fOXX Phaser Stages

[njkmonty]

sounds great  !
you did well fitting all that into a 1590b!

[nickbungus]

I know this guy who keeps forcing me down the SMD route.

[nickbungus]

I've had a bit of info leaked to me from non other than Nigel Knight from Knight Audio Technologies.

Apparently, he has fixed Brian Mays at some point in time.

Any way heres what I now know.

They are all series cascaded, but there is no regenerative feedback loop. The output of the phase shift stages are just mixed back with the input signal. LFO is pretty much the same as the second gen unit.

So I asked what the keys differences are and he said

Yes. Lose the feedback and mix with an op amp.

Great help!!  I've got somethings to go on..  Not sure exactly what mix with an op amp means

[anotherjim]

I'd assume the opamp mix is the same as the 6 stage where the last stage feeds into the wet/dry Z2 pin9 mixer via C21.
As it stands, the wet/dry mix is 50:50 because R45 & R48 are both 100k. It occurs to me that the phase stages might not give perfect unity gain all through so many stages, so 50:50 mix might not be optimum. I'd be tempted to make R48 a 47k in series with a 100k trim pot, so you can tweak for the best mix.

[nickbungus]

Thats great Jim.  I'll be building one this week to test it out.

What about losing the Feedback Regen loop?  Just omit R2 and R7?

[Kipper4]

I have a 6 stage Causality with wet dry and wet only switch.
Suits many things.

[anotherjim]

What about losing the Feedback Regen loop?  Just omit R2 and R7?

Yes, that's all there is.
As Rich mentioned, there are a few tricks to make these more versatile - if you don't want to be one of those bad people who just copy schematics

[nickbungus]

I don't know if you can remember Jim but you recommended adding a 100nf into the feedback loop to stop dc.  Is this now not an issue or do I need to put it somewhere else?

[anotherjim]

Ah, it was the input op-amps bias voltage getting in the output I was worried about. Without the feedback connection, it isn't a concern so no extra cap there.

[nickbungus]

I noticed on the gut shot photo there is still a pot next to the rotary switch.  Seeing how the feedback has been taken out I asked Nigel Knight what the second pot was for.  He replied

You only need two pots. Speed and Depth. The depth controls the amount of LFO that is mixed with the DC bias to the FET's.

Any help on how to implement this would be much appreciated?

Is this R43?

[nickbungus]

Hmm R29 or r43?  I'm struggling as to where to put the depth pot, is at the gate or the source of the fets?

[anotherjim]

Depth of LFO, I'd go for R38, try replacing with a 1M pot, but if the control range isn't perfect, then some value of pot in series with some fixed resistor to give the best minimum to maximum range. In fact, you could just replace R38 with the R2 + R7 from the original feedback circuit.

[nickbungus]

Well I was planning to do this once I had finished building a stock layout.



So i thought I would run it by Nigel Knight and he came back with the following:

I wouldn't put the depth control in the opamp feedback loop as that will also change the bias to the FET's. Just put it in as a level control before the 2M2 input resistor. There is no point in having a wet/dry mix as both signal sides need to be the same amplitude for the phaser to work. I'd also lose the two 1 meg pots and feedback loop at the top

So after a bit of confusion on my behalf, he said the way to do the depth mod is as follows:

take the LFO to the top of a 100K pot, bottom to ground, wiper to the 2M2

So I've got lots of stuff to play with and find out what works best

[anotherjim]

Ahhh now...
The wet/dry I suggested, as drawn above, was meant to be an internal trimmer, not a panel pot. The intention is to get maximum phase depth which is when wet & dry are equal amplitude, and you only need to tweak one of them to even it up.

The matter of adjusting sweep depth and affecting FET bias. The bias voltage is dominated by the bias trimmer setting the voltage on the +in of the opamp. If the feedback resistance between the amp out and -in pin was zero, the output to FET's will be exactly the bias setting and nothing to the left of the 2.2M will have any effect. As it is, there is a fixed 1M feedback resistor, which means there is a built-in offset on the bias coming from everything to the left of the 2.2M input resistor and half of that offset voltage will be subtracted from the bias setting. Reducing the feedback resistance with a pot will not only lower the LFO depth but reduce offset, not increase it. It should work as drawn and, as before, the range of LFO sweep should still depend on the bias trimmer setting.

The option of fitting a pot to the left of 2.2M acting as a volume control on the control will move the offset towards 0v depending on its setting - which will also change the bias. Of course, this might have been what Foxx did. Manufacturers pick whatever works well enough, even if there is another way to do it.

Doesn't seem to be much love for having feedback. But adding extra stages doesn't mean that the output has to be fixed at the last stage, and neither does feedback.  For all I know, feedback might work well in a 12 stage if it comes from the original 6th stage?

[nickbungus]

Thanks Jim.  I'm soldering a board right now.   I'm going to try it all and report back my findings.  Maybe do some YouTube videos as I go.

[nickbungus]

Doesn't seem to be much love for having feedback.   

That's what I'm not getting.  On the 6 stage, to get the Brian May sound you need to crank the feedback.  To not have it seems odd but he's adamant the v1 as used by BM doesn't have it.

[nickbungus]

Its getting there.  I've had to make a couple of changes.  Firstly I wasn't getting a nice clean sweep.  I changed the notorious R34 from 2M2 to 6M8.  The 6 stage schematic has it down as a 2M2 but Larry (Armdnrdy) reverse engineered one and it was a 6M8.  The sweep is now much more even.

Secondly, I tried the depth mod as suggested by Nigel Knight.  It just works but I would like more depth so I will be trying Jims suggestion next.

Lastly there was such a volume drop and it also sounded a bit too tinny.   Normally my memory is shocking (usually when my wife tells me things) but I remembered this from the days of yore.  (Jim, can you believe you told me this four years ago!).

There is an easy way to boost the output. R49 could be increased. 220k would be a bit more than 6dB boost. There is a tone change with that though. R49 & C20 form a low pass filter and increasing R49 may well make it sound darker. As designed, the filter is cutting over 7khz, which doesn't really affect guitar tone, to keep it similar, C20 would need to be halved for every doubling of R49.

Not only did it boost the volume to parity, but it got rid of the tinniness. 

Jury is still out on the feedback/regen loop

[anotherjim]

Well, my idea for the depth control can only reduce the sweep. Increasing that 2M2 is also reducing sweep. So it might well need to be not so large as 6M8. We are limited by the size of control pots too, I think the highest I ever had was 2M2.
It will always interact with the bias setting so it's going to be a judgement call.
To experiment with feedback, I would use a 100nF cap in series with the Q control and try feedback via that from each of the phase stage opamp outputs in turn.

[nickbungus]

I remember once trying a 4M7 and getting a really nice sweep but that was with 6 Stages - Will give it a bash too!

[nickbungus]

Going to try replacing R34 with one of these and play with the biasing to see what I can get.