stuck debugging flanger BBD

[BillyBoy]

Hey, flanger and MN3007 experts out there.  I've goofed something up and am stuck trying to find the fix.  I'm looking for any hints on where to go next.

I have been debugging on this for days, so I have tried many, many things and have a lot of interesting notes and details I could share.  But I keep ending up stuck at the same place, so I need to get kicked out of my rut.  I'll try to leave out as much as possible here but the specific details I'm providing are all a part of my problem.  Feel free to ask anything else.  I will not be offended - embarrassed quite possibly, thankful most definitely, but not offended :^)

First, I'm building a Deluxe Electric Mistress.  I have my own boards.  The LFO is on a separate board from the main FX circuit.  I have built a few of these recently, and they all worked first time.  Then I made some board changes to try some mods in the mixing and output section.  While I was at it, I switched from using a single bias trimmer for opamps and BBD to separate ones, adding the requisite bias resistor, dc blocking cap, etc.  I also added ground plane pours on both sides of the board.  Now, when I build the board, I get no MN3007 output no matter how I try to bias it.  I also built a new LFO board to go with the new FX board.  All I did on the new LFO board was add a CLR and blinking LED, and ground plane pours on both sides of the board.  Those are the only changes between the working and non-working boards.

Since I have working versions using earlier versions of the board, I can try out MN3007 chips.  I am using known good MN3007 chips that work fine on other boards.  I can also switch to one of my earlier known good LFO boards, to eliminate the entire new LFO as a problem.  However, my new LFO board does work fine with an earlier FX board.  An earlier known good LFO board does NOT make my new FX board work.  Just FYI, the LFO is used by some Electric Mistress or Deluxe Electric Mistress, I can't remember which.  It is very similar to the one in Current Lover, and has the CD4049 on the end to drive the BBD.  And the timing cap is adjusted to get the speeds I want for my MN3007.  Anyway, because of this info, I believe I can say that neither the LFO or MN3007 are the problem.   

Here's the data that has me stuck.  These are the measurements taken directly on the pins of the MN3007 in my non-working board.

                pin 1 (ground):                  14.50V 
                pin 2 (clock 1):                   aprox 8.0 – 9.5V as the LFO varies, with Filter Matrix switched to manual, I get 30KHz at the extreme slow end and 340K at the extreme fast end (sometimes I go as wide as 25-440K).  I can hear the clock throbbing.
                pin 3 (input):                      my bias lets me adjust from 0v to 14V (I measured it).  I have clear, strong audio signal at this pin
                pin 4 (Vgg):                         goes to a 10uF cap to ground (oriented correctly), and usually has about 1.0 – 1.5V, sometimes as low as a few 100 mV if I've been measuring it and discharging the cap
                pin 5 (V+):                           0V
                pin 6 (clock 2):                   identical to pin in all measurements
                pin 7 (output):                   0V
                pin 8 (output):                   0V

When I search for output signal while adjusting the bias, once in a great while I get a little heavily distorted clock noise when the bias is somewhere near 4.5V.  But it goes away very quickly, like it is being drained to ground or fading out somehow.  I cannot make it happen twice.  Once it stops, then I can't get it back until "later" – letting it sit, power cycling, something will eventually "reset" it and I'll hear it again briefly.

This always reminds me of the new ground planes I added.  I've gone back and visually followed every trace and hole looking for ground plane issues on the FX board (new unpopulated one).  I've even done continuity checks on every single hole.  All seems well.  On my board, there is a trace connecting pins 7 and 8, which then goes on to the rest of my circuit through a cap.  Not only is there no sound at pins 7/8, there is nothing at or past the cap (taking all my new board mods out of play except for the biasing and ground plane changes).  Right now I am testing a board that has nothing populated past that cap. 

Being suspicious of a bad ground plane connection, I tried 2 things.  First, I removed the MN3007 completely and put a jumper from pin 3 (input) to (pin 7).  I get audio at pin 7 and after the cap.  The board is not grounding it out.

I also reseated the MN3007, but with pins 7 and 8 bent out so they did not seat in the socket.  Then I tried to bias.  Almost, but not quite, the same thing happened as before.  I actually got audio on the output pins very briefly as I turned the bias voltage down toward 4.5 – 5V.  But it faded away just like the "distorted clock noise" did with those pins fully socketed.  Could be a fluke, or it could be that whatever is happening just happens a little faster when those 2 pins are socketed.  I expect the "heavily distorted clock noise" is really faint, heavily distorted audio.

I also replaced the 10uF cap from Vgg to ground with a jumper.  Since I've done that, I haven't heard even the little bit of fading clock/signal noise that I used to have.  But I didn't always have that noise before, so this may not mean much.  The only difference I noticed after this change is that instead of 0V on the output pins, I have around 150mV.  But that may be coincidental.  I think I've seen tiny voltages like that from time to time, which keeps me thinking that I'm somehow draining to ground somewhere.  But that somewhere seems to be between pin 3 and pin 7/8...

I also backed out my biasing change.  I took out the resistor on the input line to my bias trimmer and replaced the DC blocking cap I had added with a jumper.  My bias is now coming in on the output from an opamp, as it does in my working board.  No difference at all, other than which biasing trimmer I use to search for signal.

Also, note that all my MN3007 measurements (except the missing output on pin 7 and pin are identical to measurements I've taken from one of my working builds on the previous board version.

I've built a lot of pedals, so I'm careful and reasonably proficient, but still have much to learn.  I've checked all my usual methods of goofing up, many times, carefully.  I've removed/replaced components, reflowed solder, looked for bridges/bad joints, and lots and lots of debugging.  I've even built multiple of these non-functioning boards to further reduce the chances of soldering errors and bad components.  I'm only populating the minimum part of the circuit needed to test for MN3007 output.  Everywhere I have probed or tested, I seem to have good/expected values with the exception of those two output pins.  This is true for each board and experiment I've tried.  I haven't found anything wrong anywhere.  That just means I'm overlooking or misunderstanding something, of course.  I've even been suspicious of my probe, and tested it :^)

So that's the hole I'm stuck in.  Everything seems good for my known good MN3007, yet I can't get it to give me any output.  Everything feels like some sort of grounding problem, or just a lack of effort on the part of my MN3007 :^)  I'm sure there is something I'm not measuring that matters on the MN3007 pins.  What could I look for there that would shed light on why the MN3007 refuses to pass signal?

[Scruffie]

Schematic please.

Also, you say Deluxe Electric Mistress, people call the 9V and Reissue that but do you mean the actual vintage version? If so as you lack any output voltage i'd hazard you've left the gain trim as pull down instead of pull up.

[BillyBoy]

Hi Scruffie, and thanks for having a look.

The DEM in question is the late 90's v5.  Here's a link to the schematic I used (except the LFO) at Ralf Metzger's Electric Mistress site: 

http://www.metzgerralf.de/elekt/stomp/mistress/images/deluxe-electric-mistress-v5-schematic.gif

The only differences I have from the front end of this are: 
using 15V regulator on 18V power feed for about 14.5V instead of 12V, and instead of the (2) 1M resistors to the bias voltage on the input, I have a 100K going to my bias voltage trimmer that lets me adjust from 0V to 14.5V.  Component values all the same.  The output of the second opamp stage goes directly to the input pin on the MN3007.  This is what works on my previous builds on my original boards. In my new board I added the 100n cap and bias setup like Current Lover has on the input to the BBD, but it doesn't change my problem if I jumper that cap and remove the bias. 

Right now on the board I don't have anything populated past the BBD output pins, so it is just the front end and MN3007 - no mixer, output, gain trimmer, feedback, etc.  Just the two opamp stages feeding the MN3007.  Trying to keep it simple and isolated.  Connections to the BBD are as described above.  I think the LFO is the same as the Current Lover.  I think you and I traded messages about Electric Mistress LFO's a long while back and I settled on that one.  I've used it successfully ever since.

I'm guessing I've done something bad when I added the ground planes to the board, since that is the only change from the working board to the partially populated, non-working board.  I have also tried populating everything after the BBD, and that works as well as it can without wet signal coming out of the BBD.

[BillyBoy]

Oops - I think I hit a "tab" key or something while I was trying to make that more readable and it prematurely posted on me.  Other than messy formatting I think the info is OK.

[Scruffie]

The BBD wont produce an output with the pins just floating, they both need a resistor pulling them up to V+.

The LFO is a bit different to the current lover version in that actually, the swing changes with rate.

I really need to see how you've implemented the MN3007 to see if you've done it right to tell you if the problem is there or elsewhere.

[BillyBoy]

OK - so that's my first problem.  I'm testing the wrong way with my output pins floating right now.  Now I have to 'fess up about what started me down that path.  When I moved some stuff around on the layout of the new board, I reconnected to bias power instead of V+ after the resistor on my output pins, plus I was trying to remove my mods after that point from the problem by not populating that part and introducing unknowns. 

So now I'm guessing that my only problem is that incorrect connection to bias instead of V+ (which I disconnected since it was wrong).  Then I inadvertently created an insurmountable fake problem since all subsequent measurements were taken with no V+ on the output pins    I'll never get anywhere that way, right?  I tricked myself into thinking the problem was before the output pins, not after.

For the record, barring the errant V+ connection, my MN3007 is implemented as it is in Current Lover, except I tied the output pins together and used a single resistor on the combined output.  My other mods are optional things which I can jump through or around to get things going.

If that's correct, then I'll see if I can hack my board to get to V+ instead of bias power in an ugly way.  If that clears the problem, then I'll get new boards with that fixed.

Does that sound about right?

[Scruffie]

So you used the NPN buffer from the current lover, not the PNP set up from the V5? Should be fine then yeah.

[BillyBoy]

Yup, 2N5089. 

Thanks for quickly kicking me out of my rut.  In a previous life, I was a software engineer.  Debugging software and electronics often benefit from some of the same skills and approaches, but once in a while "software thinking" simply doesn't work in the world of electrons.  Like this time.  And when you're thinking the wrong way, it doesn't really matter how hard you think :^)

[BillyBoy]

Your first comment got me thinking about something separate, but related...

You said I might have left the gain trimmer connected to ground instead of power.  Since I don't have a gain trimmer, I'm guessing that refers to the gain trimmer in the earlier DEM models (v1 – v3 at Metzger's site).  If I built that version, would I still want to add the pullup resistor to V+ on the output pins and the NPN buffer, followed by the gain trimmer connected to ground to get a nice bias level for the opamp?  Or skip the buffer and hook the gain trimmer to V+ for pullup on the output pins?