A-150-8 Octal Switch Bleed Mods?

[gtrmstr53]

Hi there,

I recently picked up a used Doepfer A-150-8 Octal Switch, and noticed that it has a noticeable amount of bleed/crosstalk. For example, if I use a switch in 2-input/1-output mode, and then patch an oscillator to input 1 and connect the output to my headphones, I'll hear the oscillator even when the switch is set to input 2. I looked at it on a Jones O'Tool o-scope, and the scope barely registers a signal when it's set to input 2, so there shouldn't be much of an issue with triggers/gates/cvs. However, it would be nice to reduce the bleed so I could use it for audio and V/oct signals too.

Anyways, does anyone know of any mods I could do to this module to reduce the bleed? It comes with 4x DG409 ICs, which the datasheet says are multiplexers. Are there any different ICs that I could replace these with?

Let me know if you have any thoughts on this, or if there's any other useful info I could share. I'm pretty new to switching circuits, so any general resources could be helpful too.

Thanks!

[EATyourGUITAR]

definitely the single sided PCB with no ground plane that is the problem. the DG chips are good chips. you need more shielding and more distance between traces. reducing the current through the traces also helps. this is all in the PCB design. the only way to fix it is to make a new PCB design.

[guest]

as EATyourGUITAR points out, it is most likely capacitve coupling between the signals, and this is probably not happening inside of the IC (although there is capacitive coupling that occurs within the IC). the only thing that can be done about this (assuming youre not going to lay out a new PCB!) is to reduce the impedance on the other side of those capacitors. capacitive coupling is just a bunch of high pass filters, so if you can push the cutoff frequency way up by having smaller 'R' values (you cant change the 'C' values) you can attenuate that bleedthrough.

unfortunately, without a schematic, i cant make any more specific reccomendations.

[gtrmstr53]

definitely the single sided PCB with no ground plane that is the problem. the DG chips are good chips. you need more shielding and more distance between traces. reducing the current through the traces also helps. this is all in the PCB design. the only way to fix it is to make a new PCB design.

as EATyourGUITAR points out, it is most likely capacitve coupling between the signals, and this is probably not happening inside of the IC (although there is capacitive coupling that occurs within the IC). the only thing that can be done about this (assuming youre not going to lay out a new PCB!) is to reduce the impedance on the other side of those capacitors. capacitive coupling is just a bunch of high pass filters, so if you can push the cutoff frequency way up by having smaller 'R' values (you cant change the 'C' values) you can attenuate that bleedthrough.

unfortunately, without a schematic, i cant make any more specific reccomendations.

Thanks so much for the responses. I was assuming it wouldn't be something easy, and unfortunately I haven't been able to find a schematic. I have reached out to Doepfer to see if they know of anything, so I'll report back if they know of anything. Otherwise, I'll probably plan on sticking to trigs/gates/CVs with it.

Best!

[guest]

can you tell me what other ICs are on there? i found schematics for the A-150 (the older version of this) and its all just passive switching, no opamps, which makes it harder.

[guest]

also, try this for a test:

plug one channel into your vco, and plug the other channel into a CV source set to 0V (or just ground the tip of the inserted cable). then see what the crosstalk is like when you listen to the CV channel. if the crosstalk is between the input and output jacks, then there isnt much that can be done. if its between the input jacks, that might be fixable.

[gtrmstr53]

can you tell me what other ICs are on there? i found schematics for the A-150 (the older version of this) and its all just passive switching, no opamps, which makes it harder.

For the main board, the only other obvious IC is a PIC18F27K40 microcontroller, which I would assume I wouldn't want to replace (at least not unless I have access to the source code to edit and reflash). There's also a 1x9 pin component that I don't recognize visually. The pcb marking for it says 'RN1 8x10k', so my guess is it's a 'resistor network' or 'resistor something'. Other than these and the 4 DG409s, there's a 78L05 voltage regulator on the main board.

For the controls board, there's a BC547 transistor above each of the switches along the right-hand edge of the faceplate.

I've attached a picture of the top of the main board. Let me know if you could use any other pictures.

[gtrmstr53]

also, try this for a test:

plug one channel into your vco, and plug the other channel into a CV source set to 0V (or just ground the tip of the inserted cable). then see what the crosstalk is like when you listen to the CV channel. if the crosstalk is between the input and output jacks, then there isnt much that can be done. if its between the input jacks, that might be fixable.

I'll try this when I get home later today, and let you know how it goes.

[EATyourGUITAR]

good news, it is a two layer board. bad news, it has super long traces in all the wrong places and that board to board connector can not possibly be shielded. everything perfectly parallel for maximum bleed. when you want to reduce costs, follow doepfer.

[gtrmstr53]

also, try this for a test:

plug one channel into your vco, and plug the other channel into a CV source set to 0V (or just ground the tip of the inserted cable). then see what the crosstalk is like when you listen to the CV channel. if the crosstalk is between the input and output jacks, then there isnt much that can be done. if its between the input jacks, that might be fixable.

I'll try this when I get home later today, and let you know how it goes.

So, I did like you said and plugged a VCO into a mixer into I/O 1 and plugged O/I into my headphone outs. Then, I set an output from my Fonotronik Cascade to 0V and connected it to I/O 2. I had the switch set to listen to I/O 1 and set the gain on the VCO's mixer to a loud but listenable volume. When I switched to listen to the CV on I/O 2, I could notice the slightest bit of bleed from I/O 1 if I listened really attententively, but it wasn't anything worse than the noise floor on some of my other modules. Then, I unplugged the CV from I/O 2 and the bleed was significantly higher.

If I'm catching the drift of your original post, I'm guessing this behaviour means that the two I/Os are bleeding into one another and not into the O/I, because grounding one of the I/Os pulls the 'bleed' signal down to ground (or approximately so).

Anyways, let me know if that's on the right track or not. And any pointers about what to look into now would be much appreciated.

[gtrmstr53]

good news, it is a two layer board. bad news, it has super long traces in all the wrong places and that board to board connector can not possibly be shielded. everything perfectly parallel for maximum bleed. when you want to reduce costs, follow doepfer.

So all things considered you're saying the bad news wins out, at least when it comes to using shielding as a fix? Is that following you right?

[Joe_D]

...it has super long traces in all the wrong places...

Wasn’t that a country hit?

It might have fit in on this record:

AFC61B01-463B-4018-BCD5-E4DA65DF5CE1.jpeg (72.33 KiB) Viewed 71 times

https://bobmoogfoundation.myshopify.com ... l-trythall

(Apologies, Gil Trythall was one of my modular teachers a very long time ago)

I am glad to report that my A-150-8’s bleed rarely bothers me. Of course, after this thread, I’ll be on the lookout (er, listenout) for it and will consequently probably be bothered more.

[guest]

thats good news, because that seems fixable. so check the input jacks, and see if they have normalling switches on them. if they do, connect those switches to ground.

[gtrmstr53]

thats good news, because that seems fixable. so check the input jacks, and see if they have normalling switches on them. if they do, connect those switches to ground.

That makes sense. The jacks have 4 solder lugs, so I'd guess they have normalling switches. But it'll probably be a week or two before I have time for that much soldering so I'll let you know how it goes. Thanks for the help!

[mskala]

Are you sure that the "bleed" is even happening internally to the switch module at all? Try disconnecting the patch cable from the oscillator to the switch module and just listening to the switch output in that state. If you can still hear the "bleed," then it's coming through some other path instead of the patch cable, most likely through the power system. That scenario is common enough that it's the first thing I always want to rule out when the topic of "bleed" comes up. Coupling between traces on a PCB is a real effect, but it's not such a big effect at audio frequencies as people sometimes imagine, and not the first thing I would suspect.

[gtrmstr53]

Are you sure that the "bleed" is even happening internally to the switch module at all? Try disconnecting the patch cable from the oscillator to the switch module and just listening to the switch output in that state. If you can still hear the "bleed," then it's coming through some other path instead of the patch cable, most likely through the power system. That scenario is common enough that it's the first thing I always want to rule out when the topic of "bleed" comes up. Coupling between traces on a PCB is a real effect, but it's not such a big effect at audio frequencies as people sometimes imagine, and not the first thing I would suspect.

Yea, it definitely seems to be internal to the switch. I tried connecting nothing to the two I/O jacks and routing the O/I jack to my headphones, and didn't notice anything. Then, when I plugged the oscillator into the switch and switched away from it, I heard some bleed. The bleed was the same pitch as my oscillator, and it reacted like you'd expect when I fiddled with the oscillators's course tuning knob.