Clocked Delay Issues

I installed my ER-301 last night and so far I am loving it. However, I’m having two issues with the clocked delay that I would like to get sorted out. In all cases, I’m feeding the clock input with pulses from my Cirklon.

The first issue is that, occasionally during playback, the delay time will suddenly fluctuate, similar to what occurs when the delay time is initially calculated after receiving the first couple of clock pulses. As far as I can tell, the Cirklon is putting out constant and very steady pulses, so I’m inclined to think the issue is with the ER-301.

A second issue is that if I clock the delay with relatively rapid pulses (say, 16th notes), the delay time fluctuates very noticeably. Is this expected behavior? The issue seems to go away with less rapid pulses, but perhaps it’s just less severe. I have a recorded example of what I’m talking about, but since this is my first post I’m not able to upload it.

Hello miminashi.

I just sat down with the clocked delay unit and a simple saw tooth wave (provided by an external VCO) and couldn’t get any fluctuations, so I need more information.

Here is was my test setup:

  • External VCO sawtooth into A1 --> left and right input of stereo chain OUT1+2.
  • Stereo chain contains a CLOCKED DELAY unit with left ratio as 1/3 and the right ratio as 2/3, so that I get clear triplets: LR, L, R, LR, L, R, and so on.
  • Patched the square pulse from the same external VCO into G1 and patched that to the clock input of the delay unit.

I was able to vary the external VCO’s frequency from 0.333Hz to 2kHz without any noticeable hiccups or fluctuations to the triplet sequence. Of course after 20 or 30Hz it just sounds like an oscillator but you would still be able to hear any fluctuations if there were some.

So some thoughts that came up while I was trying to reproduce:

  • Currently all delay units have a maximum delay time of 1s. So if you divide the clock down so that resulting delay is more than 1s you will get inconsistent results (well, you will get a 1s delay).
  • How much fluctuation are we talking about here? Perhaps my test by ear was not accurate enough?
  • What are the low and high voltages of the Cirklon’s gate outputs? Is the delay’s clock input threshold set correctly for the Cirklon’s levels?

The CVIO on the Cirklon produces 0-5V gates, which I have just verified with a multimeter. I’m using a 0.1 threshold for the clock. The strobe graphic is in time with the expected clock pulses and doesn’t seem to be missing any strobes.

I just ran a little experiment with the following settings: 1PPQN clock from the Cirklon, 16 numerator and 8 divisor in the clocked delay; and a 4PPQN clock from the Cirklon, and a 4 numerator and 8 divisor in the clocked delay. These settings should produce the same results, if I’m understanding the interface correctly. The 1PPQN setup produces nice, clean repeats, but the 4PPQN setting has a fairly dramatic “wobble” of at least several semitones.

Here is a link to a recording which illustrates the problem: https://clyp.it/1m44egmf Lemme know if you would like me to email it to you instead, as apparently the link will self-destruct in 24 hours.

P.S., the Cirklon gates are high/low, no intermediate voltages.

Ah, now I understand. Your recording helps a lot.

Hmm, I think I know what it could be. At the moment, trigger timing is only sample-accurate when really it needs sub-sample accuracy. This sample-accurate timing causes jitter in the final clock period estimation which is then amplified by the delay tracking algorithm in the delay line. The SINE OSC unit’s sync input could also benefit from this.

Implementing a variable delay line whose pitch doesn’t change when the delay changes is also in the works.

1 Like

That makes sense. I wonder if it might explain the other issue, as well.

On a related note (one that perhaps needs its own feature request topic), it might be handy if there were a configurable limit on the maximum time between two clock pulses, with the goal of avoiding delay time hiccups when, e.g., stopping and restarting the sequencer.

Have you tried processing your incoming clock with a TAP TEMPO unit?

I did, but I can’t remember the results. I’ll try it again and report back.

Okay, I tried the tap tempo. The start/stop behavior is much nicer, but the delay modulation persists.

Edit: Actually, I may have spoken too soon, regarding the start/stop thing. I was feeding the clock pulse to both tap and sync in the tap tempo, which was masking the effect to some extent…

Still experiencing the first issue from my initial post (spurious delay time fluctuations) with the latest firmware.

I noticed that sending clock pulse to clocked delay seemed to introduce crackly noise into the signal… I noticed this a couple of times, but did not investigate too thoroughly. I wonder if anyone had similar observations though?

1 Like

I haven’t noticed anything as fine-grained as a “crackle”. Here is a sample of what I’m talking about.

The square wave bleep is on the first beat, the clock is on every quarter note, and the division setting in the clocked delay is 3:2.

May I have more information about the clock source, wave shape and frequency please?

1 Like

Clock source is a gate output from the Cirklon CVIO. The waveshape seems to be a 50% duty cycle pulse, 0-5V. The frequency is 2,6Hz (quarter notes at 156 BPM). I’ll attach a recording of the clock source made using the ER-301.

clock.zip (1.5 MB)

Discourse didn’t like my clipped wave so I’m attaching a zip of the minute-long recording.

2 Likes

I tried clocking the delay with a saw unit, and found that randomly modulating the frequency of the saw by a relatively small amount (~1%) induces the delay time fluctuations. There is some jitter between clocks coming from the Cirklon (less than 1ms), but the magnitude of the fluctuations does not seem proportional to the amount of jitter.

As a side note, at one point in my testing ,G1 (which I am using for the clock input) stopped receiving input and I had to power cycle the 301 to get it back.

1 Like