Hey Joe, a happy user of your mk6 here in Brissy, even met you once at Mike Lenehans when you brought your speakers up from Sydney! All this talk about amplifiers has me thinking about an upgrade. What do you look for in a Amp that suits these speakers? I know from experience they work with anything but is there a sweet spot?
I. 4 drivers as in the Elsinore design have only a slight excursion during normal playing (90dB volume), and even at 100dB it will only be + - 1mm. Solving changes in inductance and impedance at excursion + - 5mm is only an academic matter, it will only concern exceptional cases of playing in large rooms at concert volume and music with sub-bass below 30Hz II. Impedance compensations are also unnecessary when using a transistor amplifier with a low output impedance as confirmed by my listening test. III. The use of the Vifa driver, on the other hand, is very suitable for the Elsinoire design, it has a smooth top end, compared to the previously used SBA and Purifi converters, on the contrary, it is clearly suitable for a first-order crossover.
Hello,
Watch what you say, some here may consider it blasphemy.
However you are mostly correct.
The capacitive Impedance compensations are there to correct the inductive load on the amplifier. Think of it like Power Factor Correction. Both the amplifier and crossover will function better.
Thanks DT
Watch what you say, some here may consider it blasphemy.
However you are mostly correct.
The capacitive Impedance compensations are there to correct the inductive load on the amplifier. Think of it like Power Factor Correction. Both the amplifier and crossover will function better.
Thanks DT
How about this one, the amp that was designed with Elsinores in mid:
Mike heard an earlier version and compared it to a $20,000 amp made in Italy.
Hello,
Looks like amplifiers are fair game again.
I like tube amplifiers as much as the next person. By today's standards they cost too much and they make way too much noise and distortion, THD+N or IMD.
I have a Benchmark that I use and three Topping LA90 power amplifiers that I use in a tri-amp setup. I use an old NAD 216 that I paid less than $200 powering the mini 2-ways on my bench. No hiss or hum/buzz out of any of them.
Try an Aiyima a07 for a few dollars to see how you like class D.
Thanks DT
I just compared current distortion on a 18sound PA mid driver using three different amps. The result was the same for all three amps, same distortion levels. However, adding series resistors made a clear improvement. Went from 0.11% to 0,07% IMD (number generated in ARTA) with a 6,8ohm resistor in series. Even lower with 10ohms. Might be something to play with in active setups I'm thinking.
I used Arta, a 0,6ohm current sensing resistor on the negative lead, and fed the voltage over the resistor to a balanced input on the sound card. Used dual tone 300Hz & 2kHz (this driver will be used as a mid within those frequencies). Before anybody asks, I did adjust the current with the resistor to the same level as without the resistor.
I also tried some inductors,1mH air coil, was a minor improvement, 1.5mH iron core was actually worse.
Interesting experiment, thank you Joe for the inspiration!
It started with when I was listening to music through the speaker, I thought it sounded a bit smoother with a resistor in series.
I used Arta, a 0,6ohm current sensing resistor on the negative lead, and fed the voltage over the resistor to a balanced input on the sound card. Used dual tone 300Hz & 2kHz (this driver will be used as a mid within those frequencies). Before anybody asks, I did adjust the current with the resistor to the same level as without the resistor.
I also tried some inductors,1mH air coil, was a minor improvement, 1.5mH iron core was actually worse.
Interesting experiment, thank you Joe for the inspiration!
It started with when I was listening to music through the speaker, I thought it sounded a bit smoother with a resistor in series.
You do know that amplifiers make less distortion at higher outputs. Putting the resistor in series makes the amplifier operate in a lower distortion range of its' output. Yes I do this with my tri-amp JBL D2 compression driver. It is more about lower amplifier noise and distortion than anything to do with the driver. If you have watts to burn this works.
Thanks DT
Yes, that is what I'm thinking too, might be worth burning some power.
IMHO it seems the current distortion has very little to do with the amp distortion, more to do with the driver characteristics and the impedance the driver 'see'. The distortion levels seen on the current are maybe 10-fold compared with the distortion generated by the same amps when feeding a resistor.
IMHO it seems the current distortion has very little to do with the amp distortion, more to do with the driver characteristics and the impedance the driver 'see'. The distortion levels seen on the current are maybe 10-fold compared with the distortion generated by the same amps when feeding a resistor.
The important measurement is acoustic, what the microphone hears.
What you are measuring now is voltage across a resistor. It may be amplitude differences and not distortion at all.
In the scream of things what you are measuring across the resistor may be small next to the distortion output of the driver.
Thanks DT
NOT THIS ONE!
BELOW: Low noise, low distortion, and designed to deliver current so that the amp produces low distortion on both voltage and the current side:
Note the AC voltage ratios listed on the left, that the EL34 output stage, the tube itself, and how close to unity gain it is at 1.5:1, and also the input gain is 756 or 58dB. This high input gain is required because the EL34 basically has no real gain. By tweaking two small capacitors and with half-decent output transformers, the response can be flat to 100KHz, but even without those caps tweaked, the response is dead flat at 20KHz without any negative loop feedback from around the secondary of the transformer.
Is this the lowest distortion amplifier that uses tubes in the output stages? I think there is a good chance. Only local feedback is used. Finally, the power Fets are current drive via their Sources and there is no signal on the Gate, which means there is no damaging Gate to Source capacitance that we normally see in power Fets circuits. No capacitance also means extraordinary bandwidth because it does not get rolled off.
There are a LOT of tube hybrid amplifiers out there. But they have tubes only on the input stages and solid state on the output. This is a reverse hybrid amplifier. The output transformer in the above circuit is not thought of as a step-down transformer. In fact, it only makes sense that this is a step-up transformer current wise. Connect an ideal single driver to the secondary and it will only see the current of the transformer and not its voltage.
Indeed a close analysis of the circuit and getting an understanding of it, shows that the circuit above is more based on current than voltage, indeed the first input stage are balanced Darlingtons are transconductance stages, where voltage on the input is immediately turned into current. The way the output tubes are driven, there is a current component. The above circuit and its development go back to 1983 working along with the late Allen Wright. Certainly, the Vacuum State DPA-300B amplifier is a precursor to this circuit but with the significant point that the tubes are operating beyond triode and the DPA-300B was not. I did the early proto-typing of the DPA-300B (there are numerous witnesses), but the above circuit and its performance are on another level above that. I only wish Allen would still be with us because he would have been excited.
This amplifier reduces what Esa Merelainen calls "EMF derived distortion" induced by the non-linearities of the driver. I believe that the point that has been missed is that this current distortion is produced by the amplifier and not the driver. That means each amplifier will not necessarily copy and produce the same distortion as another amplifier. That is a big deal to say this. Although they will both be producing current distortion unless the driver is perfect. Alas, it is not. So two amplifiers will produce distortion, but their distortion profiles may not be the same.
Cheers, Joe
What I did yesterday comparing three amps, the current distortion profiles were as good as identical between the three amps. I used the overlay function in Arta so I could compare the FFT's. These were all solid state amps with EF outputs, two of them DIY and one HK. Tube outputs with transformers etc might be something entirely different.
DT: I get that the acoustic output is what we hear, but I believe output is affected by several factors, and the current distortion is one of them so, studying that is of some interest I think. However, I don't think spikes on the FFT of the current can be considered amplitude swings.
DT: I get that the acoustic output is what we hear, but I believe output is affected by several factors, and the current distortion is one of them so, studying that is of some interest I think. However, I don't think spikes on the FFT of the current can be considered amplitude swings.
Can you detail the setup for measuring current distortion? I just want to know if we are talking about the same thing.
Can you elaborate on this, why would amplifier take the distortion created by driver for it to own and claim its mine? If you analyze the circuit from loudspeaker side, its easy to understand its the driver creating its own distortion current and emitting it acoustically. All the amplifier does is it provides low impedance path in the circuit for the distortion current to flow, unless series resistance (like high output impedance of current drive amp) or impedance with passive network (like a series coil) reduces it. This is the issue what makes your explanations on this subject hard to understand for me, much easier just to look at it this way from speaker side and it all makes sense. Peace
I wrote it in my earlier post https://www.diyaudio.com/community/threads/the-elsinore-project-thread.97043/post-7116385
Please ask if you have more questions.
Yes, the series resistor is an improvement. But how is this for a thought, if the speakers are 8 Ohm, then put an 8 Ohm resistor in parallel with the speaker, this presents a four Ohm load which should not be a problem for the amp.
The 8 Ohm resistor forces the amplifier to produce more constant current with frequency variation and flatten electrical current phase angle. I call it "poor man's current EQ."
MANY OF YOU GUYS READING THIS SHOULD TRY IT AND LISTEN!
Also, it is important as a reference to show both voltage and current distortion. The voltage should show the ideal that we never get and the current distortion will also be greater.
What signal stimuli did you use? 1KHz is not quite enough, which I suspect you are using?
Did you use a current sense resistor? It should be low in value. Here is a suggested setup:
An 8 Ohm driver usually has a DC resistance somewhere around 6 Ohm, so using a current sense resistor of 0.1R use s small very low distortion opamp like LM4562 or similar.
May I suggest that you try combine 150Hz and 1250 Hz, an example is shown below:
Disclosure, this is Esa Merelainen's measurement.
The 8 Ohm resistor forces the amplifier to produce more constant current with frequency variation and flatten electrical current phase angle. I call it "poor man's current EQ."
MANY OF YOU GUYS READING THIS SHOULD TRY IT AND LISTEN!
Also, it is important as a reference to show both voltage and current distortion. The voltage should show the ideal that we never get and the current distortion will also be greater.
What signal stimuli did you use? 1KHz is not quite enough, which I suspect you are using?
Did you use a current sense resistor? It should be low in value. Here is a suggested setup:
An 8 Ohm driver usually has a DC resistance somewhere around 6 Ohm, so using a current sense resistor of 0.1R use s small very low distortion opamp like LM4562 or similar.
May I suggest that you try combine 150Hz and 1250 Hz, an example is shown below:
Disclosure, this is Esa Merelainen's measurement.
Where would you put the measurement resistor in the above circuit? If you measure current going to both, it will naturally be nicer, since the resistor is a linear load, but if only the current going to the driver is measured, I bet it will be the same. I see the problem as being in the un-linearities in the driver causing distorted current, and the distorted current 'reflected back' into the driver causing distorted sound. If the driver 'sees' a higher impedance, less is 'reflected back'. The relationship to distortion in the acoustic output is another story, but it seems the Purify-guys think it matters. Maybe I will try measuring acoustic distortion from a driver fed via a series resistor vs no resistor.
I loaded the measurement setup with a resistor ('ideal load') to get a baseline distortion level, which should be the same as voltage distortion from the amps, and it seemed comparable in distortion level as when I measured the distortion on the amps normally (-80dB or more). Connecting the speaker really made the FFT ugly in comparison. I briefly tried to connect a Vifa C17 too, and that was slightly worse in numbers, but the distortion profile (harmonics) was notably different. The main driver of interest was a 18sound 8NMB420, nominal impedance 8ohms.
As I wrote earlier (post #4989), I used 300Hz & 2kHz, and a 0,6ohm sensing resistor, other than that it is the same as your setup. I did not need an opamp, since the balanced input on the sound card is pretty sensitive (Focusrite). I did not screenshot all the FFT's and make a complete report, I just noted the IMD % values shown in ARTA. I did however visually compare the FFT with 'overlays' in ARTA. It was just a quick test out of curiosity, and I think I am a it wiser from it.
Yeah this is how I see it happening as well, from Esa book and from Purifi papers.
I'll shortly comment on the 'reflect back' description you used. As its all happening simultaneously 'reflect back' is kind of a misnomer as it denotes something that needs to go somewhere and then back, which is not simultaneous
And here is the long thought process if someone wonders: Voice coil moves because amplifier output voltage makes current over the amplifier load impedance, basic function of loudspeaker. We need voice coil to move as this makes into acoustic sound and current through the voice coil makes it move.
Voice coil movement makes back-EMF voltage in the driver motor, which results current over driver load impedance that also flows through the voice coil and makes it move, all happening simultaneously, its all in the same circuitry. This is current on top of the current amplifier is supposed to be making through voice coil but its generated by the driver itself as byproduct, distortion current, and if let to flow makes voice coil move it makes distortion appear in acoustic domain as well.
As long as driver moves the same the back-EMF voltage stays the same because its property of the driver and we cannot do much about it, but current resulting from it depends on impedance between driver terminals that completes the electrical circuit where current can flow. Reducing driver generated distortion current reduces distortion in acoustic domain.
Amplifier can also make distortion if its load varies, but that is fixed with the conjugation network, it is separate issue.
If resistor is parallel to driver as in Joes image it makes amplifier load more resistive but does not do nothing for driver load. Driver generated current flows through amplifier output impedance instead which is almost a short circuit with voltage amplifiers. This very low impedance path enables max distortion current in the circuit which realizes max driver distortion acoustically. Put the resistor in series instead and now also driver load impedance increases, driver generated current decreases, acoustic distortion decreases as result. This driver generated distortion seems to be much more than what amplifiers generate, so its backwards thinking to try fix the amp or look it from amplifier perspective when all we need to fix is impedance between driver terminals.
Last edited: