new prototype
I was wondering after all this reading about the Williamson if i could make myself such transformer.......
I choose a different tube setting for the transformer: Kt88, Kt120 or 300B must be nicer.
I hoped to get 2Hz to 200Kz (-3dB) with full power at 10Hz. Power up to 40W.
5000Ω primair and 8 secondair and a small airgap so it can tolerate a small dc difference between tubes (up to 10mA).
So i made a prototype, did some measurement with a tone generator (load 2x 725Ω and sec 8Ω): not 2Hz but 5Hz
but this could be lower in a real setup with real tubes with a larger signal. High is even better then 200KHz.
It's only a proto type......
I was wondering after all this reading about the Williamson if i could make myself such transformer.......
I choose a different tube setting for the transformer: Kt88, Kt120 or 300B must be nicer.
I hoped to get 2Hz to 200Kz (-3dB) with full power at 10Hz. Power up to 40W.
5000Ω primair and 8 secondair and a small airgap so it can tolerate a small dc difference between tubes (up to 10mA).
So i made a prototype, did some measurement with a tone generator (load 2x 725Ω and sec 8Ω): not 2Hz but 5Hz
but this could be lower in a real setup with real tubes with a larger signal. High is even better then 200KHz.It's only a proto type......
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
As for building a Williamson output transformer, it's quite possible to wind your own close to original design specs, or redesign it for better specs using modern materials.
Using push-pull 300B tubes would be far closer to the original topology. Using all triode directly heated tubes as a lab reference amplifier. If you want more power, simply parallel more 300B devices, OR use 2 suitably selected larger triodes that work with the pre-existing driver stage. I understand the original design was based on triodes ( such as L63 and 2A3 ) not tetrodes ( KT66, 807, etc ) which came a bit later.
In the case of input triodes, unless you have a swag of L63 tubes laying around, using separate active devices would be easier to match than double triode tubes. Output power would be close to the original with a pair of 300B.
All something to think about.
Using push-pull 300B tubes would be far closer to the original topology. Using all triode directly heated tubes as a lab reference amplifier. If you want more power, simply parallel more 300B devices, OR use 2 suitably selected larger triodes that work with the pre-existing driver stage. I understand the original design was based on triodes ( such as L63 and 2A3 ) not tetrodes ( KT66, 807, etc ) which came a bit later.
In the case of input triodes, unless you have a swag of L63 tubes laying around, using separate active devices would be easier to match than double triode tubes. Output power would be close to the original with a pair of 300B.
All something to think about.
Last edited:
Specially for transformers: its almost impossible to wind as original because materials from the past are not available. Secondly you not want too because modern materials are much better.
Just a few general remarks from me as the experts have already spoken.
As to history, D.T.N. Williamson was an employee of Marconi-Osram in 1946, and the amplifier was designed mainly to promote the then new KT66 made by them, for audio applications (remarks by Williamson himself). I am not aware that it was ever used with 2A3s. The much later 6L6GC is similar to the KT66 within normal manufacturing spreads, but with a marginally higher Pa of 30W compared to KT66s 25W.
Regarding the much criticized l.f. stability, it should be taken into account that the power supply filter capacitors were only 8µF at the time. L.f. stability is easily achieved by higher value filter capacitors and using different coupling capacitors, and not a worthwhile reason to disqualify the design as such.
I find that OPT designers tend to accentuate low leakage reactance at the expence of internal capacitance; the latter quite often is the major cause of h.f. limitation, depending on the topology. In a typical example of my own, I use only three secondary sections. The response is still limited by mainly the interwinding capacitance. H.f. started rolling off at about 60kHz (somewhat countered by a resonance at about 85 kHz) , while the 4,5 mH leakage reactance itself only started acting at >120kHz.
As to history, D.T.N. Williamson was an employee of Marconi-Osram in 1946, and the amplifier was designed mainly to promote the then new KT66 made by them, for audio applications (remarks by Williamson himself). I am not aware that it was ever used with 2A3s. The much later 6L6GC is similar to the KT66 within normal manufacturing spreads, but with a marginally higher Pa of 30W compared to KT66s 25W.
Regarding the much criticized l.f. stability, it should be taken into account that the power supply filter capacitors were only 8µF at the time. L.f. stability is easily achieved by higher value filter capacitors and using different coupling capacitors, and not a worthwhile reason to disqualify the design as such.
I find that OPT designers tend to accentuate low leakage reactance at the expence of internal capacitance; the latter quite often is the major cause of h.f. limitation, depending on the topology. In a typical example of my own, I use only three secondary sections. The response is still limited by mainly the interwinding capacitance. H.f. started rolling off at about 60kHz (somewhat countered by a resonance at about 85 kHz) , while the 4,5 mH leakage reactance itself only started acting at >120kHz.
Many years ago I built a pair of Williamson clones using Hammond 1650P's. I adhered to the original circuit (but wired for Ultra-Linear operation) with Talbot Wright's modifications and slightly different GFB cap/resistor values.
It is stable and sounds good. I still have the pair and listen to them from time to time.
Nick
It is stable and sounds good. I still have the pair and listen to them from time to time.
Nick
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.