Morley Sapphire Flanger meets A/DA control circuit

[DiyFreaque]

Hi,

This is my first post to the stompbox forum, I've been lurking for about a week now.  I must say, this is an *excellent* forum, and I've learned quite a lot.

I'd like to thank Mike Irwin for his adaptation SAD1024 of the A/DA flanger, and to Stephen Giles for drawing it out and making it available.  I used to own an A/DA, and I loved it.

Anyway, I've been toying with building a flanger for my synth, using the MN3207.  I based the design off of the Morley Sapphire Flanger, but quickly figured out that the sweep range of the Sapphire wasn't anywhere close to what I wanted.  I tried a number of different schemes to increase the sweep ratio, then finally tried out the 4007/4047 configuration from the A/DA circuit - bingo!  That was exactly what I was looking for.  That thing will clock the heck out of that MN3207 - I was getting just under half a millisecond delay at the high end of the clock, and it'll go as low as you would want (actually a lot lower).

I've posted a write-up about what I've done so far, and a few prelim samples of my experimentation at my site at this address:

http://mypeoplepc.com/members/scottnoanh/slsdiy/id18.html

This page is my 'Birth of a Synth' page, and the write-up and samples are at the top of the page under '11 November 2004'.

Actually, I've amassed all of the parts for the DC-2 chorus, so this will go on hold until I've played with that a bit, which won't be until next weekend.

Cheers,
Scott Stites

[Mark Hammer]

Welcome on board, Scott.  Mike deserves all the accolades you are willing to throw at him.  My personal patron saint of modulation.

Your Birth of a Synth page, and indeed the entire website, is fabulous and a real find.  The samples are wonderful and excellent illustrations.  (Mike:  I'll download them and burn them for you).

A pleasure to have someone who straddles the synth and FX world so nicely.

[DiyFreaque]

Thanks for the kind words, Mark.  Actually, visiting your hammer.ampage.org page a couple of years ago was one of the things that got me going in DIY again.  The great documentation you make available is priceless.  It was there I first learned about the importance of hypertriangular modulation in a flanger.  Also, your posts help me to understand the reasons why the A/DA is so phenomonal and how to go about trying to get the same thing out of a flanger circuit (I still need to explore the importance of filtering regeneration I think).

I'd started DIY in the late 80's when life got in the way.  I started out as a guitarist, got interested in effects, then got interested in actual synthesis.  Only a couple of years ago, I was inspired to get back into it.  What a shock to see that the CEM's, SSM's, BBD's were essentially a thing of the past!  

The nice thing about stompboxes is that they are not only great for guitar, but they also can be used to good effect for synthesis.  A phase shifter is an all-pass filter, a flanger is a comb filter, a big muff is a non-linear waveform shaper   ,etc.  Modulating a phase shifter or a flanger with an envelope generator is a real kick.

As a sidenote, I've followed the TZF threads with great interest - the all pass stages look to be a good idea.  Also, Small Bear has the 128 stage MN3206.  I've done some preliminary graphing of where the clock frequencies would coincide if one were to use this for the delayed 'dry' signal, and it certainly would be possible to avoid the audio range interaction of the two BBD clocks during the all-important TZF range.  Of course, at a different point the clock freq's would cross. But I'm wondering if one took the same pains that were taken in the Boss DC-2, the interaction could be kept to an acceptable level.  After all, I assume the BBD clocks on the DC-2 cross on each and every sweep cycle.  Use of the MN3206 could also (possibly) result in an interesting stereo effect?  IE, dry+swept BBD in one channel and 'delayed dry' and swept BBD in the other channel.  BTW, 256 stages would probably be more ideal (higher clock freq) - two of the 128's could be chained - whoops, now that's three BBD's .

But, that's probably another thread, and I'm starting to wander here.

Cheers,
Scott

[Mark Hammer]

Well thanks back at ya!

Some of the TZF discussion has centred around the disadvantages to having multiple HF clocks sitting in the same circuit.  One of the things I've pondered about recently is whether there would be any advantage to running something like an MN3206 or MN3009 or MN3004 off the same clock as an MN3207 or MN3007, and manipulating LFO waveshape to sweep in a manner to produce relative time differences.  It wouldn't get you TZF, but it could get you a much wider sweep if it was possible to chop the minimum delay by 25-50% via a sort of tracking time-offset.  With both BBD chips run off the same HF clock, presumably the sorts of problems associated with multiple HF clocks would be absent.  Whether the extra effort needed modify the LFO shape would be less than the amount of effort needed to keep two different HF clocks from driving each other crazy is another thing.

Case in point.  Using a common clock with an MN3009 (256-stage) and an MN3007, when the MN3007 is at 500usec delay, the MN3009 is at 125usec delay, making a relative difference of 375usec.  When the MN3007 is at 10msec, the MN3009 is at 2.5msec, making a relative difference of 7.5msec.  A couple of quick calculations indicates that this doesn't actually change the sweep ratio.  Sweeping from 0.5-10msec is 20:1, and sweeping from 0.375-7.5msec is also 20:1.  What it does do is change how close to zero you can go.

Again, if the amount of work involved is no different than what it takes to keep heterodyning in check, then obviously it provides no advantage.

[Vsat]

Hi Scott & Mark,
One of those Morley schematics (Sapphire or Emerald?) looks almost identical to the A/DA -also uses the 4047 and 4007 for the clock.

One of the more interesting suggestions for mitigating TZF-heterodyne noise (I first heard it from JH on the sdiy list, a couple of others have come up with the same idea since then on this list) is to use two different-length BBD's... when the delay times are equal the clock rates will be different... you will still get the hiss 'n pops 'n squeals but at least they won't occur right at the zero-point.

I've tried using the DC-2 for TZF and it certainly has the noise as well... good that Boss chose to put a compander on board!
Regards, Mike

[DiyFreaque]

Hi Mark,

I hadn't considered taking that approach - running one clock certainly would prevent the heterodyning.  As you mention, it wouldn't get you TZF, but it's certainly a very good direction to go in experimentation.  For one thing, it would keep me from abusing the specs of the MN3207 - in particular, I'm exceeding the maximum rated clock speed by a multiplication factor of 7 (though it can be seen that OEM flange devices routinely exceed these specs, too, perhaps by not so much).  I'm trying (unsuccessfully) to picture in my mind how two delayed signals separated by a varying time factor would sound being swept - if it would be the same effect as a perceived wider sweep, or a more intense/different type of flanging sound, either of which I wouldn't mind having.  

And suppose you had a separate polarity control and regeneration loop for each? The possiblities for variation would certainly be quite expanded.  For example, one could have the 'high band' and 'low band' with different levels and/or polarities of regeneration.

It could be that the LFO would require little to no modification to work.  The log response of the clock frequencies should apply to both the BBD's equally....hmmm, but how about to our perceptions?

[StephenGiles]

I'm all ears! The WEM Hyperflanger used 2 MN3004s using separate clocks modulated by the same LFO, but inverted to one clock. Being stereo, the through zero effect through headphones is amazing. I'm sure the circuits are posted somewhere, maybe even on this site.

But going back to your post Mike, would there be any mileage say in using a TDA 1097 for the fixed delay or even SAD 4096, and SAD 1024 for the modulated delay.
Stephen

[DiyFreaque]

Hi Mike,

Right, actually the Sapphire flanger and Crystal chorus both use the 4047 and 4007 for the clock.  However, the component values and interconnections vary enough to make a huge difference.  When I breadboarded the Sapphire, it didn't have nearly the sweep range that I was after, and I felt it started a bit lower in clock freq than I wanted.  Granted, I didn't have the MC3403 for the LFO (forgot to order it) and used an LM324 instead.  But even if I tried a different offsets and LFO amplitudes from my LFO module with the Sapphire 4007/4047 configuration, it just turned into a train wreck - a dreadful range that oscillated wildly at the top end and still didn't go low enough at the other end.  I'd had better luck with optoisolators and current sinks on the MN3102.  I knew you were getting great results with the interconnection/values used in the A/DA circuit, so the parts were there, and I figured I might as well try it, rather than reinvent the wheel.  Much to my extreme delight, it worked *perfectly* - especially given that I restricted the circuit to the 9V supply required by the MN3207.  If that hadn't worked, my next step would have been to try a 4046/4013 arrangement, but this works just fine for what I'm doing.

I'd wondered about the DC-2.  I'll be trying it out shortly (though it won't be til next week-end).  Do the BBD clocks "cross frequencies" on the sweeps as I'm assuming?  And if so, how is the heterodyning kept from totally mucking up the sound?  I know you mentioned the tactic of using a separate regulator for each BBD, but, even so, the original circuit doesn't sound too bad at all, judging from the samples on Modezero.  

I'm sure companding has something to do with it (I do intend to compand this circuit before all is said and done).  Speaking of that (and JH), JH seems to think soft clipping of the compressed signal via germanium diodes as done in the Quadra phase shifter is a good thing.  Have you ever messed with that?

Take care,
Scott

[StephenGiles]

Scott, companding may, and I only say may, eliminate some of the grunge which makes the ADA such a fine sounding flanger. Even mine, built on stripboard, here:

http://img.photobucket.com/albums/v317/StephenGiles/adafl.jpg

.....is not noisy enough to think about compansion - unless someone came up with a less fierce companding circuit!
Stephen

[StephenGiles]

Just a thought, Mike - you suggested using different quad opamps with better performance than the 3403s. Was it MC34070 or something like it? I can't find the post, but I couldn't find any in the UK after your suggestion.
Stephen

[Mark Hammer]

One of the peculiarities you will note that is common to both the A/DA and the Sapphire is that they both have LESS lowpass filtering on the signal coming out of the BBD than they do on the regen signal.  This is so idiosyncratic in flanger designs, that, between the unusual filtering and the 4007/4047 combo for the clock circuit, you'd swear that whoever designed the Sapphire for Morley was aiming for A/DA cloning territory, minus the MN3010, of course.

Uh-oh, are we going to start a rush on these things, now?  This place (http://www.prepal.com/data/Morley.htm) indicates average 2nd hand price at around $50.  If it can be easily tweaked into A/DA-ness, that's not a bad deal at all.

[StephenGiles]

Are you still at work Mark, it's 10.51 pm here, what time do you have at the moment?
Stephen

[Vsat]

Steve, Scott, Mark and list,
I'd say worth trying TDA1097 in combination wth MN3007 or MN3207 or SAD1024... note TDA1097 is 1536 stages which is exactly 1.5x 1024 stages. Could also use three MN3004, or 1 1/2 SAD1024.

Try using two separate flangers first though, or an analog delay plus a flanger... you may find the  noise level not to be much of a problem.

Possible test:
use MN3007 for fixed 10 mS delay - this requires a 51.2 KHz clock

use TDA1097 for variable delay - sweep it from 5 ms (153.6 KHz clock) down to 15 mS (51.2 KHz clock). This will give an effective sweep of +/- 5 mS relative to the fixed 10 mS delay of the MN3007. The clock rates are well above 20 KHz in all cases to eliminate audible whine from the clocks, and to maintain good bandwidth.  The zero-point occurs when the TDA1097 is also producing  10 mS delay... the clock will be at 76.8 KHz which is 25.6 KHz above the MN3007 clock... the sum and difference beat frequencies are ultrasonic.  Also need to consider beating between harmonics of the clock signals, although they will be lower in amplitude than the fundamentals.

Dimension-C:
The DC-2 does not even have clock null trimmers on the MN3207s. The heterodyne hiss and pops is not a real loud effect... but it is there and can be annoying when you've learned to listen for it. What the compander does is hide the noises when there is no input signal present, when a note is being played the noises will (usually) be masked. The compander is really needed to make the effect professional-quality, even when separate regulators and null trims are added. The BBD's by their very nature have a great deal of clock signal present in the outputs.

ARP Quadra: the soft limiter does a nice job in the ARP Quadra phaser in allowing high-regen near-self-oscillation to be achieved while retaining a fairly mellow sound without blowing speakers. In contrast, the Boss Jet Phaser uses a single diode directly in the regen loop which adds a lot of harmonics and grit to a relatively clean guitar signal to accentuate the phasing effect... and there is also an optional distortion circuit in front.
Regards, Mike

[Mark Hammer]

Are you still at work Mark, it's 10.51 pm here, what time do you have at the moment?
Stephen

It's 6:49PM as I read this.  Not to worry.  I'm not working hard,merely catching up on correspondance.  I have no net connection at home anymore, and am heading out of town on Sunday to give a talk in Toronto, so it's e-mail tonight or not at all.

But since I think this is the last correspondance, I bid you all a good weekend.  I think I'm going home.

[DiyFreaque]

Hi Stephen,

Good point about the companding.  My ultimate goal is to have a fairly wide range in functions (including chorusing, and *maybe* doubling) rather like Craig Andertons flanger/chorus thing (dang, can't remember the name - it's at Ampage).  However, it may not be terribly difficult to rig up a bypass for companding if one wanted to do that - maybe being the key word there .  I think dropping down to the lower chorus/doubling range might require the compansion, unless I keep the signal going into the front end fairly muted.  Then again, I'm liking this flanging alone so much, maybe I'll do another 'all the tricks' module and keep this one the way it is, after I explore the A/DA tricks a bit more - the diode arrangement in the input stage is a bit different than on the Sapphire.

Mark,

I sure wouldn't recommend buying a Sapphire and trying to make it do this.  A major difference in what I'm doing now is in the LFO.  After I read your comment about modding a Sapphire, I thought - hey, yeah!  But, I tried it (I still have the LFO going and the darlington transistor set up).  The range of values on the LFO I'm feeding it and what the Sapphire's puts out is very different.  On the Sapphire LFO, going into the Darlington pair, the voltage ranges from ~7.4 to ~6.4 volts thereabouts.  That translates into a clock frequency of so low, the clock squeal and aliasing pretty much get in the way of everything - though it would make a great noise box .

The voltage range I'm feeding the A/DA clock circuit goes from 0V, which translates into a clock freq of around 1.43 MHz(!) at the 4047 output to around 3.5 V where the clock frequency goes low enough that one wouldn't want to go any lower (the noise box described above).  So, if one bought a Sapphire, the LFO would have to be seriously modified as well, which would require the services of someone on the order of Mike Irwin   .  I'd say this is a project that would done if one were actually building a flanger and had MN3207's on hand rather than SAD1024's.

Getting back to that high clock freq - the level straight out of the 4047 drops fairly significantly, and loses the square edge to boot, so I imagine the MN3207 must be on the ragged edge, though I don't hear any 'drop out'.  I need to get some 4049 buffers.  Wonder if I can get away with just one more Jameco order.....hmmmmm (you could never guess I'm married, could you ).

Take care,
Scott

[DiyFreaque]

Thanks for the info, Mike!  I've always had a plan somewhere in the back of my head to try the Quadra phase shifter.  The companding in it alone looks pretty interesting.

Have you noticed that the MN30XX bbd datasheets tend to show clock null trimmers, and the MN32XX datasheets tend to show fixed resistor values?  In fact, clock cancellation shows up as a feature in the initial blurbs of the MN30XX datasheets, and is not mentioned in the MN32XX datasheets.  I've wondered about that from time to time.

BTW, I'm not sure if it's been mentioned on this forum, but in case it hasn't, a very nice Panasonic guide to BBD's can be found here:

http://www.sdiy.org/BBDHell/

Cheers,
Scott

[Vsat]

Scott,
I hadn't taken particular note that clock null trims weren't shown in the MN32xx datasheets... but my experience with MN30xx shows that they are necessary... at least for lower clock rates where the clock is at audio freq and you depend on nulling and filtering to remove as much of the whine as possible. Also something very puzzling ..one of the delay chips (either MN3214 or MN3012) consists of three PARALLEL BBDs with separate outputs, and a common input. One BBD is 3 stages, one is 5 stages, and the other is much longer... around 1024 stages.

Anyone have any idea what this is for (or seen it used in a commercial fx unit)?
Mike

[Maneco]

Congratulations for your work,Scott,your site is a tresure of information and inspiration...
I have a dc2 clone built on perfboard,and is my favourite chorus...i even recorded some vocals through it...maybe in a precise studio enviroment in theory some noises would occur,but i love it...

Welcome!

[DiyFreaque]

Hi Mike and Maneco

Maneco,
Thanks for the compliments on my site, but there sure isn't anything as spectactular there as your micro-looper and nano-looper.  Wow!  Those are fantastic - I remember drooling over the EH 16 second delay ads in Guitar Player magazine years ago, and it looks like you've not only resurrected the concept, but improved considerably on it.  The samples are nice, too - I especially like the reverse stuff.  

The Rendher stuff is out of this world, too - seriously.  So far I've downloaded 'Cuarto Creciente' (I'm on dial-up, so it takes a while).  Man!  Nice song, nice production, nice effects - great stuff!  The stereo effects are really nice.  My 'lab assistant', Matthew (my 5.5 year old son) really flipped for it, too, BTW.

Mike,
I agree, a clock null trim sure makes a lot of sense.  I think the MN32XX apps from the data sheets are probably being a bit 'optimistic'.  I'll definitely follow your lead on putting the nulls in as well.  I tell you, I've learned more on this forum about BBD's and operational theory than I have anywhere else by far.  Thanks a lot!

I'll let you all know how the DC-2 goes.  I'm leaving town tomorrow, so it won't be til next weekend that I get to it.

Cheers,
Scott

[Maneco]

Thanks for your feedback!
I have a lab assistant too,Lucía,she's 5.5 as yours