Steve Deckert
|
I'm sure many audiophiles think everything is fun and games in a business like Decware, but it has it's moments believe me. Somehow those are always moments tied directly to troubleshooting bizarre problems which after 30 years of millions of self-inflicted bizarre problems with circuits, I have come to believe in the spiritual realm because some problems have been so complex that they opened a portal into other non-friendly dimensions.
I could write an entire book on just this subject, and perhaps I will some day if I can still remember any of it.
For example, and yes it only happens when you are too busy to have it happen, a preamp comes across my bench for QC and the perhaps most prominent feature of the preamp, the tube VU meters don't look the same. One is slightly fuzzy and the other is crisp and sharp like a hi-res image.
These kind of things short me out, so immediately I dive in to investigate why the hell is one tube fuzzier than the other. No matter what I tested, I could find nothing wrong with the fuzzy side. Hours of this torcher passed with no results. Both sides measure the same but look different. Replaced the offending side part by part. No results. An entire damn day has now passed.
Day two, this MF is going to give up it's secret today or die. Mornings are the best time for this attitude. Another day of greatly enhanced suffering ensued as the nonsense continued. One tube displays flat at any frequency with sharp focus, the other rolled off after about 3K which was actually desirable because it visually goes with the music better. Still we can NOT have two that do not match.
I finally figured out that the fuzzy one that had the rolled off response was actually the spec, and my carefully done sample was also this way. I then tested all 50 other components for this module that I had in inventory and they also were all exactly the same as my sample, so what I had is one rouge board that worked twice as good as the standard.
This is what suckered me in.
So why was it different? Turns out the trim pots, of which there are two, were all labeled wrong. For example, 500K linear tape pots were actually 260K. The one that actually worked differently/best was reading 380K. Not one was anywhere near 500K. Which brings me to my main point about troubleshooting. When you have been building the same handful of amps for 20 years, the only thing that usually ever goes wrong is BS with parts and sadly those parts can always be traced back to their source. In this case who knows how many 250K trim pots were miss-labeled as 500K and went into the market, were sold and found their way into products.
As I said, the error actually produced a desirable rolled off response that make the movement of the VU display look like it goes to the music better, so in the QC process it would never be found unless you watch the meter at full signal, no signal, and then sweep them from 20Hz to 20kHz and see if they match the whole time, which is what we do.
If I had my money back for all the defective parts we've discovered and the time lost correcting the problem over twenty years, I could pay cash for a new home.
|