I was actually thinking about making a medium-high efficiency 2 way. SB acoustics has fantastic tweeters measuring incredibly flat with 90-96dB sensitivity. Only problem is finding a quality midwoofer with that efficiency. The best I've found so far is the Scanspeak Illuminator 12MU/4731T with 90dB efficiency. Maybe a MTM with 93dB efficiency? I really want to make a MT though... MTM's are just ugly to me.
Apparently open reel master tape playback is also third order dominant. May be why many people associate it with sounding more detailed than digital, etc. I'll see if I can find the paper.
The third order is the dominant distortion in analog tape replay. However in a studio mastering system it should be inaudible. Most analog master tapes are pretty old now though. The treble fades away first. The extra equalisation on playback of old tapes inherently increases distortion so that it becomes audible.
1940's and 1950's speakers intended for quality sound sound ok, superficially (if the cone is not worn out). But put a 1950's speaker side by side with a modern quality speaker, and usually the better clarity of the modern speaker will be immediately apparent.
1940's and 1950's speakers came in essentially three types.
The first type was intended for cheap radios and potable radios. They were designed for fully enclosed but not sealed radio cases and have cone resonances around 110 to 200 Hz.
The second type was intended for table radios and radiograms. These were designed for open back baffles and have a cone resonance typically around 50 to 60 Hz, some up around 150 Hz.
The third type was intended for the "hi-fi" market and nearly always designed for bass reflex baffles, and have a cone resonance usually around 35 Hz. Sealed baffles, also known as acoustic suspension, did not become popular until the late 1960's. Sealed systems are less efficient than bass reflex and give much higher distortion. But they are a lot smaller. Their arrival roughly coincident with solid stage hi-fi is part of the reason why people think that tubes sound nicer than transistors.
In American speaker industry terminology, also used in Australia and other countries, the first type used mostly F62 or F63 cones.
The second type used mostly F25, F26, F50, or F69 cones.
The third type used F20 or F59 cones.
If you see an old 1940's, 1950's, 1960's speaker with a code number Fnn or endeing with Fxx, you can thus tell what the speaker will perfom like. The F numbers are like the type number of tubes.
The cone resonance frequency approximately determines the lowest frequency the speaker can reproduce efficiently.
Western manufacturers of tube radios and TV's often requested their speaker suppliers to supply speakers optimised for the size and shape of the radio or TV cabinet. Or to give a particular sound. These would be given other F-numbers. You can encounter two speakers, say one from an old radio, and one form a TV set, that were from the same speaker maker, and look identical. The radio speaker will have say an F63 cone and the TV speaker an F14 cone and sound completely different.
Not true. Not even remotely true.
Of course, you cannot expect an 8-inch speaker from a 1940's portable radio to do as good as a 1940's 8-inch speaker intended for quality hi-fi.
Last edited:
mine does 50 wpc into 4 speakers, but peaks rarely go over 3 watts when loud.
the .001 thd and 115db signal to noise ratio are more important then raw power
the .001 thd and 115db signal to noise ratio are more important then raw power
But what do you listen to?
Different types of programme make different demands on your system.
1960's Motown, just for example, was engineered to sound good on 1960's car radios, which were by today's standards, not all that good - just one channel and AM radio. Motown is so compressed it doesn't really matter what you listen to it on, so long as it not a transistor portable.
A lot of 1950's and 1960's rock was recorded on substandard equipment - the famous Sun Studios for instance used standard AM radio station equipment, not proper recording studio equipment. Equipment absolutely incapable of reproducing anything outside the 50Hz to 10KHz range and a signal to noise ratio specified at no better than 60 dB. A fair quantity of British music was mastered onto tape recorders no better than good domestic standard eg WeairWright decks. Only the top money spinning acts or acts with recognised talent eg Beatles, Cliff Richard, were consistently recorded on fully professional EMI equipment.
I listen to mostly rock, jazz and some classical and country. some of the recordings are much poorer then others, however the 12ax7 preamp stage really improves the overall smoothness.
Your general statements may be relevant to many vintage speakers but there are many quality German manufactured drivers which are sought after for the quality of their designs.
I point you towards the Isophon range in particular.
Shoog
No dissagreement there. Quality British and from other European makers as well. Not made to American standards or nomenclature, obviously, but very good. Also Philips.
I was replying to to your statement that 1940's and 1950's speakers could not reproduce bass as well as modern - not true as the American "F-code" nomenclature I posted shows; and your claim that these speakers almost all required open back baffles - not true, most quality speakers then were intended for bass reflex.
I have original 1940's, 1950's, 1960's, and 1970's datasheets on American and Australian speakers (Australian production stopped in the 1970's, except for a couple of small niche market companies), and I designed products using them back in teh 60's and 70's, so I can write about what they made. I don't have any info on European speakers of the same eras (except for some basic Philips catalog listings). I have no reason to believe they were inferior.
I am aware that the trend to sealed boxes/acoustic suspension started in Europe and was stronger in Europe, as their small dwellings compared to American and Australian houses made small things much more desirable - and sealed boxes are much smaller than bass reflex boxes for the same bandwidth.
There were also cultural factors. Dutch and German folk preferred a response that was flat or a few dB down in the bass region. American and Australian listeners have always prefered their bass somewhat boosted (ie what a typical American thinks is flat isn't flat). In Australia, Philips could not sell their top high quality lines for that reason - even though those of us with technical qualifications and the ability to test with instruments knew they were as good as anything from America.
Last edited:
Incidentally, I have recently designed a transformerless singled ended push pull tube tube amp (40 W into 800 ohms from parallel push pull TV line output tubes) prompted by a request from the editor of a certain electronics/audio magazine.
I do have a full range of test instruments, but I need an 800 ohm speaker system so I can audition the amp. I don't have one so I purchased some used Philips 800 ohm 1960's production woofers. Testing shows they are as good as new and will reproduce bass down to about 32 Hz. But I am having trouble sourcing good 800 ohm tweeters (400 ohm will do). I purchased some Philips 800 ohm midrange/tweeters (half a doxen from various sellers) on ebay but all turned out to be faulty. I may have to use a modern 8 ohm tweeter with a wacking great custom transformer.
I do have a full range of test instruments, but I need an 800 ohm speaker system so I can audition the amp. I don't have one so I purchased some used Philips 800 ohm 1960's production woofers. Testing shows they are as good as new and will reproduce bass down to about 32 Hz. But I am having trouble sourcing good 800 ohm tweeters (400 ohm will do). I purchased some Philips 800 ohm midrange/tweeters (half a doxen from various sellers) on ebay but all turned out to be faulty. I may have to use a modern 8 ohm tweeter with a wacking great custom transformer.
Last edited:
Keit,
Here's your chance to build those line arrays you've been planning - put 50 16 ohm drivers in a vertical line and wire them in series. Voila, 800 ohms!
I actually thought seriously of doing that.
But I can't justify the expense of 100 drivers (for both channels), for something I don't need other than as support for a magazine article, which may pay me no more than a couple of hundred dollars. The amp is no problem - it's mostly made of old tube TV parts I had on hand.
I'm short of space - I already have several good stereo systems - ones I had designed and built over the years. So when the editor has got whatever photos he wants, and has sent someone independent of me round to have a look and a listen if he wants, the amp will be dissassembled and probably the completed 800 ohm speaker systems sold on eBay. They'll be unique.
Last edited:
you could use a 70 or 100 volt line transformer to drive an 8 ohm tweeter with that woofer
just place the crossover cap on the primary. the Transformer can pass more power at higher frequencies.
just place the crossover cap on the primary. the Transformer can pass more power at higher frequencies.
Member
Joined 2009
Paid Member
For me it's different. I have an amplifier you would describe as hi-fi, very low distortion. But I have other amplifiers that sound different and I don't consider them to be bad designs since I like their sound.
The whole idea of transformerless push pull (SRPP & other tube topologies) is essentially twofold:-
1) you avoid the distortion inherent in an iron-core transformer
2) There is no transformer, with its leakage reactance, shunt reactance, and interwinding capacitance, in the negative feedback loop, so the loop can be a lot tighter, so even less distortion.
So using a transformer to enable use of an 8 ohm midrange/tweeter needs to be considered with care and to a certain extent defeats the idea of having an 800 ohm system. A PA line transformer will not do. Pa line trasformers are not intended for high quality use.
Not that unlike a normal tube amp output transformer, a midrange/tweeter transformer will be outside the feedback loop, so its iron core distortion (which is somewhat like cross-over distortion) is more obvious.
A physically large custom transformer will be needed, based on a ferrite core - physically large even though its not going to be passing bass. And, really, I would like no transformer, else some readers might say "well that's why it sounded no better than his equivalent standard tube amp", even if that's actually wrong. If it does sound no better - I actually think that will be the case.
Outside a few reports in professional journals that I have found, there seems to be no comprehensive description of critically evaluating SRPP power amps.
Last edited:
I have some tweaked Bottlehead Straight 8 speaker clones that I like very much. They are 8 speaker line arrays with a tweeter, and are 96dB efficient. Everyone who has heard them have had very good things to say about them.
I want to build a PP ultra linear amp using cathode feedback windings, looking for about 30 watts out of them. I've heard these speakers with 30 watts, and the dynamics are great. They run more than loud enough with my current 8 watt 300B amps, but looking for better peaks. Anyone know of a source that would wind a transformer for such a beast? Softone out of Japan makes one, but it only has 8 ohm outputs, and I need 16 ohm.
twystd
You could consider using a transformer with a secondary with 4 ohm and 15/16 ohm tappings. Such a 4/16 ohm secondary is centre tapped. You can earth the centre tap and connect the cathodes to the ends of the secondary.
This means you can use a standard output transformer, you don't need one which special feedback windings.
On the face of it, doing this is sub-optimal - the voltage on the secondary is too low to give optimal 3rd order cancellation like ultralinear. But in practice, as Neville Theile found, it's not so bad. The gain is higher than for an optimal cathode feedback winding, so the demands on the phase splitter or driver stage are less, so it will distort less.
Mr Thiele also showed that even if you use a 4/15 ohm secondary (which means the centre tap is not an exact centre tap, it still works very well.
However, should you wish to get a custom transformer with a special cathode winding, I would be surprised if you can't find a local transformer company to do the job. Even in my city, a city with almost no electronics industry, there is a company that can do it. But you will need to deal direct with the company's engineer, and not go thru regular sales staff - or you request will be mangled and you'l get a "power transformer", not an output transformer with low leakage inductance etc.
Last edited:
Not really following you there. here's a link to the Softone OPT with a winding schematic (and a lot of other info as well): Softone RX-80-5 Push-Pull Audio Output Transformer That I understand, bear with me as I am entirely self taught (was a printer for most of my life), so kinda slow on this stuff.
I should have made myself more clear, I'm thinking of Ultra Linear windings from the primary connected to the screens of the tetrodes, and running cathode feedback windings on the cathodes as well. Some people call this super triode.
One more question if you don't mind, In the case of the Softone transformer, it applies 8.2 dB of feedback to the cathodes, since 8.2 dB equals a voltage gain of 2.57, does this mean that my driver stage has to swing 2.57 times the voltage to drive the power stage to full power, as compared to no feedback? In other words, if in class A it would take a driver stage to swing 38V peak, or 76V swing peak to peak in each phase to drive the power stage to full power with no cathode feedback, would my driver stage have to supply 2.57 times that voltage for full power with 8.2 dB of cathode feedback? Any "feedback" on this would be very helpful🙂
twystd
What I was suggesting is the same as the webite shows under the heading "Cathode Feedback" - the only diffrence being is that their transformer gives you a double 16 ohm winding (ie 64 ohm end to end), which is better than using teh 16 ohm end-to-end secondary of a standard transformer.
I see. To secure optimum performance you will have to get a special custom transformer.
You are correct. That is why I said in my earlier post that cathode feedback at a less than optimal setting can actually perform better, or at least just as well, as the more swing you demand from the driver, the more it will distort. A simple rule applies: for every db of more output demanded, 2nd harmonic goes up approx 1 db, 3rd goes up approx 2 db, and so on.
Cathode feedback is quite rare, precisely becasue it makes the driver distort more. You even completely run out of headroom, if the cathode feedback is high enough.
Combining cathode feedback with screen feedback is even rarer still. That's because both do the same thing - reduce 3rd order distortion in the output stage by a grid curvature cancellation mechanism. If the screen tapping on the transformer is an optimal tapping, ie at the point which gives minimal distortion, adding cathode feedback will not make it any lower.
If you look at things from inside the tube, ie look at what the anode and screen are doing with respect to cathode voltage, both screen feedback and cathode feedback give the same conditions. Thus both use screen grid curvature to oppose control grid curvature. It is of course the curvature of the grid voltage/anode current relationship that causes distortion. So by playing off one grid against the other, we get to cancel out distortion. It is not just simple negative feedback.
The advantage of cathode feedback, and why it was used in the Quad II monoblock, Trimax broadcast radio studio monitors, and by Neville Thiele in luxury EMI (HMV brand) audio products, is that it needs a cheaper transformer than does screen feedback. It doesn't perform better (because it makes more demands on the driver), it's cheaper to make.
Cheap as chips if you use a standard output transformer and wire the cathodes to the speaker winding as I said - and the performance is then only slightly worse than with genuine ultralinear screen tappings.
To get the best possible performance with ultralinear requires a carefull and detailed transformer design & construction, as complex interactions of leakage inductances and inter-winding capacitance in the tapped primary can make the output stage tend towards instability. Rarely does it actually oscillate (though it does happen), but the phase shift causes the distortion to be higher than basic theory would predict. To prevent this, the best UL transformers are have complex winding segregation & backwinding schemes that are expensive.
The problem of interwinding capacitance and leakage inductance causing phase shift and thus increased distortion is much reduced with cathode feedback, because the cathode or output winding is low impedance. So the transformer can be cheaper.
To make a transformer that gives both options as with the Softone seems rather silly.
Last edited:
- Status
- Not open for further replies.