Help me diagnose an op amp reverb driver circuit I built for belton "digilog" module

[teemuk]

I don't know how to bias an op amp.

The opamp needs to have a certain DC voltage potential both in its inputs and output. This voltage potential should be approximately half of opamp's total power supply "rail" voltage so that the signal has highest headroom towards positive and negative voltage excursions.

The Opamp chip has two supply rails: Vcc and Vee, Vcc being the positive and Vee the negative. If you power opamp from a single-ended +5VDC supply then your two rails are 5V (Vcc) and the ground (Vee). Bias voltage for the opamps needs to be half of that, 2.5VDC, and you must reference each opamp's input to this potential accordingly. Now your signal can (in theory) swing from 2.5V to 5V at positive halfwaves or from 2.5V to 0V at negative halfwaves. If in doubt, you can google search for something like "single supply opamps". There's plenty of documentation about this. Since DC offset for the signal will be 2.5VDC AC coupling becomes mandatory.

The schematic on the other hand does not show the generic power supply arrangement. It does, however, show overall reference to zero volts (ground), which means the opamps must be biased for the kind of configuration where half of total rail voltage is zero volts. This practically means a dual supply with positive and negative voltage rails. If you have 5VDC in in your access then Vcc becomes +2.5VDC, Vee becomes -2.5VDC and bias, half of total rail voltage, conveniently centers to zero volts, which is also the ground potential. Signal can now swing from 0V to +2.5V at positive halfwaves and from 0V to -2.5V at negative halfwaves. (Higher if you use higher supply voltages, like +/-15VDC). AC coupling is not mandatory but in practice good practice (no pun intented) to prevent amplifying DC offsets.

I'm pretty sure you have either configuration messed up: Either you use dual rail supply with other bias than zero volts, or you use single rail supply biased to zero volts. Both errors result into serious asymmetry of amplification and DC offsets in outputs.

Forget the +5VDC supply shown in schematic. It's documented there because 5VDC is mandatory for the reverb IC. The opamps, they can use any other suitable supply and in the schematic the opamps are clearly drawn to use dual, positive and negative supply voltages. No, it's not shown in documentation because it's the sort of basic knowledge that everyone embarking on opamp projects should have already been familiar with. If bias is ground then supply voltages are symmetric with positive and negative potentials, if bias is a DC offset then supply is single ended.

[teemuk]

When you talk about protection diodes do you mean that they would be arranged to clip the input signal at (maybe at .7V) to protect the op amps? Why don't op amps like big signal voltages? What's big? Will they brake or just clip?

Yes, basically the idea is to limit peak voltage value of the input signal with clipping.

OpAmps do not like input signal voltages that exceed their designed operating range, which is closely determined by their power supply voltages. Look at the datasheet and find the spec called “common mode voltage”. A rough rule of thumb would be that you do not want higher input voltages than what the OpAmp’s supply voltages are. Way before exceeding this limit OpAmps have likely started to clip, but exceeding these limits may cause the OpAmp to behave with various unlinear, and very nasty ways; Sudden phase inversion of the output voltage, or output latching to one of the supply rails come first to mind. Generally good design practice is avoiding this.

However, what comes to limiting the input to safer magnitudes you may find a) .7V threshold too low and b) that input current needs to be limited as well. So basically you hook up diodes to another low impedance source but ground, one with higher voltage potentials. The power supply of the opamp will be just fine for the purpose: One diode connects from signal path to Vcc, another diode connects from signal path to Vee. You want these diodes to be reverse biased under normal operation so mind the orientation. The diode(s) turn conductive if input voltage exceeds power supply voltage plus diode forward voltage, diode current is sinked by the power supply and signal gets clipped. If you looked at those spec sheets you usually find that this is ample enough “clamp” to keep input signals within the common mode voltage range. Then, before this clamp, you add a series resistor to input that limits the input current to safe values. Few kilo-ohms is usually ample enough for the job.

Adding such protection circuit to the input is indeed advisable. In fact, you should probably add one to the output too, to protect the final OpAmp output from surges coming from the following circuitry. Naturally current limiting resistor is placed on the side of the clamp where it limits clamp’s current from fault current. Basically, the orientation is reversed. If you look at some amplifier circuits that have solid-state FX loops or reverb circuits you probably find few good examples of real-life implementations of such circuit. The Belton documentary has just the rudimentary stuff that is needed for the sake of example. Protection schemes, power supply arrangements, etc. are additional “know-how” and not included.

By the way, that schematic is a poor "ripoff" of the simplified circuit diagram in BTDR-series datasheet. The datasheet's circuit diagram and following documentation, on the other hand, are presented in a manner that would have actually solved a great deal of your issues beforehand.

As for the power supply for this thing, I would recommend the existing regulator board you have PLUS another regulator board with +/-15VDC regulation at any method you see fit. Insert the +5VDC regulator to the +15VDC supply so you need only one rectified power supply in total. Use the +5VDC supply portion only for the reverb module and make sure the grounding layout follows this arrangement in sensible way that results to least noise.

Do not power from the filament circuit! That is the most unreliable method in comparison to alternatives such as the bias supply or possible extra taps in the secondary. Galvanic isolation of having a separate winding for the solid-state circuit’s power supply is best but using the power tube bias supply is usually a good compromise. Tube doesn’t often short circuit to its grid but it can often short circuit to filament, not to mention the filament power supply is a hog for huge surge currents inherently. Your safest bet is to tap to the bias supply with another rectifier that provides positive and negative voltage sources. These power supplies can eve be half-wave rectified (so it’s a simple job of two diodes), you filter and regulate them anyway, and OpAmps have magnificent power supply rejection ratio so little ripple in the supply won’t harm anyone.

You probably need to design a new PC board for this but it’s worthwhile. When you’re at it, you can probably combine the whole thing, including all the power supplies to this single board. If you want those few bits of extra safeness in marrying delicate solid-state circuits to high voltages put some kind of surge voltage protector, like MOV, to input of the SS power supply section (basically before its rectifier diodes). Oh yeah, and try to fuse that power supply line if possible at such low current draw. Then another one of those design tricks not shown in rudimentary example schematics: Add 100nF filter capacitors as close to each power supply terminal of each OpAmp as possible. This will help a great deal in stabilizing the ICs. Some of the OpAmps in the rudimentary example schematic may need additional, about 22 – 100pF, capacitors that connect between the output and the inverting input. Sonically these should be transparent but they help to prevent any inaudible high frequency oscillation.

Then, I suppose you didn't study the documents at
http://www.neunabertechnology.com/support/BTDR
Do it now.

The circuit on the Belton data sheet mixes the dry and wet signals. It is intended to be inserted inline with the signal chain.

And this, accentuated as a very important point. Also, that circuit is designed for instrument level signals with very low amplitudes.

[elipsey]

This power supply should get the job done. The +/- 15V is good for about 15mA, the +5 more than 100mA.

The LM358 is a very slow opamp not suitable for audio of any quality. The TL072 is more than 10 times faster. It's used in lots of solid state guitar amps.

The circuit on the Belton data sheet mixes the dry and wet signals. It is intended to be inserted inline with the signal chain.

Thanks, looks great. Maybe I could just connect my 7805 to that +15V supply and stay away from the filament supply altogether?

Do you think +/-10V is close enough? I just bought a big ass sack of 10V 1W zeners diodes....

[loudthud]

Thanks, looks great. Maybe I could just connect my 7805 to that +15V supply and stay away from the filament supply altogether?

The problem with that is now you need 100mA from the bias supply. The rectified bias supply is roughly 50V so 50V * 100mA = 5 watts. Not impossible but it's a can of worms. I did a quick simulation to come up with the 100uF filter caps on the +/- 15V and the 2.2K resistors dissipate just over 0.5W, over 1W if the supply shorts to ground. You would need probably 470uF and 330 ohm 10W resistor, then you have to worry about the Zener 'cuz it's gonna be burnin 1.5W if there is no load. If the zener burns up, the 50V is gonna kill everything else on that line including the 7805 and the Belton module.

The filament supply in my diagram uses parts that will take up less room than the one 10W resistor and they don't get hot. The only real issue is where you ground the center tap. Don't ground it to the chassis directly but tie it to signal ground where the Belton module is. Speaking of ground... the ground for the first two 100uF caps (for the +/- 15V) should tie back to where the center tap of the high voltage winding is grounded. The rest of the grounds for the zeners and the other caps should ground to the opamp board and signal ground there. The 2.2K resistors will isolate the noise between the two grounds.

Do you think +/-10V is close enough? I just bought a big ass sack of 10V 1W zeners diodes....

They should work with no changes comma but dot dot dot the headroom of the reverb circuit will be less.

There is still the issue of how do you turn the reverb on and off. Belton doesn't address that issue. Can the output of the module just be shorted to ground? They don't say. You could switch the whole reverb circuit out of the signal path, that will take a relay.

[elipsey]

Apologia and Mortification

When I supplicate myself to the Wisdom of the forum, with the spirit of the Humble Acolyte, I will not presume to quarrel with the Fair, Just, Gentle, and Instructive reproaches of it's Esteemed Members. However, I would be most insufferably mortified to imagine that the Esteemed Members had perceived, through some fault which is certainly my own, that grave and outrageous insult which would be implied by Sloth, should it seem that I have failed to conscientiously, and in advance, endeavor to study all the subject matter which I am able to obtain by Searches of the Internet, though flawed and imperfect my searches can only be, pertaining to the content of my Original Post before submitting said post. Therefore, I submit that before posting I did indeed Search the Internet, but that my searches were hampered by the Imperfect Virtue that is possessed of all men, who are all Fallen from Grace, and that I am therefore guilty as much of Ignorance as of Sloth, and I pray that the Esteemed Members will only provide me with documentation or tutorial subject matter which might correct my ignorance, which I would eagerly study in a spirit of Abject Penitence before posting any further questions pertaining to the aforementioned subject matter.

Yes, basically the idea is to limit peak voltage value of the input signal with clipping.

OpAmps do not like input signal voltages that exceed their designed operating range, which is closely determined by their power supply voltages. Look at the datasheet and find the spec called “common mode voltage”. A rough rule of thumb would be that you do not want higher input voltages than what the OpAmp’s supply voltages are. Way before exceeding this limit OpAmps have likely started to clip, but exceeding these limits may cause the OpAmp to behave with various unlinear, and very nasty ways; Sudden phase inversion of the output voltage, or output latching to one of the supply rails come first to mind. Generally good design practice is avoiding this.

However, what comes to limiting the input to safer magnitudes you may find a) .7V threshold too low and b) that input current needs to be limited as well. So basically you hook up diodes to another low impedance source but ground, one with higher voltage potentials. The power supply of the opamp will be just fine for the purpose: One diode connects from signal path to Vcc, another diode connects from signal path to Vee. You want these diodes to be reverse biased under normal operation so mind the orientation. The diode(s) turn conductive if input voltage exceeds power supply voltage plus diode forward voltage, diode current is sinked by the power supply and signal gets clipped. If you looked at those spec sheets you usually find that this is ample enough “clamp” to keep input signals within the common mode voltage range. Then, before this clamp, you add a series resistor to input that limits the input current to safe values. Few kilo-ohms is usually ample enough for the job.

Adding such protection circuit to the input is indeed advisable. In fact, you should probably add one to the output too, to protect the final OpAmp output from surges coming from the following circuitry. Naturally current limiting resistor is placed on the side of the clamp where it limits clamp’s current from fault current. Basically, the orientation is reversed. If you look at some amplifier circuits that have solid-state FX loops or reverb circuits you probably find few good examples of real-life implementations of such circuit. The Belton documentary has just the rudimentary stuff that is needed for the sake of example. Protection schemes, power supply arrangements, etc. are additional “know-how” and not included.

So I ought to know that:

1. The range of an op amp's input signal should not exceed the limits of it's power supply.
2. The input of an op amp should be biased to a voltage that is in the middle of it's power supply range, which may require AC coupling especially when using a single ended power supply.

OK, I got it

By the way, that schematic is a poor "ripoff" of the simplified circuit diagram in BTDR-series datasheet. The datasheet's circuit diagram and following documentation, on the other hand, are presented in a manner that would have actually solved a great deal of your issues beforehand.

Are we sure we are talking about the same schematics, and the same datasheet? The schematic I posted came from the BTDR series datasheet, and it's the only schematic in the datasheet, which I linked in my original post.

As for the power supply for this thing, I would recommend the existing regulator board you have PLUS another regulator board with +/-15VDC regulation at any method you see fit. Insert the +5VDC regulator to the +15VDC supply so you need only one rectified power supply in total. Use the +5VDC supply portion only for the reverb module and make sure the grounding layout follows this arrangement in sensible way that results to least noise.

Do not power from the filament circuit! That is the most unreliable method in comparison to alternatives such as the bias supply or possible extra taps in the secondary. Galvanic isolation of having a separate winding for the solid-state circuit’s power supply is best but using the power tube bias supply is usually a good compromise. Tube doesn’t often short circuit to its grid but it can often short circuit to filament, not to mention the filament power supply is a hog for huge surge currents inherently. Your safest bet is to tap to the bias supply with another rectifier that provides positive and negative voltage sources. These power supplies can eve be half-wave rectified (so it’s a simple job of two diodes), you filter and regulate them anyway, and OpAmps have magnificent power supply rejection ratio so little ripple in the supply won’t harm anyone.

You probably need to design a new PC board for this but it’s worthwhile. When you’re at it, you can probably combine the whole thing, including all the power supplies to this single board. If you want those few bits of extra safeness in marrying delicate solid-state circuits to high voltages put some kind of surge voltage protector, like MOV, to input of the SS power supply section (basically before its rectifier diodes). Oh yeah, and try to fuse that power supply line if possible at such low current draw. Then another one of those design tricks not shown in rudimentary example schematics: Add 100nF filter capacitors as close to each power supply terminal of each OpAmp as possible. This will help a great deal in stabilizing the ICs. Some of the OpAmps in the rudimentary example schematic may need additional, about 22 – 100pF, capacitors that connect between the output and the inverting input. Sonically these should be transparent but they help to prevent any inaudible high frequency oscillation.

Then, I suppose you didn't study the documents at
Digi-Log Modules - neunaber audio effects
Do it now.



And this, accentuated as a very important point. Also, that circuit is designed for instrument level signals with very low amplitudes.

Actually, I visited that link before posting and read everything in the "Design Resources" section. It includes four documents:

1. The Belton datasheet, which I linked from another site in the original post, and which I have thoroughly studied. I used it to build the circuit I have now, and I think that circuit would probably work (although perhaps poorly) if I were to simply connect a -5V supply to the LM358 GND pin. Incidentally, I notice that in the LM358 datasheet(s) this pin is always called GND. If it had been called Vcc- as it seems to be called in the datasheets for some other op amps, I might have taken the hint.
   
   
2. An "Example reverb pedal schematic". This schematic is copyrighted by Brian Neunaber who I presume is the proprietor of the domain from which it is served: neunabertechnology.com, a hobbyist website. I looked at this schematic and found it to be interesting, but also very different from my application. The schematic is for a pedal, so it assumes the presence of an external power supply, includes switching, expects a very small input signal, &c.
   
   When you said the schematic I posted from the Belton documentation was a "ripoff", did you mean it was a ripoff of Mr. Neunaber's reverb pedal schematic?
   
   The Neunaber schematic didn't seem like the right solution for me, but maybe I should have learned more from it. On further inspection, I see that it has a single ended 9V power supply, that it employs a pair of 10K resistors as a voltage divider to derive a 4.5V bias reference, that AC coupling is used where necessary, and that it is implemented with TL074 op amps. Perhaps this is a Quad packaged version of the TL072 that loudthud mentioned. The fact that this pedal is designed for a low level input signal, and yet provides nearly twice the op amp power supply range I used in my driver circuit ought to have alerted me to the fact that my 0V-5V power supply would be grossly inadequate, even if it were properly biased.
   
   Are these the sorts of things that you wish to bring to my attention?
   
   
3. Some sort of parts list that is stored in a text file with no character encoding specified. So it's all Greek to me (also, Cyrillic, simplified Chinese and cat faces).
   
   
4. A archive of documents called "Tube amp integration". It seems to be the case that a few years ago, a fellow called Dan Rhodes designed a "carrier board" for the Belton module. The carrier board seems to include the power supply and op amp circuitry that I am trying to design. However, the documents in the linked archive, as far as I could tell, only explain how to install the board in your amp, but do not include a schematic of the carrier board itself, presumably because Dan wished to sell it, and did not want to give away the recipe for his secret sauce.

Thanks for replying. I'm learning

[elipsey]

The problem with that is now you need 100mA from the bias supply. The rectified bias supply is roughly 50V so 50V * 100mA = 5 watts. Not impossible but it's a can of worms. I did a quick simulation to come up with the 100uF filter caps on the +/- 15V and the 2.2K resistors dissipate just over 0.5W, over 1W if the supply shorts to ground. You would need probably 470uF and 330 ohm 10W resistor, then you have to worry about the Zener 'cuz it's gonna be burnin 1.5W if there is no load. If the zener burns up, the 50V is gonna kill everything else on that line including the 7805 and the Belton module.

Yikes.

The filament supply in my diagram uses parts that will take up less room than the one 10W resistor and they don't get hot. The only real issue is where you ground the center tap. Don't ground it to the chassis directly but tie it to signal ground where the Belton module is. Speaking of ground... the ground for the first two 100uF caps (for the +/- 15V) should tie back to where the center tap of the high voltage winding is grounded. The rest of the grounds for the zeners and the other caps should ground to the opamp board and signal ground there. The 2.2K resistors will isolate the noise between the two grounds.

Ok, I see what you're driving at... teemuk seemed to have some good reasons for staying away from the filament supply, but I guess I will probably have to make some compromises somewhere to shoehorn this into a tube amp, as it were. Does PSUD model solid state stuff? Or did you use something else?

There is still the issue of how do you turn the reverb on and off. Belton doesn't address that issue. Can the output of the module just be shorted to ground? They don't say. You could switch the whole reverb circuit out of the signal path, that will take a relay.

Well, right now I just have a volume pot connected to the return, which seems to work ok. Is there anything wrong with that?

If I add very much more stuff to my circuit, it seems like I might need a printed circuit board, but I'm not well equipped to etch my own PCB, and my girlfriend gets angry when I pour muriatic acid in the sink. It looks like there are lots of places on the net where you can submit a design and purchase a PCB. It seems like this might take a while, and if don't get it right on the first try, it might take several whiles. If anyone has experience with a seller they would care to recommend, I might try it anyway. I'll probably try to find a TL072 and a couple of 15V zeners, and then use a dual-ended supply with a circuit that's close the one I have. Maybe I'll try connecting a -5V supply, (batteries?) to the circuit I have now, just for laughs. However I slice it, I guess I'll be waiting a while for parts in the mail. I would be delighted to pass the time with some educational reading about op amp design patterns, or solid state audio in general, so as I have said before I would be grateful if any of you would like to recommend some.

Thanks again for your assistance, everyone.

[Dave H]

elipsey, does your PT have a real centre tap on the 6.3V winding? I read above that it has an artificial centre tap (2 x 100R?). I don't think loudthud's circuit will work with that.

[elipsey]

Nope. I thought i just hadn't connected it, but on closer inspection, it doesn't have one. I'm still a little fuzzy on whether and how this is a problem. Is it because too much ground current would have to flow through the resistors in the virtual center tap? Some other reason?

My 5V supply is derived from the filament supply via a bridge rectifier, which is grounded to the chassis, and it seems to be working fine... Is that bad?

[Dave H]

Nope. I thought i just hadn't connected it, but on closer inspection, it doesn't have one. I'm still a little fuzzy on whether and how this is a problem. Is it because too much ground current would have to flow through the resistors in the virtual center tap? Some other reason?

Yes, it's the current in the CT.

My 5V supply is derived from the filament supply via a bridge rectifier, which is grounded to the chassis, and it seems to be working fine... Is that bad?

It's bad. It can only be working because of the resistance of the artificial centre tap. If it had a real centre tap one of the diodes in the bridge would be a short across half the 6.3V winding.

[loudthud]

Good catch Dave H!!! That circuit needs the center tap to work. The circuit you are using now will probably be ok, just check the output with your scope set to AC coupling to make sure that there is very little ripple on the 5V. The 7805 needs some extra voltage to work properly and the data sheet doesn't spell it out very well. You'll start to see little negative going spikes on the output from the ripple if there is not enough voltage from the rectifier and filter cap.

You can use 9V batteries to power the opamps. They don't draw much will last quite a while if you disconnect them when the amp is turned off.

[nickb]

Nope. I thought i just hadn't connected it, but on closer inspection, it doesn't have one. I'm still a little fuzzy on whether and how this is a problem. Is it because too much ground current would have to flow through the resistors in the virtual center tap? Some other reason?

My 5V supply is derived from the filament supply via a bridge rectifier, which is grounded to the chassis, and it seems to be working fine... Is that bad?

If you have CT of the transformer filament winding connected to ground then it bad as you have created a short circuit, exactly as Dave says. OTOH, if you have a virtual ground using resistor it in effect supersedes the resistors and making them redundant so might as well remove the resistors.

Here is the five volt only version that I was thinking about in post 13 Help me diagnose an op amp reverb driver circuit I built for belton "digilog" module.


I chose that opamp as it will work off a low power supply voltage and it's than an LM358 and so more suitable to use on the input. Note I made the resistor values high to give a higher input impedance. If you want this higher than 180K then we'll may have to add a buffer. I'm sure this could be refined further...

You might find this easier power supply wise. Anyway, if nothing else you can see there is more than one solution, it's food for thought and you're getting a free education too

[elipsey]

Good lookin out Dave, I see what you mean. If I drew a picture of a 6.3-0-6.3 filament supply connected to a bridge rectifier, two of the bridge diodes would obviously be shorting the AC leads, but I just plugged it without thinking about it. I did put a probe on the filament supply the first time I tried I connected the rectifier so I would know if I shorted something or screwed up. Hopefully I could have got away with that for one or two seconds before I wised up and switched it off... Lucky I don't have a physical center tap.

Thanks for the schematic nickb, that's nice work!

It looks like one consequence of using a small power supply would be that the input signal must remain very small. newb question here: Does 270mV pk mean .540Vpp, and that would look like about ~.2Vrms on my meter or something?

Anyway, .2V or .5V, or whatever isn't very much, so I guess the input would have to come from after the first gain stage at most, maybe right off the input jack like loudthud was saying. does that mean that the reverb would distorted be cascading gain stages later in the amp?

I thought this would be more complicated than I thought, but it's actually more complicated than I thought I thought it would be.

[nickb]

It looks like one consequence of using a small power supply would be that the input signal must remain very small. newb question here: Does 270mV pk mean .540Vpp, and that would look like about ~.2Vrms on my meter or something?

Anyway, .2V or .5V, or whatever isn't very much, so I guess the input would have to come from after the first gain stage at most, maybe right off the input jack like loudthud was saying. does that mean that the reverb would distorted be cascading gain stages later in the amp?

Yup, 0.27Vpk =0.54Vpp. Meter reads rms which is 0.191V

One limitation you have is the Belton module input at 1.5V pk. I picked the values so it would work on the input to sidestep overloading issues. Using bigger voltage rails will help with the input to the opamp range limits but you then have to reduce the signal to feed the Belton module. The 270mV is what you get if you work back from the Belton 1.5vpk and divide by the gain of the input opamp. I chose the gain to get 270mV which I felt was a decent number - that's debatable for sure I didn't want to have too small a gain under and so underdrive the Belton as that might make the digital nature ('quantizing noise') of the Belton apparent.

Yes, if you were to place this after the first stage you will overload it. The first stage could easily output 5 volts AC but we still don't know what we are dealing with do we, hint, hint? If you want to insert after the first stage then you will need the bigger output voltage swing and that means bigger voltage rails. OTOH the noise will be better. That I suspect is the nub of the issue. After first stage means complex power supply but less noise. Before first stage simple power but more noise. Make your choice...!

[nickb]

Oh. BTW if you haven't looked at the the link teemuk supplied, you really should. A completely fresh perspective ( how does no opamps or power supplies sound). Check it out!

[elipsey]

Oh. BTW if you haven't looked at the the link teemuk supplied, you really should. A completely fresh perspective ( how does no opamps or power supplies sound). Check it out!

Are you refering to the following link?

Then, I suppose you didn't study the documents at
Digi-Log Modules - neunaber audio effects
Do it now.

Ahem... Either teemuk has posted some other link, which I have overlooked, or I am not taking the point. As I discussed at some length in Post #18, I did not notice that any of the documents which are linked from that page, other than the Belton datasheet whose only schematic I have already used to implement my circuit, seem to directly address my application. If I have overlooked a helpful or relevent document, please tell me which one so I can pull my butt out of my ass right away and go read it.

Which of the several documents on the page teemuk linked do you refer to?