Monday, 16 November 2020

Clean Jazz Guitar Amp Builder Notes — Part 3: Pre Amplifier

This is part 3 of a series about a complete prototype guitar amp.

Part 1 lies here
Part 2 lies here

Preamplifier Stage Notes:

I performed most of these experiments last year and better type these notes now as I'm forgetting some details.

I sought a low noise, clean amplifier. In the head of most jazz guitarists an ideal guitar tone sits —
it might be Jim Hall in 1965, or Tal Farlow in 1957, or Julian Lage in 2020.  I've discussed this topic previously and it essentially defines subjectivity. Tube versus solid state? 12 inch versus 10 inch speakers? Single coils versus humbuckers? The microphone!  Was EQ applied by the recording engineer? And so on. Let's glide past this dribble and assume the 'ideal guitar sound' in your head sounds as good as mine.

I chose to power 2 speakers bolted in an open back cabinet. I feel open back cabinetry sounds more natural and previously felt 100% confident that sealed, ported speaker cabinets were the only way to get my jazz tone. Bah non, t’as fait n’importe quoi. Thus my tone shaping lies orientated to driving an open-backed speaker cabinet while still getting some thumping bass response at low, practice-level volume if desired.

First up comes the DC power supply for the op-amps. For maximum op-amp headroom, I ran the split supply DC as high as I could safely manage by choosing 17 volt zener diodes.


Above — I built the voltage regulators on the Stage 1 preamp board. These hulky transistors seem excessive, but in an earlier version, a BD139 failed and smoked up my lab. Don't get me going about how bootleg parts have polluted our parts drawers and reduce some of the joy. These 20 year old, Motorola TIP BJTs won't fail now or ever. My original design had diode current limiters, but I dispensed with them when I hurriedly rebuilt the voltage regulators after the NPN part failure. BP is ground; please refer to Part 1 for details on how I grounded the various stages to avoid hum and reduce noise.

Above — Block diagram of the preamplifer. I'll present each stage separately, but may show other circuits under each section. The final versions get labelled 1, 2 and 3 for clarity.

STAGE 1

Above — First preamplifier stage.

The 10K and the 220 pF shunt capacitor form a low-pass filter at 72 kHz which attenuates high frequency signals such as AM radio stations that might sneak into the amplifier input.  You may go as high as 470pF if needed; or may raise the resistor value to boost HF filtration.  

Ideally your entire preamp should lie in a sealed metal box to reduce capacitive pickup, however, I designed as I built and fell short with shielding.  Any noise arising in the first preamp goes down the chain and Stage 1 sets your amp noise floor.

The 10K input resistor also serves  to protect the op-amp from large signal overload such as a static spike or a high amplitude signal. Further anti-parallel zener diodes clamp excessive amplitude signals. Most engineers avoid zeners rated below ~ 6.2 volts as they may conduct prior to signals reaching the zener or knee voltage and might cause signal distortion.

How much gain and how to distribute it proves a vexing design consideration. Running lots of gain often means lots of noise.  Ideally, we want just enough preamp gain for the power amp to reach its maximum clean power, but not a drop more. Distributing some of the gain before and after volume controls is 1 way guitar amp engineers manage the noise. The main goal is to avoid amplifying noise at low volume control settings. I spend most of my practice time at low volume settings so I don't drive my family nuts with repetitive practice routines.

Imagine if you built a 15 dB gain preamp stage and then placed a passive volume control pot after it. Your preamplifier stage is always running at 15 dB gain and its noise and the noise of the potentiometer will go down the preamp chain. If you put another fixed 15 dB gain stage after this potentiometer, then even at low volume control settings, the system noise after the pot is still amplified by 15 dB.

Applying active volume controls  (an inverting op-amp stage with the volume pot in the feedback loop) is 1 way to keep noise down at lower volume control settings.  Your stage gain goes up & down with the series resistance of the volume control pot. You'll see this in countless guitar amplifiers as it avoids amplifying noise at low volumes while still affording decent headroom for the input.

Over several decades, many of the guitar amplifiers I've played were high gain "rock star specials" and quite noisy. By noise I mean Johnson thermal noise, shot noise, flicker noise, plus voltage and current noise from amplifier stages.  Crank your amp to maximum gain by turning all the pots to "ten"; unplug your guitar cord from the amp input, or turn the volume pot on your guitar to zero and listen in a quiet room. You'll hear the noise floor of your amp at it's worst. Advance the tone controls to boost and then cut. If 1 exists, rotate the reverb control from minimum to maximum  and see what happens to the noise you hear in the speaker. 

I liken this sound to frying electrons in a skillet. It reminds me of listening to galactic noise in an astronomy receiver. Pretty much, they are the same thing.

Many home brew and some commercial guitar amps are really noisy.  Reducing noise serves as a major design goal for me.  In this amp, I tried to keep noise down by using low value resistors (as possible) , quiet op-amps, active volume control and hopefully wise gain distribution.  Some noise inevitably arises in the tone circuitry and for me this is where most of my amp's noise gets generated.

Currently my home brew amp offers less noise and louder maximum room volume than my Polytone Megabrute and my blackface Fender Reverb. I still want it quieter in future versions however.

Referring to the stage 1 schematic, my first stage unity gain arrangement buffers the input from the 5K volume control pot. The buffer allows you to avoid the noisy 50 or 100K volume pot you often see in guitar amps.  The Baxandall active gain stage perfected by Doug Self gets employed.  I strongly recommend Doug Self's book: Small Signal Audio Design, now in 3rd edition. Self, an adroit author and wise clinician, plys measurement based advice that can't be found anywhere else.

The U1B buffer allows use of quieter low value resistors in the parallel U2 op-amp stages. This amplifier exhibits very low noise on lower volume settings and the only downside is the volume control isn't perfectly logarithmic across the rotation of the knob. You get used to it however.

I prefer to not go above a setting of "9" on the gain pot as the noise performance at "1-9" seems stellar.  

Another big concern is tone shaping the input. The capacitors C1 and C2 achieve this; especially C1. I placed a pair of alligator clips and tried capacitors in the C1 slot ranging from 1 nF to 10 nF. I settled on 4.7 nF as shown. C1 sets how much of your guitar's low frequency you want to high pass filter.

If you apply too low a value, your guitar may sound thin and tinny. To large a value for C1 might cause you guitar to lack highs. The effect maybe subtle, but still important. I agonized over choosing the correct C1 for 1 year  — pick something and stick to it ( for awhile anyway). You may help fatten a bright guitar with your C1 and C2 choice.

For C2, I've got a 1 µF cap in place currently. Between 1 and 2.2 µF seems ideal for me. This helps prevent a flabby bass response when playing loudly. If you play distorted guitar, your C1 and C2 choices would likely be very different than mine. Down the chain I use 4.7 to 10 µF signal capacitors. My small, mostly donated  collection of signal capacitors are older, mostly metallized polyester film types rated at 200 volts plus, and thus are big and unwieldy. 1 day, I will order some smaller size caps.

Guitarists hotly debate whether plastic film capacitors insulated with polyester, polypropylene, polystyrene or some other exotic dielectric material "are the best". It's quite laughable. Doug Self tackles this myth with gusto in his Small Signal Audio Design book. I just prefer caps that don't distort the signal, leak DC, nor break my bank account.

Finally, I employed the lovely LM4562 op-amp in Stage 1. Without  the 100 nF COG/NP0 bypass caps on pins 8 and 4, I've seen this part oscillate at between 3.5 and 4 MHz.  The 5532 would also prove a solid op-amp choice.

STAGE 2

I spent a lot of time studying and testing tone circuitry. I'm still uncertain what I'll eventually stick in my final amp version, however, here's my current circuit:

Above — Second preamplifier stage.

As you can see, it's an equalizer. I've had this in place for over 1 year to learn what frequencies I seek to boost or cut. Upon reflection, it seems I mostly like to cut lower middle frequencies. I grew up listening to guitar amps with scooped lower mids and it's ingrained. 

I actually built a partial 1/3 octave equalizer  (50, 63, 80, 100, 125, 160, 200, 250, 315, 400, 500, 630, 800, 1K, etc.) but stopped at 6.3 KHz to experimentally see what frequencies I like to adjust in a clean guitar amp. The frequencies I seemed to care about are those I picked for the Stage 2 equalizer.


Above 2 photos — My outboard equalizer with 12 possible pots ( 2 were missing at the time of the top photo). On my guitar amp, I placed a line out after the first preamp — and a line in going to the PA stage with a bypass switch (if needed) so I could test various external tone circuits. On the outboard EQ, I tested up to 12 frequencies at once.

I'll discuss my final Stage 2 EQ, going from left to right; or low frequency to high starting with the bass boost.  

My speaker tests get covered in Part 4, however, occasionally, I practice with 6 or 8 inch speaker mounted in open-back cabinets.  A switchable, low Q, 75-80 Hertz bass boost makes a huge difference for these 2 speakers.  It's also good for low volume playing when you really wish a super fat bottom end for a change in pace.  A front panel switch enables or kills this filter.

I really like the main bass control to be wide (low Q) and centered at 125 Hertz. A center frequency of 150-160 Hz is too high for me.  A slightly lower bass center frequency sounds OK, but since I have the bass boost at ~75 Hz , a 150 Hz bass control worked well.

Low mid range frequencies

For many, the lower mid range tone shaping defines the guitar amplifier.  Some of you have studied the transfer function graphs of classic Fender, Marshall and other famous amplifiers to see the importance of lower mid range shaping to set the signature tone of an amplifier.

From making the partial 1/3 octave, EQ, I learned I strongly dislike 250 and 315 Hz. These along with 400 and 500 Hz (respectively to a lesser degree)  make up the so called mud  (or mud range) frequencies to my ears. Thus I fixed cut 250 and 315 Hz. I found that 400Hz should be adjustable (to vary the cut)  and that 500 Hz is OK, but should never get boosted from my experiences.

800 Hz ranked important in my listening tests. Generally a little bit of cut helps my tone at this frequency.

I can't handle 1 KHz at all.  For me, its a nasally, festering, fingers down a chalkboard frequency. For some reason, 1.2 KHz seems a  bit more palatable.  Depending on the room and guitar, I may slightly boost, set neutral, or slightly cut this frequency. Since I favor the 1960's Jim Hall sound, I added 2K and 3K2 Hertz but nothing higher in frequency. The Hair control is more just a general treble boost or cut — very subtle.  Although I omitted 5 KHz, it might belong on some guitar amps.

For the lower mids and highs, a Q of 5.1 sounded better than lower Q versions. I think I tested a Q of somewhere around 1.7 and then 0.85 for some of these frequencies.

Since there is DC on the pots, I stuck to the format employed by countless amp designers to avoid scratchy pot adjustments: I built with the TL072 op-amp ( 6 of them ). I just ordered some OPA4134 for future exploration.

Other tone circuits

In jazz amplifiers, it seems that the Baxandall  tone circuit reigns supreme. A few older Fender solid state amps ran their classic R-C passive tone circuit ( the Fender scooper ) plus later in the preamplifier stage, a Baxandall type tone control circuit to keep the classic Fender midrange but still allow some boost/cut of the bass and treble etc..

I first built the old bass and treble control employed in the early Polytone amps such as the Mini Brute or Model 104.

I reworked it with 10K pots to lower Johnson noise plus op-amp current noise in keeping with the solid advice and influences from Doug Self's book:

Above — The basic bass & treble tone circuit that reasonably keeps the time constants of the early Polytone tone circuits.  This stage works well for all its simplicity.  I've put a version of this circuit in radio receivers, and a code practice oscillator. I prefer this circuit over the bass, middle, treble Baxandall tone circuit employed by Polytone in later amps including the Mega-Brutes.

You will find countless versions of 3 Baxandall frequency tone circuits in guitar amplifier service manuals. I built a few and found a 4 frequency tone control suits me better than a 3 frequency circuit.

Above — My favorite 4 frequency tone circuit. 20 K pots prevent the input impedance going too low when boosting heavily, You find interaction between some of the peaking filters, but this is pretty sweet for 1 op-amp. There are also scores of 4 frequency variants in guitar amplifier service manuals that cut/ boost about 12 dB.


Above — The board I put the 4 frequency tone circuit in. I added gain controls and 5532 op amp to turn this into a simple, complete jazz guitar preamplifer. It sounded pleasant.

Further, from his aforementioned book — I built Doug Self's low noise, variable frequency, variable Q, state variable,  mid range parametric equalizer. Of the 3 parametric EQs I built, his worked the best in terms of noise performance and function.  I just never feel parametrics work the mid range the way I seek for a jazz guitar amplifier.

I also tried some high-pass filter circuits. 1 stood out.

Above — A high-pass filter for guitar input. This filter works well and seems to enhance bass frequency tightness. We have no use for frequencies below 60 Hz. I built my guitar input stage from stage 1 right on the first unity gain buffer.

Stage 2 Conclusion

After using the Stage 2 equalizer for ~ 1 year, I learned that in a future circuit, I will probably only place a quad FET op-amp and make a 4-band equalizer for  150, 250, 315 and 400 Hz with 250 and 315 Hz in a fixed cut.  

Perhaps I will add a separate 1 op-amp stage bass boost as well. The high mids and highs will likely get handled with a single op-amp with a switchable peaking/ shelving EQ circuit.  Only time and experiments will tell.

I seek to keep the parts count down and employ lower value resistors in the equalizer as possible to keep noise at a minimum.

STAGE 3

Above — Third preamplifier stage. This is similar to the stage 1 input amplifier section. The last buffer is made from the left-over TL072 op-amp stage on the equalizer.
 
The master volume pot dramatically  lowers the noise when gain is reduced. 
 

Miscellany

 
Above — The voltage regulators with current limitation. The zeners shown are 16 volt jobs. 1 diode compensates for the BE of the transistor, while the other diode limits the voltage across the emitter resistor to the diode ON voltage of around 0.6 volts.  The current limit runs a bit over 200 mA per rail.
 

Above and below — The amp chassis after testing. No hum!


The index for this project is on my guitar-related page. Click here.