From: Al Davis (
[email protected])
Date: 2002-08-04 22:09:39 PST
I don't normally read this group, but this thread caught my eye, and I
did some snooping in the archives. I figured I would leak a few more
API secrets.
I am responsible for a few API products, pre-Datatronix, including the
original 560.
In the comments on eq's there was mention of "proportional-Q" and
"fixed-Q" designs, in reference to the variants of the 550. It was also stated (more than once) that the 550A-1 is essentially 3 bands of the 560. This is not true. The 550A-1 is fixed-Q. The 560 is proportional-Q. The 560's Q actually varies more than the original 550 does. The 550A-1 design is really more like the 554, but with significantly lower Q, and single op-amp filters instead of 3 op-amp filters. Also, the 550A-1 has first order shelving, contrast to the 554's second order shelving.
Most graphics have some ripple in the response when you boost or cut.
This is because the Q is too high. Although it was not specified, I considered it to be a requirement that at most reasonable settings there would be no ripple. The idea is that if you boost two adjacent bands you get a single hump, not two. You can boost a band and use the adjacent ones to broaden it. If you really want peaking, use a parametric where you can tune the frequency.
Another factor is that with any tuned circuit, if the Q is sufficiently high, there is ringing. The higher the Q, the longer it takes for the ringing to damp out. If the Q is low enough, it is really more like a pair of RC circuits, and there is no ringing. The 560 is designed so that unless the setting is full boost or cut, the Q is low enough to not get ringing.
The 560A was not intended to be an API product. It ended up being one
because of the way Datatronix got into the scene. It began during the
period between when the original API went bust and the acquisition by
Datatronix. That one was my design, too. Ideally, it would have been
the same, but there were some key parts that were not available. The
change from the 2520 is well known. The reason for that change is
simply that the company that made them went bust.
There is another change that I have never seen mentioned, that is more
important than it might seem at first. Both the 560 and 560A use
special S-taper controls. The resistance curve has an "S" shape. It
is symmetric but is steeper in the middle, and relatively slow at the
ends. If it used linear pots, the control would be all scrunched toward the ends. The use of the S taper also allowed a higher resistance, resulting in lower noise and the "proportional-Q". The 560 used Alps pots. The 560A used Noble pots. The Alps pots were not available in the small quantities we needed. The taper is a little different, but the same at the calibration points. The Alps taper really looks like an "S". It is all curved. The Noble taper looks more like piecewise linear with rounded corners. This was not a conscious decision, and was considered to be a negative at the time, but we really didn't have a choice. It points out an issue that is often overlooked in products like EQ's ... A production change beyond the control of anyone who should know can significantly affect the sound.
Now, back again to the fact that it was not intended to be an API product ... Since it was unknown what would happen to API, and we didn't own the design, we could not just use the same design. This meant doing it over without access to the original drawings. Even though the goals were the same, and they use essentially the same topology, this makes for some subtle differences. One of these is that the Q in the 560A is a little lower than the 560. Also, it varies a little more, or in other words it is a little more proportional.
In the actual business deal, I was not "hired to do it" but rather it was a joint venture between me and Studio Consultants, where we split the profits and risk.
Now, back to the 550A-1 ... I wanted to call it something else, maybe 551, because it obvious to me that nothing could really replace the 550A. It should be marketed as a lower cost alternative. The original design was incredibly expensive to produce, and a significant money loser. Something had to be done. The final reason for not calling it something else was that they had a huge stock of front panels that said 550A.
API did not have good listening facilities, and the people we sent the
prototype to really liked it, so the problems slipped by. I became aware of it during the dead period between the bust and Datatronix, and made a modification to the design that would make it proportional-Q without raising the cost. Datatronix wasn't interested, and they continued to make the fixed-Q design. Most of them were made after they were aware of the problem and the fix they didn't use.
The 550A-1 design used filters in the feedforward or feedback path, with a summing node, giving a fixed-Q design.
The 560 (and 560A) uses a completely different gyrator based design. The "filter" simulates a series R-L-C circuit. When the pot is near the middle, its resistance is in series with the gyrator's R-L-C, lowering the Q. It was actually designed as an LC equalizer, then an active circuit was substituted for the L.
;D
