Looking for information about this pedal. I have the schematic. I have a layout. I have even built a working pedal. So.... To my question.
The BBD called for in the schematic is a Reticon RD5106A. I have some R5106 (notice the subtle difference) that I used in the build but there is one big problem.
THEY ARE TERRIBLY NOISY!! :o
So my question is....
Does anyone have this pedal or pics of this pedal that can tell me the EXACT part number that was used for the BBD?
I am rationalizing that the small difference in part numbers makes all the difference in the audio quality.
Any help would be appreciated ;D
Link to schematic if it helps. :-\
http://www.metzgerralf.de/elekt/stomp/mistress/images/deluxe-electric-mistress-v4-schematic.gif
In my personal experiences all the R510X BBD's are notoriously noisy/hissy. This is a chip that begs for companding or some form of noise reduction in it's chosen circuit.
Just my 2 tin cents...
Quote from: oldschoolanalog on August 30, 2012, 03:18:44 PM
In my personal experiences all the R510X BBD's are notoriously noisy/hissy. This is a chip that begs for companding or some form of noise reduction in it's chosen circuit.
Just my 2 tin cents...
Thanks Dave...
In my research, I have found the same thing. I just cant see how EHX would have used it in this circuit for so long with the noise issues though. There is no companding in the DEM V4 which is what leads me to believe that the RD5106A is a better chip for audio quality. I would just like to get some reassurance on that idea.
I'm not really sure if the "A" suffix really has much to do with fidelity. Compare SAD1024 to SAD1024A. I have found them to both sound the same. Maybe it's in the layout and use of ground planes in the EM design? I am just throwing some thoughts out there as I really can't say for sure. Maybe Puretube or Theehman can shed some light on this as they both have a vast knowledge of EH products.
Ton? Ron?
A few thoughts:
You could parallel a pair of BBD's, this will decorelate the noise for a theoretical impovement of 6dB.
Filter the DC input for the BBD further to prevent power supply ripple generating noise (if you've been using a battery and found the noise unchanged, ignore it).
Make the signal much hotter on input to keep above the noise floor. This may require larger power rails!
Decrease the circuit impedance to reduce johnson noise (again, at input this would be a far bigger factor). TL074's can drive a 2K load, not sure if you'd want to go lower.
Filter above 10Khz on the wet signal(bit of a bodge, but it'll help!)
Use a less noisey BBD! The MN series are supposed to be the cat's pyjamas!
Quote from: Jazznoise on August 31, 2012, 06:23:57 AM
A few thoughts:
You could parallel a pair of BBD's, this will decorelate the noise for a theoretical impovement of 6dB.
Filter the DC input for the BBD further to prevent power supply ripple generating noise (if you've been using a battery and found the noise unchanged, ignore it).
Make the signal much hotter on input to keep above the noise floor. This may require larger power rails!
Decrease the circuit impedance to reduce johnson noise (again, at input this would be a far bigger factor). TL074's can drive a 2K load, not sure if you'd want to go lower.
Filter above 10Khz on the wet signal(bit of a bodge, but it'll help!)
Use a less noisey BBD! The MN series are supposed to be the cat's pyjamas!
I have tried filtering the DC input AND I have tried powering it from a steady source (the circuit I built runs on a charge pump but I also tested with a regulated PS) and nothing gets rid of the noise. The layout I did is basically the same configuration as the original unit except that I minimized the unused open space, added an on-board charge pump circuit, and shrunk it good enough to fit into a BB size enclosure.
I still believe it has something to do with the BBD part numbering. I highly doubt that EHX would have put out a pedal with THIS MUCH noise :icon_eek: It flanges fine and the controls respond great. It simply is unusable because of the noise.
Still hoping that someone with a working pedal will see this and be able to tell me the exact part number from their BBD. :-\
Hmm, is the charge pump close to the BBD? I'm wondering your noise is just switching noise now...but the PS idea contradicts that! Can you or have you tried to bypass the charge-pump with a power supply and test the results?
Is the noise spectral or is it "white"? A clip or even an FFT shot would greatly help me understand what kind of noise you are experiencing!
Quote from: Jazznoise on August 31, 2012, 09:17:47 AM
Hmm, is the charge pump close to the BBD? I'm wondering your noise is just switching noise now...but the PS idea contradicts that! Can you or have you tried to bypass the charge-pump with a power supply and test the results?
Is the noise spectral or is it "white"? A clip or even an FFT shot would greatly help me understand what kind of noise you are experiencing!
I am using an LT1054 chip in the charge pump so it can handle the higher mA draw of the circuit. The only down side of that is that it DOES NOT have the same frequency doubling/boost capability of the MAX1044 and other chips :-\
I am going to try to remove the 12V regulator from the circuit and just directly inject the 12VDC from my regulated PS into the circuit and see what happens again. I am 90% sure I already did this and it was still noisy but I will try again.
I cannot do any video/audio clips as I do not have the tools to do so however, if I could explain the noise it is much on par with AM radio/Tune in Tokyo type noise. Almost like you are rolling the dial on a manual tune radio back and forth. This changes with the Rate knob as well.
Poop, that tells me nothing other than the noise undergoes comb filtering/gets flanged. My bad! Set it to filter matrix and take a tap from the output of the BBD and turn the color knob to 0. White noise? Ticking?
I'm hoping this is PSU noise or a ground issue :icon_confused:
Quote from: Jazznoise on August 31, 2012, 10:38:01 AM
Poop, that tells me nothing other than the noise undergoes comb filtering/gets flanged. My bad! Set it to filter matrix and take a tap from the output of the BBD and turn the color knob to 0. White noise? Ticking?
I'm hoping this is PSU noise or a ground issue :icon_confused:
Thanks. Will try that later this evening and post results.
Quote from: Govmnt_Lacky on August 31, 2012, 09:25:01 AM
I cannot do any video/audio clips as I do not have the tools to do so however, if I could explain the noise it is much on par with AM radio/Tune in Tokyo type noise. Almost like you are rolling the dial on a manual tune radio back and forth. This changes with the Rate knob as well.
I had a similar problem with my DEM from EHX with the RD5106A chip and made a video. Is your problem anything like this:
I made a thread here a while ago about it: http://www.diystompboxes.com/smfforum/index.php?topic=87204.msg732560#msg732560. I haven't figured out what the problem was, it seems to have gone away by itself. It only would happen with my telecaster that had active pickups and not with my strat with passives. No charge pump in this circuit, it's being powered by the provided EHX wall wart.
@blueduck
THAT'S IT!!!
From about the 12 second mark of the video is exactly how my pedal sounds. It still flanges but it has that constant AM radio noise.
I also run active pickups (18V :icon_twisted:) but sadly I do not have a passive guitar.
For those interested... The video is exactly what I am hearing. Any advice to get rid of the noise with active pickups?
EDIT: actually if I remember correctly, it makes the noise without my guitar plugged in. More to follow after work!
Try with a capacitor, 10nF or more, from BBD pin 6 to GND.
If it's a clock-reaching-input issue, then it might be dimmed or better...
If noise comes from BBD's guts, you now know what's the purpose of double outputs in other BBDs... :icon_cry:
Quote from: Fender3D on August 31, 2012, 03:43:56 PM
Try with a capacitor, 10nF or more, from BBD pin 6 to GND.
If it's a clock-reaching-input issue, then it might be dimmed or better...
If noise comes from BBD's guts, you now know what's the purpose of double outputs in other BBDs... :icon_cry:
Thanks for the tips Fender.
I don't have any caps with me to try tonight but I can give it a go tomorrow.
Tonight I am going to try bypassing the charge pump with a regulated 12V supply and also try to audio probe the BBD output like instructed above.
More to follow later :icon_wink:
It sounds like it's oscillating. Blueduck I beleive 12bass in the previous thread got it right and it's due to some very high frequencies in the feedback loop! Try increasing the 47nF (C13 in the other thread) capacitor in parallel with the colour knob to 100nF and see what happens.
Quote from: Jazznoise on August 31, 2012, 04:21:17 PM
It sounds like it's oscillating. Blueduck I beleive 12bass in the previous thread got it right and it's due to some very high frequencies in the feedback loop! Try increasing the 47nF (C13 in the other thread) capacitor in parallel with the colour knob to 100nF and see what happens.
The ring-modulator-esque sound in the video sounds like aliasing noise. It is caused when frequencies which are greater than 1/2 the sampling rate get into the BBD path. What is produced is a bunch of sum and difference frequencies. If the charge pump runs at anything close to the clock frequency, it may be injecting high frequency noise into the delay path, causing the aliasing noise. Moving the charge pump farther away may help. As suggested above, a steeper low pass filter may also help. I've experienced the same thing with my SAD1024A A/DA clone, but with much less severity. Basically, I can only hear aliasing with test tones, not with recorded music or an instrument. BTW, if it is aliasing, the noise should go away with very short delay times (high in the sweep), because the BBD sampling rate will be too high for aliasing to occur.
Quote from: 12Bass on August 31, 2012, 05:29:17 PM
The ring-modulator-esque sound in the video sounds like aliasing noise. It is caused when frequencies which are greater than 1/2 the sampling rate get into the BBD path. What is produced is a bunch of sum and difference frequencies. If the charge pump runs at anything close to the clock frequency, it may be injecting high frequency noise into the delay path, causing the aliasing noise. Moving the charge pump farther away may help. As suggested above, a steeper low pass filter may also help. I've experienced the same thing with my SAD1024A A/DA clone, but with much less severity. Basically, I can only hear aliasing with test tones, not with recorded music or an instrument. BTW, if it is aliasing, the noise should go away with very short delay times (high in the sweep), because the BBD sampling rate will be too high for aliasing to occur.
It's actually a mirror frequency, but yes! (Going 100hz over SR/2 produes a tone of SR/2-100). Going VERY far over the nyquist produces very low frequencies until you hit the lower nyquist and they begin to reflect back upwards. Interesting and fun stuff, but obviously nobodies goal here!
From the video the more I listen the more I think it's aliasing. Hearthe way at the start it's at the end of the LFO cycle at, presumably, the longest delay time on that setting? Resetting the input capacitor might help, but I'vea feeling it's when he messes with the colour knob that the intensitie varies the most so my moneys there...until it doesn't work, of course! :icon_twisted:
Two clocks in the same box is a recipe for trouble, especially if they are not that far apart in frequency.
This is why many digital pedals can sound dead quiet when powered by their own unique power supply, and noisy as hell when two or more are sharing the same wall-wart without any power filtering between the two pedals.
Skip the charge pump and just power the thing with an appropriate supply.
And when powering with a regular PS, be sure to UNpower the chargepump IC. It makes me wonder btw, that there is not a single small cap (100nF) to be seen anywhere across the power rails in any of those mistress schematics. Did EHX use those IC-sockets with built-in ceramics by chance, which might got lost in the tracing process?
Sorry guys. Didn't get a chance to try and debug last night. Will be able to try out everyone's suggestions either later tonight or tomorrow morning. More to come.
OK fellas... here is the report and it is not good so far :'(
Setup: Color at minimum. Range set variable during testing. Speed set to minimum. Mode set to Matrix. NO guitar/signal input. Input left open and output connected to amp.
- Bypassed the charge pump circuit and directly input 12V regulated to circuit. Still have noise. The noise changes with the Range and Speed pots. Still like the video.
- Inserted 100nF cap from V+ input to ground. No change!
- Inserted 10nF cap between BBD pins 6 and 2. No change!
If someone has the time and Express PCB, I would be willing to send them my layout to examine for possible problems. Otherwise, I am at a dead end.
Ideas?
Thanks ;)
Did you remove power to the charge pump, to ensure that it is not still generating noise?
Quote from: 12Bass on September 02, 2012, 08:55:58 AM
Did you remove power to the charge pump, to ensure that it is not still generating noise?
Basically, I removed the 12V regulator from the circuit (socketed) and injected the 12V from the PS at the regulator output. So yes, I did not even have power on the charge pump circuit as the charge pump circuit was removed from the equation when I removed the regulator.
OK, the charge pump IC has been removed, so it cannot be the source of the noise that is still present. I'm going to see if I can make some recordings of aliasing with my A/DA clone for comparison. From what I recall, the DEM design has fairly minimal low pass filtering on the delay path, so that might be contributing. Still, I've minimized the filtering on my A/DA and it is not particularly noisy, nor does it produce much audible aliasing (unless it is deliberately fed a high frequency signal which produces aliasing). Are you using any other pedals before the flanger? I'm wondering if there might be some high frequencies (>10 kHz) being produced by something upstream which are causing the aliasing to occur?
Did you try swapping the 47nf feedback cap for a 100nf?
I think, with the idea in mind that the noise is aliasing, that that's probably the money shot!
@12Bass
I will be anxiously awaiting your findings.
@Jazz
I will try swapping the cap this afternoon. I see that cap is in parallel with the Color pot. The color pot does not seem to effect the circuit negatively. The problems seemed amplified more with the Range pot. When I turned the color pot I only noticed the flange effect and the overall noise floor was amplified.
Could this perhaps be a problem with running the LFO through a quad op amp? I seem to remember talks about circuit issues with this configuration. Possible solutions were to separate the LFO to its own dual op amp.
Will report on the cap swap later this afternoon.
Thanks to all who Have helped thus far ;)
Greg,
if you have it on breadboard yet, you may try with different power supply for each circuit block (at least for clock section and BBD).
If the noise occurs with color CCW, I wouldn't mess with that block...
Although, to my ears, it seems being generating inside the chip...
Quote from: Fender3D on September 02, 2012, 01:26:31 PM
Although, to my ears, it seems being generating inside the chip...
This is what I fear as well. The only thing that makes me think otherwise is the fact that ALL of the chips sound identically noisy. That includes (2) 5106 chips with the clock set to 25uS and (12) R5107 chips with the clock set to 50uS.
I just dont want to face the fact that I might have 14 bad chips :'(
Will try the 47nF cap swap in a bit and post results.
Swapped the cap in parallel with the Color pot from 47nF to 100nF and to 220nF : Still noisy :'(
This may be the first shelved project I have had in a LONG time!! :icon_eek:
Only other options I can think of is either a whole lot of bad BBDs -OR- problems with the LM324 running the LFO instead of 2 dual op amps.
Can you picture the clock waveform when the noise occurs?
Since you get it for a small delay frame, it may be BBD loading clock generator strangely...
Quote from: Fender3D on September 02, 2012, 04:43:18 PM
Can you picture the clock waveform when the noise occurs?
Since you get it for a small delay frame, it may be BBD loading clock generator strangely...
I cant do that at the moment as my oscope is currently at the office :'(
I will try to do it sometime next week.
I am really confused about this as EHX put this circuit out as is. My layout is almost an exact replica of theirs except for eliminating the vast unused space and the inclusion of the charge pump circuit (which was already bypassed)
Before I can record clips, I'll have to retrieve my flanger from a friend. Also, it uses the SAD1024A in a different circuit, so I'm not sure if the comparison will be that useful. But it may help to illustrate what aliasing noise sounds like.
OK folks... back to this again AND I think I found something promising in the quest to keep this OFF of the shelved list! ::)
Take a look at this link. Particularly the schematic at the bottom AND the component-side board pics:
http://www.metzgerralf.de/elekt/stomp/mistress/deluxe-electric-mistress-v4.shtml
Pay close attention to C11 (220uF electro). If you notice, on the schematic it shows the + side of the electro at the junction of C9 and the 100K trimmer. On the actual board itself, it shows the + side of C11 at the junction of the 680R resistor and the 5K trimmer ???
Could this be another FUBAR on the part of EHX?
Could this be what is causing all of the "Tune in Tokyo" noise that I am getting?
Thoughts??
In case anyone is interested...
Swapped around the 220uF cap and VIOLA!!!!
NO CHANGE!!!! :icon_evil: :icon_evil: :icon_evil:
Still noisy as HELL!!!
Back to the shelf I guess :'(
I'm still willing to email the Express PCB layout to ANYONE who can look it over and see if I am missing anything :'(
As this is my ONLY shelved project, I would love to get this up and running... or at least find out if my R5106 chip is indded the culprit.
Hmm, my only final suggestion would be lowpassing the input. Try changing the 1n cap between pins 6 and 7 of IC1b for a 2nF should still keep you above the 20Khz mark. If it improves but is still a problem, bring it up to 4nF. After that you're really gonna start ruining your bandwidth (If we consider losing some of the top octave not ruination).
Quote from: Jazznoise on November 13, 2012, 12:15:16 PM
Hmm, my only final suggestion would be lowpassing the input. Try changing the 1n cap between pins 6 and 7 of IC1b for a 2nF should still keep you above the 20Khz mark. If it improves but is still a problem, bring it up to 4nF. After that you're really gonna start ruining your bandwidth (If we consider losing some of the top octave not ruination).
Thanks Chris.
I have already found 1 error on the EHX factory schematic (the 220uF electro cap between the Bias and Mix trimmers is backwards compared to pics of an actual unit)... who is to say they didn't goof up a cap value?
Would you be willing to look at my Express PCB layout to verify?
Cheers ;D
I haven't built it, so I can't verify if the Schematic is correct. If you're unsure as to wether it correlates to the schematic, send it on and I'll look. If you are sure, I won't be of much help. Email it on and if I get a chance this week I'll check for errors, for now leisurely brainstorming will have to do you.
Unless I'm crazier than I thought, from reading the Datasheet I see the Max bandwidth for a 30mS delay is 8.6Khz. I just simulated the input circuit and my results are odd - nowhere near unity gain. Using a TL074 for now. Changing the cap to 2n does make a difference, however, and brings 20khz down about -3dB. Can't hurt. Get another 1n cap, touch its legs off the other 1n cap and see if the noise goes down, I say!
@Jazz
I will send the Express PCB layout to you as soon as possbile. Thanks for the help offer ;D
Anyone else want to step up and give this layout a look over? Don't be affraid... it wont bite! :icon_mrgreen:
OK...
As I am really frustrated that this is on the shelf, I decided to look at it one more time. :icon_rolleyes:
Something that I did notice in my design is that the V- rail appears to form a big loop around the outer egde of the PCB. In other words, it looks like a big complete ring around the outer edge of the board with the needed V- points tapping off at certain spots of the "ring."
Could this cause possible noise issues? Does this ground "ring" need to be broken at some point?
The best example I could give is that it looks like the following Tonepad layout except that MY outer V- rail is not broken like it is on the right side of this Tonepad layout
http://www.tonepad.com/getFile.asp?id=122
Thoughts??
You'd always better avoid rings...
Either on PCB or when "requested" by women... :icon_mrgreen:
Quote from: Fender3D on November 15, 2012, 10:50:40 AM
You'd always better avoid rings...
Either on PCB or when "requested" by women... :icon_mrgreen:
Got it! ;)
After posting this... I went back and looked at pics of a factory DEM V4 (earlier in this thread) and it looks like EHX ran their ground in a big loop around the PCB as well ???
Did you try the capacitor thing?
Haven't gotten to look at the PCB, I reckon the schematic may have issues - what I heard in that YT video still sounds like Aliasing to me.
Quote from: Jazznoise on November 15, 2012, 03:03:22 PM
Did you try the capacitor thing?
Haven't gotten to look at the PCB, I reckon the schematic may have issues - what I heard in that YT video still sounds like Aliasing to me.
Havent gotten to try the cap. Probably wont get time to until the weekend or maybe even next week :icon_redface:
Ill post what I find with it though ;)
Quote from: Jazznoise on November 15, 2012, 03:03:22 PM... what I heard in that YT video still sounds like Aliasing to me.
I tend to agree. Going to try to get my flanger back soon and record some clear examples of aliasing noise.
Here's a sample of deliberately produced aliasing noise from my A/DA flanger clone. It is the result of the BBD clock of the flanger aliasing with a 20 kHz square wave sent from a frequency generator. For this sample, I set the flanger at the very bottom of its range, which should have a BBD clock of around 34 kHz. If what you're hearing sounds like this, then the BBD clock is aliasing with another signal.
http://members.shaw.ca/kevinmacza/20_kHz_aliasing_sample.wav (http://members.shaw.ca/kevinmacza/20_kHz_aliasing_sample.wav)
@12Bass
That is "some" of the noise that I am hearing. Just not that pronounced. It has been a while since I have ran up the circuit. I need to get a sound sample recorded and uploaded/linked here.
Any recommendations for services? ;D
YouTube is an option, or SoundCloud. My ISP offers free webspace, which is what I use.
BTW, I think the frequencies in my aliasing sample are incorrect due to a mix up when recording. The frequency generator was set to output at a 96 kHz sampling rate, but the recorder was set to record at 44.1 kHz; so I think the square wave sent to the flanger may have been around 8.4 kHz, not 20 kHz. Also, the noise is really loud because it was deliberately produced and normalized to maximize the volume. In practice, with normal music signals, the aliasing noise is not audible. I just wanted to provide a clear example of what aliasing sounds like for reference.
My guess is that the BBD clock signal may be leaking into your audio path, or perhaps there is a problem with the LPF in the audio path, or both.
Good god that aliasing is horrible, well done! :icon_lol:
I agree, 12 Bass, I think the input filter is wrong - my suggested mod to the cap on the input of the LPF before the bucket brigade might help. I think the value listed has a cutoff far too high for the BBD. 2n or even 4n of input capacitance should get you closer to a trouble free life (At least with flangers, anyway).
I haven't forgotten the layout you sent me, but I'm currently feeling like a bag of warm poo due to a cold so you'll have to be patient!
Quote from: Jazznoise on November 19, 2012, 03:31:44 PM
Good god that aliasing is horrible, well done! :icon_lol:
I agree, 12 Bass, I think the input filter is wrong - my suggested mod to the cap on the input of the LPF before the bucket brigade might help. I think the value listed has a cutoff far too high for the BBD. 2n or even 4n of input capacitance should get you closer to a trouble free life (At least with flangers, anyway).
I haven't forgotten the layout you sent me, but I'm currently feeling like a bag of warm poo due to a cold so you'll have to be patient!
What do you think the range could go UP to? I was looking over the pics of an EHX board and, I know this isn't an exact science but, it appears that the cap you are referring to is about the same look, shape, and size of some other 100nF caps on the board.
I am wondering if this is REALLY another screw up on the schematic and the actual units used a much bigger cap in that spot. :-\
I think anything 2n would cut anything above 20khz, 4n 10khz and 8 in 5khz. 8 would probably be the upper limit before your flanging began to lose it's sparkle.
Of course that's me looking at the filter in isolation, I'm not sure how the loading from the previous stage will effect it - that's more of a PRR thing! :P
OK...
Ran parallel caps with the 1nF in the filtering section. From 1nF to 100nF :icon_eek:
Some of the noise dissipated however, other parts of the noise were CLEARLY unaffected :'(
Cut the loop trace in the grounding AND verified ALL component values, placement, and orientation. All is correct. STILL GOT NOISE :icon_evil:
I think I may need to get some cheezy audio sample posted so you can hear my pain.
Any recommendations for a good place to upload audio samples?
Quote from: Jazznoise on November 19, 2012, 03:31:44 PM
Good god that aliasing is horrible, well done! :icon_lol:
Perhaps this dates me... but I think that the aliasing noise sounds a lot like the sound effects in Galaxian. Those who have ever played the game will understand. :icon_wink:
Dropbox is fine for hosting! Did you notice any treble loss on the "Wet" side as you went through those values? 100n should be muffled blanket!
This is perplexing. Honestly, I'm starting to think your BBD's delay times are just becoming too long to be useful in the audio range. That said, due to the pitch of the noise being around 2-3Khz I can't see the sampling frequency going below 5khz. The only other question is is there a large noise source onboard the pcb.
Those crazy inharmonic noises are always very sci-fi to me, 12bass! :icon_lol: Ever listened to Melt Bannana?
http://www.youtube.com/watch?v=yOEILhkXv5c
Quote from: Jazznoise on November 20, 2012, 06:02:31 AM
Did you notice any treble loss on the "Wet" side as you went through those values? 100n should be muffled blanket!
Absolutely! As I went up through the values, I noticed a treble cut with each increment. Like I mentioned above, SOME of the noise dissipated however, MOST of the noise remained.
Quote
This is perplexing. Honestly, I'm starting to think your BBD's delay times are just becoming too long to be useful in the audio range.
In this circuit, I am using an R5107 for the BBD instead of the R5106. I adjusted the clock trimmer to 25K in order to double the clock frequency as the 5107 is a 512 stage BBD and the required R5106 is only 256 stages. From the research I have done, this is what needs to be done in order to support the extra delay stages. (Double the delay = Double the clock frequency)
Am I wrong?
I will try to put something up from Dropbox as soon as possible. It may not be the best quality but, I will try to capture what I am hearing.
EDIT: @Jazznoise
Were you able to look at my PCB design? Did you see ANYTHING that looked wrong compared to the schematic?
With a 512-sample device (RD5107A), the BBD clock should not drop below 34 kHz. Even that would be really low for a flanger, producing a 15 ms delay. 1.024 MHz would provide a 0.5 ms delay, which would be reasonable for the maximum clock. In short, the BBD clock should be high enough that aliasing with the frequencies normally generated by guitar signals should not be a major issue, unless there's a problem in the circuit.
Melt Banana sounds pretty crazy!
A highpass filter on the clock leg could help to keep it well outside of the audio band. I'm not sure how that would interfere, as its the rising edge that the clock input is really looking for, but if the -6dB point was 20Khz then problems should be minimised. I'm wondering is there crosstalk between the clock and input, but how can we reduce it?
Are you using the CD4013 to generate your word clock? I'm not au-fais with some of these things by any means but I've been told CMOS 555's generate less ticking than some other models. Is your D Type CMOS?
What program did you use to generate the .pcb? It's not opening in Ultiboard!
Read this entire data sheet first. Do the math. The R5107 is the equivalent of a 1024 stage BBD. It has specific clocking requirements in regard to frequency and rise/fall times for the clock you are using. If these are out of spec maybe this is contributing to noise issues. Maybe.
Just an observation.
http://thmq.mysteria.cz/em1022/RD5107.pdf
Note: the data sheet for the RD5107A states that it is a 512-sample device, not a 512-stage device. The SAD1024A is a dual 512-stage device, which can use the two BBD sections in series to produce a combined 1024-stage device. Perhaps the difference is due to the fact that the RD510x series only uses a single clock (tick, tick, tick, tick), while the SAD1024A employs a two-phase clock (tick, tock, tick, tock). The result is that each 512-stage section of the SAD1024A produces 256 samples - or one sample for each pair of complementary clock signals.
In summation, the 512-sample RD5107A can be viewed as equivalent to a 1024-stage device, with the BBD clock set accordingly. If my math is correct, the above frequencies should be in the ballpark.
Quote from: oldschoolanalog on November 20, 2012, 09:43:15 AM
Read this entire data sheet first. Do the math. The R5107 is the equivalent of a 1024 stage BBD. It has specific clocking requirements in regard to frequency and rise/fall times for the clock you are using. If these are out of spec maybe this is contributing to noise issues. Maybe.
Just an observation.
Was in the middle of composing my post.... :P
If the RD5107A is anything like the SAD1024A, it will be pretty noisy at low clock frequencies. Plus a ~ 10-15 kHz clock would be much more likely to produce aliasing byproducts. Would be funny if the problem was nothing more than the clock set too low. Of course, it shouldn't be producing very useful flanging if that's the case.
>In summation, the 512-sample RD5107A can be viewed as equivalent to a 1024-stage device, with the BBD clock set accordingly. If my math is correct, the above frequencies should be in the ballpark.
>If the RD5107A is anything like the SAD1024A, it will be pretty noisy at low clock frequencies.
+1 totally agree. I am wondering if the clock arrangement used is within the spec for rise & fall times. To check this one would need a 'scope.
Scope would be nice for sure.
However, audio samples should give a pretty good idea of the BBD clock. If there's double-tracking or slapback echo, it's way too slow and needs to go up, way up....
The single phase clock input of the R510X BBDs is divided by 2 internally. So... to get the ~34kHz low clock f you need for max delay one would have to have their external clock input @~68kHz.
I agree w/12Bass. It might just be clocked way too low.
Get out the f counter and adjust the clock trimmer for 68kHz or so at the clock input pin for the low end of the sweep. With the EM in "filter matrix" mode this should be an easy task.
My 2 cents.
OK gents... my head is spinning a bit so I just want to clarify.
Using the 5107 BBD in this circuit....
What clock frequency should I set at Pin 1 of the BBD?
@oldschool
According to your post, I should have the circuit in "Filter Matrix" to accomplish this correct? Also, should I set the Color, Range, or Speed controls to any particular setting for this adjustment?
Really appreciate the assistance with this! It has been driving me nuts just starring at me on the shelf! :-\ ;D
RD51s have fs = fc/2, and BW = fs/3, check your filters...
Start w/68kHz @ pin 1.
In filter matrix mode turn the range pot fully to the end that is the low f side. Speed and color shouldn't matter but put them at minimum to be on the safe side.
Hope this helps.
Quote from: Fender3D on November 20, 2012, 10:30:45 AM
RD51s have fs = fc/2, and BW = fs/3, check your filters...
Could you elaborate on the filter checking?
Still learning about these darn filters... ::)
We've address both your input and feedback filters at this stage so it's unlikely that this is still the issue. Grab a scope, try and determine the speed you're clocking the BBD at.
Quote from: Govmnt_Lacky on November 20, 2012, 11:03:48 AM
Could you elaborate on the filter checking?
the max allowed frequency should be (less than) fs/2,
schematic on page 1 shows basically no filtering (1k is -3dB then ~1dB/oct)
we still don't know what frequency your VCO runs at...
IIRC you somewhat solved this issue lowering the clock capacitor in your micros (thus raising the clock freq.)
When I designed the layout and assembled the circuit, the only mod that I made was to increase the clock trimmer from 10K to 25K in order to dial in the "what I thought was correct at the time" higher clock frequency. I was going by the 5107=512 stages=twice the frequency mentality. I am seeing now that my initial thoughts were probably wrong :icon_redface:
I also trimmed the clock frequency according to the info on the factory schematic which calls for 25uS at Pin 1 of BBD with circuit in Filter Matrix mode and Range fully CW. Seeing as I had the 5107, (instead of 5106) I set it to 50uS. There is no mention of frequency in KHz on the schematic. Just rise/fall time.
Tonight, I will connect my Fluke meter and read the FREQUENCY at Pin 1 of the BBD on each end of the 25K pot rotation and report ;D
Sorry guys.... no oscope at home at the moment and the circuit is not at work :'(
Ok campers... for those who are following along ::)
Range pot set to full CW
Clock trim full CW : 91KHz
Clock trim full CCW : 18KHz
Range pot set to full CCW
Clock trim full CW : 934KHz
Clock trim full CCW : 258KHz
I currently have the trimmer set to 68KHz on Pin 1 of the R5107 with the Range pot at full CW.
Gonna attempt to upload a sound/video clip to hear the noise. I know it will be there as I tried adjusting the clock trimmer before and the noise was there throughout its travel. :-[
OK, massive fail on the sound clip upload :icon_redface: Wont get into it!
So, I will give my best description of what is happening and if it doesn't get fixed I may just re-etch a newer, better board and start over!!!
With the clock set as above (68KHz) at Pin 1 of BBD, with Color at max, Speed at minimum, and switch in Filter Matrix.
As I cycle the Range pot back and forth it sounds EXACTLY like I am turning the dial on an AM/FM radio with no stations in the area!! :icon_eek:
When I turn the Speed knob up, I can hear the LFO tick mixing into the signal.
Quote from: Govmnt_Lacky on November 20, 2012, 06:37:53 PM
I currently have the trimmer set to 68KHz on Pin 1 of the R5107 with the Range pot at full CW.
What does happen if you raise that 68kHz setting (higher flanging aside)?
Does Tokyo stop broadcastings? :icon_mrgreen:
Quote from: Fender3D on November 21, 2012, 12:17:20 PM
What does happen if you raise that 68kHz setting (higher flanging aside)?
Does Tokyo stop broadcastings? :icon_mrgreen:
I did notice that when I adjusted the clock trimmer to the higher side (91KHZ) that the sound does start to subside. I looked into making the adjustment however, according to the datasheet, the R510X BBDs have a max clock of 1MHz. Seeing how this adjustment would put me up to 934KHz with the Range pot full CCW, I didn't follow through with it.
EDIT: Just to add to the notes... the test test above was done with NO input. The noise was strictly from the circuit operating at voltage with an output directly to amp.
>I did notice that when I adjusted the clock trimmer to the higher side (91KHZ) that the sound does start to subside. I looked into making the adjustment however, according to the datasheet, the R510X BBDs have a max clock of 1MHz. Seeing how this adjustment would put me up to 934KHz with the Range pot full CCW, I didn't follow through with it.
Why not? You would still be within limits. Go for it!
@oldschool
I don't see how I would remain within limits. If I increase the trimmer range then my max clock frequency is going to break 1MHz.
Could you explain a bit?
Sorry Greg, I guess I misunderstood what you were saying. :icon_redface:
I have been doing a lot of that lately. Think I'll just shut up for a while. Don't want to create any more confusion.
Apologies.
Crank it and see if you notice the BBD's flange going static at some point. The main worry here is wether doing this removes the noise - if it's still there then the problem remains regardless.
Quote from: Govmnt_Lacky on November 21, 2012, 05:13:58 PMI don't see how I would remain within limits. If I increase the trimmer range then my max clock frequency is going to break 1MHz.
"1MHz max clock" is not a harmful warning as "max. 15V",
you may raise clock well over its max. declared value, downsides are gain loss and static at a certain point
BTW +1 on what jazznoise said
@Jazznoise
Were you able to look over my layout?
Any feedback?
Quote from: Jazznoise on November 20, 2012, 09:39:53 AM
What program did you use to generate the .pcb? It's not opening in Ultiboard!
Did you try increasing the clock?
Quote from: Jazznoise on November 27, 2012, 11:59:14 AM
Quote from: Jazznoise on November 20, 2012, 09:39:53 AM
What program did you use to generate the .pcb? It's not opening in Ultiboard!
Did you try increasing the clock?
I used ExpressPCB to gen the layout.
I have NOT attempted the clock increase yet. I will need to remove the existing 25K trimmer and/or series 3K9 resistor to do this and I am a bit weary :-\
The board I etched was not the "best quality" and I have already had signs of traces lifting during mods (nothing I couldn't fix on the spot and shore up) so if I go in with a trimmer change, it will most likely need to stay permanently. Or, until I re-do the layout with thicker traces and re-etch :icon_eek:
I don't see much space for crosstalk between the 4013 and the R5106. Pin 8 maybe, but the cap should be helping there. A scope on the V rails of the R5106 would tell you how dirty they are, but they should be fine.
It seems fine, I'd use your layout but we're 5 pages of stumped in on this one.
Well, like I mentioned before... the layout I sent you DOES have some errors that have since been corrected (220uF cap backwards between schematic and actual PCB pics, transistors backwards, wrong input cap value, etc.)
Bottom line is that these R510X chips (I have 6s, 7s, and 8s) were "found" at my work in a discard pile. I really hate to say it but, it might just be that ALL 18 of them are just plain noisy! :'(
If you look at the factory schematic, it calls for RD5106A (which could only be RD5106ANP-011) I have since raised the question if whether there was a difference in tolerances between the RD510XA series and the plain old R510X as far as audio quality.
I REALLY DONT want to face the fact that this awesome find was really a bust! :icon_redface: :icon_evil:
Hmm, it's a reasonable question. I think the clock mod might be the next thing, and at that point if there's no luck then I think blaming the BBD stage becomes a meritable answer.
How to deal with that will have to follow, I guess. Learning experience! :P
Just to follow up on this thread for reference....
Using a charge pump configuration with the circuit (LT1054) was causing the problems. Once I connected a usable transformer to feed the proper voltages, practically ALL of the noise disappeared. :icon_cool:
I dont know if was due to "sewer ground" cross talk or if the circuit was just drawing a bit too much current from the pump but, it works great with the transformer as per original spec from EHX.
Ah ha! ;D
Glad you got it working! I'd love to hear it if you make any clips!