|
Components and Circuits For discussions about component types, alternatives and availability, circuit configurations and modifications etc. Discussions here should be of a general nature and not about specific sets. |
|
Thread Tools |
26th Aug 2021, 1:05 pm | #21 |
Dekatron
Join Date: Sep 2005
Location: Seaford, East Sussex, UK.
Posts: 5,997
|
Re: Single-Ended Partial Cathode Loading Output Stage?
Surely the Quad arrangement with a separate winding should make little difference if the coupling in the transformer is good? The Quad arrangement is required for push-pull output stages though.
|
26th Aug 2021, 1:37 pm | #22 |
Banned
Join Date: Mar 2021
Location: Morpeth, Northumberland, UK.
Posts: 936
|
Re: Single-Ended Partial Cathode Loading Output Stage?
The Mozart has long been a favourite of mine in terms of valve amp design. So clever, and works so well! THD and damping factor better than most rival push pull amps due to the very high feedback.
Surprises me that it hasn't been reborn for modern use but I guess the rather unusual transformer design has much to do with that. |
26th Aug 2021, 1:44 pm | #23 | |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,801
|
Re: Single-Ended Partial Cathode Loading Output Stage?
Quote:
In reality, the feedback loop gain doesn't drop below unity until far above the audible frequency range. So stability requires good behaviour right into the zone where transformer strays are starting to get nasty. There is more freedom to avoid leakage inductance into a single feedback winding of light wire than there is into a multi-tapped heavy wire speaker output winding. David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
|
26th Aug 2021, 8:12 pm | #24 |
Heptode
Join Date: Dec 2020
Location: Swaffham, Norfolk, UK.
Posts: 582
|
Re: Single-Ended Partial Cathode Loading Output Stage?
I would assume that if you had a transformer suitable for your design, it wouldn't need much space in the winding window of the core if the wire could be a fairly fine guage, and few turns. It may be an easy modification to do if so.
|
26th Aug 2021, 8:51 pm | #25 |
Dekatron
Join Date: Jan 2004
Location: Newcastle upon Tyne, Tyne & Wear, UK.
Posts: 8,172
|
Re: Single-Ended Partial Cathode Loading Output Stage?
The problem then arises about both flux distribution within the windings and stray capacitance. Placing a few turns of fine wire for optimum results will never be easy and sufficient voltage (or current) needs to be generated in this winding for the feedback to be effective
Ed |
27th Aug 2021, 2:33 am | #26 |
Nonode
Join Date: Jan 2009
Location: Papamoa Beach, Bay of Plenty, New Zealand
Posts: 2,943
|
Re: Single-Ended Partial Cathode Loading Output Stage?
Across a broad range of valve-era products from the setmaker to the hi-fi level, it seems that taking the main feedback loop from the OPT secondary winding, rather than from a separate (tertiary) winding, was the modal choice. Possibly the transformer design difficulties along that vector were perceived to be more easily managed than those arising from the separate winding vector. Or perhaps the transformer manufacturing industry had effectively made that decision. I recall that one argument mustered in favour of taking the feedback from the secondary was that it was the actual output to the speaker that was being sampled, not a once-removed facsimile. But I have not seen any quantification of the in-practice difference.
Given that, then the use of the OPT secondary as the source of cathode feedback was a oui-sequitur, so as to speak. As already said, in the Quad case, the need for a push-pull feedback source for the cathodes pretty much dictated that a separate winding be used. Maybe a biphase secondary could have been used, with the speaker fed from one side and the cathode feedback from both, but that would have been lopsided, with different currents in each leg. Quad also used global feedback from the secondary, so the cathode feedback was another loop, internal to the global loop. Cheers, |
27th Aug 2021, 3:04 am | #27 | ||
Nonode
Join Date: Jan 2009
Location: Papamoa Beach, Bay of Plenty, New Zealand
Posts: 2,943
|
Re: Single-Ended Partial Cathode Loading Output Stage?
Quote:
Quote:
Cheers, |
||
16th Jan 2022, 2:22 am | #28 |
Triode
Join Date: Oct 2009
Location: Masterton New Zealand
Posts: 17
|
Re: Single-Ended Partial Cathode Loading Output Stage?
Great to read that single ended cathode feedback is not a recent invention.
What I have not seen yet is single ended with cathode feedback, ultra linear and global negative feedback. Hopefully we will find out soon - I have ordered a pair of ISO (Tango) FC-20S output transformers which have 40% UL tap and a series of secondary taps in order to make it a "multi impedance" output transformer. Using the 3.5K setup leaves an extra tap on the secondary beyond the 16 ohm connection. The 16 Ohm connection will be going to ground and the extra tap will be used for global negative feedback. And the ground connection of the secondary will be going to the cathode of an EL506 (a premium 7868) running at 17.3 W Pa. Efficiency will be around 35% giving around 5.5 - 6 watts output at less than 0.5% distortion at onset grid current. The 18H inductance gets slightly increased by adding the speaker winding and hopefully using 60mA through the output transformer increases the inductance somewhat more. Fortunately I do not need ground shaking bass since the model suggests quite a bit of rolloff below 40Hz. Fingers crossed the LtSpice model holds up near these numbers. AM Last edited by AmadeusMozart; 16th Jan 2022 at 2:52 am. |
16th Jan 2022, 2:55 am | #29 |
Triode
Join Date: Oct 2009
Location: Masterton New Zealand
Posts: 17
|
Re: Single-Ended Partial Cathode Loading Output Stage?
And the proposed design - no design survives unscatched the build: Building undoubtly will reveal that modifcations might be needed for stabilisation. E.g. As per Langford-Smith a capacitor in series with a resistor between plate and screen to tame the output transformer self-resonance.
Last edited by AmadeusMozart; 16th Jan 2022 at 3:07 am. |
26th Apr 2022, 2:45 am | #30 |
Triode
Join Date: Oct 2009
Location: Masterton New Zealand
Posts: 17
|
Re: Single-Ended Partial Cathode Loading Output Stage?
The amplifier is now finished.
As mentioned earlier: No design escapes unscatched. I reverted back to pentode mode with CFB. The distortion is about double of UL+CFB at maximum output. But at 1 watt (and lower) output the Pentode+CFB had about half the distortion of UL+CFB. GNFB plus CFB proved parasitics at approx 30kHz at full output in one channel and although possible to get rid of it I was not happy with the resulting wave forms, square and triangle. A couple of Spectrol 162 multiturn potentiometer to adjust the output tubes to close to 17.3W Pa (close to design center rating, rather than the 19 W maximum rating) and a couple of linear Vishay P16 potentiometers to adjust for differences in amplification and balance the channels. A 45 second thermal delay relay to switch NTC in the primary out of the way (slow start filaments) and a damper diode for slow ramp up of B+ help to extend life of hard to find NOS 7868 / EL506. Square wave testing showed that square wave was still nice and clearly defined at 50 Khz, no leading spikes or ringing. It changes to sinusodial around 85 kHz - 90 kHz. Power levels are around 1.5 dB down at 90 kHz and very flat from around 45 Hz to 85 Khz. The chassis was split in two: a "permanent part" for the PSU with feedthroughs to the amplifier section which has a removeable side and top in case a total redesign was needed. Layout is a bit of a cross between Quad II and McIntosh and I can either have the connections on one side and the on/off on the other for on a narrow shelf or turn it 90 degrees with the switch to the front for next to a computer. At 6.5 watt (maximum) output I measured with my HP 339A 1.9% THD, by 5 watt it had gone down below 1%, by 1 Watt is 0.09% one channel and 0.07% in the other channel. Noisefloor was -82dB one channel, -83dB other channel versus 5 watts. With open input I measured on the 14 Ohm output connection less than 0.4mV residue. Mains transformer is mounted on silicon rubber grommets to avoid mechanical chassis hum. Listening, totally subjective, give an impression of noticeable more dynamic range: adjustments by sound engineers in volume in the recordings are very noticeable, sometimes even annoying as volume has to be adjusted. I'm now hearing background stuff I never noticed before - a train passing by, a phone ringing, a car honking, some talking. Switching back to a SE EL84 with GNFB some of these are not noticeable at all, others are barely noticeable. My wife also noticed this without pointing it out. I feel I've squeezed about the maximum out of a tube amplifier that I've been capable off, others may do better. Ofcourse I do not know how much the $$$ ISO (Tango) OPT's are contributing to the result. I did notice a change in results if using just "run of the mill" ECC81 versus carefully selected ECC81's that were properly matched between sections and with each other. (Telefunken that have a painted dot on them). I'm going to try some Brimar E81CC next, they are on their way from Langrex to New Zealand. edit: Modelling an output transformer in LtSpice is a next to impossible undertaking but taking the losses of the OPT into account then the actual results are not that far off from what I expected. Last edited by AmadeusMozart; 26th Apr 2022 at 3:01 am. |
27th Apr 2022, 8:30 am | #31 |
Heptode
Join Date: May 2017
Location: Konongo, Ghana
Posts: 510
|
Re: Single-Ended Partial Cathode Loading Output Stage?
That is going to be a beautiful build!
About not having seen single ended with cathode feedback, ultra linear and global negative feedback (your post #28): Attached are three schematics I had on my PC.
__________________
Robert |
27th Apr 2022, 3:01 pm | #32 | |
Triode
Join Date: Oct 2009
Location: Masterton New Zealand
Posts: 17
|
Re: Single-Ended Partial Cathode Loading Output Stage?
Quote:
Unfortunately CFB+UL+GNFB or CFB+GNFB was showing parasitic oscillation around 30kHz in one channel above 6 watts. Likely due to the OPT. PS HTS Leeuwarden, afterwards at Philips for a while before moving to the other side of the planet. Glad to have made our life down under. Cheers! |
|