Welcome to the next level. Now that we have seen how tube amplifiers are designed in a conventional way, we are ready to make a leap forward. Our article today is Ankoru Design Notes by Andy Grove, from sound practices issue 12. It starts off a bit ropy, so bear with me for a page. It really gets better from the second page, where Andy outlines some goals, beliefs and the importance of taking design choices that complement each other. The third page is where the gold is: it describes the solid reasons for using an interstage transformer, leading into how challenging it is to design and make a good interstage. Next, tube selection and circuit are discussed, followed by (the compromises in) the power supply. The article wraps up with an impression of the resulting amp.
teaser quote: ‘[…] to recreate a musical event, an amplifier must work on both a macroscopic scale as part of a communication system between the performance and listener and on a microscopic level as a collection of valves and parts which must be tamed and optimised for the task at hand.’
my take
Amplifiers are made by hooking up resistors to the plates of tubes to make them amplify, then coupled with caps to the next tube until you get to the output. Then you have to use a transformer to match the tube to the speaker. When you get really tricky, you try some direct coupling. This was the received wisdom in the western tube audio world from the 40s through the 90s. It was how I knew things were—around 1997—and what we have seen up to now at the reading club. Then things changed.
Around that time more than just a few people started experimenting with interstage transformers. Japan was again where this originated, the west followed decades later. Suddenly it was on the radar, in 1997. In no way do I want to portray this week’s article as ‘the one that started it all’ in the west. But it does do a good job of explaining why we should bother to have second thoughts after 50 years of ‘doing it right.’
It is all on that third page. First, avoiding waveform distortion through loading the driving tube with an high AC impedance. If you check again figure 3 in last week’s Crowhurst article and follow the load line from B to A, you see that the distance between the points where it crosses a grid line becomes shorter and shorter. This is the compression, the waveform distortion. This is also present between points E and F in figure 4, even when staying away from the bendy bits at the bottom right of the plate curves. If however, one has a horizontal load line similar to the one that goes through point B in figure 3, you see that one can easily swing past 250 volts on the plate with little–to–no compression. Same tube, more voltage swing and an order of magnitude less distortion.
Second, loading the tube with a low DC resistance leaves in the Ankoru just about all of the 360V of the driver B+ available for the 300B (plate will be a few volts under 360, its filament biasses at 60V). All of that can be used for swinging down—up is no problem for a interstage loaded tube: the choke action of the primary winding makes that the plate can go up as far as it wants to go. Apropos choke action: plate chokes (we will meet them another time) and interstages have these two traits in common high AC impedance, low DC resistance.
But a transformer is two chokes on the same core, so the story continues. Grid current is discussed next by Andy, and it is in my opinion the key to success. The grid is a slight rectifier of the signal applied to it (get close to zero volts and forget about slight) and this is why DC is part of the story, why DC resistance of the grid circuit matters (converts current to volts) and why capacitors can store it (shifting bias and smearing waveforms with a time-delay). With low DCR (up to a 1000 times less than a grid resistor) in the circuit that connects the kathode of a tube to its grid, there is very low current to volts conversion. With no coupling cap, even this tiny effect does not get stored. Note that the AC part of the grid current always has to be supplied by the driving tube.
With an interstage transformer (or a grid choke) the low DCR comes with high AC impedance that does not load the driving tube: all drive goes to the grid itself. And there is no coupling cap. But wait, note that it is the whole kathode–to–grid path. In the Ankoru the bias supply (bottom-centre of the schematic) is included in that. If that had been designed with 10k resistors or pots then the low DCR in the interstage would be all for naught. The 1k or so DCR of this bias supply is added to the total DCR and the 22µF cap does store any bias variation caused by grid current, however small it is. So watch those kathode–to–grid paths.
pop quiz
And this time I do not know the answer: what does Andy mean with ‘the original and best sounding configuration’? I always thought just a choke or interstage loading the 7044, but checking now (with last week’s method) we see that out of a tube with about 2k plate resistance, loading it with 20k///270k (= 18k6) you get 90% of the gain. Close enough to the 100% of choke/interstage loading. So I think the hint is in the ‘simple, single stage RC coupled affair’ [emphasis mine] fragment. I still chuckle about ‘dealer proof.’
You see in the schematic a transformer before and after almost every tube. ‘Ideally’ there would have been one after the 7044 as well. That is because I am convinced that not only big output triodes, but every tube along the chain from cartridge/dac chip to the speaker benefits from choke loading and low DCR on its grid. Yes, the tubes and voltage swings are much smaller towards the front and so are the non-linearities, but these get amplified proportionately more along the chain. And note that high µ and/or high gm tubes (see warning last week) have relatively high grid current issues.
Finally: in the power supply, note the great similarities between the driver and the bias supply. That is really classy, taking bias supplies seriously.
Now go and read the article, see you next week.
ps: more about the 845 in issue 1: Meet the Tube: the mighty 845.
No comments:
Post a Comment