Damping factor is a highly overrated spec. The damping of a loudspeaker's mechanical resonance is largely effected by the back emf causing dissipation in the voice coil resistance. For the Q of this system to be as intended, the residual series resistance (amplifier and cables) should be low compared to that of the voice coil.
This is one reason why we want to have a low output impedance on the amplifier and why it is sometimes referred to as damping factor.
Now, whether this residual resistance is one tenth or one thousandth of the voice coil resistance has very little effect on the final Q.
There seems to be an incorrect intuition among many audio folks that a source with a very low output impedance will have superlative control over the speaker cone's movements. In fact, even if you put a dead short across the speaker's imput terminals, you can still move the cone quite easily. There is a limit to the amount of control the source can have and that limit is... the voice coil's resistance!
Actually this is a good thing, otherwise it would be completely impossible to tune the loudspeaker for a flat response. Closed speakers would roll off at relatively high frequencies and ported ones would have an undamped resonance at the port frequency.
What does this mean for the concept of "damping factor"? That the formula DF=Znom/Zout is completely meaningless.
A meaningful formula would be DF=Rdc/(Zout+Rdc). This formula correctly reflects that the damping factor approaches, but never exceeds unity as amplifier output impedance drops.
So why then do amps with ultralow impedance sound so well-controlled in the low-end? Simple coincidence. Such amplifiers have very high DC loop gains causing their distortion to increase with frequency. The colouration at high frequencies fools the ear to believe that the bass is tight - tighter than is physically possible. Bass control freaks are actually listening to a form of euphonic defect.
This is one reason why we want to have a low output impedance on the amplifier and why it is sometimes referred to as damping factor.
Now, whether this residual resistance is one tenth or one thousandth of the voice coil resistance has very little effect on the final Q.
There seems to be an incorrect intuition among many audio folks that a source with a very low output impedance will have superlative control over the speaker cone's movements. In fact, even if you put a dead short across the speaker's imput terminals, you can still move the cone quite easily. There is a limit to the amount of control the source can have and that limit is... the voice coil's resistance!
Actually this is a good thing, otherwise it would be completely impossible to tune the loudspeaker for a flat response. Closed speakers would roll off at relatively high frequencies and ported ones would have an undamped resonance at the port frequency.
What does this mean for the concept of "damping factor"? That the formula DF=Znom/Zout is completely meaningless.
A meaningful formula would be DF=Rdc/(Zout+Rdc). This formula correctly reflects that the damping factor approaches, but never exceeds unity as amplifier output impedance drops.
So why then do amps with ultralow impedance sound so well-controlled in the low-end? Simple coincidence. Such amplifiers have very high DC loop gains causing their distortion to increase with frequency. The colouration at high frequencies fools the ear to believe that the bass is tight - tighter than is physically possible. Bass control freaks are actually listening to a form of euphonic defect.
maxlorenz said:
I recently finished a transformer based passive volume control (AKA TVC) wich is wonderful, at least to me.
1) Are balanced and unbalanced input impedances different?
2) Are there versions with higher input impedance to mate better with passive pre's? Or, how can I increase input impedance, without compromising sound?
3) Will balanced input sound better than unbal?
4) Is it wise to separate the PSU box from the power amp box?
5) Is it comparable if I use one 220VAc to 2x32VAC transformer with separate rectifiers/caps instead of two completely separated PSU's, in terms of noise and channel crosstalk, etc..?
1) Both input lines are tied to ground with 10k.
2) If you want higher input impedance, an op amp other than the NE5532 should be used (e.g. OPA2134 or AD8620). Then the input "termination" (bad word) can be upped to 100k or higher if you want.
The 5532 has rather high input bias current, and 10k is installed to counter this.
3) The modules' inputs are differential. They do not require to be driven by opposite-phase signals. The most important thing to do is to use the differential inputs to extract the output signal from your preamp referencing the preamp's ground. Outside audio this is known as Kelvin sensing.
There is some discussion on this thread explaining how to best connect an unbalanced output to a differential input.
4) No.
5) Either use two transformers with each their own rectifiers and storage (dual mono) or one transformer with only one rectifier/storage. Trying to use one transformer with two sets of rectifier/storage will invite some very hideous hum.
Bruno Putzeys said:
just a side note - I have no "cone speaker" nor a speaker cabinet. There's quite some "voice coil" to control, about 5 square feet of magnets, which it drop to 2.5ohms somehwere around 100hzl. Magnepan speakers eat amps for breakfast. I know of people who finally found what they have been looking for using 2x700w/channel to drive them. If I get anywhere into that ballpark, I should be very happy with the UCD amps as well.
What you describe about the HF distortion of low impedance amps - I could theoretically live with that in my planned biamp setup where the amp I am mostly concerned about will be driving the bass panel and not see more than 250hz (and rolling off at 18db/oct above that).
From what I have seen so far about the modules, I think the UCD400's will just fine with my speakers. What size of transformer would you suggest for a 2-channel amp to drive just the bass panels up to 250hz with two UCD400s. This is a 2.5-4 ohm load and I would like to have a good margin of needless extra current capacity - I guess 1000 or 1200va would do, or did I misinterpret one of your earlier posts about what it takes to significantly increase the 8-ohm rated power at half the load?
Sorry if my questions sound silly at times - this will be my first power amp project. I've spent most of my DIY time with DACs and preamps so far. Also - I do have great amps - I just want even better amps at half the price. I want to learn as much as possible before I begin in order to do this right the first time
Peter
pburke said:What size of transformer would you suggest for a 2-channel amp to drive just the bass panels up to 250hz with two UCD400s. This is a 2.5-4 ohm load and I would like to have a good margin of needless extra current capacity - I guess 1000 or 1200va would do, or did I misinterpret one of your earlier posts about what it takes to significantly increase the 8-ohm rated power at half the load?
Sorry if my questions sound silly at times - this will be my first power amp project. I've spent most of my DIY time with DACs and preamps so far. Also - I do have great amps - I just want even better amps at half the price. I want to learn as much as possible before I begin in order to do this right the first time
800VA would already be sufficient for what you are planning. In "consumer audio" we'd be using a 500VA transformer to power 800W worth of amplifier, so 800VA would already provide the margin you're looking for.
Keep in mind that class D amplifiers always deliver higher output current than the power supply current they're drawing. (Pin=Pout but Uin>Uout)
To all Customers who bought a UcD180,
I have some importend points;
First of all I recommend when you like to use an other op amp to watch the maximum voltage of this device for instance by the AD8620 you have to reduce the voltage from 15V to 12V. However we don't us a voltage regulator (in an earlier post I made this mistake) but a transistor with a zenerdiode, so you have to replace this SMD zenerdiode. For details which diode must be replaced please send me an email.
Second, if you like increase the inputresistance you can go to 100K but because we use a NE5532 in the frontend you can not remove the coupling caps!!! For that you have to replace to a lower offset op amp as an OPA2134 or an AD8620 (change also the zenerdiodes from 15V to 12V)
We are testing the AD8620, when this is a major improvement we will offer a UcD180 version with the AD8620.
For some detailed questions please mail me direct.
Best regards,
Jan-Peter
www.hypex.nl
I have some importend points;
First of all I recommend when you like to use an other op amp to watch the maximum voltage of this device for instance by the AD8620 you have to reduce the voltage from 15V to 12V. However we don't us a voltage regulator (in an earlier post I made this mistake) but a transistor with a zenerdiode, so you have to replace this SMD zenerdiode. For details which diode must be replaced please send me an email.
Second, if you like increase the inputresistance you can go to 100K but because we use a NE5532 in the frontend you can not remove the coupling caps!!! For that you have to replace to a lower offset op amp as an OPA2134 or an AD8620 (change also the zenerdiodes from 15V to 12V)
We are testing the AD8620, when this is a major improvement we will offer a UcD180 version with the AD8620.
For some detailed questions please mail me direct.
Best regards,
Jan-Peter
www.hypex.nl
Jan-Peter, thanks for the updates, have you considered putting some of this info on your web site along with some of the other interesting stuff that's been written here? I know you have a link to this thread but its now very long and it becomes difficult to find info is such long threads.
Things that I think would be useful :-
Description of power supplies recommended, possibly some actual designs, eg dual mono, single; tranformer ratings etc
Description of signal connection, balanced vs. unbalanced.
Protection information, Switch on etc.
Cooling recommendations.
Opamp discussion
Info re cooling requirements, eg making it clear that although they run cool some heatsinking is needed.
Regards
TC owner of module 70 and 71 !
Things that I think would be useful :-
Description of power supplies recommended, possibly some actual designs, eg dual mono, single; tranformer ratings etc
Description of signal connection, balanced vs. unbalanced.
Protection information, Switch on etc.
Cooling recommendations.
Opamp discussion
Info re cooling requirements, eg making it clear that although they run cool some heatsinking is needed.
Regards
TC owner of module 70 and 71 !
Wytco0,
We are working on a paper where all this info is written. But the info on the Forum goes fasters as we can write it on paper
At the moment I am doing a temperature test on the UcD400.
The UcD400 is mounted on an aluminum plate of 190x220mm in free air.
Dummyload is 4 Ohm.
UcD400 delivers 150W.
Total eff. is 91.5%.
The whole setup is now already playing for 1,5 hour and the temperature next to the fets is 61oC.
Regards,
Jan-Peter
www.hypex.nl
We are working on a paper where all this info is written. But the info on the Forum goes fasters as we can write it on paper
At the moment I am doing a temperature test on the UcD400.
The UcD400 is mounted on an aluminum plate of 190x220mm in free air.
Dummyload is 4 Ohm.
UcD400 delivers 150W.
Total eff. is 91.5%.
The whole setup is now already playing for 1,5 hour and the temperature next to the fets is 61oC.
Regards,
Jan-Peter
www.hypex.nl
Not if you attach a source.
The input resistors are in parallel with the source. Noise is determined by the combined impedance at the input. This includes the preamp's output impedance.
Input resistors are added in order to prevent accidents when nothing is connected, and to insure that a known DC resistance is seen in case of capacitively coupled preamp outputs.
The input resistors are in parallel with the source. Noise is determined by the combined impedance at the input. This includes the preamp's output impedance.
Input resistors are added in order to prevent accidents when nothing is connected, and to insure that a known DC resistance is seen in case of capacitively coupled preamp outputs.
Output Inductor
Hi Bruno,
I think you mentioned the o/p cap was 0.1uF, does this mean that the o/p inductor is about 20 to 40uH?
Would it be better to have an air cored o/p inductor? It seems a bit of a waste to have foil coils in my speakers when the o/p inductor is ferrite cored (at least that's what it looks like).
The IR apps note mentions PWM amps have zero PSRR - any comments?
Regards,
Dave
Hi Bruno,
I think you mentioned the o/p cap was 0.1uF, does this mean that the o/p inductor is about 20 to 40uH?
Would it be better to have an air cored o/p inductor? It seems a bit of a waste to have foil coils in my speakers when the o/p inductor is ferrite cored (at least that's what it looks like).
The IR apps note mentions PWM amps have zero PSRR - any comments?
Regards,
Dave
Re: Output Inductor
Air coils are problematic due to the stray field. In one design (known here colloquially as the mother of UcD amps) I used "ferrite shielded air coils", basically a coil wound on a pot core that has been hollowed out completely (only ferrite outside the coil, not inside). This is for all intents and purposes an air coil but without the stray field problem. The improvement is noticeable, but not shocking.
The UcD400 modules will have foil wound coils. The advantage of these gets more pronounced at high powers, which is why it was not necessary for the UcD180.
The reason why you like foil inductors in the loudspeaker cross-over is because of their mechanical stability. The small size of the UcD180 coil makes it inherently stable.
The note in the IR app note concerns "school-book PWM amplifiers". The PSRR of such devices, like the vast majority of linear amplifiers, is zero in open loop, with the loop gain alone standing in for rejection (refer to the data sheet of any op amp: over most of the frequency range, PSRR of the "worst rail" tracks loop gain). Methods of improving this in linear amplifiers are known but rarely put in practice. In addition, the schoolbook circuit has power-supply dependent gain.
In full bridge class D amps, PSRR in the real sense is no longer an issue, because any error is imposed on both outputs simultaneously. In the schoolbook amp amplitude modulation still happens. In fixed-frequency amplifiers, this problem is readily solved by making the tri-wave carrier track the power supplies. This fixes the gain and removes modulation.
In self-oscillating amplifiers, a similar mechanism takes place automatically, because the feedback HF signal scales with power supply. Self-oscillating amplifiers (bar the most naive implementations) don't exhibit amplitude modulation at all. In fact, this was a source of some disappointment by another poster to this thread, who was hoping to use the power supply as a volume control.
So, UcD has no rail modulation problem, but if no action were taken the half bridge version would still have a PSRR equal to loop gain only (obvious if you consider that a common-mode error on the power rails is equivalent to the same error source appearing in series with the output). The problem is tackled by injecting the common mode voltage of the power rails, divided by loop gain, into the negative terminal of the amp. The resulting cancellation is good, with final PSRR hovering in the 65dB region over the full audio range.
The output filter is 30uH, 680nF.Dave S said:
Air coils are problematic due to the stray field. In one design (known here colloquially as the mother of UcD amps) I used "ferrite shielded air coils", basically a coil wound on a pot core that has been hollowed out completely (only ferrite outside the coil, not inside). This is for all intents and purposes an air coil but without the stray field problem. The improvement is noticeable, but not shocking.
The UcD400 modules will have foil wound coils. The advantage of these gets more pronounced at high powers, which is why it was not necessary for the UcD180.
The reason why you like foil inductors in the loudspeaker cross-over is because of their mechanical stability. The small size of the UcD180 coil makes it inherently stable.
The note in the IR app note concerns "school-book PWM amplifiers". The PSRR of such devices, like the vast majority of linear amplifiers, is zero in open loop, with the loop gain alone standing in for rejection (refer to the data sheet of any op amp: over most of the frequency range, PSRR of the "worst rail" tracks loop gain). Methods of improving this in linear amplifiers are known but rarely put in practice. In addition, the schoolbook circuit has power-supply dependent gain.
In full bridge class D amps, PSRR in the real sense is no longer an issue, because any error is imposed on both outputs simultaneously. In the schoolbook amp amplitude modulation still happens. In fixed-frequency amplifiers, this problem is readily solved by making the tri-wave carrier track the power supplies. This fixes the gain and removes modulation.
In self-oscillating amplifiers, a similar mechanism takes place automatically, because the feedback HF signal scales with power supply. Self-oscillating amplifiers (bar the most naive implementations) don't exhibit amplitude modulation at all. In fact, this was a source of some disappointment by another poster to this thread, who was hoping to use the power supply as a volume control.
So, UcD has no rail modulation problem, but if no action were taken the half bridge version would still have a PSRR equal to loop gain only (obvious if you consider that a common-mode error on the power rails is equivalent to the same error source appearing in series with the output). The problem is tackled by injecting the common mode voltage of the power rails, divided by loop gain, into the negative terminal of the amp. The resulting cancellation is good, with final PSRR hovering in the 65dB region over the full audio range.
Straying wildly OT
Over here we call it Product Manager's Power Output.
There's an anecdote attached to this. When I started work, PMPO ratings were already en vogue and hovered typically around 16 times rated output. It was basically a race between my employer and Aiwa to see who could lie most without blinking.
A colleague was commissioned by Marketing to design a "measurement" that could be performed on these "mini systems" as sold such that PMPO ratings could be proven in case any complaint of false publicity were filed.
So what he did was: set the mains input to maximum (10% above nominal). Place the speakers (system and speakers are always sold as a package) in parallel across the amplifier terminals. Feed the set a single 1ms impulse, synchronised to coincide with the peak of the AC mains, in opposite phase to the two aux inputs.
The peak voltage per output thus became 40% higher than under rated conditions. Being bridged, the output voltage was twice that (ie. 2.8 times normal output). Then the load impedance was half of nominal load. Presto: a single pulse of 32 times rated RMS output power!
JohnW said:For our love of marketing Guys (& Girls),
Following on with some discussions we had earlier about output power ratings – MikeB posted a perfect description IMHO for P.M.P.O “Pure.Marketing.Power.Output”
Over here we call it Product Manager's Power Output.
There's an anecdote attached to this. When I started work, PMPO ratings were already en vogue and hovered typically around 16 times rated output. It was basically a race between my employer and Aiwa to see who could lie most without blinking.
A colleague was commissioned by Marketing to design a "measurement" that could be performed on these "mini systems" as sold such that PMPO ratings could be proven in case any complaint of false publicity were filed.
So what he did was: set the mains input to maximum (10% above nominal). Place the speakers (system and speakers are always sold as a package) in parallel across the amplifier terminals. Feed the set a single 1ms impulse, synchronised to coincide with the peak of the AC mains, in opposite phase to the two aux inputs.
The peak voltage per output thus became 40% higher than under rated conditions. Being bridged, the output voltage was twice that (ie. 2.8 times normal output). Then the load impedance was half of nominal load. Presto: a single pulse of 32 times rated RMS output power!
Hi Bruno,
You'll find all my data here
If you would send me a short Email, I could send my questions to you directly.
Thank you in advance!
If you discuss about the funny P.M.P.O. (just now I understand the professional meaning of it): Does anybody know, how to extract the (approximately) real power handling capability of tweeters and midrange speakers from their nominal power handling relating to the given frequency range? E.g. 90Wrms @2kHz-20kHz with a 24dB/Oct. crossover. As I understood until now, it depends heavily on the audio material. The question is related to the UcD more or less, because I want to use 3 modules, each for one speaker in the cabinet and want to protect the chassis from overload.
Thanks!
edited:
Regards, Timo
That's what I see, if I try to do what you preferred.
You'll find all my data here
If you would send me a short Email, I could send my questions to you directly.
Thank you in advance!
If you discuss about the funny P.M.P.O. (just now I understand the professional meaning of it): Does anybody know, how to extract the (approximately) real power handling capability of tweeters and midrange speakers from their nominal power handling relating to the given frequency range? E.g. 90Wrms @2kHz-20kHz with a 24dB/Oct. crossover. As I understood until now, it depends heavily on the audio material. The question is related to the UcD more or less, because I want to use 3 modules, each for one speaker in the cabinet and want to protect the chassis from overload.
Thanks!
edited:
What I think about that: you enlarged the air gap only. Because the magnetic flux is everytime closed, the flux outside of the coil goes through the ferrite. The stray field is represented by the amount of the magnetic flux, which cannot be covered by the magnetic conductor (ferrite). Of course it will lower the coil's influence on the sound because of it's B/H curve linearisation (better: straightening) and the reduction of the hysteresis influence. And I agree with you, the effect is comparable to an air coil thanks to the very large gap.
Regards, Timo
tiki said:
Hi Timo,
I don`t think you can do much to prevent blowing up your tweeters when you feed to much signal to your amp or the wrong signals. If you have a less powerfull amp, then you will get clipping, tweeters don`t like that. If you feed your tweeters with too much low frequent signal, they may not like that. I think the only thing that prevents your tweeters or other drivers from blowing up is to use sensible crossover frequencies and slopes and be sensible with the volume control. I`m in the same situation as you, in the beginning it feels tricky to connect a tweeter directly to a poweramp. However, up to now, no disasters have happened. If you want to play a bit safer, you could use a capacitor in series with the tweeter to block DC and low frequencies to a certain extent, that will be at least some sort of protection (I`m not doing that but my tweeters are not that expensive (vifa ringradiator) so blowing them is not such a big financial disaster). I can tell you that the UcD modules behave very decent when switching the power supply on and off. I have no concerns connecting a tweeter directly to an UcD180.
Best regards
Gertjan
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.