Before all that, though, i needed to take care of a slightly more practical issue: the level meter LED string. Measurements with test-signals showed that the output (ie. the 1/4" jack line output) would clip, as expected, in the region approaching 30Vpp. The adjustable output buffer has a gain of anything between -∞ and 1.5x or so, which means that's definitely not the limiting factor. Regardless, the signal which supplies the level meter circuitry is tapped off from the input of said buffer.
Browsing through the AN6884 datasheet, and comparing the application example to the actual circuitry present on the daughterboards made things somewhat straightforward. The first thing the "investigation" showed, was that the numbers associated with each LED were somewhere between meaningless, and fraudulent; "inaccurate" would be a gross understatement.
The AN6884 has five current-sink pins, good for up to 18mA or so, tops.With the 8-LED level meter in this unit, they use only four from each chip (two per channel). For each chip, the third of the five LEDs is the "0dB" indicator, the other four becoming active at -10dB, -5dB, +3dB and +6dB, relative to the signal level required to light up the "0dB" LED.
The front panel claims that the levels indicated by each LED are -25, -15, -10, -5, 0, 3, 8 and 18. No idea whether those are supposed to be dBu levels, or dBV, or what - your guess is as good as mine. Now, the lower four LEDs are connected to the -10, 0, +3 and +6dB pins; right off the bat, assuming the front panel markings are in SOME sort of dB, you can see a discrepancy right there. The higher four LEDs are connected to the -10, -5, 0 and +6dB pins of the second chip. More make-belief number-magic, and i won't stand for that.
My pet peeve with the stock setup was that, according to measurements, the top LED (which is red, i assume to hint at clipping) lit up at a mere 2.2Vpp or so (0dBu / -2.2dBV), which is a whopping 22dB before anything approaching actual clipping (of the opamps). I think you'll agree, that's entirely unreasonable, and utterly useless. Off to some number-crunching...
Long story short, i couldn't be bothered to fiddle with the "lower" set of LEDs, and decided to focus on the higher-level indicating half. The associated AN6884 has a 150K / 8.2K voltage divider on its input (to pad down the signal before the internal +26dB gain stage). Some spreadsheet number-juggling showed that a 22K in parallel with the existing 8.2K would bring the sensitivity down to a more usable level. Since i wasn't willing to scrounge through my metal film axial resistors, i decided to series up two small 0603 10K resistors, and leave it at that.
True, there is now a bit of a larger "gap" (in terms of signal level) beetween the fourth and the fifth LED out of the string of eight lighting up, but that's in no way an issue. Since the output buffer is as flexible as it is (from totally killing the signal up to its 1.5x gain), there's plenty of fine-tuning range available. A somewhat happy accident / coincidence is that now, after the mod, the top LED lights up just as the triode starts to noticeably round off the lower half of the test-sinewave - a quite handy "tube drive active" indicator!
Okie-dokes, back to the initial "issue", the channel daisy-chaining. Since in this mode, i only wanted the second channel to be the "overdriven" one, i had no qualms about tapping off the signal from the output of the CH1 triode. Since, in order to be able to achieve the switching with an SPDT switch, i needed to inject the signal into an existing inverting amplifier, the only viable option was the second half of the input NE5532. I had initially hooked up the signal to the instrument input of CH2, but found the signal of CH1 would significantly drop when connected.
Since one needs a series resistor in order for an inverting opamp to act as a mixer, i found a previously-discarded 7.5K on my workbench and went with that. In all fairness, this still results in a slight drop in the CH1 signal level, but i'd consider it to be "next to nothing" (1.6Vpp on the brink of CH1 tube clipping, versus 1.4Vpp, which is a mere 1dB of difference). With the CH1 gain just below the red LED in the level meter lighting up, ie. just before the CH1 tube clipping, maxed-out gain on CH2 flattens out the top half of the sine wave enough to be noticeable. And that's just the "baseline"; i don't yet know whether this thing can go all the way to eleven, but i'd say it's in with a shout.
Hell, it might even be interesting to see how a guitar sounds through it (with the output going either into an interface and a cab-sim in a DAW, or into the effects return of an actual guitar amp). I'll just grab it with me the next time we have band rehearsals.
Oh, and in case you were wondering what's with the two switches above the compression knobs, well... That'll be the topic of "part 4".
No comments:
Post a Comment