DOD Flanger/ Doubler r870 volume drop issue

[trixdropd]

My Dod R870 flanger / doubler sounds awesome, but using it in a true bypass loop shows that the unit has a volume loss issue. It has pots for in and out, but they cannot solve the issue.
i read about a fix for another flanger on here with the same issue. Any help would be wonderful.

if the unit has the mix knob all the way dry, the sound is proper, but as you make it wetter it gets quieter. my goal is to use this in a true bypass loop with my perfect setting at bring it in and out, which results in bringing it in being quieter than when it was out. It needs like an ouput gain adjustment or something.

here's the schematic:



Internal shots:

[Mark Hammer]

First, neat unit!

Second, the noise performance could probably improve by subbing lower-noise dual op-amps for the 4558s in the audio path.  Given the non-use of batteries, and the input resistances, you can probably expect to hear less hiss from an NE5532 or LM833.

Third, your question.  Since the intent of the circuit IS to introduce notches, which will, by definition, reduce the overall amplitude of the output, the goal is to find some way of compensating for that inherent loss of signal by adding gain somewhere.  The use of companding improves the performance of the device, but limits the points of intervention.  So, increasing the gain of the input stage will tend to have little impact on the blending, and adding gain to the op-amp stage used to invert the delay signal (the one with the 100k feedback resistor just to the right of the compander, where the regeneration signal is tapped from) will monkey with the tracking of the compander.

I'm stumped as far as simple component-value changes to the basic circuit go.

[oldschoolanalog]

It needs like an ouput gain adjustment or something.

Couldn't you just add some gain to those 2 output op amps?

[Mark Hammer]

That wouldn't affect the level difference between effect in/out.  However, if the bypassing is done elsewhere (as the OP suggests), then that  might do it.  Where I was stumped was in trying to think of something that would accomplish the balance using the switching on-board the unit itself.

[aziltz]

it looks like theres a simple buffer at the end before each out put.  perhaps you could build a sub-circuit that could replace that opamp and add some fixed boost to the output.  ie, perfboard a clean op-amp boost that plugs into the buffer sockets.

[trixdropd]

Thanks for the help so far guys... I jammed on this thing today and it needs a substantial boost in the output to make it useable in my situation. I am using it in a midi controlled looping system. I need either an output pot for channel a and b or a pot for each that lets me set the effected level to my dry sound. I'm not a wizard at modding these things so give me good details on what you mean on some of these things.

As far as getting lower noise, I'll worry about that in the future as the unit sounds great, but the volume issue is a huge one.

It would probably be to complex for me, but a real solution i think would be an output gain that raises as the mix of the effect goes up, or the effect itself needs a gain boost so that when it's mixed with dry it doesn't get quiet.

[Mark Hammer]

If the switching is external to the unit, then there is no reason why you could not adapt the output stages on the 870 to add a bit of gain.  Right now they are set for unity.  A simple gain of 1.5-2x would fix matters easily.

So, a 10k resistor in the feedback loop of each op-amp stage, and a 15k-22k resistor from the inverting pin to ground.

[trixdropd]

If the switching is external to the unit, then there is no reason why you could not adapt the output stages on the 870 to add a bit of gain.  Right now they are set for unity.  A simple gain of 1.5-2x would fix matters easily.

So, a 10k resistor in the feedback loop of each op-amp stage, and a 15k-22k resistor from the inverting pin to ground.

I'm not an expert here but this sounds doable. I'm gonna think on this a while and post what I think is the solution. I'd love to learn something out of this and get it fixed rather than just getting it fixed... Thanks for your help, it's helping me learn!! 

I'm still researching. opamps are hard for me to understand.

the pinout states:
1 out              1       8      vcc+
1 in -              2       7      2 out
1 in +             3       6      2 in -
vcc-               4       5       2 in +

So am I correct in thinking that I need a 10 k resister from out 1 to in 1 - and from out 2 to in 2 -? As far as the inverting pin resister to ground, which pin is it? I can't figure it out for sure. I'm guessing the vcc-?

[Mark Hammer]

Okay, here's what you do.  In this picture, you'll see the RC4558 that forms the two output stages.  I can tell its them because they go to a pair of 1k resistors, then to an electrolytic cap (which I can't see clearly, but I'll bet you it says .33uf) and 100 ohm resistor to ground.  The yellow caps just to the right are likely the 4.7uf units.  I can assume that because they a) have a 100k resistor to ground, and b) have blue wires from there running to the output jacks.

The output and "-" pins of the relevant op-amp pins are identified in red.  The nearest identifiable ground point is also identified in red.
To do the mod, you would do the following:
1) Pull the 4558 out....gently.
2) Cut the trace between pins 1 and 2, and between pins 6 and 7, on the copper side.
3) Solder a 10k resistor between pin 1 and 2, and another between pins 6 and 7, on the copper side.  If you use 1/8 resistors, you should be fine, and they will easily fit the available space.
4) Run a 15k-22k (I've shown 18k, but it's not written in stone) from pin 2 and pin 6 to that ground point.\5) Make sure the leads of the added resistors are not going to short out against anything.  You can do that with some heat shrink, or some insulation stripped off wire, or even a piece of tape.
5) Replace the 4558 with a better dual op-amp like a 5532 and you're good to go.

Note that the gain produced by that stage is determined by (Rf+Rg)/Rg, where Rf is the feedback resistor and Rg is the resistor to ground. (10k+15k)/15k will give you a gain of 1.67x, and 18k will give you a gain of 28k/18k n= 1.55.  So, use your judgment about just how much additional gain is needed to bring the output up to the level of the completely bypassed signal, and select an appropriate value for Rg.

[trixdropd]

Mark, you're dead right on all assumptions. I have checked the traces and verified everything you said. Money's tight at the moment, so i'm gonna wait on the new opamps, but I will upgrade them soon though.
I may give this a go later tonight. I have some 10k resisters and will just try 20k on the inverted outs. Is is possible to substitute  25k trimpots on the pcb I can adjust and forget in place of the 15-22k resisters?

I'll post some pics of the underside of the board for anyone curious. This unit is really cool and sounds just as good as the mxr flanger/ Doubler but much cheaper.

I'm probably gonna build bypass setup into this via a relay so you switch between effected and non effected sound. While you're around the thread could you tell me a suitable place to steal 5-9 v and maybe 20-30 ma on the board?

lastly,
thanks again for your awesome help Mark, and all the others. In addition to making my unit better, I believe i understand a lot better what op amps can do. I learned that different resister values between the out and the -in and ground and in- can amplify signal, or voltage depending on the limitations of the op amp. I have a ton to learn in my understanding, but guys like you help a ton, I can't thank you enough for that!!

Cheers!!

Jeremy

[Mark Hammer]

Is is possible to substitute  25k trimpots on the pcb I can adjust and forget in place of the 15-22k resisters?

Absolutely.  I was actually going to suggest a trimpot, but figured it might make things more complicated than necessary.  If you're up for it, it's the right way to go.

The principal reason why its right is not just because it will avoid trial and error resistor replacement, but because the two outputs are sum and difference, and these generally have different amplitudes from each other.  Consequently, one will need juuuuuuuust a little bit more gain than the other to have equal-level outputs.

The better op-amps certainly can wait.  I only mention them because they CAN make a difference, because there are so many reasonably priced alternatives these days, and because they are socketed.  But first, let's get this puppy working!

[trixdropd]

Is is possible to substitute  25k trimpots on the pcb I can adjust and forget in place of the 15-22k resisters?

Absolutely.  I was actually going to suggest a trimpot, but figured it might make things more complicated than necessary.  If you're up for it, it's the right way to go.

The principal reason why its right is not just because it will avoid trial and error resistor replacement, but because the two outputs are sum and difference, and these generally have different amplitudes from each other.  Consequently, one will need juuuuuuuust a little bit more gain than the other to have equal-level outputs.

The better op-amps certainly can wait.  I only mention them because they CAN make a difference, because there are so many reasonably priced alternatives these days, and because they are socketed.  But first, let's get this puppy working!

I'm holding off for a few days Till i can get the trim pots and then i'll do it right one time hopefully!!

i'm highly confident that this will work just like I had planned. I like the fact I can adjust the outputs to taste rather than point and shoot....  I will see this thread through to it's conclusion with an aftermath with soundclips and pron!!

[markusw]

Wow. Looks very tasty 

One question: why are two BBDs?

[Mark Hammer]

Why two?  One is used for flanging, and the other is used for the longer-delays required for "doubling".  Notice that there is only one clock, which feeds the SAD512 (512 stages) and the MN3005 (eight times as much delay with 4096 stages).

[markusw]

Why two?  One is used for flanging, and the other is used for the longer-delays required for "doubling".  Notice that there is only one clock, which feeds the SAD512 (512 stages) and the MN3005 (eight times as much delay with 4096 stages).

Thanks a lot! That's what the 10/75 ms switch is for, I suppose 

[trixdropd]

Mark, I finally tried messing with this tonight. Number 1, I'm gonna add the parts to a socket, plug that into the existing socket, and then plug the opamp into the new socket.
When it comes to cutting the traces between 1,2 and 6,7 I assume pin 1 and 7 stay connected to the resisters as marked in my pic and pin 1 and 7 goto ground via resister? 

[trixdropd]

Mark, I finally tried messing with this tonight. Number 1, I'm gonna add the parts to a socket, plug that into the existing socket, and then plug the opamp into the new socket.
When it comes to cutting the traces between 1,2 and 6,7 I assume pin 1 and 7 stay connected to the resisters as marked in my pic and pin 1 and 7 goto ground via resister? 

Bump, I wanna finish this tonight!!

[Mark Hammer]

Well, I hope it's still tonight where you are!

Yes, the 1k resistors stay in contact with pins 1 and 7.  The schematic shows that the outputs of those op-amps go to 1k series resistors and caps to ground, so we'll just leave them there.  All we are changing is the gain of those op-amps, which is presently set at 1x.

[trixdropd]

Well, I hope it's still tonight where you are!

Yes, the 1k resistors stay in contact with pins 1 and 7.  The schematic shows that the outputs of those op-amps go to 1k series resistors and caps to ground, so we'll just leave them there.  All we are changing is the gain of those op-amps, which is presently set at 1x.

Thanks for the replay Mark!! It still is tonight.. I'll give this a go in the next little bit. Thanks again for the help!!

[trixdropd]

Mod complete. It seems to work like a charm. Couldn't I have used just 1 trimpot though instead of 2?

I was able to mount the resisters and the trim pots to a socket, which I cut the leg of pin2 2 and 6 off of so they can't connect when i attach it. I didn't need to cut any traces, and the only mod to the dod is soldering the ground wire....