Another Volume Pedal Retrofit Question

[JRM]

Hi, I'm beggining one of this VPR projects based on Craig Anderton's famous circuit with the OTA CA3080 and the TL071.
I've seen some building reports and found out that some people aren't happy about the hiss level and others find out that the gain is too much.
Unfortunately, I haven't found clear answers regarding the solution to this problems.
I guess hiss will be hard to fix as it is mainly related to the OPAmps used in the project.
Regarding Gain, what are the resistances that control the gain of the TL071? I'm confused as I was expecting an input resistance to calculate/tweak the gain with R6=100k. I wished to know in order to replace it by a trim or even a pot to control the gain from the outside of the pedal (either that or a 3 way swithc with 3fixed resistances, giving 3 diferent top volume levels.
Schem: http://www.tonepad.com/getFile.asp?id=2

[Mark Hammer]

The hiss isn't THAT objectionable, although it may pale in comparison to some of the gear that aspires to dead-quiet, and stick out, as a result.  Perhaps more relevant is the inability of the 3080 to handle the sorts of massive signal levels some folks might want to impose on it at the end of a series of pedals.  It is the necessary attenuation of the signal at the input, to avoid challenging the 3080, that results in the fairly modest S/N ratio achieved.

Seems to me that this would make an excellent time for someone to adapt the basic VPR circuit to a more contemporary, and available, chip like the LM13600/LM13700.

Alternatively, a simple passive attenuator, using an input resistor in series with the signal, and an LDR (controlled by an LED whose brightness can be foot-controlled) may be the most high-headroom/noise-immune path to go on.

[JRM]

Thank you for your answer Mark.
For a passive volume control I agree with you that LED/LDR is the way to go.
But...
I like the idea of having an active volume control with a tunable gain as that can enable a control of the overdrive of the input of the amp.
Having searched for the LM13700 datasheet and found in the page 10 (http://www.ti.com/lit/ds/symlink/lm13700.pdf) some Applications. Of interest we have the "Voltage Controlled Amplifier" and the "Stereo Volume Control" that we may use in mono mode (just half IC). Unfortunately I don't have the knowledge to fully design one of this babies.

Edit: I've found this:

http://www.geofex.com/Article_Folders/VCA%20Applications.pdf

and it may be a lead to a full design but I still lack the knowledge to complete it.

[JRM]

Well, I decided to study a little bit and taking the design proposed by RG Keen I've drawn this DELUXE VOLUME PEDAL with scratch proof properties as the pot is out of the signal path and adjustable gain.
If I'm correct, the current gain at the OTA is given by: Iout=2.Iin.Ictl/Id, and then the voltage gain is Vout/Vin=2.Ictl/Id.Rload/Rin. In our case, Rin=10k, Rload is switchable in order to achieve several preset max gain rates (0, +3, +6, +9, +12dB), Ictl will vary from near 0 to 1mA and Id must be calculated. Again if I'm correct, Id=(9-0.7)/4.3k=1.93mA.
Please send me your comments  and hints as I'd like to see this on a "breadboardable" version soon.

[gritz]

duplicate of the following post. Damn my eyesight...

[gritz]

Hi JRM  

There's a couple little isues with that layout:

1) Linearising diodes are made of fail  . they do reduce distortion at low signal levels, but they also reduce the overall gain of the circuit, so you have to pump in more control current (or use a bigger load resistor) and this actually increases noise. OTA distortion isn't particularly objectionable (it's mostly third harmonic) and it's arguable whether it's even that detectable in a lot of guitar - related applications. A bit of crunch on high level transients is a lot less noticeable than white noise all of the time.

2) The OTA's control current input sits at about 1.2V and is a very low input impedance so you have to resort to lateral thinking to feed it the correct current for a proper volume response. I just knocked together a back o' the envelope circuit:



A1, Q1 and the 3k3 resistor form a voltage-to-current converter. Treading on the pedal in an "I want louder" manner sends an increasingly negative voltage to the non-inverting input of the opamp. The opamp responds by pulling it's output negative in an attempt to get both it's inputs to an equal potential. This causes Q1's emitter (and thus the 3k3 resistor) to follow the input voltage. Voltage across a resistor means current and this current is dumped via Q1's collector into the OTA's control current pin. D1 protects Q1's base-emitter junction from reverse voltage (otherwise there's no feedback mechanism should tiny offset voltages and whatnot cause the opamp's output to head positive of 4V5).

The voltage divider formed by the 330K and 100K (and the 100n filter cap) is a bit of a back o' the envelope job too - it should give enough headroom for Q1 to do it's stuff even if the battery gets down to 7 volts or a bit less (which is pretty spent...) A low battery will cause a small loss in max volume, but because of the logarithmic way that ears work it won't be a huge deal.

Voltage across the 3k3 resistor at maximum volume will be about 1 volt, which equates to a maximum current of about .3 mA. The 13700 will work with control currents up to a bit more than 1 mA, but be aware that if you accidentally apply a positive voltage to the control current pin without a current limiting resistor the OTA will die. It's easily done on a breadboard, but at least there's two OTAs on a chip!

If my maths is right then a load resistor of 10k should give a max gain of a wee bit more than unity. If my maths is right...

Resistor R Leak *might* be needed if you you want minimum volume to be zero and there's an unavoidable bit of offset voltage around the opamp. It's just occurred to me that you may want a minimum volume of more than zero. Sticking a resistor (or pot / trimmer) between the top end of the 100k pot and 4v5 will do this.

I don't think there's a need for C2. Similarly I've dispensed with the trimmer at the inputs of the OTA. At moderate gains the 13700 is usually pretty well balanced so it's output with no signal should be around 4v5 and so the output of it's Darlington buffer will be free to swing to a max of +- three-and-a-bit volts (which is about the same as the TL072's output can manage with a 9V supply).

Feel free to give me a shout if you have any problems or questions (or if I've made a balls-up somewhere).  

[JRM]

Hi gritz,

You've used the OTA in a different way, more like a voltage controlled resistor. I'm a mechanical engineer so my electronic knowledge is very limited. I can't fully understand your design but I can imagine that it works fine. Using the OTA as a voltage controled resistor is something that have already crossed my mind to remove the real pot of the circuit. The only thing I don't know yet is if I could design it with an adjustable gain (one of my requirements).

[gritz]

Hi gritz,

You've used the OTA in a different way, more like a voltage controlled resistor. I'm a mechanical engineer so my electronic knowledge is very limited. I can't fully understand your design but I can imagine that it works fine. Using the OTA as a voltage controled resistor is something that have already crossed my mind to remove the real pot of the circuit. The only thing I don't know yet is if I could design it with an adjustable gain (one of my requirements).

Nope, its not a voltage controlled resistor - it's a pretty standard voltage controlled amplifier (with the control voltage from the variable pot in the pedal turned into a current to properly impedance / level match the pot to the LM13700's control input). Bar a couple of minor details it's the same as the circuit you posted!

Signal is fed via a TL072 unity gain buffer to the 13700's non-inverting input. OTAs like the 13700 / 13600 and CA3080 distort like crazy if the difference in voltage between the two inputs exceeds about 113 millivolts (if memory serves), so the 10k / 220R divider network that feeds the OTA's non-inverting input reduces the biggest likely signal down to a sensible level. The 220R between the inverting input and Vsupply/2 balances the bias currents between the inputs and help ensure that the output of the OTA sits reasonably close to Vsupply/2 rather than offsetting itself. Incidentally, if you've ever looked at the LM13700's data sheet you will probably have noticed that the divider networks at the OTA inputs do a lot of dividing. This is because the applications on the datasheet are all powered by +-15 volts so the biggest possible voltage swings are in the order of +-13.5 volts. We're playing with 9 volts here, so we don't have to worry about such big signal swings, or the bit of distortion that's caused by hitting the OTA's inputs a little harder (this is guitar-land after all). Getting more signal to the OTA improves our signal to noise ratio (which we do worry about). OTA noise is proportional to control current. But I digress.

Where was I? Oh, yeah, applying a few tens of millivolts of signal to the OTA's non-inverting input. The OTA outputs a current that's more or less proportional to the diffence in voltage between the two inputs multiplied by the control current (which we're modulating with the volume pot). We need to turn this output current back to voltage (so we can hear it), so we feed it through a load resistor that's attached to Vsupply/2. This signal is then buffered by the Darlington pair built into the 13700 and sent on it's way.

It really is a VCA, not a voltage controlled resistor. Trust me, I'm a doctor, etc!  

[gritz]

A quick update:

I had half an LM13700 going spare on my breadboard so I built the circuit depicted in my woefully scruffy sketch above. It works as predicted - maximum gain is a dB or two above unity (i.e only just noticeably louder than the input). The volume control is smooth and silent. There is a tiny bit of hiss at maximum gain (which is to be expected from a 13700), but it's by no means obtrusive - in fact it's vanishingly small compared to the hiss generated by the Danelectro Fabtone that's feeding it. The noise might be a problem if the circuit was used in front of a high gain distortion pedal, or into a high gain amp channel, but stick it after a pedal chain and into a cleanish / slightly crunchy amp and it should be fine. It seems to be here, anyway. 

[JRM]

Sorry, I haven't read your schem correctly
Running the pedal in the end of the chain is one of my objectives. For controlling the volume before the high gain distortions I have the volume control of the guitar (very nice to play with the fuzz face, cleaning and distorting as going down and up the gtr vol).
In your circuit, without the clipping diodes, the gain will be given by the RLoad/10k ratio?
What do you think of a gain range from 0 to +12dB? It will be enough to induce overdrive in my amp (Laney Cub12 usually set for clean)?

[gritz]

In your circuit, without the clipping diodes, the gain will be given by the RLoad/10k ratio?
What do you think of a gain range from 0 to +12dB? It will be enough to induce overdrive in my amp (Laney Cub12 usually set for clean)?

Hi JRM: yep, increasing RLoad will give proportionally more gain, so for +12dB (which is very nearly equal to x4) a 39k resistor would be pretty close. However, it will also amplify the amount of noise created inside the LM13700 by the same amount, so if you're looking to engineer some gain into the circuit to push your amp a bit the best way to do that would probably be to turn A2 into a non-inverting amplifier with a suitable range of gain (mebbe selectable with a switch, or a pot), rather than a unity gain buffer. This will be a quieter way of doing things. Everything after A2 stays the same.

How much gain will you need to drive your Cub? I'm sorry, but I really don't know! It might be time to grab the breadboard and knock together just the non-inverting booster opamp part to see what kind of boost is required to push your amp (and to hear how it sounds) and go from there with the rest of the circuit. Sorry I can't be more help with that bit.

[PRR]

> based on Craig Anderton's famous circuit

IMHO, several of Craig's plans were ahead of their time and behind modern expectations.

> hiss ... is mainly related to the OPAmps

No; in Craig's plan hiss is partly due to very low bias in the 3080, lack of input buffer which (with guitar's need for high impedance) forced fairly large divider resistance, and way more than unity gain (about 6, which sure will raise hiss).

> what are the resistances that control the gain of the TL071?

'3080 outputs a *current*, not a voltage. The 100K around the '071 converts the current to a voltage. The base gain of the 3080/071 pair is 100K*Gm. The Gm of the 3080, at the (small) maximum current biasing the 3080, is 500 microMhos. Which can be inverted as 2K ohms, Then voltage gain over both chips is 100K/2K= 50. Times the input divider which is about 1:7.8, so about 6 overall.

So if the goal is "unity gain at Max" then 100K should be 15K-18K.

But the low current and high impedances beg to be tampered too.

And '3080 is in and out of stock. Currently Small Bear has for-sure-good '3080, but until last year supplies of the '3080 were dubious.

I started re-working it, but Jonathan is ahead of me. The LM13700 has output buffer. The dime we budgeted for out buff can now be spent on input buffer. Divider impedance can be low, 220 ohms, 1/10th the resistor hiss of 22K. _I_ would incorporate the anti-distortion diodes, but this is a matter of taste, and requires re-computation. Jonathan has wisely raised '3080 max bias 10X higher than Craig, reducing '3080 input noise voltage.

So use Jonathan's plan.

The only thing _I_ would do radically different: _I_ happen to like Craig's volume-pot plan. It is simpler. The wiper "goes the right direction" which reduces debugger confusion. The min-gain setting is offset by D1 which will semi-track the '3080's internal offset, the only strong reason to use a true Current Source. ('13700 may need two diodes.) To hit the new 0.3mA bias, Craig's R9 must now be 27K. This will work fine with a 50K pot, OK with a 100K pot, saggy with a higher pot so you might want a pot buffer. This gets half the complexity of Jonathan's Current Source, true.

With Jonathan's plan, watch Vbias as you turn the pot. Looks to me, using Craig's 10K+10K Vbias divider, Vbias will sag 1.5V at max gain. The current source could as easily be fed from +9V, though the 330K:100K divider has to be re-computed (or run from +9V to 4.5V).

[PRR]

> turn A2 into a non-inverting amplifier

Risks overloading the '13700 input. Does a flogged '13700 sound better than a flogged Cub? (I dunno.)

The diff-pair input will overload (3% 3rd harmonic) at 60mV p-p or 20mV RMS (without lin-diodes; twice that with diodes).

Your 10K:220 divider scales that 45X higher, 900mV, a good level for guitar-chain.

If you gain-up the front-end (why not just lessen the 10K:200 divider?) then normal guitar levels overload the 13700 front-end *before* its gain-action.

If you bring signal in almost to input overload, bias and load can set the output to *any* level up to (nearly) the full 9V.

Therefor I'd be thinking a 10K/39K (or 39K with 13.5K optional parallel) switchable Rload, for normal gain (less hiss) or big gain and output (accepting hiss). This might be a 50K pot plus 5K fixed, with a mark at unity and a mark at Cub-beating level.

[gritz]

Hi PRR:

I think that Craig's bias current topology tips a a hat to the fact that volume pots are of a high value, so trying to suck a sensible amount of OTA control current out of one will result in a very non - logarithmic sweep (all the volume will be at one end). You could of course swap the pot for a 10k log and futz with the filter bit, but I figured that a diode and a PNP was cheaper and less hassle (there being two opamps on a TL072 already). There is method in my madness.  

The 13700's bias pin sits at about 1V2 as opposed to 0V6 for the 3080, so if anyone was to adapt Craig's circuit for one then it would maybe need two diodes between the bottom of the pot and ground, rather than one, but that's a minor detail.

I agree that the 10k / 220R combo sails rather close to the wind with the 6-and-a-bit possible peak to peak volts output of A2, but by then A2 would be clipping  and the output of the OTA would be reaching the limits of it's swing too. The circuit was conceived to be used at guitar level throughout - i.e. the odd transient above a couple of volts pk to pk and brief enough that any bending of said transient wouldn't be audible (which is how pedals generally work). It's the usual compromise that wasn't planned to have humungous amounts of gain , so I guess the addition of a moar ooomph is a whole new subject! Perhaps 180R or 150R instead of 220R would be more "textbook", although I note that Craig's circuit runs the OTA input much hotter...

Your switchable load does sound like the simplest option, but (bearing in mind that I know nothing of the Cub, or how much crunch JRM is looking for), 12dB (wherever it's applied) doesn't sound like an awful lot to scare a clean channel into drive. I guess we'll just have to wait for JRM to try some stuff and then we'll go from there.  

Good call on the saggy bias btw - I didn't budget for that (I have a buffered Vref on my board so it didn't dawn on me). Why don't they make a  TL073?  

[JRM]

Thank you Jonathan and PRR for your expertise and inputs on my too big ambition-to-knowledge ratio. I can read a schem (or even write one) and solder but not much more.
Thinking over your remarks made me conclude that a much easy way to deal with the adjustable gain is to simply have the pot in series with a 5k fixed resistance that PRR sugested and then dial in a gain that suits my needs. If I adjust the Cub's Gain just one point lower than starting OD, I believe that multipying the signal by 16 will be enough to overdrive it (though it's difficult for me to test it accurately, I'll try a rough guess varying the input signal by playing with the volume pot of my guitar). A mild OD is my target, for more high gain I rely on pedals.
I confess that I'm overwhelmed by your discussion regarding what's best: to put the gain adjustment on A2 and then run LM13700 on a unity gain or to run A2 in unity gain and then adjust the gain of LM13700. I understand the issues (distortion vs. hiss) but I couldn't reach a conclusion. The final discussion about Vbias is about Vbias stability? Did I correctly understood that at max gain Vbias will float +/-1,5V and not kept stable at 4,5V? We could solve that with another Opamp, can't we? If so, one could use TL074 that has 4 opamps and it's the same price (isn't what you're talking about having a buffered Vref?).
Thanks again

[JRM]

I finaly had the opportunity to properly test the influence of input level on the overdrive of the amp. With my Epi Riviera with mini-humbuckers I've dialed a low enough gain not to have any overdrive and doubled the input level with the help of my TC Flashback delay (if one put the FX level to max with a very short delay, the output is doubled as it adds the delayed signal to the dry signal, as anyone can check on the pedal manual). I've also used the guitar volume knob, rolling it from 3 to 10 that would make a similar effect (with a logaritmic pot). The results were as expected: a mild and warm overdrive was immediately present. So, a voltage gain of 0 to 8 under my foot would be very nice!

[mrdorbe]

I build the volume retrofit (the Tonepad version)for me and for a friend, I put it in a diy volume pedal and for for my friend in a junior 25k earnie ball...
For me it works great (250k pot) but my friend says he experience high signal loss when he uses the pedal in the amp fx loop (mesa road king)
Can anyone help me solve this maybe I can modify the schematic to add a high pass filter or decrease one of the caps to have a better high frequency response.

I'm sorry for the nubile questions I'm a bit experienced in building guitar fx but my knowledge in electronic is not so great
So I will be great full If you'll explain in a simple way.
Thanks.

[Ben N]

I wonder if the various OTA issues can be reduced by adapting the Engineer's Thumb approach and putting the OTA in an opamp's feedback loop.

[mrdorbe]

Anyone? How can I get better treble response from the volume retrofit?
Hopefully without loosing the other frequencies ...

[ChanchoPancho]

Hey, are there any developments/news about this project? I have a stereo volume pedal with an active circuit that can give you like +9db of gain. It works great but the pot scratches a lot and it's been very annoying. I've been planing for make a "de-scratcher" circuit for a while now based on the craig andertons 3080. Since my pedal is stereo and the fact that (here) the 13700 is cheaper than the 3080, I would love to implement something like Jonathan's circuit, but in stereo.

I have a couple of question about the previous posts:
1. As I understand the 10k/220R network attenuates the input signal. Is that made to not saturate the 13700? Is there a rule? I looked at the 13700 datasheet but couldn't figured out this.
2. Jonathan: Your control circuit for Ictl is just to modify the pot response?