BAJA discrete differential low current low noise guitar preamp [documentation]

[bajaman]

Here is a little onboard guitar pre-amplifier I have been using that has a very low current draw ( battery will last for years), and very low noise. Over the coming posts in this thread I will attempt to show how it can be used in a variety of ways to fine tune the tonal response of any guitar it is installed in. Okay, so why another active guitar onboard pre-amplifier ? Well, most of the designs I have encountered quickly flatten batteries, especially if you forget to unplug your instrument cable from the jack socket when not using the guitar. With this pre-amp you can leave it plugged in for over a year before the battery is flattened ! Most low current draw op amp designs using TL06x series draw ten times as much current as this pre-amp, so your battery is going to last a tenth of the time that this one will ! The exception of course is the programmable current LM4250 op amp which Leo Fender famously used in the original Music Man guitars - especially the Stingray bass
This new pre-amplifier is a very simple discrete op amp with a non inverting input , inverting input, and output. It can connect between your guitar pickup (or selector switch etc) and use your existing 250k or 500k volume control. It can be configure to give +6dB to +18dB of gain by adjusting the value of four resistors (R1,R2,R3,R4) - very useful as a clean gain boost. Winding back the volume control does not wind back the top end response like a passive guitar setup, and you can run any length of instrument cable without losing treble because the pre-amplifier removes the effect of increasing cable capacitance interacting with the pickups top end resonance as it does in a passive system.
This pre-amp can be used with balanced or unbalanced inputs - more to follow
Ideally the differential pair of BC550C transistors should be selected for matched hfe for best possible response - most digital multimeters have an hfe test function on them.
cheers
bajaman

Baja discrete differential preamp 36uA current draw low noise 230821.GIF (9.37 KiB) Viewed 2804 times

[bajaman]

to use the preamp in differential mode, connect the pickup windings (start and finish of coil) to the non inverting and inverting inputs - do not connect the coil to ground ! The metal baseplate or any shielding wires can be connected to ground - this is known as the balanced mode. To use in unbalanced mode, just attach the inverting input to ground.
The gain is set by the ratio of R2/R1 and R4/R3 - ideally R1=R3 and R2=R4, but this is not necessarily ideal when dealing with magnetic induction pickups - more on this later
The value of R1 +R2 sets the damping of the pickup's resonance - too low will muffle the top end and too high will cause ringing - more on this in a later post

As a working example: I wound a firebird mini humbucking pickup - 2.7H inductance, 198pF capacitance and 7.18K ohms dc resistance. Normally the response curve for this particular pickup would show a 24dB resonant peak in it's response at approximately 6.8kHz - with the parallel connection of a 500k volume control and 500k tone control (250k ohms ), this peak in it's response curve is damped to a 6dB peak at a slightly lower 6k4Hz - better, but still not ideal . In fact the value of damping resistance required to damp the peak completely and thus enabling a flat frequency response pickup is closer to 90k ohms - adding a 141k ohm resistor across the coil windings would achieve this, or perhaps it would be better to use 250k volume and tone controls with this particular pickup ? Anyway, remember that 90k ohm resistance, because this is what R1 + R2 must equal to damp that pesky resonant peak. Hopefully you are all still following this .

I realised this resistance by using 22k for R1 and 68k for R2 to damp the resonance and 680k for R4 and 220k R3 to set the gain of the preamp. I know these latter resistors are a factor of ten higher than R1 and R2 - there is a reason for this that I will attempt to explain in the following posts .

The firebird pickup was connected to the non inverting input and ground - the inverting input was also connected to ground - this is the simplest unbalanced preamp configuration and yields a flat response pickup with a 3dB down low pass rolloff at approximately 7.8kHz .

Okay, so what is the voltage gain of this preamp ? - simply (R4/R3) or roughly 3 times the voltage gain or 10dB - a very useful boost. Of course, this gain could be increased or reduced by changing the values of R1 and R2, as long as the sum of these resistances adds up to 90k, and the values of R3 and R4 are adjusted by the same ratio.
cheers
bajaman

Baja firebird pickup preamp example 230821.GIF (10.07 KiB) Viewed 2774 times

[bajaman]

Here is another example, this time a Baja wound stratocaster pickup wound with 42awg polynylon covered copper wire and alnico 3 rod magnets.
The coil measured: 2.63H inductance, 6.3k ohms and 138pF capacitance (inductance was measured @100Hz and capacitance @100kHz)
Unloaded the coil has a 27dB peak at a resonant frequency of 8.3kHz - loading with 125k ohms ( two 250k pots in parallel to represent the standard stratocaster volume and tone controls) still shows a slight 1.5dB hump in the response centre around 5kHz - it took 100k to fully damp the resonant peak and give a smooth rolloff
Keeping the same R1 R2 and R3 R4 ratios and aiming for a similar 10dB gain required R1 to be 25k and R2 to be 75k (in practice i used 22k and 75k and achieved a 9dB boost with a -3dB point at 8.7kHz
cheers
bajaman

Baja stratocaster pickup preamp example 230821.GIF (10.33 KiB) Viewed 2739 times

[Jarno]

Really nice, if one was to make this in SMT it would really really be tiny (a good 1u cap will be an issue then, I think).
Would an active current source be a way to get current consumption even lower (or lower noise)?

[dylan159]

Very interesting. BC550 are definitely a good choice for low noise. My worry is for the overall large resistor values, on both inputs and elsewhere, those could compromise the low noise intention. Maybe you can run some simulations, or me for you.
Another interesting choice is to have a somewhat low input impedance to damp the resonant peak. That's completely personal and acceptable of course, but if you're going to damp it, you can do it after the cable too.
Maybe an option is to set the gain only with the feedback, using lower value resistors, thus getting rid both of those two in the feedback and the series resistance at the (+) input. I don't know if it's possible to reduce the load and tail resistors without increasing the quiescent current or using a current sink. Since it's AC coupled, maybe? Let me know.

[bajaman]

Hello Jarno - no need for complicating matters with an active current stage - there is only 4uA current through each of the BC550C differential pair - I used a 1uF multilayer ceramic (mlcc) - works well as is only a bias setting component to set operating current.
hello dylan159 - the large resistor values are necessary to run at such a low current - in practice the preamp is extremely quiet - no sign of any thermal noise at 4uA ! The "low input impedance" is actually the same as in a passive guitar setup where the resonant peak of the pickup is damped by the volume and tone potentiometers which are similar in impedance - this preamp is connected direct to the pickup and the guitars existing volume and tone controls are then connected to the preamp's output. However, now when the volume control is turned down there is no loss of treble as in a conventional passive guitar setup and the influence of the cable capacitance interacting with the pickup which effectively lowers the pickups resonant peak frequency and hence it's top end response is completely gone
Oh - I forget to mention that unlike most op amps that have class B (or AB) output stage, this one has no crossover distortion because it is operating in true class A
cheers
bajaman

[dylan159]

hello dylan159 - the large resistor values are necessary to run at such a low current - in practice the preamp is extremely quiet - no sign of any thermal noise at 4uA ! The "low input impedance" is actually the same as in a passive guitar setup where the resonant peak of the pickup is damped by the volume and tone potentiometers which are similar in impedance - this preamp is connected direct to the pickup and the guitars existing volume and tone controls are then connected to the preamp's output. However, now when the volume control is turned down there is no loss of treble as in a conventional passive guitar setup and the influence of the cable capacitance interacting with the pickup which effectively lowers the pickups resonant peak frequency and hence it's top end response is completely gone
Oh - I forget to mention that unlike most op amps that have class B (or AB) output stage, this one has no crossover distortion because it is operating in true class A
cheers
bajaman

Collector and tail resistor need to be high for low current and 9v supply, but the input and feedback ones don't, right? my idea was to remove the series resistor at the +in, compensate the input impedance, and compensate the gain in the feedback divider, all while scaling those last two down by 10 or so. On this point, maybe the two ltp transistors don't need to be matched after all, since the circuit is all ac coupled and with dc feedback, but I'm not too familiar with it so tell me if I'm wrong.
About this being before the guitar controls, I had missed that, makes sense. Still, it relaxes one requirement with input impedance, not sure if it can be exploited.
Thermal noise: do you have measurements? Again I could simulate it if you want. Not being audible doesn't mean it couldn't make a noticeable difference in front if a distortion pedal for example.
You're right that op amps usually aren't class A output, but unless it's a 358 or similar, is the distortion audible? If it's good for hifi it's 50 times good enough for guitar.

[bajaman]

"On this point, maybe the two ltp transistors don't need to be matched after all, since the circuit is all ac coupled and with dc feedback, but I'm not too familiar with it so tell me if I'm wrong."
yes - you are wrong - please read up on differential pair input design
"my idea was to remove the series resistor at the +in"
no - do not do that, just wait and all will be revealed or in the meantime, do a search for differential amplifier design
"compensate the input impedance, and compensate the gain in the feedback divider"
not sure what you mean here ?
"Thermal noise: do you have measurements? Again I could simulate it if you want"
no, i do not have measurements - feel free to simulate if you want to
"is the distortion audible? If it's good for hifi it's 50 times good enough for guitar"
perhaps so perhaps no - is it perceivable though ? totally disagree with the last comment sorry.
cheers
bajaman

[bajaman]

okay - the continuing unfolding story
And now, an example of using the preamp in the balanced differential mode
A few years ago I wrote a thread on this forum about active pickup design based on the EMG 81 pickup.
If you search this forum you will surely find it along with a few active EMG tone correcting snippets etc.
https://www.electrosmash.com/emg81 is worth checking out
I revisited my active pickup design and measured the individual two coils I wound which have both their finish winds connected to ground
I used this meter : https://www.ebay.com/itm/264863029607?e ... Sw2rNfXOyg
I measured each coil's inductance @100Hz, capacitance @100kHz and dc resistance
Feeding these figures into my simulation program and adjusting damping for each coil gave me the following preamp example

This preamp has approximately +14dB of gain and has an essentially flat frequency response from 50Hz to 5kHz - it has smooth damped rolloff above 5kHz and is -3dB down at 8.8kHz
cheers
bajaman

[bajaman]

Here is a picture of the smd version of the preamp - 32mm x 15mm x 1mm thick, so should easily fit on the back of most guitar pickups

Baja discrete differential preamp 36uA current draw low noise 230821.GIF (11.68 KiB) Viewed 2573 times

A through hole version is currently at the pcb fabricator
cheers
bajaman

[Jarno]

Oh yeah, that is interesting, so instead of passively combining two coils in a humbucker, you could do it actively, using the two inputs on the preamp.
This way, you could also use one single coil pickup and a dummy coil, Alembic-style. Interesting!

[bajaman]

"so instead of passively combining two coils in a humbucker, you could do it actively, using the two inputs on the preamp"
yes - you got it This is how EMG have been doing it for years - a major shortcoming or problem with the traditional Seth Lover designed Gibson style humbucker is that the coils are connected in series and in a PAF wind they usually measure around 2.3H each coil - one would expect that the total inductance of the series connection would then be 4.6H, however due to mutual inductance they actually measure closer to 5.1H !
In the EMG design which utilizes a differential preamp, the coils are still connected in series but with one very important difference - the junction of the two coils is grounded ! Now each inductance is treated separately, one by the non inverting input and the other by the inverting input - being a differential configuration, the common mode (or induced hum signal in this case) signal is reduced much more efficiently than merely connecting the coils passively and out of phase with each other. Top end response is improved and hum is reduced to insignificant measurements - a win win situation
cheers
bajaman

[bajaman]

here is a picture of the through hole version of the preamp - it measures 42mm x 17.2mm - not much bigger than the smd version

Baja discrete differential preamp 36uA current draw low noise through hole version240821.GIF (17.89 KiB) Viewed 2545 times

cheers
bajaman

[dylan159]

please read up on differential pair input design

I did both before the first message and after, within my limits. The fact is that this is more than just a LTP, having the following DC coupled common emitter and the DC feedback, so for certain things I still have to find out for myself using the means at my disposal.
I've run some simulations both to find out where I was wrong, and to give you additional data. Sorry I'm cluttering the thread, but this is for the sake of discussion and that's what forums are for after all
I think the list of things I want to explore is the following: bias and input offset voltage (at the bases), gain, cmrr, input impedance and noise, distortion.
To challenge my assumptions I compared in LTspice the thunderbird pickup version with one of my own.
The first point is bias, offset between the two bases and need for matching. For this I made one of the transistors in the LTP a 3904 which is pretty awful for matching. There are other differences in the schem which will be explored next:

Experimenting with this and dc Q point, I realized how important is the dc feedback from the output stage collector to the inverting input base is. A common strategy to avoid matching in a simple LTP is emitter degeneration. There there's not much choice, but here that actually worsened matters, because it reduced the open loop gain. So what if I went the other way and increased the open loop gain by grounding the emitter of the third transistor? That improved matters quite a bit, and the offset between the two bases went down to just 1.4V more than the original version, while using a 3904!
Ok, but what about CMRR, that is known to benefit from matching, at least for the differential input version, where there's common mode hum to cancel. To test this I connected in both the input signal to both inputs, and gave them both a 22k/68k divider (this in both circuits). Results are the following, with green being the version i was testing. Maybe the increase is caused by the increase in open loop gain? (roughly 1539 instead of 1468)

Next is gain. Again speaking for the single ended use, I suggested removing the voltage divider at the noninverting input and setting the gain entirely through feedback, all while scaling down the impedances in the feedback path. As in the above schematic, the input bias resistor has been made 90k to equal the series combination of R1 and R2, and the feedback set for 10dB of gain:

Just to check, the input impedance at the noninverting input is the same for both(used different sources so the inputs are not paralleled):

Then there's noise. Here it is for the firebird schematic, including the pickup model. Original:

Test version:

Finally, distortion with a 100mV 1k sine measured to be 0.68% in the original, 0.672% in the experimental version.
Now is where I ask you to explain to me, I've tried to go as far as I could on my own, but if there's something else I'm missing, please tell me and it will be a learning moment.

[dylan159]

Final post scriptum consideration: headroom. The 47k is an heavy load on the 68k collector resistor, limiting output headroom at about 1V less.
Here's a tradeoff between noise, current and headroom. We could scale up the feedback by 10, going back to the original headroom at the cost of noise, which is still lower since there's no noninverting input divider.

But there's another way, and it's rebiasing the output stage for a more symmetrical swing. This costs us in quiescent current all other things being equal, but for example changing the bias divider to 1Meg/1.5Meg, bringing the quiescent current up by 75uA but with headroom benefits: a 1V input with the same gain gives 0.5% THD, while the original is clipping (14%). I don't consider this a solution though, because it goes a bit against the low current goal, even if it's just 0.1mA total.
Then maybe, in the vein of tradeoffs, biasing with 2.2Meg/1.5Meg gives a bit more headroom than the original, still with the suboptimal feedback, while only costing 37uA more current.
I get why those choices have been made, and I hope I'm finally getting the whole picture, but my observations still seem valid and resulting in a working circuit that performs as well.

[bmxguitarsbmx]

Sweet stuff Bajaman! What size surface mount resistors are these? 0805? Many thanks!

[bajaman]

they are 1206
cheers
bajaman

[bajaman]

Thanks dylan159 for the simulations and observations
I must admit that i was not paying enough attention to headroom issues - most pickups have only small output levels in the 100mV to 200mV peak region and in practice I have not heard any nasty distortion in my own working examples. However, I admit that the main gain stage (the BC557B) is not ideally biased for the maximum undistorted swing - my main objective was to achieve a very low current drain to enable very long battery life and i mistakenly overlooked where the collector voltage was sitting - too low !
I had 3 objectives when designing this preamp - low current draw, ideal pickup peak resonance damping, and removal of capacitive loading of the guitar cable on the pickup's high end response. I feel I have achieved these objectives but overlooked headroom which may be a problem with some very high output pickups, so i have revisited the circuit to address this shortcoming.
I have run some simulations and made some component adjustments - I have removed the 1k emitter resistor as it is no longer needed, reset the biasing and adjusted resistor values for balanced currents through the long tailed pair. The amended circuit now consumes 43uA, and the output collector sits at 5.3v DC for much improved headroom
Kindly bear in mind that this preamp can be used two ways - if using unbalanced with the inverting input connected to ground then R1/R2 = R3/R4 or R1+R2 = R3+R4 - R1+R2 are chosen to provide optimum critical damping of the pickup's peak resonance and will vary according to the pickup's characteristics (inductance, resistance, capacitance), also C2 should be at least 10 times the value of C1
If using in the balanced or differential mode the ratio R1/R2 = R3/R4 still holds true BUT R3+R4 will need to be increased in value because the impedance of the inverting input is a lot less than the non inverting input - using the same value resistors will overdamp the pickup coil's resonant peak and severely restrict the preamps high frequency output level. for this reason, the values of R3 and R4 need to be increased 7.69 times in my example. I cannot give exact scaling for other pickups, but a value 8 to 10 times the value of R1 and R2 would get you in the ballpark C1 =C2 in the balanced configuration !
cheers
bajaman

Baja guitar preamp 43uA current draw 230821.GIF (8.72 KiB) Viewed 2422 times

[bajaman]

Here is the updated Baja Firebird pickup and preamp example, shown connected to the guitar's existing 250k log volume pot
cheers
bajaman

Baja firebird pickup preamp example 43uA version 230821.GIF (10.52 KiB) Viewed 2408 times

[bajaman]

I just tested the actual preamp (and firebird pickup) with the modified resistor values and noticed the onset of clipping on my oscilloscope at 480mV peak sine wave signal input @1kHz - more than enough headroom now methinks
By the way, i should emphasize that this preamp design is not suitable for connecting as an outboard unit - the input impedance is far too low if placed after the guitar's volume and tone controls - It must be connected directly to the pickup, with the guitar's existing controls connected to the output of the preamp. i may take a look at developing this circuit as a stand alone boost, treble boost or even distortion pedal - if so, it will not be a low current design and will be the topic of a separate thread

cheers
bajaman