.
The dynamic character of your amplifier is worth more than a high damping factor. Many Nelson Pass designs do not have a high DF, but sound great. Other amplifiers may have a high DF but sound really bad. There are other examples the other way, such as the Marantz 300DC. I think the DF of the Marantz is around 60.
-Chris
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.