Idea for a dedicated Rhodes preamp, need help !

[PermO]

So, I did some more study on OPA2134 and came up with this for my input stage;


Still not shure about this...

I've got a 6 pos 2 pole Lorlin, so I did not draw all the RG A and B positions, but I figured this setup would allow for adjustable gain setting...

A is the output for the balanced linedriver, B is the output for the pedals, I put a coupling cap there.

As there will be different gain requirements for the A and B output I figured that the RGA resistors can be a different value then the RGB resistors to keep the outputs providing the right level for either the 1646 and the pedals.
Or just keep them the same and create an output pad for the pedal output ?

Am I overcomplicating things ?
Will this setup cause the smoke again ?

When my chip arrives this week I will breadboard it with just RGA1 and RFA, but I would like to know if this configuration makes any sense for a instrument preamp design.


Thanks !

 

[zzzzz]

Just posting because I'd like to follow this and mess around as I've been thinking of doing the a similar thing for my Rhodes stage for years to get a good sound for home recording without the need for an amplifier or lugging it around to my studio.

 

[abbey road d enfer]

I've got a 6 pos 2 pole Lorlin, so I did not draw all the RG A and B positions, but I figured this setup would allow for adjustable gain setting...

I doubt you need so many gain options. Three seems all you need.

A is the output for the balanced linedriver, B is the output for the pedals, I put a coupling cap there.

As there will be different gain requirements for the A and B output I figured that the RGA resistors can be a different value then the RGB resistors to keep the outputs providing the right level for either the 1646 and the pedals.
Or just keep them the same and create an output pad for the pedal output ?

I Think one of the outputs should follow the other; the first in the chain being the global sensitivity control, the other being a slave. And you'll probably want attenuation for the pedal output.

Will this setup cause the smoke again ?

Shouldn't.

When my chip arrives this week I will breadboard it with just RGA1 and RFA, but I would like to know if this configuration makes any sense for a instrument preamp design.

You don't need to load the input twice. The unmarked resistor should go.

 

[Newmarket]

I Think one of the outputs should follow the other; the first in the chain being the global sensitivity control, the other being a slave. And you'll probably want attenuation for the pedal output.

I reckon that's "six of one, half a dozen of the other" really. Functionality vs Gain then Attenuation.
Depends on user's priorities I guess.

But more than that I'd say - you need a series resistor on the Pedal Output - to avoid instability with cable capacitance. Then duplicate that to 0V on the 'Cold Leg' for an impedance balanced output. (or "Approximately Impedance Balanced Output" for those who insist on objecting to this approach) Works with unbalanced patch lead into pedal and you're laughing if you feed it to a balanced input with a balanced cable.

 

[5v333]

I agree with having some buildout resistance at the output going to a pedal/amp. But impedance balanced out doesnt seem necessary..?

The other output is aiming for balanced gear so that will cover in thoose situation.

 

[PermO]

@5v333 that is correct.

Also got a message from CCaudle to not blow anything up yet as he would like to draw me something, so I'll be patient.

The chip is not in the mail yet, so it's save

I like the idea of adjustable gain setting as it would make the box more versatile, I found this concept used in a hifi amp and liked the idea of a "VCA like" control instead of running the output trough a pot.

The unmarked R makes no sense indeed, and 3 or 4 gain steps would probably be plenty, but I've got this cheap 6 pole switch so...

Loading the pedal output with a resistor, check... I also noticed in many designs the input is loaded with a series resistor, those where mostly aimed at guitar.

 

[Newmarket]

I agree with having some buildout resistance at the output going to a pedal/amp. But impedance balanced out doesnt seem necessary..?

The other output is aiming for balanced gear so that will cover in thoose situation.

Yeah - but you might at some time want to feed two balanced inputs and it's essentially a very very low cost option with no real downside for unbalanced connections. Since resistor value can be low at, say, 68 Ohm.
Personally I'd probably recommend another cross coupled output like the THAT or a Ground Sensing Output but the Impedance Balancing has the advantage of ultra simplicity.

 

[Newmarket]

@5v333 that is correct.
Loading the pedal output with a resistor, check... I also noticed in many designs the input is loaded with a series resistor, those where mostly aimed at guitar.

Yes. But guitar application isn't the important thing. It is isolating the output from the cable capacitance and stopping the feedback getting out of phase to the degree that it caused the output to oscillate - basically chasing its own tail at high speed !

 

[5v333]

Yeah - but you might at some time want to feed two balanced inputs and it's essentially a very very low cost option with no real downside for unbalanced connections. Since resistor value can be low at, say, 68 Ohm.
Personally I'd probably recommend another cross coupled output like the THAT or a Ground Sensing Output but the Impedance Balancing has the advantage of ultra simplicity.

In that case i would recomend wiring up a trs jack in a way that unbalanced connections would short the ring to ground. While balanced connections would make use of both resistors at the output.

However i would just use one output
unbalanced for pedals etc and simulate higher impedance coming from a guitar for ex by having a series 10k resistors.

 

[PermO]

Yes, I would like the FX output to be a dedicated output to drive pedals with the right level and output impedance so it will work with any pedal you hook up.
The 2134 is perfect to buffer without adding any substantial noise.

Unbalanced jack for pedal interfacing.

The XLR output with the THAT balanced line driver is all you could wish for when recording or driving a (transformer) balanced input of a compressor / EQ / Mixer.

So you can drive a professional recording chain, and a bunch of pedals and record both.
To have a nice clean signal from the piano in the DAW is also great for reamping and processing in a virtual enviroment.

 

[Newmarket]

In that case i would recomend wiring up a trs jack in a way that unbalanced connections would short the ring to ground. While balanced connections would make use of both resistors at the output.

Exactly - that's what I meant.

 

[Newmarket]

Yes, I would like the FX output to be a dedicated output to drive pedals with the right level and output impedance so it will work with any pedal you hook up.
The 2134 is perfect to buffer without adding any substantial noise.

I see. Bear in mind that a high output impedance is required only for specific fx pedals - basically Fuzz Face and derivatives - where the intended effect relies on a high Z source. Meaning that this pedal needs to be first in any chain of fx. There's a noise penalty for using high Z series resistance so I'd think about making that switchable if practicable to do so.

 

[ccaudle]

So, I did some more study on OPA2134 and came up with this for my input stage;

I would like to first point out some small things which are conventions that will make it easier for other people to help comment on your design.
Schematics should be drawn in such a way that they allow understanding the circuit function as quickly and easily as possible, so that the least amount of mental effort is spent on small details such as which pin is which function, and avoiding having to consult the datasheet for every detail. As such, the pins are usually labeled in the schematic symbol with the function. For amplifiers that is just the inverting and non-inverting inputs if there is only one output, or if the amplifier has differential outputs the inverting and non-inverting outputs would also be labeled. For multi-function devices, the separate functions are typically drawn as separate schematic symbols so that the components associated with each section can be arranged in the easiest to read manner.
With that in mind I have redrawn your original schematic from Saturday using those conventions (and also US style resistor symbols just because that is my habit):




I believe that should be equivalent unless I made a mistake. That is really just an aside so that it makes comparison with the alternate suggestion easier (which I created in that conventional style).

As suggested by "Abbey Road" above I have drawn an input stage with switched gain, and a second buffer stage to drive the pedal output.
Discussion below the picture about some components which may not be needed, and alternatives to achieve similar.



I have included the component C2 as a leading bypass around R2, "lead" in this context referring to the phase response in the feedback network. It may not be needed, depending on the op-amp used and the parasitic capacitance at the inverting input node.
To reduce the parasitic capacitance at that node, I have shown the gain setting resistors split into an R3 component connected directly to the node, and resistors R4-R6 which are actually switched, on the assumption that the switch will be located on a panel and so connected with wires at least multiple cm long. Note the "make before break" designation for the gain switch, if the switch is of a style which allows the traveling contact to temporarily move to a position with no resistors connected, the gain will drop to unity (gain of 1, aka 0 dB) and then suddenly change to the selected gain when the switch contact closes. That is likely to make large "click" or "thump" signals at the output. Even a make-before-break style switch might do that, but my assumption is that the disturbance will be less if the gain does not change to 1 suddenly. That assumption would need to be verified to make sure it is not actually worse sounding transitioning temporarily to a higher gain when the two resistors are connected in parallel for a short time.

Capacitor C3 is to cause the gain to drop to 1 at DC. The offset voltage of the op-amp is an intrinsic feature of the amplifier, caused by imperfect matching between the devices in the input stage. That offset voltage will be amplified by the gain of the stage, so for a 30 dB gain in the first stage, that could multiply the maximum 2mV offset voltage by around 30 and result in 60mV DC offset at the output. The 1646 has a gain of 2 (6 dB) so would be 120mV maximum at the output of the 1646.

Possibly you do not care about 60mV or 120mV, in which case C3 could be removed. Another alternative could be to put a coupling capacitor on the output of the first stage, blocking any DC offset from the first stage to the buffer and the 1646. Since the buffer is configured as unity gain (gain of 1 by direct feedback connection of output to inverting input) the offset voltage of the second stage is not amplified, and so should be 2mV maximum (possibly up to 3mV at elevated temperatures, see electrical characteristics in section 6.4 of the datasheet). The second amplifier section of the OPA2134 will need a resistor to GND in that case, the input to an op-amp must always have a DC path for bias current to flow. The DC path can be from the output, which is why the non-inverting input does not need a resistor, any bias current needed can be provided from the output. The non-inverting input as drawn can get bias current from the output of the first amplifier section, but if a coupling capacitor is inserted that DC path is then broken, and another DC path (i.e. through an input resistor) will be needed.


The gain of a non-inverting op-amp circuit as drawn is 1 + Rf/Rg (using the previous naming convention for feedback resistor and grounded resistor), where the Rg component is the combination of R3 and whichever resistor is connected by switch, or just the switched resistor if you omit R3.

We had previously discussed gain in terms of decibels, which is a common designation for pre-amp gain setting, but is not directly translated to the resistor ratios. Decibel scale is logarithmic so that you can choose between e.g. 0 to 60 dB gain rather than 1 to 1000.
For voltage gain, the conversion between raw gain and dB is:
db_gain = 20 * log(gain) (that is base 10 logarithm, not the natural logarithm).

So the previously discussed 30dB gain would be:
30 = 20 * log (gain)
gain = 10^(30/20) = 31.6

So using the gain equation above:
32 = 1 + Rf/Rg
31 = Rf/Rg

You will probably have to adjust that slightly based on resistor values you can easily get. Since you do not have to match exactly some existing gain range you can pick whatever works easily (or even use a pot in place of the switched resistors for prototyping).

Section 7.3.3 of the datasheet discusses effect of source impedance on distortion performance. My understanding of the pickup configuration of a Rhodes piano is that the DC resistance is about 2.5k ohms, presumably higher impedance at higher frequencies due to coil inductance.
Based on the recommendation in section 7.3.3 I would aim for the parallel combination of the gain setting resistors to be somewhere around the range of 2.5k ohms. That is not really critical if it is difficult to achieve, it only affects distortion performance a little.
At the gain ranges you need that will probably not be practical, so for now just consider that something to keep in mind for future designs where you might want lowest distortion.

If you have a 10k ohm resistor conveniently available for the feedback resistor:
31 = 10k/Rg -> Rg = 10k/31 -> Rg = 322 ohms.
If you are using 1% resistors 324 ohms is a standard value, if you are using 5% resistors then 330 is probably the closest standard value.
That is assuming you are just connecting the resistors directly for prototyping, and not using R3 as drawn to help isolate the op-amp input from the switch wiring. If you use a pot in place of the switched resistors with the wiper connected to GND, then R3 serves the purpose of setting the maximum gain, and the minimum gain is determined by the combined value of R3 and the pot when the wiper is at the top of the resistive element.

That is probably a lot to take in, hopefully it is a useful balance between being too basic on one extreme, and assuming you already know how to design circuits on the other.

 

[PermO]

Thank you very much for this comprehensive reply.
This is all the info I need in one place.

My opamp has to come from UK and is delayed, also ordered the "small signal design" book.

I think I will go with the "Abbey Road version" as there is less complex switching involved.
And I was also wondering about my Lorlin switch, they're cheap and pretty robust but indeed the switching introduces a 'no contact' moment and the switching 'pop' that goes along with that.

I'll have to read this several times to make sense of it all, but it is very helpfull to have all the info in one place.

I'll draw a full schematic and see if I can figure out the correct values.

 

[PermO]

Well, I finally got my OPA2134 in.... that took a while.

And this weekend I received the Douglas Self Small Signal Audio Design book, I have not finnished it, lol, but that's really a great book, what I've seen so far it's very well written (for a layman like me to understand) and it comes with lots and lots of schematics.
Last monday I had had a friend over for dinner, he was browsing my book and ordered it later that evening.

That was a great tip, money well spend !

I prefer paper over a computerscreen anytime...

 

[PermO]

Ok, so I'm still not getting this...

I hooked up a simplified version like this;


And what I get is a very loud signal / noise as soon as I switch it on.. so I switched off right away, no smoke, checked all the connections, they all seem to be following the schematic, no shorts... try again, an get the same loud noise at the output.

So I don't get it,... still

 

[abbey road d enfer]

And what I get is a very loud signal / noise as soon as I switch it on.. so I switched off right away, no smoke, checked all the connections, they all seem to be following the schematic, no shorts... try again, an get the same loud noise at the output.

With 10 Meg input impedance, the circuit picks all sorts of interference. It must be enclosed in a conductive shield connected to the circuit's ground and the input must be shorted for the preliminary tests. Then you must connect a suitable source, such as an electric guitar or bass and start with the volume off and slowly increase the volume.

 

[Brian Roth]

There is right at 30 dB of gain in the first stage.....

Bri

 

[PermO]

With 10 Meg input impedance, the circuit picks all sorts of interference. It must be enclosed in a conductive shield connected to the circuit's ground and the input must be shorted for the preliminary tests. Then you must connect a suitable source, such as an electric guitar or bass and start with the volume off and slowly increase the volume.

Ah, thanks,...
The top of the box is still open indeed, the PSU is shielded though.
I have 4 input resistors on a switch, I'll select the lowest value of 47K and short the input and see what happens.

There is right at 30 dB of gain in the first stage.....

Bri

Yes correct, the idea is to have a selectable gain by adding extra resistors via a rotary switch.
The 330R should be the maximum amount of gain...

Maybe not the best way to test this setup.. I'll replace that for a higher value to start with.

 

[PermO]

Yup, that fixed it... went for a 2.2K instead of the 330R and tried again... that provided a very decent level with my (passive pickups) Tele...

Sounds pretty good for a DI at first notice...

Now I have a baseline to work from and select my resisitors on the rotary switch.

Another thing I can't figure out is the C2 in the schematic by ccaudle (wich I am using as my lead) in post #73, do I need it ? is it good practise to put it in anyway, and what value are we talking about, is this a nF thingy or more than 1uF ?
I have enough small capacitors, I don't know what will do the trick