Hello everyone, I recently purchased an example of the Cook Ultimate Amplifier, a spare Langevin 316A output transformer, and a copy of the magazine article in question, as well as a follow-up article that answers some questions about the design. I will post a picture of the amplifier and scans or pictures of the schematic soon. I plan on powering up the assembled amplifier (which is supposedly working) and then producing a second example based on the schematic, the article, and the amp that I have.
The amplifier was designed by Emory Cook, a pioneer of stereo sound reproduction and a prominent figure in the early "high fidelity" scene. Of course, this does not necessarily mean the design is good, but interesting to note. Apparently Cook's company began selling a later iteration of this amplifier using 6550s, but it seems that very few examples survive.
I can post square wave shots and frequency response of the working example if anyone is interested. Unfortunately, I do not have an audio analyzer or suitable soundcard for THD yet.
I can post square wave shots and frequency response of the working example if anyone is interested. Unfortunately, I do not have an audio analyzer or suitable soundcard for THD yet.
The "ultimate" of this amp is that it is designed to deal well with reactive loads such as speakers instead of a pure resistive load.
Most ( all) amp performance measurment are done with pure resistive load and is somewhat misleading when a real speaker is used. This may be the reason different amps "sound".
Most ( all) amp performance measurment are done with pure resistive load and is somewhat misleading when a real speaker is used. This may be the reason different amps "sound".
Very interesting. I used to live in Stamford Ct. and hiked over at Pound Ridge Reservation often. (mentioned in the obituary article). Would be interesting to know where his barn was at in Pound Ridge.
Some issues I see with the "Ultimate" design are too many stages (5) and too many coupling caps in sequence (4). It needs to reduce these, possibly using a frame grid tube to get rid of all those not so linear 12AU7s. Some CCS loads or tails could help linearity too. Some adjustments for bias matching and AC balance would be helpful.
All of the various Fdbks come from the secondary side there, where OT phase issues come into play. Shorter local loops from the primary side (or UL taps) for the tube linearization portion at least, would allow the use of a cheaper OT, without so many windings on it. Common UL taps can provide some Fdbk points, with close coupling relation to the secondary, for accuracy. (the global Fdbk loop can provide the final touches on linearity and output impedance still.)
Then we have some new techniques, which weren't available then, like Crazy drive or Unset, which if properly implemented should provide impressive ability to tailor the results with either linear gm or high quality "triode" emulated outputs. And, using bigger tubes that don't need to be paralleled.
The impedance adjustability can be done using just a low value current sensing resistor at the secondary with global Fdbk, as done by ElectroVoice.
Some issues I see with the "Ultimate" design are too many stages (5) and too many coupling caps in sequence (4). It needs to reduce these, possibly using a frame grid tube to get rid of all those not so linear 12AU7s. Some CCS loads or tails could help linearity too. Some adjustments for bias matching and AC balance would be helpful.
All of the various Fdbks come from the secondary side there, where OT phase issues come into play. Shorter local loops from the primary side (or UL taps) for the tube linearization portion at least, would allow the use of a cheaper OT, without so many windings on it. Common UL taps can provide some Fdbk points, with close coupling relation to the secondary, for accuracy. (the global Fdbk loop can provide the final touches on linearity and output impedance still.)
Then we have some new techniques, which weren't available then, like Crazy drive or Unset, which if properly implemented should provide impressive ability to tailor the results with either linear gm or high quality "triode" emulated outputs. And, using bigger tubes that don't need to be paralleled.
The impedance adjustability can be done using just a low value current sensing resistor at the secondary with global Fdbk, as done by ElectroVoice.
Last edited:
Voltage sets voice coil velocity. Current sets voice coil force, or acceleration. I recall mention that velocity sets sound pressure. Which seems to make sense. Air Pressure on a truck windshield would depend on speed, not acceleration.
Current drive would be very susceptible to speaker suspension and cabinet back pressure forces.
Voltage drive will simply pull whatever current is needed to get the velocity or sound pressure requested by the voltage. However, the DC/permanent magnetic field must be uniform across the VC extension range, or distortion will result.
Some VC drivers use an extended voice coil so that interaction with the DC field is constant, no matter where the position of the VC. That does raise the VC resistance however, which will subtract from the drive voltage depending on drive current. A better solution is to keep the VC short and extend the magnet gap across the full extension (expensive).
The info I've seen from ElectroVoice was that the purpose of speaker impedance matching was to damp out speaker resonance. Critical damping occurs when the Z loading matches the wL and 1/wC factors of the speaker at it's resonance. However, VC DC resistance is in series, and that changes with VC temp. Ferrofluid in the gap can help cool the VC, but probably causes problems at HF.
Current drive would be very susceptible to speaker suspension and cabinet back pressure forces.
Voltage drive will simply pull whatever current is needed to get the velocity or sound pressure requested by the voltage. However, the DC/permanent magnetic field must be uniform across the VC extension range, or distortion will result.
Some VC drivers use an extended voice coil so that interaction with the DC field is constant, no matter where the position of the VC. That does raise the VC resistance however, which will subtract from the drive voltage depending on drive current. A better solution is to keep the VC short and extend the magnet gap across the full extension (expensive).
The info I've seen from ElectroVoice was that the purpose of speaker impedance matching was to damp out speaker resonance. Critical damping occurs when the Z loading matches the wL and 1/wC factors of the speaker at it's resonance. However, VC DC resistance is in series, and that changes with VC temp. Ferrofluid in the gap can help cool the VC, but probably causes problems at HF.
Last edited:
One could differentiate the voltage signal coming into an amplifier to get a current drive signal. Changes in input signal voltage require accelerations. The speaker suspension would need some working on to make it constant or zero force.
The cabinet would need to be extra big to remove back pressure effects. Or a voltage driven speaker immediately behind the front current drive speaker could be used to remove cabinet back pressure effects.
The cabinet would need to be extra big to remove back pressure effects. Or a voltage driven speaker immediately behind the front current drive speaker could be used to remove cabinet back pressure effects.
Last edited:
I thing that is very speaker dependent.
Most ( all?) speakers are optimized for voltage drive , which is kind of strange since it's the current through the coil that makes the coil move.
Classical tube amps tends to be in the middle and deliver not constant voltage nor constant current but more to constant power. Which suit some speakers very well.
See Current-Drive - The Natural Way of Loudspeaker Operation for a theoretical discussion of why current drive makes the best of a speaker.
Here are some pictures of the amplifier in case anyone is curious. Decided to share using Gdrive rather than uploading to get around the filesize limits.
There are some pretty interesting construction techniques here. Never seen a big variable wirewound resistor like this, although I guess they are still made. I'm not sure how closely this example matches the original schematic, as I haven't gotten around to tracing things out yet.
There are some pretty interesting construction techniques here. Never seen a big variable wirewound resistor like this, although I guess they are still made. I'm not sure how closely this example matches the original schematic, as I haven't gotten around to tracing things out yet.
I will be able to sleep better tonight . . .
knowing that there is an Ultimate Amplifier out there somewhere in HyperSpace.
36 Posts, and we are 1/4th of the way there.
Now, if I knew that there was an Ultimate Speaker, Ultimate Signal Source, and an Ultimate Recording . . .
Just saying.
knowing that there is an Ultimate Amplifier out there somewhere in HyperSpace.
36 Posts, and we are 1/4th of the way there.
Now, if I knew that there was an Ultimate Speaker, Ultimate Signal Source, and an Ultimate Recording . . .
Just saying.
Last edited:
"Akashic"
Wikipedia: "In theosophy and anthroposophy, the Akashic records are a compendium of all universal events, thoughts, words, emotions, and intent ever to have occurred"
Obviously that would have to include all contradictory and contrarian voices and sounds at once. A true HiFi would have to superimpose all speaker motions at once, exploding the speakers and burning out the tubes. But if it does pass this destructive test, you will be assured that it -was- truly HiFi.
Need to send one of these Akashic records over to George (Tubelab) for his extreme Gestalt testing .
Wikipedia: "In theosophy and anthroposophy, the Akashic records are a compendium of all universal events, thoughts, words, emotions, and intent ever to have occurred"
Obviously that would have to include all contradictory and contrarian voices and sounds at once. A true HiFi would have to superimpose all speaker motions at once, exploding the speakers and burning out the tubes. But if it does pass this destructive test, you will be assured that it -was- truly HiFi.
Need to send one of these Akashic records over to George (Tubelab) for his extreme Gestalt testing .
Last edited:
The password is not needed to Read or Print the document. Only to Change, Comment, Extract, and some other features.
Attachments
