Rothwell Love Squeeze mods. Threshold and Attack

[nonost]

Hi. I want to mod the Rothwell Love Squeeze. I would like to add threshold and attack pots. I've attached the schematic.

If I'm not wrong I can put a 10K pot instead of R10(10k) for a threshold control. Maybe a 5K resistor in series with a 5K pot would be better, I will do that if the threshold goes to high with just the 10K pot.

For the attack pot I've been told that I can change R9(22r). But I don't know how to do the maths. What value gives me 100ms? I know C9(10uF) is involved in that formula but I'm lost.

Also I've noticed how the stock values dull a little bit the overall sound. Any way to prevent that treble loss?

I would really appreciate any help, cheers!

[Mark Hammer]

The harmonic content of a plucked string resides in the first few moments, and disappear quickly, leaving primarily fundamental with a bit of lower-order harmonics; varying with the type of string, gauge, bridge, body type, etc.  The actions of a compressor are to pull back the gain in response to peaks, and then bring the gain back up again after the initial attack, so as to provide a more constant amplitude.  The perceptual result is that the strings sound dull with compression because it shoves the harmonic content out of the spotlight.

The contemporary strategy for dealing with the treble loss is to blend a bit of the uncompressed input signal with the compressed one.  This lets the initial treble peak be heard pretty much as is, while still permitting the constant level of everything after to be heard as compressed.

A question: are your inquiries for mods based on some degree of use of the stock circuit, and specific dissatisfactions/needs arising?

[nonost]

Hi Mark. I've been using this comp for a while and I'd like a not that fast attack response, playing bass at high settings is more noticeable. Having an attack pot would make it more bass friendly as well.

Also the "comp" pot acts as a ratio&threshold control to my ears and I would like to make them independent in some way since I like a high ratio setting with a low threshold.

The "treble loss" thing is something that I've felt with others comps before and I wondered if I could get rid of it.

[PRR]

R9 (and opamp limits) controls attack.

But it may take major re-design to get a full-feature limiter out of this. There may be better starting points than this plan.

[nonost]

The threshold mod works very well. I swapped R10 with a 100klog pot (50k looks more that enough but didn't have any).

However the attack mod didn't work. Tried low value pots instead of R8 and it only gave more output (light overdrive over unity). After cleaning the output with the volume I didn't hear any change in the attack at all.

By the way, R11 (10k) and C10 (10uf) set the decay to 100ms if I'm not wrong. Will 10k and 22uf give 220ms?

Thanks guys, just trying to learn everyday a little bit.

[anotherjim]

Bear in mind that IC1b is only amplifying signal enough to pump a negative control voltage for the FET onto C10.
R9 serves to control the rate C10 can get charged up while R11 sets the rate it discharges. However, R9 (Attack) and R11 (Release) are effectively a voltage divider and the cost of messing with the resistance values is that the maximum control voltage on C10 becomes too small if R9 is too large compared to R11.
Note that the reference for the opamp Vb has been set a little higher than 1/2 of VA to get a little more negative swing out of IC1b with the hope of being able to produce a more negative voltage on C10. Changing to a CMOS opamp like TLC2272 would give a little more range because the output can swing more negative than for the TL072. However, that kind of improvement is only slight in practice.

Changing the gain of IC1b will be an effective sensitivity/threshold control, but it won't really change the interaction between Attack and Release. However, R10 should not just be a variable resistance - it shouldn't go to zero ohms. Fit a small "minimum" resistance in series with the pot, say 1k. You don't want the compressor to become insensitive to bass content, so the cap C7 doesn't want to be too small. For bass guitar, it should be larger than 220nF at 470nF, but that is with R10 at design 10k value. If you reduce R10 below 10k the bass response will reduce. Happily, it can probably have a C7 value that is extra-large (C3 can be chosen to reduce any low frequency "rumble" getting in, it probably doesn't need to be larger than 4.7nF). Make C7 4.7uF with negative lead to 0v.

From what Mark said above, the input has a lot of treble content in the pluck/attack phase. What if we filter that treble out? Then it would only be the longer "body" of the input that the compressor responds to. A simple way to do that would be to put a capacitor across R8 to form a low-pass filter. If R8 is 39k, a 10n cap will start to cut above 400Hz leaving the low mids and bass content (where most of the guitar signal energy is anyway) to set the compression.

Having done that, you don't want the release time too long or the compressor will act on the attack of the next pluck! So don't increase R11 or C10 too much.

[bool]

+1 on forming a LP with adding a cap across R8. This "just works" with such style compressors. But a corner freq. at 400Hz may be too low. Positioning it at 800Hz or even higher (like 1kHz) may be a better idea, so a cap. value should be approx. 4N7 or 3N3.

[nonost]

Thank you Jim! Very insightful. I've read it plenty of times already.

The threshold mod works great. I encourage any builder to put a 25k pot instead of R10 (I was wrong about de 100k pot, I had it wired backwards). With a 25k pot you can go lower and higher.

Tried 10uf||10uf and 10uf||22uf with the C10 cap (switch). To my ears it slows the attack and release very nicely. The 10uf||22uf is more obvious though.

I'm going to try a bigger value for C7 as Jim suggested in order to improve it with bass guitar. But the LP filter idea is what I like the most. Definitely trying it. I'll try values that Jim and bool suggested.

I will let you know how it goes! 

[snk]

Hello,
I built a Rothwell compressor clone today.
I am enjoying it, and playing with some components values to tweak attack and release to taste.
I have added a 25K threshold pot, and it's a good addition.

My little issue with is that I plan to use it also on drum machines, and noticed that this compressor is very, very sensitive to transient spikes : hi-hats, snares and sharp percussive material triggers the leds too quickly, which leads to harsh and unwanted distortion (it's short, but noticeable).

I have read the advice to tame the highs in the detection circuit, by putting a capacitor across R8 (39k). Thank you for the tip, PRR, I will try this.

I would have another related question : if it makes the diodes clip too early to my taste, should I try with yellow or white leds (which clip at higher voltage), or are the leds also placed here in order to protect some components afterward (and therefore, using yellow or white led might create additional issues)?

[iainpunk]

changing the clipping diodes is owkay, they do not protect anything as far as i can see.
you could also 'stack' diodes to get higher Vf, like an 1n4148 and a red LED in series, gives about 2v.
i think you would like asymmetric clipping, it gives a more gradual descent in to distortion.

cheers, Iain

[snk]

Thank you, Iainpunk.
In the meantime, I have already tried running it at 18V, in order to get a bit more headroom. I have also increased C7 to 680nF, and put a 1.5nF cap across R8 to remove the treble (above 2.7khz) from the sidechain.
It seems slightly better (but it's not a scientific test, just a quick listening session).

I will try with yellow or white LEDS, and look to your suggestion.
You mean like putting a yellow led+ a diode on one side, and a white led on the other side?

At the moment, the distortion sounds ugly (like clipping a DAC, a short spike of distortion at transient. At first, it is barely audible, but once you hear it, you can't un-hear it).

[snk]

Ok : 2x1N4148 + 1x3mm red led on one side, and 1x 1N4148 + 1x 3mm green led on the other side cured all the clipping issues I had (and, when it occurs on extreme settings -like high input level, and compression & threshold set to max, with also extreme attack & release values-, it seems slightly less harsh).
Thank you, Iainpunk!

Now, it is so transparent that i am wondering if it is even ON 
Is there any elegant way to add a reduction indicator led?

[snk]

Further report, for anyone interested (now, or for future builders) :

1 - At first, I intended to have slow/fast settings for attack & release (using dpdt switches). But after playing for a while with the values and feeding the compressor many kinds of signal, I am considering using pots instead. Indeed, sometimes the slowest release is too fast for the incoming material, but sometimes it's just the ticket, and the same applies for longest release times. So, I think it's good to have a rather wide range of options, as well as the in-between values.

2- One thing which surprised me (and which is not an issue, but I'd be curious to have some technical views about it), is that this FET compressor is very fast, but I noticed that with the compression and 25k threshold pot both maxed, the release envelope had a recovery time longer than 4 seconds. I wasn't expecting this (default release time is around 100ms), and it doesn't annoy me , but i am wondering : is such drastic timing changes caused by the mods I did (increased C7 to 680nF -from 220nF-, C1 to 100nF -from 33nF-, C3 to 33nF -from 1.5nF annd using a 25K pot for threshold)?

[snk]

... There is something I still quite don't get with the Love Squeeze :
Now I have threshold, attack & release pots, and I can hear their action clearly.
The only parameter I can't hear working is... the COMP pot 
At first, I thought it as because it's supposed to be "very transparent", but now i am wondering and trying to understand...
Is the COMP pot actually a ratio parameter as I thought?
What I don't get is that i can turn it fully clockwise or anti-clockwise, it doesn't seem to increase or decrease the output volume, so it doesn't seem to compress more or less.
I have tried monitoring the effect processing level using my mixing desk, and if I feed the compressor a -6dB signal, it will output a -6dB signal, if i feed a -12dB signal, it will output a -12dB signal and so on... So, with the attack & release, i can hear envelope shaping, but with the COMP pot, i don't hear dynamic processing... Is there something i'm missing?

[iainpunk]

The harmonic content of a plucked string resides in the first few moments, and disappear quickly, leaving primarily fundamental with a bit of lower-order harmonics; varying with the type of string, gauge, bridge, body type, etc.

interesting, never thought of the darkening of a compressor in that way.
my hubcap banjo loses is fundamental after a few seconds and only rings out high order harmonics, focused around 750-ish Hz, i haven't tried it with a compressor yet, might be really great.
the reason the fundamental falls off is the aluminium hubcap body, the way the hubcap is tensioned, and that there is no direct connection of the strings to the wooden through-neck construction,

a great addition to a compressor circuit is a 'filter-mixer':
original thread link

i haven't experimented with your particular compressor, but i do have experience with the "really cheap compressor" and i used a modified BMP tonestack (also a tilt control, but without the mid scoop) as a "mixer".

where you would normally have the input, i fed the compressed signal and where normally the ground would be is where i put in the (boosted up to 2 times) clean signal, so i could have clean high and compressed low or vice versa. both clean and compressed had volume controls as well.

here is a simplified drawing, i think this is more useful than a normal mixer

cheers, Iain

[nonost]

Ey snk! I don't know if you fixed your problem, but I'm sure it's the FET not working properly. I've been there, chinese copies don't work. CEN brand works great. When the FET is working you will hear pretty clear.

Cheers!

[Fancy Lime]

Somehow I missed this thread both times it has been around the block. So maybe now is as good a time as any to chime in with my favorite Love Squeeze mod. Better late than never, I suppose.

The thing that I like best about the Love Squeeze is that it has the gain controlling element (the FET) in the ground leg of an amplifier. That would be the R13, C11, Q1 arrangement. This mean, you can very easily add a second ground leg parallel to that. Because ground legs with caps in them in non inverting opamp stages act as a kind of high pass filter (or really more like a treble boost), you can use this, to selectively bypass the highs from being compressed past a predefined ratio, by choosing the R and C of that second ground leg right. This makes it kind of a poor man's dual band compressor. You can of course even make it adjustable by using a pot instead of a fixed R. A "transparency" pot, if you will. I find it really odd that I could not find any reference to this mod anywhere. I'm sure someone must be doing that, right?

Also, I would use a J112 or J113 instead of a 2N5457. They are much easier to get in non-fake because they are still in production afaik. They are also better suited for the purpose. If you can, use one with a Vgs(off) between 1.5 and 3V for good clean squish. Lower will lead to more distortion with loud input signals, higher will mean less available maximum compression when used with a 9V supply. If you use it at 18V or higher, just throw in any old J112 and you should be fine.

Andy

[nonost]

Ey Andy! That sounds very nice. I will try it next time. Right now I'm adding a led compression indicator. It works well, but the ultra white led is too bright even with a 100k resistor. I'm using this little circuit:



Would it be a good idea to reduce transistor gain by increasing the 33r resistor?

Point A goes right after the 10uf cap at the output of the second op amp. Where it meets the two diodes.

Cheers!

[Fancy Lime]

Hi Nono,

for your purpose, you don't need R2 at all. I would leave out R2 and connect the emitter directly to ground. Increadse R3 to reduce brightness of the LED.

That being said, the circuit is not really showing you the compression but the AC signal level that goes into the envelope detector for the compression. Those two are of course related but it is important to note that they are not the same. You can still use your circuit as a useful compression indicator, if you tune the series resistor R3 right, so that the optically perceptible brightness correlates well with the audible compression. If you tune this in a quiet environment under ideal listening conditions, then you have a good indicator to use on stage where it can be difficult to hear details of your own playing. Depends on how disciplined your drummer is, of course

Cheers,
Andy

[nonost]

Yeah, I know. My idea is to record a really brief loop and meassure the ac signal right at these point. Knowing the diodes threshold (0.240v) and the ac audio signal voltage I think I can adjust the led brigthness pretty good.

So far is working great. It isn't affected by the vol pot, only the comp one.

Thanks a lot!