This week, we go back to sound practices issue 1: It Will Stand by Herb Reichert. We have already read an article By Herb and since he contributed to the first 15 print issues [corrected —ps], we’ll see him a few times more. This article shows where Herb is coming from. It starts off with some childhood memories, then some observations on the value of music. Next, it recalls how He got bitten by the hi-fi bug and it turned out that DIY was the only way to get there. True triodes are discovered and feedback is eliminated. Then the question comes up: what has gone wrong with high-end audio? The “fork in the road” is traced back and the intrinsically linear way of doing things is rediscovered: low parts count, efficiency and headroom. Then, the final de-complication: single ended triodes. Herb finishes the article with observations about the value of hi-fi systems and the value of building them.
teaser quote: ‘I never really wanted a “Hi-Fi” system, just something rugged and honest that puts me in sync with the music. I wanted to feel the Stratocaster and Deluxe reverb. I wanted to respond to the weight and tone of a Steinway and Bosendorfer. I liked to hear the tubey sound of a Hammond organ.’
my take
‘Don’t take this article too literal (especially if you felt yourself somehow disagreeing with it).’
It may be Herb’s personal audio journey, but actually this article is also the manifesto for what what became the mainstream for ultra-fi and sound practices magazine: goosebumps type music reproduction is found in no-feedback, single ended triodes, on high-efficiency speakers. Funny enough, the Marantz–Quad system that got Herb hooked is the exact opposite of this…
Meanwhile, there are shades of things to come: paper cones and alnico magnets; low source impedance, high transconductance; passive phono EQ, transformer coupling, first order crossovers; tube/choke/capacitor power supplies, or a simple pi filter; a vintage Altec system.
One topic I really want to highlight today is headroom. There are two forms of headroom that Herb discusses.
The first one is overall system headroom. This is the fraction of maximum amplifier power are you using on peaks in the music. Having 6dB of headroom means that you are using at maximum a quarter of the Watts your amp can deliver. This can be 0.25W out of one Watt on your horns, or when using today’s audiophile speakers, 25W out of 100. The peaks refer to when you really got it turned up, because you are rocking out or just listening to—usually not so compressed—classical music. Something as simple as the leading transient of a string being plucked, or a flute or piano note, can mean a full-scale signal. Banging on some drums always works.
Having system headroom makes for effortless reproduction. Even the output tube/power supply/output transformer conglomerate is working in the best, linear part of their range. But quite a few of us—including me in one of my systems—have to live with seven-or-so Watts on 90dB speakers. That is probably not even zero headroom (just enough Watts), but worse: clip. That is where smooth clipping behaviour and agile recovery help. There was plenty said about that in the Flesh and Blood article.
The second type is the relative headroom of the driver stage and all stages before that, all the way back to the cartridge or the dac chip. In each system one has to live with the amount of system headroom, but it is totally avoidable—and hence unacceptable—that any other gain stage is close to clipping at full power output. That would be a pileup of strain, degrading the effortlessness of the system. So each gain stage needs it own additional headroom, relative to the output tube.
The stage where this is the hardest to achieve is the driver stage just before the output tube, because it needs to deliver the highest voltage swing of any gain stage. Having 6dB of relative headroom here means that when the output tube needs 100V peak on its grid for full power output, that the driver stage is designed to deliver exactly 200V peak (or 400V peak–peak, do not forget up and down swing). I say exactly because having more and more gain stage headroom becomes increasingly difficult: 12dB mean 800V p–p, 20dB means 2000V p–p in our example. As you design driver stages yourself, you will see that for increased headroom there is a real cost: in quality of tubes that can be used (max plate voltage limits); driver B+ (it may be even higher than the output tube’s B+); and linearity of the stage, over the voltage swing range that will actually be used. I design myself for 6dB headroom in the driver stage.
When looking at the gain stages before the driver we see that the required voltage swing for full power drops quite a bit and it should be possible to aim for 12dB headroom in those.
I thought I highlight the headroom issue here because I do not see it discussed much in online forums nor in magazine articles. Worst example was a recent, too-clever-by-half direct-coupled design that managed to have exactly zero headroom in the driver stage. This was both published in a DIY magazine and presented at the last ETF. I feel the editors of the magazine dropped the ball on this one, it should never have been published.
bonus tracks
An extra shot of complementary reading can be found online. Some more ‘early days in nyc’ anecdotes (the colleague who opened the SoHo store is Don Garber, still building amps under the Fi name, as the store was called). There is also quite a bit more about values and the value of hi-fi systems.
ps: I found one more fitting bonus track: Playing It Over and Over.
Now go and read the article, see you next week.
No comments:
Post a Comment