A/DA Flanger does TZF?

moosapotamus · November 16, 2006, 11:14:58 AM

Nice work, analog kid! Very glad to hear you're getting to the bottom of things. Congrat's!

Quote from: markusw on November 16, 2006, 04:35:12 AM
Some more crap posted by me
Obviously, it's the other way round. 27k gives higher gain than 4k7

I was thinking the same thing... But, you also have that 5K in parallel with the 4k7. So, if you increase the 4k7 to 27k, doesn't the overal feedback resistance for IC2b go down, giving you less gain?

RE: the PCB layout going out to fab...
I'm thinking I should start a new thread just for verifying my final layout (seems to have gotten burried in this awesome mammoth thread). I'm certainly going to comb through it a few more times myself to make sure I didn't botch anything. But, I'd like to enlist the help of some others for the same purpose. If I can get a thumbs-up from a few others, I'll feel a lot better about getting a batch of PCB's fabricated for folks.

Thanks!
~ Charlie

oldschoolanalog · November 16, 2006, 11:25:27 AM

The 5K and .01u are the de-emphasis components. They are there to provide gain reduction at the freqs that were pre-emphasized in IC 1a (4K7 & .01u).

analog kid · November 16, 2006, 02:18:24 PM

"27k would create higher gain than 4.7k "
DANGIT Markus!!

that's was thinking first too but when you said that I firgured I was wrong and never tried fooling with that one. Figuring I had as much gain as i 'should' from that resistor. LOL!
So maybe I will give that value a try since my unit seems to be working VERY well ( still with the GATE disconnected) except for a very large volume drop! I have already tried fiddling with the mix R's to bring volume up but it's not creating enough IMO.

analog kid · November 16, 2006, 07:53:22 PM

I found when probing that the signal was dropping(in EVEN) on the threshold side of R34. (a 10k) So I replaced with a 820ohm and flanged output is very close to equal. ODD+EVEN, maybe still just a bit weaker, but this makes me wonder just how much lower it is ok to go with R34 in order to keep Odd/Even at equal volumes without just Jumpering as in the No-Gate verion and STILL not have a negative effect on how well the Threshold control functions?? it seems R34 is only there to put some resistance between IC3C's output and the drain of the FET in the gate. and jumpering it disables the Threshold/gate completely. so I don't know enough to understand Exactly what that R value does for the N.Gate , any outlook here. With the 820ohm in right now I can notice a boost in the Flanged output/ enhance when between the Resistor or Jumping it but ONLY in EVENmode, ODD is same no matter. So I guess it' s just finding the balance of how low adjusting this resistor for strong Even vol. / and still having the noise gate function well. If so I think I about got it so long as the gate's gonna work well still if I drop r34 much more than 800ohms. '
according to Gripp , Threshold Should STILL function and gate not disabled when R34 is Jumpered.
PS: I am very anxious to see you get your final vesion pcb layout finished and ready to have sent off Moos'!! Even though i surely will have my A/DA finished , running as it should and boxed by then, I am very interested in the progress and maybe even having one made!!

toneman · November 16, 2006, 09:28:57 PM

Moose wrote:
RE: the PCB layout going out to fab...
I'm thinking I should start a new thread just for verifying my final layout (seems to have gotten burried in this awesome mammoth thread). I'm certainly going to comb through it a few more times myself to make sure I didn't botch anything. But, I'd like to enlist the help of some others for the same purpose. If I can get a thumbs-up from a few others, I'll feel a lot better about getting a batch of PCB's fabricated for folks.

Hey charlie,
How about makeing just a few, say 3 or 5. Then having 3 or 5 folks assemble the proto-pcbs.
U know Murphy's Law?.....well...... $:-\$
rather than cutting a bunch of traces and running a bunch of jumpers....

If U think it's better to get a larger run of pcbs, to keep the individual pcb piece/price low,
then, i guess, do final de-bugging after...well....you're the ramrod ....

head 'em up; drill 'em out;
...i'm wondering what your "target price" is

T

analog kid · November 16, 2006, 09:59:15 PM

now that the trace going to pin 7 instead of pin 6 of the 4049 was caught allowing Moos' to fix it in his version of the layout , which was the only thing for which I had to modify the pcb, I can't imagine it'll be anything other than VALUES needing to worked out for builders. since charlie is sticking with the basic spirit of this layout, though Of course he is changing the routing around a bit and all those added pads,so maybe that would be a good idea? I don't know just a thought

markusw · November 17, 2006, 01:36:41 AM

Quote from: analog kid on November 16, 2006, 02:18:24 PM
"27k would create higher gain than 4.7k "
DANGIT Markus!! that's was thinking first too but when you said that I firgured I was wrong and never tried fooling with that one. Figuring I had as much gain as i 'should' from that resistor. LOL!
So maybe I will give that value a try since my unit seems to be working VERY well ( still with the GATE disconnected) except for a very large volume drop! I have already tried fiddling with the mix R's to bring volume up but it's not creating enough IMO.

Did you try that 27K??
According to some sims I did, with the 27k the overall frequency response (pre-emphasis/de-emphasis) is pretty flat while with the 4.7k there is quite some loss in bass and mids...

Markus

analog kid · November 17, 2006, 04:07:10 AM

Wow Interesting indeed. Since this is a symptom I've been noticing in my unit since i got it up and running, along with a vol. well below bypass (even with 27k'sfor 68k mix r's in the output amp, incidentally the same amp where the pre-de emph. is) but had written til now figuring a lack of gain somewhere in the crkt. . I've definitey been concerned with a lack of full frequency response which my ross has (as well as sound samples) and thin sounding even with Neck pos. w/ a strat and settings that should be the fullest. Also some hint of grittiness in the high end.
so , YES , I tried this value in there (finally) and absolutely noticed a fuller, and importantly a probably very close to unity output. Based on this I have decided this is a place to pop a couple sockets in to toy with. But I put the original value back for now because , as I have the noise gate working and troubleshooting now, And this resistor also had an affect on the noise gate( I believe, causing) a bit of gated staticly hiss with the Threshold control even when turned off / CW. When changed to 27k
so I am fixin to do some experimenting here based on the result I got as well as your invaluable simulations to back it up with something!! Also i found another schematic value difference with a 10k across the FET s/d, it's not in the same shem w/ the 27k value in Ic2b BUT it does connect right into that amp, so I wonder If that's an error

analog kid · November 17, 2006, 04:18:52 AM

BTW , Markus, since you're so good with those sim's. As I said, not coincedentally, the 68k mix r's are in the same opamp (actually connected directly to) the 4.7k/27k resistor. So would you be able to sim *(or IYO) would there be any REASON to change the values of those Mix resistors WITH the 4.7k going up?? Or for me to change them from 27k back to 68k (or any other value ) n my case? I would imagine they'd have some direct affect on each other. Thanks

markusw · November 17, 2006, 05:46:16 AM

QuoteBTW , Markus, since you're so good with those sim's.

I'd say LTSpice is good in doing the sims. I just enter the schem

I simulated the dry path only (i.e the pre-emphasis stage IC1a, the lo-pass filter stage IC2a, and the output stage IC2b) and with the 68k plus the 27k for the IC2b stage there is a slight boost with the overall frequ response being pretty flat. Looking at the stages for the delayed path I believe that this will apply for the delayed path too.

Quotewould there be any REASON to change the values of those Mix resistors WITH the 4.7k going up?? Or for me to change them from 27k back to 68k (or any other value ) n my case? I would imagine they'd have some direct affect on each other. Thanks

Personally, when looking at the sims I'd give the 68ks plus the 27k in the feedback loop a try because this should give the flat frequ response plus the higher gain.
Don't know how the threshold side chain influences the gain though.

Markus

moosapotamus · November 17, 2006, 09:56:54 AM

Quote from: toneman on November 16, 2006, 09:28:57 PM
How about makeing just a few, say 3 or 5. Then having 3 or 5 folks assemble the proto-pcbs.

If there are willing subjects who can cover the cost of fabrication for a short run like that... I guess that's not a bad idea.

For a cost comparison, ExpressPCB has a deal that they call their ProtoPro Service - 4 boards, 21 square inches (or smaller), with solder mask and silkscreen for $179 ($44.75 each).

But, since the ADA layout is a bit smaller than that (5.5" x 3.5" = 19.25"), I think the cost could be less than that with their Standard Service - double sided PCB with plated through holes, no silkscreen, no solder mask. Maybe under $40 each. I can't get an exact price right now, but I'll check it out later tonight.

Quote from: toneman on November 16, 2006, 09:28:57 PM
...i'm wondering what your "target price" is

For a run of at least 10 boards, somewhere between $25 and $30 for a double sided PCB with plated through holes, no silkscreen, no solder mask (ExpressPCB Standard Service). Adding the silkscreen & solder mask adds ~$5 per PCB.

~ Charlie

analog kid · November 17, 2006, 03:44:12 PM

QuotePersonally, when looking at the sims I'd give the 68ks plus the 27k in the feedback loop a try because this should give the flat frequ response plus the higher gain.
Don't know how the threshold side chain influences the gain though.

Yes that is a big question at hand for me right now! and yes I'd say that's correct way to go since I have the 27k's in for all the above right now and there's a pretty big boost from bypassed

I hae the gate working in both Even / Odd. as long as r34's value is as low as 1k or so they are even volume. I'm afraid to go too low here as it put resistance between IC1c output and Fet drain. I am gonna put the Mix R's back to 68k to see if it impacts the gain to the gate whether the sim will say that or not.
It seems I'm down to getting the right FET for the job. I've found some close but still there's a bit of gate hiss in time with the LFO even when Thresh full CW. Which is why I wondered how the Mix R's on the Inverted input of the output amp may impact on the Gate . can they.

Sounds like you're getting near making a few people happy who don't want to bugger with an etch such as this!! Imagine that I even shrunk the layout another 10% or so (to fit the biggest piece of Clad I had

, talk about impatience!) so if those traces were tight enough in some areas....! ie; under the IC's ,etc.. WHEW!

boogietube · November 17, 2006, 08:13:20 PM

I have been following this thread and have nothing to add, but if you did a run of 10 boards with screen, I'd buy one and build it.
I'm in awe guys.!! Keep up the work. Maybe soon I'll understand 25% of it. lol

StephenGiles · November 18, 2006, 06:20:33 AM

I have a working veroboard unit and don't understand half of it!!!

moosapotamus · November 20, 2006, 11:24:25 PM

Group Buy Info for ADA Flanger PCB's

Thanks!
~ Charlie

StephenGiles · November 21, 2006, 07:55:55 AM

Another work of art Charlie! I printed this out over 2 A4 sheets from Excel, joined them together and copied on to an A3 sheet - looks very nice indeed. For some reason I can't get Excel to print to A3 paper.

oldschoolanalog · November 21, 2006, 08:13:12 AM

Beautiful, man! "Techno-Porn" at its finest

StephenGiles · November 21, 2006, 09:31:57 AM

Was the R38 question resolved? I wound my ADA Flanger up last night as I had the house to myself

, and listened particularly to the odd/even sounds and the regeneration. I have a 10k for R38, as previously mentioned, but there is bags of fat resonance when I turn up the regeneration pot. It seems that all the various ADA versions used 30k for R38, and included R34 as 10k, whereas I do not have R34.

I don't have my orignal ADA Circuit with me but it was the MN 3010 version without part values, and the value given for R34 was 1meg. Mike Irwin pointed out that this was an error, suggested it be omitted altogether and R38 be made 10k in order to provide unity gain in the odd/even amp. I suppose the effect of including R34 will do just that with R38 at 30k. Funnily enough, that MN 3010 version is not on the adadepot site!

StephenGiles · November 21, 2006, 09:52:48 AM

But R35 is 1meg so I am mystified now. I'll check when I get home.

Gripp · November 21, 2006, 10:44:20 AM

This comes both from simulation and what's in Analog Kid's unit right now, R38 30k, R34 10k and R35 1M.
In the sim, this was about the only way to get unity between both inverting and noninverting configs when the noise gate is in place (R35 1M, that is). Seems to be that way in the real world too.
1M for R35 makes the gate work good too. The problem with Analog Kid's unit squeeling when R38 was 30k seems to be gone by now, must have been something else.
If one is omitting the gate, then jumping R34 and making R38 10k is the way to go (not having R35 or the jfet in there at all).

/Pelle

News:

A/DA Flanger does TZF?