Old phaser circuit help

[edvard]

There's a phaser circuit I've seen in a hundred cookbooks over the years and I've even attempted to build one, but the official parts I'm missing are only labeled as "9709" and "9705".
They appear to be monolithic P-channel arrays, but I cannot for the life of me find any info on them.
Anybody heard of 'em?

I suppose I could just use P-channel FETs (tried it once but I think I got the pins wrong...), but I'm curious how the original parts would do.
I also wonder if a 4016 or 4066 would work?
The advantage would be better consistency, making FET matching less of a chore, whaddya think?

[Nasse]

Yeah I believe they are arrays, perhaps not available, was that national semi application?. I can look later this evening if I can find part code. Perhaps discrete is better these days

Been thinkin doin light dependent resistor phaser, or ota version.

[edvard]

Since a phaser is one of the last things I have yet to build, I've often considered building a LDR or OTA version, but this particular circuit has been bugging me for a while.
I'll upload a scan when I get the time.

[B Tremblay]

Does the schematic look similar to 2nd Ross phaser: http://home-wrecker.com/ross3.html  That one uses a P-channel FET array.

[edvard]

Kinda...
I think all phasers kinda look a bit like each other due to the nature of the beast.
Here, I gots a scan:
http://www.aronnelson.com/gallery/main.php/v/edvard/phaser.jpg.html

[B Tremblay]

The chip in my Ross has a letter within the number (AM97C11), so maybe searching for that will help.

[Nasse]

Just looked and my reference shows AM9709CN for that quad fet, perhaps that gives hits too

[Mark Hammer]

The 9709/9711 is likely replaceable with a quartet of JFETs.  The problem is that you can't find a datasheet for that chip on-line so it's difficult to know what JFET would replace it.

On the other hand, since ALL of those similar designs place a fixed resistor in parallel with the drain-source pins to set the maximum resistance, the challenge would seem to be simply that of matching the JFETs you use instead (the advantage of using a quad-matched set on a chip, and something that Jeorge Tripps took advantage of in the BJT domain for the Swollen Pickle), and tinkering with the bias.

Apart from that, nothing really magical, just convenient.

[Nasse]

" Each stage shifts 90 degrees at 1/(2*Pi*RC) where C= pos input cap and R=resistance to ground.... Jfet resistance varies from 100 ohms to 10 k ohms while gate voltage is adjusted from 5 to 8 volts (optimum for the 8709) ... six phase sifter stages has been used each octave apart from 160 to 3,2 kHz..."

[edvard]

Just looked and my reference shows AM9709CN for that quad fet, perhaps that gives hits too

Bingo!!
One mystery cleared up, thank you!
Search for "am9705 fet" turns up absolutely nada.

The 9709/9711 is likely replaceable with a quartet of JFETs.  The problem is that you can't find a datasheet for that chip on-line so it's difficult to know what JFET would replace it.
...

Home-wrecker.com says P1087e:
http://home-wrecker.com/phaserguide.html

On the other hand, since ALL of those similar designs place a fixed resistor in parallel with the drain-source pins to set the maximum resistance, the challenge would seem to be simply that of matching the JFETs you use instead (the advantage of using a quad-matched set on a chip, and something that Jeorge Tripps took advantage of in the BJT domain for the Swollen Pickle), and tinkering with the bias.

Apart from that, nothing really magical, just convenient.

Exactly.
Anything to save me from the torture of sorting a pile of FET's would be great, although I'd be willing to do it if I wanted to build a FET phaser that badly.
I've wanted to build this circuit for a long time...

Second question (from OP):
Would a 4016 or 4066 work as a FET array?
This forum has produced an overdrive/distortion circuit using those, so I know they're not limited to purely switching purposes.
I guess it's time to get breadboarding...

[Mark Hammer]

My guess is that obsessing about the FETs is unlikely to lead anywhere particular critical or productive.  In my books, a notch is a notch is a notch.

There have been a couple of designs/projects over the years that used the MOSFets in a 4049 hex invertor as variable resistances.  While certainly convenient, with respect to footprint, cost, and matching, the 4049 does not fare well as a variable drain/source resistance in such applications.  It has a lower threshold for distortion, which means that the input signal needs to be attenuated  (complemented by a gain recovery stage on the output, which serves to boost hiss), and that any feedback loops need to be judiciously managed.  I imagine the 9709/11 was the same way, which is why you see the RC network between gate and drain (that raises clipping threshold).

With the clear success that a 2N5952 provides, the ease of getting them and low cost of arriving at matched sets, I don't see any real advantage to pursuing the 97C09/11 or even mimicking it.

Now, if it was possible to have a matched LDR array on a chip, THEN we'll talk!!

[Taylor]

Would a 4016 or 4066 work as a FET array?
This forum has produced an overdrive/distortion circuit using those, so I know they're not limited to purely switching purposes.
I guess it's time to get breadboarding...

It might be possible, since as you say Gez made a 4016 amplifier with his "Scuzzbox". I've been playing with these kinds of ideas lately, haven't found anything to work yet.

But, you can use 4016 or 4066, paralleled with a resistor, driven by a PWM LFO, to get a variable resistor. Not a drop-in replacement in a FET phaser, but it can work and it's easier to get matched resistors than matched FETs.

Now, if it was possible to have a matched LDR array on a chip, THEN we'll talk!!

I actually think that something like this must exist in the jillions of exotic optoelectronics in the Mouser catalog, which we never use because no prior design uses them. So it's going to take somebody smarter than I am to find the right part, but I feel like it has to be in there.

Edit: Just as an example of the kind of stuff hidden in there. This wouldn't work for our purposes exactly but it's not too far from what we want, with some creative circuit design.

http://www.clare.com/home/pdfs.nsf/0/4E36B2775A9C9B8B85256A2C0068D9DB/$file/LOC210_R4_0.pdf

http://www.vishay.com/docs/83654/83654.pdf

[PRR]

It's an FET switch array.

It is apparently in 1976 National Semiconductor Linear Data Book, but I can't find mine tonight, and it is not in the 1982 edition. National does not make anything so mundane today; AFAICT, nobody does.

Use plain FETs.

> a matched LDR array on a chip

No such thing. The duals are as close as it gets, and they have wide spread.

LDRs use different stuffs than transistors and ICs. The advances of the 1970s which made Si arrays as consistent as they are were not followed by the guys brewing CdS with handed-down secret-art recipes.

CdS is cadmium yellow pigment. This suggests easy kitchen cookery. I suspect that what you need is not pure CdS or CdS-as-paint, but secret impurities in CdS.

And it may not be that easy. Here's a paper on growing CdS:
http://www.springerlink.com/content/q85j2604252187h0/fulltext.pdf (800K PDF)
ZnCdS thin-film via chemical bath deposition:
http://www.acadjournal.com/2006/v18/part5/p1/chemistry%20combined%20all.pdf
Here's an apparently simple screen-print and sinter process:
http://iopscience.iop.org/0022-3727/25/12/026/pdf/0022-3727_25_12_026.pdf
Looks like cooking-time is critical.

[Harold]

It's an FET switch array.

It is apparently in 1976 National Semiconductor Linear Data Book, but I can't find mine tonight, and it is not in the 1982 edition. National does not make anything so mundane today; AFAICT, nobody does.

Use plain FETs.

Which FETs should have about the same characteristics as the AM97C10, AM97C11 or the P1087E's? Is the 2N5952 a viable candidate?

I can't seem to find the specs on those old chips to compare the V-gate and I-drain specs. (not that I know what I'm talking about )

[edvard]

It's an FET switch array.

It is apparently in 1976 National Semiconductor Linear Data Book, but I can't find mine tonight, and it is not in the 1982 edition. National does not make anything so mundane today; AFAICT, nobody does.

Use plain FETs.

Which FETs should have about the same characteristics as the AM97C10, AM97C11 or the P1087E's? Is the 2N5952 a viable candidate?

I can't seem to find the specs on those old chips to compare the V-gate and I-drain specs. (not that I know what I'm talking about )

When using FETs as variable resistors, I think the variability even within the same part number trumps any particular benefit that might be sought after by matching characteristics.
In other words, bias it until it works and it's good enough because that's as good as it's going to get no matter what part you use.
Some might be more or less "touchy" depending on what they were designed for (I think I'd stay away from "switching" FETs), but any plain-vanilla FET like a 2N5485, BF244, or even your friendly little model will do.
Just remember to read up on R.G.'s article on matching FETs before building, it's very informative:
http://www.geofex.com/article_folders/fetmatch/fetmatch.htm

[Harold]

any plain-vanilla FET like a 2N5485, BF244, or even your friendly little model will do.

Thanx! 

...but i just discovered the old Ross phaser users P-Channel FETs instead of the proposed N-Channel FETs...

[Harold]

any plain-vanilla FET like a 2N5485, BF244, or even your friendly little model will do.

...but i just discovered the old Ross phaser users P-Channel FETs instead of the proposed N-Channel FETs...

On that thought, what will be the most suitable of these two P-Channel Fets? BS250 or 2N5460?

[edvard]

The BS250 is an enhancement-mode P-channel MOSFET, so I'd be clueless as to how it would work in the Ross circuit.
You'd have to get comfy with the datasheet (http://www.datasheetcatalog.com/datasheets_pdf/B/S/2/5/BS250.shtml), read up on how to bias it so it operates in the 'Ohmic' region of it's characteristics (http://www.electronics-tutorials.ws/transistor/tran_6.html) and tweak your phaser circuit to fit.

On the other hand, the 2N5460 is a general-purpose part that just might work if you can find some that fit the specs.
Read R.G.'s article again, substituting negative voltage on the gate for positive, and test your FETs for ones that fall in that range.
The datasheet (http://www.fairchildsemi.com/ds/2N/2N5460.pdf) gives a very wide Min-Max spec for gate-source cutoff voltage, so I hope you've got a nice handful to run through.

[Harold]

On the other hand, the 2N5460 is a general-purpose part that just might work if you can find some that fit the specs.

I've searched and found some info on the old FET arrays used in the Ross Phaser (AM97C10 and AM97C11) but I have no clue on how (or what) to compare that with the specs of the 2N5460.

The geofex article describes the "gate-source cutoff voltages". How can I find out what they are on the obsolete FET arrays? Is that the only spec that must be met?

[edvard]

In general, yeah.
Other attributes affect how it performs as an amplifier and in turn how it operates in the ohmic region, but Gate-Source cutoff value is critical to knowing if it'll make a good variable resistor within the voltage range we have (in this case, 9v).
Because FETs are famous for having very wide tolerances, sometimes all that just goes out the window and we use what happens to work that day.
See the trimpot marked "RV3" on th Ross circuit?
That's exactly what that's for... to set a bias due to wildly variating tolerances, even in the monolithic FET arrays.
If FETs were spot on, we wouldn't need that (well, mostly).

Like Mark said, chasing down exact matches in FET circuits like this isn't all that productive.
I'd say find 4 of your 5460s that match up and hit the specs close enough, plug them in and adjust the bias until it phases.