orange squeezer issues

[addnoize]

(this project uses specs from general guitar gadgets).

guys i have just built an orange squeezer with external volume pot, it works. - BUT the thing is so transparent it is hard to know whether it is on or off.  

here is a jpg of the board to show the components i used.
http://www.users.bigpond.net.au/googgyegg/powercuts/graphics/os.jpg

i wasn't able to get able to get an 1N100 diode so i put in an alternate germanium (RAT). (i have not tried a 1N914/1N4148 or 1N4004 go)

also the caps are tantalum but not sure if they are the correct tantalums. the transistors are MPF102's and the IC is a TL072CP.

also adjusting the 10k trim makes minimal difference apart from changing the overall volume and affecting a little bit of waver/ripplely sounds.

can anyone help to get this thing singing, thanks - addnoize

[RDV]

You mean you don't know what value the Tantalums are? They should be 4.7uF.

I think that it also will need 2N5457 or similar JFets for it to squeeze well.

Diode polarity is important also. As long as the diode is Ge and not Si, the type won't matter.

If you're getting

waver/ripplely sounds

, it's probably working at least a bit.

Regards

RDV

[Milothicus]

leave the volume at about halfway. turn the trimpot all the way down, and start playing. slowly turn the trimpot up just until you hear some noise. that's max compression. the more you turn up the trimpot, the less compression you have. if you turn up the volume on the guitar, and down on the squeezer, after strumming a chord, you should hear a swell in volume as the compressor releases.

[dubs]

You should get pretty decent gain from it and turning the trimpot at one point should get some gnarly overdrive.
I used 2n5457, R4558 op amp and a germanium diode from DSE.
Check you don't have any crossed tracks from soldering.

[addnoize]

thanks for all the info everyone -

i have swapped the MPF 102's for 2N5457's. still waiting to get a R4558 op amp, and i will replace the diode with a new germanium. (just in case i was sold a silicon, doesn't look like a silicon tho').

i have set up the onboard trim pot as suggested and i can definately hear the volume increase some time after the string is plucked quite noticeable on the bridge position on my tele (humbucker)............and yet still subtle.

what i did notice tho is that when used in conjunction with my crunch pedal (electra distortion) it really gives it a more unpredictable wavering fuzz - nice.

[RDV]

The diode was your problem then, that is the most critical part, the opamp was fine.

Regards

RDV

[MarkB]

TL072 sounds good in a squeezer, leave it there.
"-)

[Mark Hammer]

The 2N5457 is being used as a voltage-controlled resistor in the OS.  The op-amp cranks up the signal, and the envelope follower (half-wave rectifier) derives a signal proportional to the loudness which is used to make the resistance of the FET lower so that when you slam the strings, Q2 and the 82k resistor tied to its drain (or is that source?) form an attenuator/voltage-divider that gets turned down.

Why am I telling you this?  Because not every FET will demonstrate the exact same gate-voltage-to-drain-source-resistance-value relationship.  If, for example, based on the FET used and the envelope signal amplitude, you "forced" the FET to change its resistance from, say, 10megohms to 500k, that would be a substantial change in resistance, but in tandem with the 82k resistor, that wouldn't constitute much attenuation, would it?

There are 6 elements/factors that will impact on how much attenuation, hence squeezing (and in this case it is not "gain reduction" like you'll see some LED-meters labelled on expensive rackmount compressors, since the amount of gain applied is fixed):
1)  Signal amplitude - hotter input signals = more squeezing
2) Gain setting for the op-amp:  Stock unit fixed at gain of 23 via 220k feedback resistor and 10k resistor to ground.  Decrease 10k or increase 220k to increase gain.  Increase 10k or decrease 220k  to reduce gain.
3)  Signal loss through diode -  Ge diodes will eat up less envelope signal, silicon more, LEDs even more.  If you have several of each type, you can probably use your meter to select the one with the lowest voltage drop so as to maximize the envelope signal amplitude.  E.g., if you have a bunch of 1N60s or 1N34s, ranging from voltage drops of 190mv to 270mv (not unreasonable), select the one measuring 190mv for the strongest envelope signal.
4) FET number and item - 3819, MPF102, 2N5458, 2N5952, J113, SK30A, BF245, etc will all "work" but will show different relationships between gate voltage and drain-source resistance, as will different transistors within each type.
5) Trimpot setting - Trimpot forms part of the resistance-to-ground network that produces attenuation
6)  Input resistor - the 82k resistor forms part of the voltage divider network (along with the trimpot and FET) that "adjusts" the input volume going to the gain stage.  If you make it higher in value, the FET/trimpot network will have more impact on the amount of attenuation, although making it too high will likely produce more input noise for the gain stage, which you probably don't really want to amplify x23.

All of these things can be played off against each other.  E.g., envelope-signal loss through the diode can be compensated for by trimpot adjustment, op-amp gain, or input resistor value.

If you are going to up the value of the input resistor, a BiFet op-amp like the TL072 will help to achieve less hiss.  More standard bipolar op-amps apparently generate more hiss when used with higher value input resistances (and 82k may already be in that ballpark).

I cannot recommend highly enough, the usefulness of adding a second gain stage with the unused op-amp so that you can have some boost no matter what degree of "squeeze" you have set it to.  The second gain stage can also be used as a simple lowpass filter to remove a little more of whatever hiss may have seeped in.

[Mark Hammer]

Oops, I also forgot to mention the 1k5 resistor, which can also be tinkered with.  Obviously increasing it will reduce the apparent "sensitivity" of the envelope follower.

[addnoize]

with reference to the debugging page on general guitar gadgets, i have noticed that some of my IC voltages are a little different to the stated ones. so i think it will be a while before i get this thing straightened out

it is definately squeezing but based on mark's observations/comments i think i might socket some of the caps and resistors on the board. i still suspect the germanium diode (0A91) i have installed may not be best suited to this project, as it is the only place where i have deviated from the bill of materials?

everyone thanks for the info, addnoize

[brett]

In my experience, the OS will only ever be subtle.

A Dyna or Ross compressor has more serious "grunt".

A possible alternative to a Ge diode is a Schottky, such as a 1N5819.