Choral Reef with a MN3005?

[Aentons]

So back to the other thing... If I wanted to mod it to play around with in the future, it sounds like I should socket the 3 timing caps (C3-5). I already have sockets on the mn3008. What else should I do to get it closer to a multi mode type deal (flange,chorus,slapback) on the switch?

See pic to get more of an idea of what I'm going for (just not digital).

[Scruffie]

You just need to socket one of these in it https://lectric-fx.com/product/dandy-horse-v-1-0/ 

[Aentons]

You just need to socket one of these in it https://lectric-fx.com/product/dandy-horse-v-1-0/ 

I know it'll fit

[Aentons]

MN3009 = 0.64ms @ 200kHz - 12.8ms @ 10kHz = Flange
MN3008 = 10.24ms @ 100kHz - 102.4 ms @ 10kHz = Chorus
MN3005 = 20.48ms @ 100kHz - 204.8ms @ 10kHz = Delay

So... a daughter board down in the battery compartment with one of each of these. Could it work?... why not?

[Scruffie]

I was only half joking about the Dandy Horse...

How would you accomplish the BBD switching? And then the noise problems I mentioned, the BBDs biases will be suboptimal without each having its own bias trimmer, each will need its own supporting resistors and for longer delays you'll definitely need a cancel trimmer for the MN3005, plus with that much capacitance on the clock you'll almost definitely need a buffer to get the 200kHz for the 3009. The filtering for each will be sub-optimal, there's a very good reason the Dandy Horse is so complicated.

That's got to be $50 of chips to have a lot of drawbacks... just... give in.

[Aentons]

I'm flyin' by the seat of my pants here... it's mostly speculation on my part. 

How would you accomplish the BBD switching?

I wasn't assuming switching between each them. Was thinking more like the typical stacking/combining them like is normally seen to use 2 3008 as a 3005, but not sure how that works. Does it use them in order... so short to long?

And then the noise problems I mentioned, the BBDs biases will be suboptimal without each having its own bias trimmer, each will need its own supporting resistors and for longer delays you'll definitely need a cancel trimmer for the MN3005,

I think I have seen daughter boards with bias trimmers, so idk id have to look into more. I'm not expecting max delay... just want to see what it can do.

plus with that much capacitance on the clock you'll almost definitely need a buffer to get the 200kHz for the 3009.

I dont think i need it to go to 200k... I'm hoping 100k will give me about 1-2ms which is about where flanging starts, right?

The filtering for each will be sub-optimal

Good enough for rock and roll? idk.. Im willing to try it out

That's got to be $50 of chips to have a lot of drawbacks... just... give in.

MN3008 = $10 (from forum )
MN3009 = $10.50 (small bear)
MN3005 = $21.95 (small bear)
Total = $45.45 to be exact

Thanks again for being a sounding board!

[Scruffie]

When you run them in series like on a daughterboard, they all work together so with a 3009 + 3008 + 3005 you'd effectively have a 6,400 stage BBD, you'd need some way to switch from the output of each. So you'd get 3009, 3009 + 3008, 3009 + 3008 + 3005.

To get reasonable flanging you need at least a 10:1 sweep ratio which this circuit wont give so it will just sound like higher octave chorus, nice, but not flanging. There are so many things to consider designing a new BBD based effect which is effectively what you're trying to do, I didn't even mention gain losses.

If you're flying by the seat of your pants, you need to learn a lot more about BBD's first so get out the breadboard before you start modding.

[Aentons]

When you run them in series like on a daughterboard, they all work together so with a 3009 + 3008 + 3005 you'd effectively have a 6,400 stage BBD, you'd need some way to switch from the output of each. So you'd get 3009, 3009 + 3008, 3009 + 3008 + 3005.

I don't know enough about this yet so please bear with me but, I am confused... if they are all effectively working as a single bbd, why would the outputs need to be switched. Wouldn't it effectively be a a single bbd with a unified delay range of roughly 1ms@100k to 205ms@10k from a single output, and then you could modify the 3 timing caps to divide that range into 3?

To get reasonable flanging you need at least a 10:1 sweep ratio which this circuit wont give so it will just sound like higher octave chorus, nice, but not flanging.

I did not consider sweep range. I'll have do some research there. Do you know what ratio it will do?

There are so many things to consider designing a new BBD based effect which is effectively what you're trying to do

I get where you are coming from but I would argue that I am attempting to explore the possibilities of new sounds within the limitations of the existing circuit

I didn't even mention gain losses.

Wouldn't the internal level trimmer help with that?

If you're flying by the seat of your pants, you need to learn a lot more about BBD's first

I total agree with you and thank you for helping me learn about them

so get out the breadboard before you start modding.

I try to when I can but there's a waiting list and my kids keep any extras all full up with their blinky light switches... I need to buy some more.

[thomasha]

And I trying to build a low parts delay...

I think you should test the 3 circuits separately and check where you can compromise.

Another thing would be a programmable filter at the output of the chips and a programmable clock. I saw something like that for a pt2399 delay and the clock in the flangelicious. That could work. But then again it would be easier to go all digital wouldn't it?

Getting the chips to work separately would already be n interesting experience

[Aentons]

I found a good explanation of what Scruffie is talking about... for those following along at home:

https://guitarscanada.com/index.php?threads/looking-to-have-a-hartman-flanger-repaired-fixed-by-bduguay.80680/

"The delay chip takes a succession of samples, right? It takes those samples in alternating fashion and passes them along. The faster they are taken and passed along, the shorter the delay time. The clock generator provides a tick and a tock. When it's tick the sample is being taken and stored. When it's tock, the sample is being passed on to the next stage. Two parallel paths inside the chip are busy taking samples and passing them along in alternating fashion such that when one is sampling, the other is passing, and vice versa. The two paths then get stitched together at the output so that there is a continuous seamless stream of samples. It's a little "stair-steppy", but some filtering fixes that and crerates theillusion that it's just like a continuous analog signal rather than a sampled one.

The clock "instructs" the chip to take the sample or pass it along. The immediacy of the chip's response to that "instruction" requires that the clock pulse occur immediately, rather than gradually. That is, the clock pulse hitting the delay chip has to be square, rising sharply and falling just as sharply. Since there is a threshold voltage that the clock pulse has to reach for the delay chip to do its thing, you do NOT want the clock pulse to rise gradually, in triangular form, since that means there will be short gaps between the taking of each sample. It's like the clock generator instructing the delay chipto take a sample.....nnnnnnnnnnnnnnnOW, and pass it along nnnnnnnnnnnnnnnnnOW! The audio quality will be terrible.

The pins where the clock pulses are fed on the delay chip act like treble cut, except REALLY high up. So, the same way your guitar tone control has no impact on bass frequencies, the "treble cut" on the clock input pins has no impact if the clock rate is under 100khz, but starts to "roll off the treble" and round off those nice square clock pulses when the clock rate exceeds that, with the resulting decline in audio quality.

HOWEVER....that is only true if the clock generator circuit is the MN3101 or MN3102 chip Matsushita/Panasonic designed to make delay circuits easy and small to build. It can provide enough "drive" for the pulse to remain unaffected and perfectly square, but craps out in that regard above 100khz. If one adds some drive and buffering, then the delay chip will happily be "clockable" at rates/speeds much higher than 100khz.

The now out-of-production Reticon SAD1024, that was used in the DEM and so many other pedals, was designed such that it introduced much less "treble cut" on those clock input pins. How much less? Roughly 1/7th, such that you could feed it 700khz without any outside assistance, and achieve REALLY short delays. The truly great flangers will sweep down to under 0.2msec. I know it may not seem like much to go down to a fifth of a millisecond from 1msec, but trust me, you hear it. Dramatic flange sweeps need a sweep range of at least 20:1, and more often at least 30:1. A 20:1 sweep that starts from 1msec, would go from 1-20msec, and delays beyond 15msec tend to yield chorus effects not the "notch infestation" that great flanging provides. So if one wants a big sweep ratio, but doesn't want to crossover into chorus territory, you HAVE to start low and that requires a very high clock frequency and a chip that can handle it. Sweeps from 0.2msec to 10msec have a 50:1 ratio, and stay well within flanging range. I mention this to emphasize why folks make a big deal of Reticon chips and pedals that use them.

HOWEVER....it is possible to clock Matsushita/Panasonic chips like the MN3007 really fast as well. I've heard one clocked well above 1mhz, and the flange effect sounded like it descended from the stratosphere. Companies like A/DA learned that if you provided the extra drive to the clock signal, it reached the clock pins on the delay chip unscathed. That's one of the reasons why the old A/DA Flanger soundedso magnificant; it was able to successfully clock REALLY high, and achieve ultra-short delays.

In a way, it is a bit like the effect of an onboard preamp or buffer on a guitar when the signal has to travel long distances. In the absence of buffering, the capacitance of the cable shaves off high end. But, buffer the signal and add some drive, and it can travel great distances unscathed. The 4049 chip on the Hartman is used to provide that buffering and extra drive so that the MN3007 can be made to sweep down to uber-short delay times."




And then this in reply to the other:

https://www.thegearpage.net/board/index.php?threads/calling-all-pedal-builders-theres-a-hole-in-the-market-vintage-electric-mistress-clone.1791531/page-8

"You would think the above would be the case, but time and time and time again I see posts and pedals being built where the assumption is: "Just get the supporting circuitry correct and then adjust the clock speed".
This is incorrect. I've built an electric mistress clone into a wah shell so I can foot-control the range. This is especially useful in "filter matrix" mode since you can then manually control the sweep. The differences between BBD chips in this mode becomes overwhelmingly apparent. When using an MN3006, the flange sweep is very smooth and "vintage" sounding, just like old tape flanging. With the MN3006, I run the clock at the original speed, using a 47P capacitor.
With an MN3007, I bump the clock speed way up with a 15P capacitor, yet the flange sweep in manual mode sounds "watery" like chorus, and it doesn't sweep smoothly. The sweep is bunched up on one end of the potentiometer. It's because there are too many stages in an MN3007. No matter how fast you clock it, the signal still travels through 1024 stages in an MN3007. This introduces it's own "character" to the resulting sound. There is no way to avoid that.
To illustrate: Imagine there is a machine that punches holes into a piece of paper. One setting punches 128 holes into the paper, the other setting punches 1024 holes into the paper. Speed doesn't matter. You can run the paper through the 128 setting really slow, and the paper still gets 128 holes punched in it. You can run the paper really fast through the 1024 setting, and the paper still gets 1024 holes punched in it. Now compare the two pieces of paper. Are they the same? No. One has 8 times as many holes in it! Does it affect the paper? Absolutely!
A perfect example of this fact comes from Electro Harmonix themselves. Back in the 1999 when they reissued the Deluxe "big box" Electric Mistress, they used a Reticon RD5106A (256-stage)BBD, which was very scarce even back then. The MN3007 was a more readily available chip (they were putting them into their Small Clone reissues), so why didn't they just use the MN3007? Because it's not the right tool for the job! I would much rather trust the people who invented the effect to know what to use than the opinions of people who try to clone them."

[Aentons]

With a little research, I think I am starting to get it a little better now...

The output pin switching thing makes sense now. The more stages you have the longer it takes you to get all the way thru them at a given frequency... (duh, sorry im a real moron sometimes)

...it is possible to clock Matsushita/Panasonic chips like the MN3007 really fast as well. I've heard one clocked well above 1mhz, ....... if you provided the extra drive to the clock signal, it reached the clock pins on the delay chip unscathed

So as Scruffie said, it sounds like step 1 is to buffer the clock signal. 4049 does not look to be a drop in as there are additional pins. Is there another pin compatible clock with a built in buffer? If not, would connecting a simple transistor boost between the pins work?

(Dropping the multichip/daughterboard idea for the moment) If a dropped in a single mn3005 and somehow overclocked it, how low could it go?... 10ms like a 3008? So maybe just a chorus/delay is all that's feasible.

The filtering for each will be sub-optimal

Also, for the post filtering, it seems like I could tack on a couple of extra poles to the existing ones with just a couple of leg lifts or sockets and take it from 12db to 24db

[Scruffie]

Well done on the research.

You have more to do though, find a book on filters because you can't just stick random extra poles in, they have to be designed quite precisely and the maximum from one filter section is really 3 poles... 4 is possible but to have them work and not oscillate is a real challenge and not worth it.

And no, there's no secret buffered clock and you can't stick an LPB-1 in between, CMOS clocks are effectively digital. You can use transistor based logic buffers but it'll probably be bigger than a 4049. Again, IIRC the 4047 will drive a single 3005 to about 200kHz which is not anywhere near flanging and never will be. A BBD also works as a low pass as you drive it to ever higher frequencies, we get away with it with a 3007 but a 3005 will be a weak outputted dull mess by the time it got to flanging.

No, you can't get a 10:1 sweep from that clock, IIRC it gives about 2:1 at best. It's not the clock actually, it's the variable resistance element which in this case is the diode.

But yes, you can use a 3005 and have delay and chorus from the same chip.

And I trying to build a low parts delay...

That's not so hard, you just have to sacrifice headroom (boost the input a lot to compensate for the lack of companding) and bandwidth (heavier filtering in exchange for less filtering stages).

Those filtering chips are great except they usually run at 100:1 clock frequencies so you need a clock for the chip and then a divided clock for the BBDs. Works with digital because that's already at a high clock frequency.