Analog Delay - Modifying the AD-3208 BBD Stages

[Kevin Mitchell]

I'm back with more silly ideas for you guys. I still have a heap of mn3204 BBDs and I've decided to use many of them for an analog delay project.

I've found the AD-3208 project where 2 mn3208 BBDs are used to get 4096 stages.
AD-3208 Schem.

Here's a little editing to show where my head is;


I believe that would work as is so we'll continue.
I'd really like to double it up for 8192 stages of BBD. To do that I'll need to use another mn3102 for the next 8 mn3204 BBDs since one is only good for up to 4096 stages. I can make a loose guess on how to add the extra 8 mn3204s but not the additional mn3102.


I found this from an old thread regarding adding more stages and needing an additional mn3102;

schematic: just copy paste the 3102, and 3208 parts. Inputs into the new 3102 are the same, but outputs go to the new 3208, and the output from the 2nd original 3208 goes to the input of the new 3208, just like before.

That quote was from about 12 years ago and I can't trust what I'd make of it so I'm asking you guys! How would I go about doubling up the stages using another 8 mn3204s and the required additional mn3102?

This concept was inspired by the Maxon AD-999 and if anyone can source a schematic for that or better - the pro version please send it my way!!

[Scruffie]

...16 x MN3204 and 32 x trimmers just for the sake of using them? I'll trade you 2 x 3205 if you want.

You can make life easier and just use a 4049 to buffer the 3102 if you want though, 3 inverters in parallel per clock pulse.

If you insist on using them for a delay, might I suggest you look at the schematic for the 4 knob deluxe memory man that used 4 x SAD1024 which is 8 x 3204 effectively, you'd have to do a bit of work but it wouldn't be too hard and you should get about 400mS of delay.

[Kevin Mitchell]

Yes I know... A bit ridiculous lol 

The whole point is I have these BBDs and I would like to use them however I can. Now that I think about it... some finances..

mn3204 is about $10 each - according to the history of it's online sales.
mn3205 (cool audio clone) is about $5 each from online vendors.

So we're looking at about $160 vs $10... along with much more to dial in (trimmers).

Efficient... no. Practical... to an extent from my point. Why is there 8 3207s in the AD-999 instead of 2 3205s or 4 3208s? It made sense at the time for them? Well... for a larger production run obviously.

Please share more regarding using a 4049 with the 3102 for more than 4096 stages if you could. Any examples floating about?

I will definitely check out the deluxe memory man I've always wanted one. Though with 400ms? That's all well but I'd like to double that 

[Mark Hammer]

Maxon used 8 1024-stage chips because they HAD to.  Nobody was making 4096-stage devices at the time, or had a large-enough supply of NOS to be able to bring an 8192-stage pedal to market.

Make yourself a Dimension C, using two of the chips.  Or better yet, look at some of the older Roland synths that used a trio of MN3004s to get a really thick chorus, and duplicate that.  And if the leftover chips are driving you nuts, make 2 or 3 of the multi-phase chorus units.  I'm confident that ifthey replicate the old synths well enough, somebody will be happy to offer you a few hundred for one of those units.

The old Korg Polysix used a trio of MN3004s for its chorus.  You can see the circuit here:  http://www.oldcrows.net/~oldcrow/synth/korg/polysix/manuals/service/p6smm7.pdf

[Kevin Mitchell]

Thanks for the info!

That synth chorus is a bit much for me right now 

What I think I'll do is snag a couple mn3205 clones(v3205), tinker and modify the AD-3208 and eventually lean into working on a 9 volt Memory Man clone with some cool mods. Eventually I'd like to add an additional BBD sections with some 3204s to increase the max delay time as much as I desire.

I've read some reviews that people have had trouble with distortion using the v3205s. Though they are the best for my wallet I'll see about making the best of them.

[armdnrdy]

I've read some reviews that people have had trouble with distortion using the v3205s.

Even though the V3205s are supposed to be "clones" of the MN3205...they seem to output a hotter signal.

This is the fix for the AD-3205.

http://www.diystompboxes.com/smfforum/index.php?topic=100645.0

[Kevin Mitchell]

Thanks Larry. You're the man! 

Hey Larry, how did you go about the mn3102 section of the schematic? This question is relevant to why I started this topic. One mn3102 is good for 4096 stages of BBD. With two mn3205s the clock would have to be twice as capable.

Using the two v3205 BBDs, were you able to get over 400ms of delay as expected?

Any tips on how to go about the clock (expanding the clock) after adding whatever number of mn3204s I use?

[Fender3D]

Loading on clocks terminal is mainly a matter of pF (which you got on BBD's clock pins), so you should drive 8K BBD stages with 1 MN3102 no matter how many BBD chips you use to achieve that...

Using several MN3102 you should link 1st chip's pin 5 with other chips' pin 7 (leaving pins 5 and 6 unconnected) to avoid heterodyning, or you may use half 4013 or any clock divisor with AND, NAND or whatever logic port you may love...

[Kevin Mitchell]

I'm having a little trouble understanding. Mind you, I'm definitely not yet well enough educated to easily digest some info.

The mn3102 data sheet says "direct driving capability to 4096 low voltage BBD" which is why I guessed there's a limit along with some comments made on the forum over a decade ago.

You said one could do the job so then why would I need two? When it comes to the clock I'd like it to be small and efficient and work so I get the maximum delay possible. I just don't know what it should look like yet.

[armdnrdy]

Thanks Larry. You're the man! 

Hey Larry, how did you go about the mn3102 section of the schematic? This question is relevant to why I started this topic. One mn3102 is good for 4096 stages of BBD. With two mn3205s the clock would have to be twice as capable.

Using the two v3105 BBDs, were you able to get over 400ms of delay as expected?

Any tips on how to go about the clock (expanding the clock) after adding whatever number of mn3204s I use?

I used the "stock" 1 clock circuit. It works for two 4096 stage BBDs but...there are probably limitations.
Looking at the MN3102 data sheet, the frequency and power consumption is dependent upon capacitance.
Loading down the clock might lesson the delay time.
I didn't notice any difference in delay between the 2 clock AD-900 that I built and the 2 MN3205 AD3205 but...I never A/B them side by side.
Best to use two clocks. Look at the Maxon AD-900 for ideas for using two BBDs in series.

[Scruffie]

The issue of overloading a clock isn't a delay time thing, it's the clock square waves getting rounded off (from the capacitance Fender3D points to) and the BBDs not functioning correctly.

As I said, a 4049 buffer is a simple solution to help the clock deal with higher loading, if you look the factory schematic for the 5 knob deluxe memory man you will see it buffering the 4047 (as it notes, this was omitted on production units as it wasn't needed) you can also see it done in the Yamaha E1010 rack delay, echo flanger (although the schematic shows a 4009 when a 4049 was used in production units) etc. etc.

A couple of delay units do sync 2 x 3102/3101, can't remember which ones right now but a 4049 is cheaper and easier to implement IMO.

[Kevin Mitchell]

Thanks guys!

Scruffie, great intel! I did observe the use of a 4049 on a 4047 and the notes of it being omitted. I also happen to have some around (finding things in the parts bin is always a good feeling  )

I'm not sure how I'd use a 4049 with a 3102. But my thoughts are what would work better, what would take up the least space and especially what I could wrap my head around to avoid complications.

[armdnrdy]

This is the part of the 3102 data sheet that I was referring to:

[Scruffie]

Thanks guys!

Scruffie, great intel! I did observe the use of a 4049 on a 4047 and the notes of it being omitted. I also happen to have some around (finding things in the parts bin is always a good feeling  )

I'm not sure how I'd use a 4049 with a 3102. But my thoughts are what would work better, what would take up the least space and especially what I could wrap my head around to avoid complications.

Look at the datasheet for the 4049, 6 inverters, for CP1 you run in to 3 of the inverter inputs, connect the 3 outputs together and that's where you take your CP1 from and then for CP2 you do the same thing with the other 3 inverters. No extra parts required.

[Scruffie]

This is the part of the 3102 data sheet that I was referring to:

Power consumption goes up with clock frequency yes and the maximum clock (shorter delays) the 3102 can get to with high loading drops but the point is a single 3102 isn't specified to be able to drive a load of 8192 stages that Kevin is aiming for, it's specified to only drive 4096. Can it actually drive 8192 stages (or 5n6 of clock capacitance) without clock rounding from a long delay to a short delay? Quite possibly but that'd be for a scope to decide.

[Mark Hammer]

As I understand it, they are spec'd to go up as high as 8192 stages (5600pf total capacitance), but no higher.  At least not with anything in between the clock generator and the BBD to pump a little more current into the clock pulse.

[Kevin Mitchell]

As I understand it, they are spec'd to go up as high as 8192 stages (5600pf total capacitance), but no higher.  At least not with anything in between the clock generator and the BBD to pump a little more current into the clock pulse.

Right on. I was wondering why some MMD clones (something like the Aqua Boy) could have 8192 BBD stages and only one 3102 for the clock - nothing else.

So theoretically I could use one for 8192 stages but once I start stacking on mn3204s to increase delay time I would need to tweak the clock a little - adding in a CD4049 or another 3102.

Rock on guys. I'll have a breadboard sound demo for the AD-3205(x2) + additional stages once I have everything to dig in.

[Scruffie]

As I understand it, they are spec'd to go up as high as 8192 stages (5600pf total capacitance), but no higher.  At least not with anything in between the clock generator and the BBD to pump a little more current into the clock pulse.

The 3101 says it can do 8192 stages but the 3102 actually only states 4096 that I can see.

But yeah it's been used lots of times to drive 8192 stages.

[armdnrdy]

So theoretically I could use one for 8192 stages but once I start stacking on mn3204s to increase delay time I would need to tweak the clock a little - adding in a CD4049 or another 3102.

I personally wouldn't bother/waste using a MN3204 for a delay.

An extra 512 stages isn't going to be very noticeable.

I would go with a 3207 or 3208 but...you will probably run into sound quality degradation.

That issue is inherent in longer delays. Just my three cents! 

[Kevin Mitchell]

Don't tell me how to live my life, Larry!

Just kidding 

At ~25ms delay with each 3204 I'd notice the difference after adding just a few though it would be very subtle. Also I could keep adding them until I start to experience noticeable decrease in sound quality. Could use another 3205 or some 3208s or a couple of 3207s or a bunch of 3204s. I'll likely end up using some of each. We'll see.

I'd have to find out how many stages is too many.