Noisy reissue Deluxe Electric Mistress

[DiscoVlad]

This is the version with the RD5106A bbd chip (EC1000 Rev. B board)

In filter matrix and flange mode there is a lot of what i presume to be clock noise on the output (measures with oscilloscope at 35khz), which manifests itself as a flanged hissing when not playing.

Apart from the noise, it still works properly... i.e. makes flanger sounds.

Is there an easy way to fix this, (and the volume drop while i'm at it...)?
There's a lot of free space in the box, so i'm thinking a filter/amp board on the output should do the trick.

[DiscoVlad]

Or failing that a schematic for this version?

[blueduck577]

here's the schem:



I've been wondering about the noise also, maybe somebody could shed some light on this

[DiscoVlad]

Neat! Thanks for that.

For reference, the RD5106a Datasheet is here:
http://thmq.mysteria.cz/em1022/RD5107.pdf

The evaluation circuit in the datasheet mentions a low pass filter on the output (RD5106A Pin 4). so I'd think that would be a good start either between the BBD and Q1 transistor or possibly the emitter of Q1...

(rambling begins here)
Guessing that something in the range of 10-15khz cuttoff should be sufficient (which my quick back of envelope calculation tells me is a 1nF cap and 10.6k resistor or other similar ratio...), there's likely to be a trade off between cutting noise out, and altering the sound of the effect too much.

Trim2/R13 sets the "volume" of the output signal from the BBD which gets mixed with the dry signal at R17/R18,( and split to the feedback loop at R15/R17), so it seems the noise should be removed before that point.

The volume drop might be possibly fixed by adjusting R13, though in my experience it doesn't help much, it could be fixed by adjusting some resistor values around the 4558 (U2A/B), though I'm inclined to think a simple clean booster between the board output, and foot switch would give you volume without the potential for changing the sound too much.

(end ramble...)

Edit...
Datasheet mentions a LPF at 1/4 Fc (which is 35khz in Filter-matrix mode, and modulated around this in flange mode so filter approx 8750kHz), and 36dB/oct roll off, which is a 6th order filter

[Thomeeque]

In filter matrix and flange mode there is a lot of what i presume to be clock noise on the output (measures with oscilloscope at 35khz), which manifests itself as a flanged hissing when not playing.

Frequency of the hiss was stable 35kHz? Clock noise would be swept (by LFO in flange mode and by RANGE pot in filter matrix mode) and it's frequency would be half of frequency of clock at pin 1 of RD5106.

Maybe post audio sample and/or photo of oscilloscope's screen..

T.

[Thomeeque]

and the volume drop while i'm at it...?

How much do you think it needs to be boosted? You can adjust gain of input buffer by R4/R5 (A = 1 + R4/R5 = 1.36 now), but you should not boost it too much (it could "overdrive")..

[DiscoVlad]

Clean Signal:
1khz Sine, 400mv p-p



Effect Signal

Color = Min
Range = Max
Rate = Min
Filter Matrix = On

BBD Pin1 = 35kHz square (clean)


Settings as before but Range = Min
BBD Pin1 = 400kHz square (clean)


All settings as before except:
Range = Min
Color = Max


Probing around the BBD chip - This is in filter matrix mode, it's difficult to get stable measurements in flange mode.

Pin 1: Square wave, frequency depends on Range control.
Pin 2: 0V
Pin 4: "stepped" looking sin wave, dc offset approx 5V. apart from the steps, it looks clean...eg. no ringing.
Pin 5: Clean looking sin wave, dc offset approx 5V.
Pin 6: same as pin 5.
Pin 7: 0V
Pin 8: 11.7V DC

Q1
E: same as BBD output - doesn't appear to be any more noisy.
B: same as BBD output.
C: 0V

The delayed BBD output certainly isn't perfect, but nowhere near as noisy as the effect output.
The noise is worst around the R17/R18 Junction, where the clean and delay signal are combined. Presumably it's a product of the signals being added?
The other ends of R17/R18, are clean.

Other notes:
R7 is marked on the board, but not present.
Q1 and Q2 are BC556 (this is equivalent to 2N5087 so I don't think is a problem)
All (non-electrolytic) capacitors are ceramic disk, All resistors appear to be carbon film.

[DiscoVlad]

The volume drop isn't readily apparent in Filter Matrix Mode - with the caveat that this is a single frequency input signal!

Because the delay is fixed, the comb filter effect has some frequency bands attenuated, and others amplified.

In flange mode though, the output signal at the loudest parts of the sweep measures the same as the input so it should be unity gain, but is noticeably quieter. I'm guessing that the majority of the sweep is when the signal is more attenuated.

I'd think 10-20dB of boost should be more than plenty.

[azrael]

What about replacing the 4558s with an OPA2134?

[DiscoVlad]

Why? the output of U2a is clean, and the only dirt in the output of U2b is from the feedback signal.

[dmc777]

I've also been trying to figure out to fix the clock noise on my new DEM but can't find much info. I was also wanting to add a blend control so you can mix the amount of wet/dry signal. Bernadaurd said he'd as he does all these mods including the volume drop but I guess he go busy.