Tube Amps / Music Electronics
|For current discussions, please visit Music Electronics Forum.|
|7/26/2000 1:25 AM|
||Anyone have MXR Analog Delay Schematic?|
I'm repairing one of these gizmos. There are (3) Reticon SAD1024A inside, and curiously enough (2) CD4016 quad analog switches. Wondering what the CD4016 do - there are no "mode select" type switches on the box, so they are not for fx select/routing. Thinking they might form a switched capacitor tracking filter for the input and output, or less likely, a sample/hold to reduce spikes on the BBD outputs. Would be interesting (but time-consuming) to trace the schematic. If anyone has it scanned in, I would very much like to see it! Regards, Mike
|And now, a word from our sponsors:
|7/26/2000 8:19 AM|
I'll look tonight Mike - it rings a bell!
|7/26/2000 6:18 PM|
I've got the Analog Delay schematic if you still need it - let me know and I'll scan it in for you.
|7/27/2000 3:04 AM|
I would very much, like to have a copy!
Thanks , Ed R.
|7/29/2000 9:52 PM|
Many thanks to Robert for sending the MXR schematic! Regards, Mike
|7/31/2000 5:13 PM|
||Secrets of the MXR Analog Delay|
Thanks to Mike, as the vehicle for the generosity of a third party, I found out some tidbits about the MXR A-D which are worth passing along.
As Mike suggested, and as he was able to eventually point out to me once he had the schematic, the 4016 is indeed used for a tracking filter.
Here is the mojo...
All BBD-based devices struggle to keep clock noise out of the audio path. The standard strategy is to use fixed filters before and after the BBD chip. The problem with this is that optimizing a filter for clock rejection at the longest delays (i.e., the slowest and most audible clock speeds) leaves you with a seriously narrow bandwidth at those delays where you can afford to be more generous and let more high end through. Striking a compromise design, where you maximize bandwidth and let things slide a bit at the longest delays, is likely to result in irritating motorboat noise at long delays (which was a common complaint at the time). What MXR designers did was devise a delay-appropriate filtering system, such that the cutoff frequency was lower for longer delays and higher for shorter delays. Coupled with a compander chip, the MXR Analog Delay should have been as quiet and bandwidth-generous as was possible with BBD technology. I had always heard of the warmth of the MXR unit, and had chalked it up to fixed filters. Little did I know that it was also as capable of sounding bright as any other product at the time, when used with short delays.
The design is clever. The HF clock generator sends its signal to some 4013 flip-flops to be divided down to the usual complementary clock pulses that BBD's like, but it also sends the clock pulse to a 4069 invertor for buffering to drive two 4016 quad-analog switch chips. When you look at the schematic, you see two 4013's cascaded, and you wonder why the hell do they need to divide the clock pulse by 4 to drive the BBD's. Then you realize it was because they needed a high enough clock speed to turn the 4016's on and off fast enough to make them seem transparent. Adjusting the delay time means that any changes to the master clock has an impact on both the clocking of the BBD, *AND* the clocking of the switched filters (which are 4-pole lowpass before and after the 3 cascaded SAD1024's). Very spiffy.
A side note about switched filters to the uninitiated. If you switch a component in and out of circuit at a very high speed, its function becomes equivalent to the average component value over time. Cranking up the clock rate results in a different functional value than turning the clock rate down. (If I'm working backwards from the schematic properly, then turning the clock rate *down* results in a smaller effective component value.)
MXR used CMOS analog switches to produce dynamic filters in their Envelope Filter, and use them wisely here too. In this particular case, the delay path has a 4-pole lowpass filter before and after the delay chips. Forgive my possible misapplication of terminology, but these look like regular Butterworth lowpass filters to me. So, you'll see two op-amps, with two equal-value resistors in series with the non-inverting input, a cap from the noninverting input to ground, and another cap from the output and inverting input back to the junction between the resistors. Normally, to make it a variable filter, one would stick a dual-ganged pot in where the two resistors are (or in conjunction with 2 equal range limiting fixed resistors) and use that to tune the filter. However, tuning 8 resistors simultaneously (4 for the initial and 4 for the later lowpass filter section) is kind of tricky, pricey, and space demanding. So, MXR used two 4016's (yielding 8 analog switches) to essentially connect and disconnect eight 10k resistors from the filter circuit at very high speeds.
Although Reticon chips have gone the way of the buffalo, and long-delay Matsushita chips are soon to follow, there are still lessons to be learned for the MN3x05 crowd. In principle, it should be possible to clock the MN3101/3102 clock driver chip higher than it needs to be for the desired delay, and use the MXR strategy to drive tracking filters. The 3101 comlementary clock signal outputs could, presumably, be divided down by suitable application of flip-flops, to yield the desired delay time, and the tracking filters used to optimize bandwidth for the given delay time.
Many thanks to Mike Irwin for drawing this to my attention. We're both still curious as to whether the tracking filter idea was transplanted from the Delay to the Envelope Filter, or transplanted the other way round. In either event, it was an ingenius idea and very high tech, very clever, and very classy for its time. One wonders if tracking filters were used by any other FX designers/manufacturers.
|7/31/2000 8:52 PM|
It just occurred to me that if you had a tracking filter with a flanger, you might be able to approximate a phase shifter with it. Since flangers tend not to have as complex a filter as delay lines (because the clock frequency is generally out of the audio range), it may be possible to combine a tracking filter with manual controls.
Where phase shifters have a fixed number of notches, regardless of where they are in their sweep, flangers have more notches the lower they sweep (and the longer the delay). You can try and simulat a flanger by adding more stages and notches to a phase shifter, but that can get cumbersome and costly. Why not simply eliminate notches from a flanger?
A tracking filter imposed between the BBD chip and the straight/delay mixing stage would, in theory, filter out the delayed version of signal above the cutoff frequency, which would eliminate cancellation (and notches) above that point. In effect, you would have a flanger with fewer notches, and a relatively constant number of notches as it sweeps.
Using MXR's technique, if all you used was a post BBD 2-pole lowpass filter, then it would be a snap to use a dual-ganged pot in conjunction with the analog switches to vary the bandwidth of the delayed signal, thus having control over the number of notches.
|Page 1 of 2||Next>||Last Page>>|