Single-ended triode amps

[PeterSt]

Mani, not weird. The source of the sound is from 3 dimensions or so different (height, direction, reflections).
A 4th dimension would be how clean that sound/sine is. For example, put up a square and see how "useful" that output is. And before you try for real, I don't think you can measure the native sine as such from the speaker. So this starts right in front of the speaker itself, but you won't get any nice sine. No promise, but that's what I think and expect.
So you will get more than 50Hz and you won't be able to tell what you hear.

This story is not so much different from "spreadening" the wave because of the reflections from the wall behind the speaker. And why do I have the hunch that you don't want to believe that.

Anyway, nothing weird. Just as could be expected.

Regards,
Peter

[manisandher]

So you will get more than 50Hz and you won't be able to tell what you hear.

I'll take some measurements next week - pure 50Hz sine wave fed to speakers, recorded at seating position with measurement mic, and analyzed on Audacity. Should tell exactly what's happening.

And why do I have the hunch that you don't want to believe that.

Peter, what do you want me to say? I haven't expressed to anyone my real level of disappointment in the Orelos not working out for me. I know they're great speakers, but they way they sounded in my room...  It's a shame you never managed to make it over and take a listen for yourself. Actually, Paul's earlier account says most of it. The only thing it doesn't (can't) convey is my often deep exasperation over the last two and a half years of not enjoying music in my main room. My office  and portable systems have been my saviours in this time. I now have a system that I love the sound of in my main room. Quite frankly, I don't care if it's distortion all over the place - I'll take it in a heartbeat over the sound I had before.

Mani.

[CoenP]

I think this thread is causing confusion, at least to me so hereunder an attempt to clarify my view on power a little.  

Peak sinus power and average RMS power have physical definitions.

Rated amplifier power is always rated at a certain (restive) impedance.

This has a lot of implications.

————— tutorial break “understanding power”—————

Delivered Power is defined as (Applied Voltage)x(Required Current).

The amount of current required for a certain applied voltage is determined by Ohm's law: Required Current (Ampere) = Applied Voltage (Volt) / Impedance (Ohm).
In loudspeakers this impedance is frequency dependent and as such has a "phase angle" that defines the lag between the current and applied voltage. This last complication is not needed to understand power delivery.

Say the impedance at a certain frequency is 8 ohm (no phase angle). If we have a signal that peaks at say 40 Volts than the required peak current would be 40V / 8Ohm = 5 Amperes, hence the required peak power would be 40Vx5Amp = 200 Watts.
A sinusoidal signal averages out on the 'Root of the Mean of the Squares' aka RMS value which is root(1/2) x Max voltage. At root(1/2) x Max voltage, the required average current current would be: root(1/2) xMax Voltage / impedance = root(1/2) x Required Current. The Average Power in RMS will now be the product of those two values hence root(1/2) x root (1/2) x peak Voltage x peak Current = 1/2 x  Peak Power (sinusoidal).

Halving the impedance to 4 ohms would double the required current for a given max voltage In the case of 40V this will be required peak current of 10 Amps in stead of 5 Amps, so double the power would be required by the loudspeaker :10Amps x 40V = 400 Watt peak. Doubling the impedance to ie 16 ohms wil consequently half the required power.

The above stated ‘phase angle” makes these precise calculations of the ‘current - voltage relation’ harder and more abstract, but what happens in practice is that there can be (far) more or (far) less current required by the loudspeaker (“load”) than the resistor would predict. In other words there may be more (or less) power required than in the resistor case. The amplifiers design has to be able to cope with this.

—————— end of tutorial break — — — — — — — -

Now a power amplifier specification measurement is usually performed with a resistive (no phase angle) load of a standardized value (to be able to compare). Mostly this will be 8 ohm. Power amplifiers however have two physical constraints that determine the maximum power that it can deliver into other loads than that value; ones that are greater or smaller than 8 ohms (and/or when the phase angle requires more or less current).
One is the maximum available voltage, the other the maximum available current.

Typically transistor power amplifiers operate at voltages that are close to the voltage that is required for the 8 ohm power specification. Our 100W RMS (hence 200W sinus peak into 8 ohms) amplifier wil likely operate at 50V-60V tension. This causes an available power reduction for a 16ohm load (2,5Amp needed for 40Volts or 100W peak or 50W RMS), yet the amplifier may be able to supply the 10 amps necessary for a 4 ohm load and as such provide more power when more amps are needed. Eventually the amplifiers power supply is not able to provide any more amps before blowing the fuse or significantly degrading the supply tension.
So this amp’s real maximum delivered power will be determined by how much current it is able to supply for the maximum voltage (about 40 V). This can be much more than the rated 100Watt.

 Amplifiers with a typical current constraint are OTL designs. The tubes operate at very high voltages (150-200V), but cannot source more than the tubes allow for the 8 ohm per specification (like 1 amp per tube max). That is that a 100Watt RMS (40V, 5 Amps at 8 ohm) rated OTL amp will not be able to deliver the 10 Amps required for 4 ohms, but it is able to deliver 5 Amps into 16 ohm. That is 5 amps at 80 Volt or 400Watt peak or 200Watt RMS. This amplifier real maximum power is limited by how much voltage it is able to generate for a 5 amp load. Also this can be much more than the rated 100Watt.

Usually -Single Ended- Tube amplifiers are designed to optimally couple the tube to to the load. In such a case the maximum Voltage and maximum current occur both at the rated impedance.
So our example amplifier of 100 Watt would not be able to supply more than 5 amps and more than 40 Volts which occurs precisely at 8 ohm. Both maxed out at the same time, so theoretically one point with maximum power.
Well those are the ratings. In reality the tube amp can deliver more than 5 amps or more than 40 Volts but not at the rated impedance. This flattens the ‘peak available power’ curve somewhat. Please note that this maximum power is more like a real ‘peak/RMS power’ unlike the two examples above.

With these three amplifiers you can understand that they favor different loudspeakers. The transistor amp theoretically matches well with a loudspeaker that has a low impedance  (<8 ohm)or impedance dips, the OTL favors high impedance ones (>16 ohms) and can handle impedance peaks well and the the transformer coupled tube amp is best served with a constant impedance at the rated value.

For sure there is much more to amplifiers than maximum deliverable power , but my experience is that this rule is implicitly followed by many ’system builders’.

Lastly I wanted to observe that an increase in power does translate into an equal amount of increase in sound pressure. Most amplifiers act as voltage amplifiers so the current follows the applied voltage. As the voice coils heat up and become more resistive, less power is absorbed for that voltage. Also speakers' excursion does not follow the power (and certainly not voltage) in a linear relationship extra power is less extra excursion than you expect.

This power transfer thing is where horns have big advantages. The compression drivers provide a very resistive load to the amplifier and they use low power, hence little heating of the voicemail occurs. And lastly excursion of the compression driver is extremely small and thus very linear. They are ideal candidates for transformer coupled tube amps, especially of the DHT variety that exhibit extremely low distortions at low signal levels.

I hope this helps somewhat to be more specific in what we try to say.

regards, Coen

[PeterSt]

It's a shame you never managed to make it over and take a listen for yourself.

Hey Mani,

Can you please rephrase with proper context ?
I am sure that in English it can read positively, but I am more sure it can read negatively all the more, and I am dead sure that Dutch and maybe others *will* read this as a pure negative.

I have my response ready, but so far I feel this is not for the public. A few words to make it look more positive seems to be sufficient.

Outside of this, I fully agree (I stay neutral for now).

Peter

[manisandher]

It's a shame you never managed to make it over and take a listen for yourself.

Hey Mani,

Can you please rephrase with proper context ?

Well, I meant this quite literally. If you or Bert had heard the Orelos as they sounded in my room, I don't think you would have recognized them at all. And I think it would have required this first hand experience to really get a grasp of what might be going on.

This is in no way a criticism of you or Bert for not coming over. All your other Orelo/Orelino customers seem perfectly happy and you can't go flying around the world when something doesn't work out for one person.

Is this clearer?

Mani.

[PeterSt]

I think this thread is causing confusion, at least to me so hereunder an attempt to clarify my view on power a little.

Thank you very much Coen.

N.b.: I largely repeat myself with the text below, so maybe skip. Better read Coen't post again.

Maybe I should have added the Voltage factor, which I didn't because of simplification but also because it seems too obvious ... to me. But even with your great post we can explicitly add some text about that :

So the volume we apply, is the Voltage we apply. Very low volume is e.g. 5V, high volume could be 40V.
However, this relates to the last part of Coen's post, where high sensitive horns already play massively loud with 5V. Or if you like ... at 1W (when sufficiently efficient). So it always comes down to this one :

Required Current (Ampere) = Applied Voltage (Volt) / Impedance (Ohm).

and Ohms law is indeed everywhere. What is not in this equation is how much power the loudspeaker driver needs to move. Now though we have the eg. 109dB sensitivity "by means of a power of 1 Watt at 1 meter distance". Somehow this 1 Watt translates to 2.83 Volts.  So ... P(ower) = U (Voltage) x I (Amperes).
P = U x I is the same as I = P / U. Thus, 1 / 2.83 = 0.353 Ampere.
Aha, so 0.353 Ampere is needed to blast 109dBSPL from a 109dB sensitive speaker.

I think the larger deal of Coen's post is about that this is not realistic because of several reasons. The foremost one (I think) is the required current which is invisible to us and can be the limiting factor like Coen told (side note : this is the different approach from taking care of infinite current supply and next see your amplifer melt down because of insufficient heat dissipation - those Watts need to dissipate and fast).
More indirectly the impedance is even more invisible and as we know the less impedance the more current required. Lastly - but I repeat myself - the music itself (the frequencies) determine the level of the impedance. Thus, also when a speaker is rated at 8 Ohms, it can be so that it most often plays with 12 Ohms or more, now consuming less current than when frequencies pass by which go all the way down to the rated 8 Ohm and which is thus more difficult for the amplifier (it requires more current).

Let me lastly add that clipping of an amplifier should not happen and *if* it happens, the design is not correct.
On the other hand (big fun) if you allow for infinite current and ratings are exceeded (for example you buy a new loudspeaker and you didn't think of the fact it can go down to 1 Ohm), the melt down is underway.
Also good to see through is that while the current can be virtually infinite (just by means of the design and all), the allowed Voltage explicitly is not. Thus, when I would design an amplifier then the voltage is always limited because the parts used are voltage limited. And this limits the current required  (Ohm's law : I = U / R) ? No not really, because the R is there. That bloody impedance. So if I would design an amplifier for my high sensitivity horns with sufficient current supply, will it melt down on a 85dB inefficient speaker ? No. There will just be no sound (it now requires 40V instead of e.g. 5V).

Thank you Coen.
Regards,
Peter

[PeterSt]

This is in no way a criticism of you or Bert for not coming over. All your other Orelo/Orelino customers seem perfectly happy and you can't go flying around the world when something doesn't work out for one person.

Okay Mani. Thanks.
Btw, I would - that flying. The sad thing of this case is that it literally didn't happen, while it was surely planned. It was even planned for a second time, or at least something of that order. Ask Paul for fun, what I wrote him after your last visit. And not because you complained (heck, you should have, am I right ?).
Anyway, it is good as it is as long as you have found some peace. If you only notice that this is not per se my own situation. That this is coincidentally related to that "shame that it did not happen" could be just a bit too sensitive for me. What you don't tell, you don't tell.
Mind you, you actually once again imply that someone should have come over. I am starting to think that someone may have missed an email (to read or to write).
I suggest that if we need more of this, someone starts writing an email instead of having it on a forum.

Thanks,
Peter

[CoenP]

Hi Peter,

Thanks for the additions!

let me push yours a little further:

There is the efficiency of the loudspeaker that tells us about the amount sound pressure we get from 1 watt of power and the sensitivity that tells us about the amount of sound pressure we get from 2,83V RMS (which is of course 1 watt @ 8 Ohms, apply the math!). They are related via the impedance.

The sensitivity tells us about how much voltage the amplifier must deliver to play at a certain loudness, the efficiency purely how electrical power translates into sound pressure power.

Simply put the efficiency does not take the impedance into account, but the sensitivity does. A 100dB/watt efficient loudspeaker can have an impedance of any ohm. I have Philips 9710 AM that are 800ohms rated impedance and about 100dB efficiency, contrary to Peter's 8 ohm horns (2,83 volt) you will need a truckload of voltage (28,3V) to get a watt into them.

It takes some effort to estimate the amount of SPLs in your listening chair. As old fashioned dual channel listeners we have actually two amplifiers and two loudspeakers generating a signal. This potentially adds 3dB to the maximum attainable sound pressure.

Yet our chairs and couches are some meters from the speakers, we are usually not at the specificatied 1 meter away from them. Depending on the radiation pattern (directivity versus frequency) the sound pressure lowers with increasing distance from the loudspeaker. As a rule we may assume that the radiated surface increases with distance and as such decrease between 3 and 6dB every doubling away. So if the output is at 1m 100dB it is at 2m: 97-94 dB and at 4m 95 -88dB.  This suggests that we need more than 120dB (minus 3 ....) at the loudspeakers to get 120dB at our chairs.

But the room is not an infinite space, that is waves reflect on all boundaries: the floor, ceiling, sidewalls, but also tables closets etc. Especially the side and behind walls are acoustically close to the loudspeaker and as such are the reflections are significant. Well that is for those frequencies that the loudspeaker has a wide dispersion. Anyway a lot of energy is confined in and and interacts with the room. Signals with low directivity (lower notes) interact the most. This accounts for extra decibels for many frequencies.

In the end I do not think we need 120dB+ capable loudspeakers and amplifiers in our rooms. My 103dB Altec VOTT was intended to operate with a 15 watt amplifier. Originally they were designed for intelligible sound in a 400 seat movie theatre. I already got loudness complaints with my 2x0,8watt tube amp so in practice the "120 dB calculations"  do not really make sense.
Unlike the image industry there are little generally accepted and implemented references for sound. If this were the case we would be able listen to realistic sound levels, at least how they were recorded or intended by the engineer. Imho the 120dB case (a full-out orchestra: Mahler/ R. Strauss) can only apply with such a reference. In all other cases we would have changed the volume (=power) with the crescendo to something we can bear in our rooms.

regards, Coen

[vrao]

It is my understanding that high powered amplifiers could not be "reliably" made back in the day. Therefore an elegant solution, hand in hand with the horns.

Amplifier (also the speaker) should have enough power reserve to handle transients ... if the distortion/clipping happens, then its almost always noticed, hence the "sound signature" discussion of the equipment ... Here its mostly the high efficiency crowd, and not much of an issue. but elsewhere its another story. For example my prior Omnis were 84 dB sensitive. I would play it at nominal 78-80 dB. I don't even want to extrapolate the power requirements for lets say 20 dB headroom, with room loading included. It could not handle more than 500W for more than 3sec. Something had to give, with that low-efficiency, heat I suppose ...

I will put this across, most tweeters are actually distorting at normal listening levels.

[acg]

Mani, here is a link to the Nelson Pass stuff you mentioned in your first post.  At least I think it is that to which you were referring.

vrao, as the resident expert, what do you make of Mr Pass' writing?  I would like to agree with him because I have skin in the game (am current building my SET amplifiers), but I have never heard anything as "real" as SET amps matched to appropriate speakers.

[briefremarks]

Adding one link to contribute to the discussion.  A good discussion (with tests) of the distortion behavior of chip amps

http://www.customanalogue.com/diytubegainclone/gainclonesound.htm

[PeterSt]

I only had time for 10 seconds to look at that discussion (really not more) but with a noise line at around -90dB, forget everyting (including the discussions I suppose - but again, I did not read anything).

We are looking at -120dB after a gain of 20 (via NOS1a with a noise line a bit better than -140dB).

Peter

[vrao]

Acg:

I had a response I was working on, and ran out of time, and lost it all being logged out

I don't have any more time

But all in all I agree with NP...

Best,
VJ

Edit: Amp/speaker distortions are like Hydra .... you enderstand one and conquer it, few more arise, that didn't didn't exist before .....
 

[acg]

Thanks VJ

[manisandher]

Mani, here is a link to the Nelson Pass stuff you mentioned in your first post.

Anthony, here's the correct link: https://passlabs.com/press/single-ended-class.

Many years ago, I owned a Pass Labs Aleph 4 amp - 100Wpc solid-state single-ended. It was a very smooth-sounding amp, but didn't have enough 'kick' with the Quads electrostatics that I was using at the time, for my liking. I remember when I sold it on eBay. The buyer came over with his girlfriend to collect it and obviously wanted to listen to it first. I put on a track with female vocals and I could see massive smiles immediately on both their faces as soon as the track started. He looked at me and asked me why I was selling such a great-sounding amp.

Back to NP's article... I wonder if his thoughts are only really valid when considering a truly mass-less speaker cone. And perhaps this is why PP amps just tend to sound so much more dynamic than SE amps? High efficiency horns certainly help...

Mani.