MN3007 based analog delay project/schem?

[nelson]

I have 50 MN3007 that I need to do SOMETHING with.

I want to build an analog delay so does anyone know of a delay based on the MN3007?

All other suggestions for pedals based on the MN3007 welcome too.

[cd]

Check the analog delay at GGG.  You'll need to use eight MN3007s, each with their own output and bias trimmer.

[george]

most decent analog delays use at least 4096 stages of BBD delay (eg the DM-2 uses an MN3005 which is 4096 stage).  The MN3007 has 512 stages, that's why cd says you'll need 8.

even aside from the practical difficulties that still leaves you with 42 of the suckers.

I guess you could give them away free as a way of giving something back to the diy effects building community?

[Dirk_Hendrik]

The MN3007 has 512 stages, that's why cd says you'll need 8.

Make that 1024.

The Carl Martin Delayla uses 6 MN3207. Other than the different power supply connection this proves the idea works. In the delayla's case there is only one bias trimmer. This because the chances are high that when you use all chips from the same production batch they'll all have more or less optimal performance at the same bias level.

Take a look at the AD3208 delay at GGG and use that principle to connect the 3007's. (apart from the bias supply ofcouse)

[nelson]

most decent analog delays use at least 4096 stages of BBD delay (eg the DM-2 uses an MN3005 which is 4096 stage).  The MN3007 has 512 stages, that's why cd says you'll need 8.

even aside from the practical difficulties that still leaves you with 42 of the suckers.

I guess you could give them away free as a way of giving something back to the diy effects building community?

Dont know about giving them away free, but I may be selling them at a price thats practically giving them away.

I am in the UK though.

[StephenGiles]

Back in the 80s I was given 20 or so TDA 1022 by a kind sales rep at Phillips in London. He also sent me some info re improving the signal to noise ratio, and I duly made a delay using 16 of them. It was very very noisy indeed. Later on I built Craig Anderton's Delay/Chorus unit using just one SAD 4096 which was considerably quieter - helped by a compander of course.
Stephen

[cd]

Noise can be kept to a minimum provided you follow the manufacturer's datasheet specs.  Using one bias trimmer is possible, but obviously for best performance you want to dial in each one exactly.  Or, match them using an oscilloscope into batches.

The current Maxon AD-999 uses eight 1024 stage chips for an advertised 900ms of delay. 

[Jaicen_solo]

Nelson, if you do decide you want to sell some of those chips, put me first in line.  You can never have enough of stuff like that in my opinion. I'm gonna stock up on a load of LDR's as well, I don't think I could face a world without them!

[Mark Hammer]

There are the usual sorts of issues with regard to whether a single bias voltage, distributed to a bunch of separate BBDs, would necessarily bias them all optimally.  There are several commercial and DIY instances where this seems to work, and people give me assurances it will.  Of course if one aims for 8 BBDs in series, that's a further stretch than 3 or 4.

The thing I wonder about is the impact of all those clock input pins in parallel.  As Mike/Vsat has pointed out to me, the input capacitance on those clock pins represents a hurdle that needs to be overcome.  If you read the specs on the clock driver chips (MN3101/3102) they are spec'd to drive 8192 stages.  That seems like kind of a weird spec at first.  When you re-interpret it as "this chip cannot deliver a crisp squarewave clock pulse when faced with more than X amount of input capacitance", speaking in terms of total number of drivable stages starts to make more sense.  The input capacitance on the clock pins is a bit like the cable capacitance of a patch cable.  It acts as a capacitor to ground, rolling off highs.  In the case of the clock pins, it will take a clean 50% duty-cycle clock pulse, and turn it into a rounded version of same.  What I gather you get is that there is a brief interval between each complementary clock peak (pos peak, neg peak), where the clock pulse is not really switching anything inside the BBD because it isn't above some critical threshold for switching the FETs inside the BBD.  Remember that a square pulse is something that goes almost instantaneously to its peak value.

What is, of course, not specified in the 3101/3102 datasheets is how many BBDs that 8192 stages is supposed to come from.  My guess is a pair of MN3005/3205 chips, rather than an octet of 3007/3207s or a quartet of 3008/3208s.  Keep in mind that the input capacitance of each of those chips is placed in parallel with the others IF they are clocked directly from the same 3101/3102 witrh nothing in between.

The bottom line is that you'll undoubtedly have to buffer the clock lines going to all those BBDs.  It's not impossible, obviously, just something that needs to be attended to.  Pumping more current in via buffers will apparently offset the effect of the input capacitance.

[puretube]

iirc, that Max*n AD999 uses 2 3102`s,
and individually biases each BBD separately...
(16 trimpots for bias/outputbalancing!)

http://www.diystompboxes.com/smfforum/index.php?topic=19519.0
http://www.diystompboxes.com/smfforum/index.php?topic=20498.0

[StephenGiles]

The Yamaha E1010 used 4 MN3005, each individually biased, and the VCO clock, I think a BA617, was well buffered.
Stephen

[Hal]

you can build 50 (or at least one) CE-2... :-D

new tonepad layout!

[nelson]

you can build 50 (or at least one) CE-2... :-D

new tonepad layout!

I have an original.

I will build the FL-301.

I think building an analog delay out of MN3007 is probably too pointless. I could sell a few MN3007 and buy a couple of MN3005.

Or I could build a through zero flanger......

[A.S.P.]

information lost...