Sunday, 3 July 2022

Reading Material 2022

In 2022 I l stopped most of my subscriptions to magazines and journals. I've got 2 left:

Above — I've belonged to Radio Amateurs of/du Canada for over 3 decades. The quality of our RAC journal 'The Canadian Amateur' has grown — and the build + technical articles rank as first class.

 

Above — Having subscribed to Nuts and Volts for many years, I still enjoy reading each new issue online. It helps keep me up-to-date with the hobbyist forefront. My interests aren't confined to amateur radio; they include a spectrum of electronics-related topics and general sciences.


Above — I still scan professional EE journal databases and read abstracts on topics I favour. Some articles are open-source and free. Occasionally I purchase articles of great interest to me.

My main hobby in Spring|Summer:






Monday, 6 June 2022

Obsolete parts for an obsolete hobbyist?

The story of life as a hobbyist builder --- you get this email from a vendor from whom you buy parts:



Monday, 7 February 2022

Jazz Guitar Amp Experimenters Series — Definitely Not Your Father's Tone Stack

Hey gang !  

Let's further examine some guitar amp tone control circuitry. I'll show you results from a few of my Winter 2021-2022 experiments. Our context = clean, jazz guitar-focused amplifiers.

After publishing Gibson GA-50 Inspired Guitar Preamplifier Tribulations
I improved the basic tone circuit to reduce resistor shot noise and show the schematic below:

Above — The evolved schematic of the adjustable low and high frequency shelf guitar preamplifier. Calculating the 3 dB high + low frequency turnover frequencies gets done by the standard formula Frequency = 1 / (2 pi *  R * C ) with each potentiometer set to its minimum and maximum frequency for the low and high shelf.

Although enjoyable, after experimenting with different capacitor values in the 2 shelving circuits, I abandoned this basic circuit. Why?  Too much knob fiddling; plus I found I only liked a single 3 dB cutoff frequency for both the high and low shelf. Why bother with all this circuitry when a fixed high and low shelf frequency will do?  I also wanted to focus on middle frequency circuits.

I learned that I prefer shelving tone equalizers over peaking or resonant types for both the low and high frequency. For mid range control, I seem to prefer peaking type albeit only if the resonant circuit Q is less than ~1.5

Above — Two basic fixed-frequency shelving tone controls. These often are combined a single op-amp stage plus/minus isolation resistors. Those 2 circuits follow the familiar Baxandall topology and prove easy to design, build and use. 

Above — A bass, middle, treble tone circuit taken from the out of production Carvin Sx-2000 preamp. The trio of equalizers use a single, unity-gain inverting op-amp stage and all 3 are peaking types. Please view the series capacitor(s) coming off each tone control potentiometer's wiper. 

Thus each active tone circuit uses 2 signal capacitors that convert it from shelving to peaking. The design formulae for peaking tone equalizers gets considerably more complex than shelving designs, however 1 capacitor establishes a 3 dB point below and the other above a centre frequency which give the bell shaped response of a band-pass filter.

Above — Another circuit that provides a peaking equalizer response; the Wien bridge design. This particular design offers a different time constant for each half of the Wien band-pass filter. You'll see this often in guitar amps since it allows the potential of a slightly higher Q (and a sharper response). I really enjoyed the 753 Hz 3 dB frequency version as a low middle control on my bench test guitar amp. All 3 scaled designs use standard value caps and the 3 dB frequency may also be manipulated by tweaking the 22K resistor value.

Guitar amps often feature a single middle frequency tone control; or perhaps bling out and offer 2 middle frequency tone controls such as low and high midrange. Which middle frequencies should I choose vexes many amplifier designers. This is likely the reason we may go with adjustable midrange frequencies, however as aforementioned, I want to move away from that. I built a guitar preamp with 2 separate Wien tone circuits and painfully tried many different time constants to see what worked best for me.

Above — My current experimental guitar tone circuitry. I absolutely love this circuit and it's now my benchmark to compare new designs against. I drive this circuit with a single op-amp voltage amplifier affair identical to U1a and U1b shown in the first schematic on this blog entry.

To avoid potentiometer interdependence, each tone stage gets its own op-amp. You'll occasionally see this in high-end console mixers. Since most of us are just making 1 home brew guitar amp and not a production run where a higher parts count costs your company money, perhaps we can afford to bling out and put in as many op-amps stages as we choose?

To keep design simple, I chose identical time constants for the low-pass and high-pass circuits of each Wien band-pass stage. Thus, the standard Frequency = 1 / (2 pi *  R * C ) formula is in play.

1 design consideration = what order do I put the bass middle treble circuits in?  I did some experiments and with active tone circuits ,you may simple choose what sounds best to your ears. Some amps, however, run the middle, then low + high frequency circuits in order and this worked OK in my experiments.

Lower Middle 883 Hz    Peaking

My middle frequencies were chosen for standard value capacitors. To my ears, choosing a low middle of  800-900 Hz offers the optimum single frequency to tailor the lower midrange. Using a Fender Telecaster + a Gibson ES-175 as my test guitars, I preferred the low middle frequency slightly scooped on the neck + bridge pick up combination; or with the treble pickup alone -- and slightly boosted when playing the front/ neck pickup alone.
I tested this board through my Popcorn PA with 3 different 8 Ω speakers: A 10 inch speaker in open-back mounting, a 12 inch speaker in a open-backed cabinet + another 12 inch speaker in a closed-back, ported cabinet.

Higher Middle 1540 Hz    Peaking

I seem to prefer a higher middle frequency between 1200 and 1600 Hertz. Most often, I tended to boost this frequency in my listening tests, but occasionally left it flat. 1540 Hz is still low enough in the spectrum to adds some punch to your sound while avoiding the nasal sounding (when boosted) 1 KHz frequency. Boosting around 1500 Hz added some grit to my neck position humbucker pickup guitars, although too much boost sounded a little tinny, but not especially, since the Q only lies around 1.5 at maximal boost.

Low and High Tone Controls    Shelving

Simple shelving circuits boost or cut the low and high frequencies. The low-pass filter uses the familiar topology of the low-pass variable frequency shelving stage shown in the first schematic of this blog post. Bass response rolls on & off smoothly and capacitor values of 0.39, 0.47 and 0.56 µF were tested. I prefer the really low 72 Hz turnover frequency at this point in time. You may also tweak the 3K9 resistor value slightly, or put in a temporary trimmer resistor to find your dream low frequency 3 dB point.

Chosen for a 12 KHz 3 dB turnover, the high-pass circuit follows the standard Baxandall design & added considerable shimmer to my guitar tone when boosted. Some builders may prefer a 10 KHz cutoff. Build and test stuff! You're the King on your bench.

On most of my experiments, I ran factory original Texas Instruments 5532 op-amps. On the low and high tone circuits, the 1.0 K end-stop resistors may be changed to limit the boost or cut as you prefer. They don't have to be symmetrical. 

The 20K pots could easily be 10K potentiometers to lower resistor shot noise, however, at extreme settings of 10K control pots distortion may arise & you may have to increase your end-stop resistor values. As it goes, Baxandall circuits offer low input Z when boosting hard and heavy loading via the negative feedback path when cutting hard. Evidently, the 5532 performs better than many other popular op-amps in these extreme situations.

Further, conventional wisdom purports we use FET input op-amps since DC flows through our tone control pots.The TL072, or OPAx134 series come to mind. Bipolar input op-amps may drop noise and boost distortion performance if you don't mind adding a few DC blocking capacitors.

Other Experiments

Above — An active equalizer with all stages centred at 500 Hz with 4 different Q factors to allow listening tests. I also performed this maneuver at 190 Hz and 1.6 KHz. I placed 2 or more capacitors in parallel to get as close to each non-standard value design capacitance as possible.

I felt amazed how differently the same frequency band sounded when changing its Q. Obviously, moving between a Q of 0.85 and 1.0 wasn't staggering, however, I heard a clear difference. I liked a Q of 1 and 1.7 best. However, between these 2 values, I preferred a Q of 1 better as a boost and liked a Q of 1.7 better as a cut. I never imagined such a observation. LTSPICE did not inform me about my preferences either — gotta do listening tests.  When Q = 5.1, my boosted or cut tone sounded unnatural @ 500 Hertz.

Above — A simple middle range Baxandall design. I've shown capacitor values for 3 frequencies, but I also scaled the cap values to test at 350, 400, 573 and 4000 Hertz. I preferred the Wien bridge middle range circuits over Baxandalls for midrange when both are centred on the same frequency. Likely, peaking sounds better than shelving to my ears for middle tones. Admittedly, by adding a capacitor to any  Baxandall frequency band you may convert it to a resonant filter. The math poses a little more difficult than a Wien circuit, but its definitely an option for you to consider.

I think this might prove useful in a bass-middle-treble tone circuit sharing a single op-amp stage. You could make the low and high shelving, and then add a series cap to convert the middle control to a peaking type.

Above — The Carvin contour control for scooping the low mids (or not). No boost. Somehow this circuit, a modified Wien bridge, fascinates me. I ran it in a practice amp for about 1 year and simmered a love-hate relationship with it. I scaled it to reduce shot noise with lower value resistors and a 20K control pot. Very creative circuit.

Above —A Laney scooper. Another fixed frequency, Wien inspired, player defeatable, low middle frequency scooper. The Wien bridge circuit has spawned much cool circuitry and certainly Max Wien sits in the pantheon of electronic designers. Mid scooping circuits feature heavily in guitar distortion circuits and often serve as "mud cutters". 


 Click here for my Guitar-related index

Saturday, 1 January 2022

Guitar Amp Experimenters Bench PA — 12 Watt Popcorn PA

 

I sought a low power bench audio PA to use for guitar preamplifier development. This brick measures ~ 18 x 11 x 8 cm and delivers 12 clean Watts power into an  8 Ω load. On the front panels 3 RCA jacks provide regulated +/- 17 VDC plus a signal ground connection. Additionally, the signal input jack lies on the front panel. The rear panel sports a speaker jack, fuse and the AC mains input.

Above — A PA is built around a power supply transformer. This tall, heavy & old Hammond transformer has sat in my lab for decades: Although the label shows 25v CT, the AC voltages on each side measured 13.8 VAC under no load. Rectified and filtered this transformer provided +/- 18.89 VDC under a heavy load with no sag until pushed to the extreme. Thus it will function as a 10 -12 W guitar amp power supply — good for my purposes.

Above — The +/- 17 volt rail regulator/filter circuits. My rectifier included a full wave bridge and the main reservoir capacitors = 4700 µF.  I re-purposed The main Hammond chassis from some old project and this entire project falls under the low cost category since most of the parts were ancient specimens from my collection.

Above — Popcorn PA schematic. This is a slightly tweaked version of this Polytone-inspired PA. I recommend this PA over my original version since it offers way less distortion.

The small signal transistors = 2N4401/2N4403 - a pretty decent general purpose BJT I use from DC to HF. The input pair feature some emitter degeneration via a 100 Ω trimmer potentiometer. If you lack a trimmer pot, place a 49.9 to 68 Ω resistor on each BJT emitter instead.  

I ran out of BD140s plus sought a VAS with an fT of > 150  MHz, a low Cbe and a hfe as high as possible. VAS transistor choice seems to shrink every decade. Gone are the luscious  high voltage, high current, high beta offerings from companies such as Toshiba or Sanyo. For example, the KSA1381/KSC3503 or 2SC2911/2SA1209. Yes, these are still available from online auction sellers, but they seem very expensive and the whole bootleg part worry looms heavy. The BD139-140 seem the only low cost, readily available choice for a low budget PA like mine. The BD139/140 pair specs also widely varies - and some are just garbage. I now get mine from Digi-Key and test a couple to confirm they are OK.

In the end, I placed two 2N4403 BJTs in parallel with 10 Ω current sharing resistors for the VAS. This worked OK. Guitar amps are generally not hifi amplifiers giving ultra low distortion. Some might classify as hifi, but that is the exception. I avoided current sources, driving the VAS with a emitter follower and other distortion lowering techniques. 

The virtue of this PA = easy to build, easy to debug and sounds very good for the parts count. If you scratch build a complicated HiFi PA, you suffer a high probability of failure. Especially when current limiting circuitry and 2 current sources that work together with negative feedback go into your circuit. Often enough, your first sign something is wrong comes in the form of smoke; and by the time you figure out why, you may have destroyed some parts.

Above — FFT of the Popcorn PA. All harmonics are down at least 70 dBc. I tweaked the input pair 100 Ω trimmer to crush the second harmonic. Some of this spurious output comes from IMD in the TIP142/147 pair. In a PA, distortion can arise from many points along the signal path. I'm quite happy how this particular PA turned out.

Across the base inputs of the Darlington final pairs, I measured 2.064 VDC. Three rectifier diodes seemed to eliminate any detectable crossover distortion ( a dominant source of distortion in many PAs).

If you listen to a  guitar amp that lacks enough forward PA base bias and suffers crossover distortion, you'll hear a faint, fuzzy sideband sound along with your main guitar sound. You really hear this when playing single, sustained note phrases. This popcorn PA sounds clean & punchy with no hum except with single coil pickups.

Above — I built this version of my PA first. The BD-139/140 emitters get 120 Ω resistors so they don't need a heat sink and hopefully won't burn up if I made a mistake. I measured DC voltages, calculated current by voltage drops across selected resistors — and also tested it with a signal generator and dummy load. Only 2 diodes drops are needed to properly bias the BD139/140 pair.

With the amp working well, I then pulled the BD139/140 off the main board and wired up the chassis mounted TIP142/147 after adding 1 more diode to set the correct PA idling base bias. Finally, I added the current feedback loop. I chose the 7K5 Ω resistor during listening tests.

Above — FFT of the circuit shown in the schematic above (with the BD139/140). The distortion is about as low as I can measure with my DSO. All that is needed = a set of complimentary finals and a good PA might arise. A decent LoFi popcorn PA for guitar amps.

Above — The "brick". The Popcorn PA lies on 2 boards with carved islands for the positive, negative and 0 volt rails - and to anchor the rectifier diodes. After taking this photo, I screwed on the cover plate to seal it for safety against the pile of wires and other mess on my bench. This amp is not connected to AC mains unless I am using it and does not have an on/off switch since all my audio test equipment goes on a dedicated switched AC mains power bar.

Above — Rear view of the Popcorn PA. I'll start work on my next preamplifier circuit tomorrow and audio test it using this PA. I really like the small bench footprint this brick offers. I place the preamp circuit just in front of the brick and have small cables built for the 17 volt rails and input.

Click here for my guitar related index.


Saturday, 11 December 2021

Gibson GA-50 Inspired Guitar Preamplifier Tribulations

 Introduction

Like many, I enjoy the recorded sounds of — and feel inspired by the Gibson GA-50. That warm tone, full mid range and brown suitcase look epitomizes an American jazz guitar amp classic.
What do I like about the GA-50?  Its simplicity, the non-mega-scooped mid range, and of course, the warm, thick low and lower middle tones. Feeling inspired by these attributes, I embarked on a design journey to make a GA-50 inspired preamplifier to go with my simple experimenters PA stage.
 

I used JFETs & op-amps instead of octal tubes. Perhaps, now, even op-amps are getting outdated as digital algorithms simulating some old amp of lore push the bleeding edge of design. I’m not a fan of using tubes in jazz guitar amplifiers – give me solid state any day for clean signals. Perhaps 1 day, I'll possess the skills for digital amp design. 
I’ve studied hundreds of solid-state, analog designs and it seems that many just copy someone else's solid-state guitar amp design. Most of the ingenuity in solid-state design seems to go into the distortion circuitry.  I found a patent by someone who made a so called 'tube sounding' clipping circuit by putting zener diodes in the feedback loop of an op-amp. Really?!?  They were doing this in the late 1960s albeit for other reasons such as op-amp voltage limiters or zero crossing detectors.

It's more than the 'lack of tubes' that make some solid-state amplifiers sound poor. Design elements such as putting electrolytic caps in the signal path, along with brash, tinny-toned, picofarad coupling capacitors. Add in toxic sounding distortion — and not to mention the grinding noise from mega-high, tube-era, resistor values, and sometimes poor gain distribution that often compromises the noise performance and/or headroom.
 
The classic, passive Fender, Marshall et al. mid scooping bass/middle/treble tone stacks looms prevalent in solid state guitar amps. You will find them in countless guitar amps from as many different companies. While they do sound good in many designs, they depart from the desired GA-50 tonality and boy they exhibit loss.

My whole adventure focused on the study and testing of tone circuitry. I will blog more about that in future posts, however, as much as I wanted to keep the passive tone circuit of the original tube GA-50, it performed with lackluster results in my solid-state versions. Yes, these simple tone controls work, but such passive tone circuits seemingly lack versatility for getting a consistently warm tone with different guitars, speakers and speaker boxes.

That left active tone control circuitry and about 3 things to play with: [1] shelving bass or treble circuitry (basically these are variable low-pass or high-pass tone controls respectively) [2] peaking/resonant circuits [3] choosing a Q for my tone control circuits that gives the natural sound you hear in amps like the GA-50.

I spent ~6 weeks on the bench & computer working on my tone circuits along with the basic gain stages. I also studied non-linear design including distortion circuitry, switching, line-out circuits and speaker box emulation circuits for DI purposes. I filled 2 notebooks and learned much.

I burnt thru 3 soldering iron tips and tons of parts on my tone quest. I’ll start by showing you my latest design as of Dec 11, 2021.  I strove to keep my signal path resistor values down in value to reduce Johnson or thermal noise arising from these resistors.

Above — The complete preamplifier schematic. Click on the image for a better view since it's 1121 X 821 pixels with lossless compression. This is actually version 7 & sounds sweeter, plus uses less parts than the previous 6 versions. This design was built around the glorious 5532 op-amp. I measured no instability.

Buffers. Isolation buffers lurk everywhere !

In your lab, you're likely not making a cost-conscious design for the production line. Therefore, use as many op-amp buffers as you like to enhance stage isolation and to prevent the loading of your tone circuit at extreme settings of the potentiometer wipers. I love buffers and isolation — almost to a fault. Go ahead and remove some of these voltage follower buffers if you wish.

The input 10K resistor adds noise at a low level spot in the preamp. However, a series input resistor follows normal practices. My listening experiments yielded my personal preference for this resistor to lie between 10 and 12K ohms. That resistor and the 150 pF shunt cap are wired right on the input jack provide RF filtration in a critical spot.

The 47 nF input cap comes right from the GA-50 which uses a 50 nF cap. This capacitor value works perfectly as a high pass filter pole to attenuate the low frequency rumbling noises emitted from an arch top guitar. Further, it simplifies the design by allows readily available for a reasonable price 1 µF signal chain caps for DC blocking in a clean signal preamplifier design. My 1 µF caps = Panasonic metalized film polypropylene jobs — chosen for their low distortion.

Active volume and master volume controls help with noise and headroom management. In many solid state guitar amps you'll see a stage of massive voltage gain immediately followed by passive attenuation via a 100K or so volume potentiometer that's shunt to ground. That voltage gain stage runs at its maximum gain (and noise) all the time. Why boost — then right away attenuate the signal if you don't have to ? I tried to copy the GA-50 and just provide 1 volume control, however, found too much compromise in headroom and noise performance. Distributing the gain stages and adding an active master volume control improves the preamplifier and 1 extra volume pot on the front panel does not seem too onerous.

Tone Circuitry

I tried many tone control circuits for bass, treble, lower mid and high mid range. I made variable Q parametric stages; EQ style stages with an op-amp gyrator to simulate the inductor at Qs ranging from 0.7 to 5.1; Baxandall circuits with shelving and/or peaking; and also Wien bridge type treble and mid range circuits. Some of these Wien circuits were fixed while others went variable frequency. I also explored some passive circuitry like those found in Fender, Marshall, Hiwatt, Pignose .....etc. amplifiers as well as circuits from old hi-fi amps. I even copied an old Hughes and Ketner design that plys passive middle and treble controls, then an active, op-amp bass stage that worked pretty well.

After listening to these circuits over many weeks I came to a few conclusions in the context of a GA-50 inspired amplifier:

  • I disliked EQ style tone controls. If the Q was 0.7 to 1, they seemed a bit more tolerable
  • In general, circuits with a Q > than about 1.5 tend to sound unnatural and may trigger listener fatigue
  • I'm not a big fan of deeply scooped midrange
  • Shelving bass and treble sounded better than peaking bass and treble, however, variable frequency shelving seems quite desirable for versatility
  • Combining shelving low and high; plus peaking or resonant type mid-range controls is common in mixer boards and other professional gear and sounds OK. Again, when the Q is <= 1.5, the peaking/resonant mid range tone circuit seems more natural sounding to me

Based on my conclusions above, you now know why I went with the variable frequency shelving low and high frequency controls that forms the heart of my preamplifier. I struggled with potentiometer interdependence in early shelving circuits. When running both a high and low shelf, the low boost/cut pot may affect the high shelf and visa versa. Further, when the shelf frequency pot is rotated to lower resistance to get a higher frequency, distortion, weird behaviour, and/or noises might creep in a some settings of the boost/cut pots. Hence voltage follower isolation amps help make all that pain go away — and provides reasonable potentiometer separation. My chosen circuit still needs work.

Having 2 low and high frequency pots allows you to find a sweet spot for each respective shelf with respect to the room, guitar and whatever speaker your running at the moment.

Ultimately, I opted to not put in a mid-range peaking/resonant tone control for my GA-50 inspired amp. That just seems wrong. On the other hand, the 2 shelving frequency controls allows some alteration of the middle tones which boost versatility.  Shelving networks exhibit a Q of less than 1 and subtly, gently alter your guitar's tone. I'll blog more about my tone controls circuit experiments in the future — I've got lots of interesting material to show you.

Choosing Time Constants

In many low + high frequency shelving circuits, the designer chooses the same capacitor value but up a decade for the low frequency network. For example, .0082 + .082 µF. This allows a bit of overlap between the 2 shelves. I opted to run my high frequency network a little higher and get some calculated response slightly above the 10 KHz zone with the 20K high-pass potentiometer set to 0  Ω.

The standard calculation applies. Frequency = 1 / (2 pi *  R * C ). So, for the low-pass network with the 20K pot set to maximum resistance: Freq = 1/ (6.28 * 22200 Ω * .082 E-6) = 1/.011432 = 87.4 Hertz.  If you seek mega bass, try swapping in a 100 nF capacitor instead.

Above — An early version of my preamplifier using nJFETs as amplifiers. I found IMD and clipping on the output of the Q3 source follower. I later replaced it with an op-amp buffer, and perhaps another after the volume control before abandoning JFETs altogether. The low frequency shelf circuit is by Douglas Self in his book Small Signal Audio Design. I've got 4 of his lovely books and Mr. Self is my favourite audio design book author. I noticed that in his latest version of Small Signal Audio Design, he's added a chapter on guitar amplifiers. Yay!

This schematic includes my favourite simple passive treble network. The 20K pot is used to variably shunt high frequency to ground. I also added a "fatten" switch to roll off less low frequency for use in 10 inch and 8 inch speaker applications. I would normally switch this capacitor in or out using a front panel switch that remotely controls some DC voltage to a JFET switch using the J111 or PF5102.

Above — A whimsical idea that I tested and discarded when I tried to keep the original passive GA-50 tone circuit topology in my preamplifier. When I don't like the result of a circuit I write FAIL on it so that I never make it again.

Above —A schematic excerpt from the solid-state Gibson G-35 guitar amp. This amplifier kept the basic topology of the GA-50 tone circuit. By that time, most of their solid state amps had abandoned this style of tone control circuit and were running a passive Baxandall tone stack instead.

Modules

At first, I built my preamps on small boards that I lifted in and out of my experimental chassis to work on. I did not bolt them in — for they were just held in place by the DC voltage, ground and wires going to and from the various potentiometers. This worked OK, but soldering and unsoldering all those wires became tedious. Later, I moved to modules that sat in front of the chassis that contained integral pots. I only had to solder or unsolder just 3 wires: B+, B- and the ground wire. Much easier.

Above — 2 identical modules that contain all the circuitry from input to U1b pin 7 of the inverting amp in my Dec 11 schematic. The circuits were tested with a signal generator + DSO. Then I began installing the low frequency shelving network boost/cut potentiometer as shown in the closer or proximal module. 

Above — The completed Dec 11, 2021 module with the DC voltage and ground wires attached. An alligator clipped lead connects the output to the PA input. A Boss digital reverb permanently lives on my bench for testing circuits with reverb. I'm addicted to plate reverb. The speaker cable runs just to the left of the reverb pedal. I've currently got five 8 Ω speakers in my lab for amp testing.

Above — A closer photo of the final module. The master volume pot lies on the extreme right hand side. I chose a 2.2 µF output cap since it lies in series with a 1 µF capacitor that's soldered to the PA input. In my final, proper build, I would just likely use a single 1 µF capacitor. My 2.2 µF polypropylene caps are huge 630 volt jobs. I've got a few 100 volt caps @ 2.2 µF, but save these for final builds. Brand name capacitors are not cheap.

Speakers

No doubt some of the sonic signature tones from the GA-50 come from the Alnico magnet 12- and 8-inch speaker pair. I’ve learned that speakers and perhaps even more importantly, the box they’re mounted in prove crucial to realizing your desired guitar tone.  I’ve tried 2 speaker pairings: 12 + 8 inch, 12 + 10 inch and 8 + 8 inch. The wide sound field somehow disturbed me. The only combination I seemed to tolerate was the 8-inch pair, however, they need to go in a box designed to boost the bass response for my taste. Thus, I just use one 8-ohm speaker with this preamp + my PA.

With my preamplifier, if you drive a speaker mounted in a cabinet designed to give bass extension, the available low-end response might amaze you. This is definitely a warm, low-end focused amp – lacking the brash high-mid & raunchy treble response associated with some “transistor amps”. However, the high end is still there. Sometimes it’s pleasant to put a bit of shimmering > 8 KHz treble into your tone.

Above — An early breadboard using Ugly Construction. The power supply is a simple interface to my bench +/-15 volts regulated, 5 Watt power supply.  Islands are carved out in the circuit board for temporary connection to the power supply when developing circuits prior to putting them in the guitar amp. I measured DC voltages, DC current, AC peak-peak voltages and usually hook up a signal generator and look at the FFT also. I use the grounding techniques shown on this web page so despite all this wire, I hear no hum unless I use a single coil pickup.

Future Experiments

Above — Future things to add to my GA-50 inspired amplifier. Thanks to the books of Douglass Self and from studying many schematics, I've got confidence to work out ways to record my amp without speakers + a microphone and/or add software-based digital effects. The features found in many newer amps such as integral DI patching, speaker cabinet emulation and ground loop noise isolation circuitry will help take the GA-50 inspiration into new realms. I feel that if Gibson's Seth Lover and Ted McCarty were alive today they would embrace the latest technology including digital signal processing and of course, op-amps. 

I'll continue to work on my basic GA-50 inspired guitar amp from time to time. I've got parts coming to make a small, complete low-power amp Dec 11 version for practice. I've already tweaked the improved Polytone inspired PA from my last posting to lower its noise floor & distortion a little.

I only wish I had more time to spend on the bench.
 

Click here for my guitar-related index.


Wednesday, 3 November 2021

Clean Jazz Guitar Amp Experimenters Platform — Power Supply and PA

 

Greetings! I purchased a chassis to hold a power supply, a power amp and future preamplifiers. I drilled a bunch of holes for potentiometers, switches, plus 2 inputs and 1 output.  I'll use this basic experimenters platform in my lab and blog about my results over time.

 Prelude

My goal is to explore solid stage guitar amp preamp and tone shaping designs for clean, jazz guitar purposes. My aluminum chassis is strong & large to avoids the pains of working in a cramped box.  

I've cut enough holes for 2 separate preamplifer stages to allow A - B comparisons of 2 different preamp or tone shaping circuits. Today, I'll blog about the power supply, power amp and 1 simple preamp stage — this will set the table for future, specific, component-level experiments.

Power Supply

Above — The simple split power supply.

I had 4 usable transformers in my collection and the first 1 applied was fried. I replaced that with a lighter +/-18 VAC output  transformer purchased on eBay last year. To my delight, the no load DC output measured +/- 26.7 VDC.  A zener diode regulated portion sits on each rail for the preamplifier op-amps and JFET/BJT amps. Choosing +/-17 VDC garners maximum headroom and dynamic range from op-amps.

I went into great detail about guitar amp DC power supplies here and won't repeat myself.

This platform features no power switch since it gets connected to an AC power bar with an integral switch — this guitar amp only sees AC mains power when in use.


Above — 2 photographs of the power supply mounted in the chassis. The 2 rear panel mount fuses are not currently in use. The panel mount IEC320 C14 receptacle features a built in 20 mm fuse holder. I've got a 2 Amp slow blow fuse inside.

Each rail is "monitored" by a separate front panel LED. This proves handy. If you shunt or short a rail, the corresponding LED will dim or turn off respectively.  If your amp goes into oscillation, 1 or both LEDs will flicker at the oscillation frequency.

Power Amplifier

Only a 10 Watt or so PA is needed for an experimenters platform. This nicely keeps the weight  (from the transformer) down. Additionally, 10 Watts seems about perfect for a practice amp. I know enough guitarists with hearing impairment to feel some concern about this topic.

I built 3 PA's but eventually settled on a simple design that boasts only 5 transistors. I took the PA from the Polytone model 100, 101, 102 and perhaps others and modified it so it makes less harmonic distortion. I love this PA for its simplicity:


 Above — Power Amp stage.

With my current supply, this PA provides 12 Watts maximum clean power. Maximal power rating mostly depends on your power supply transformer, and at full power, my DC rails measure around +/-19.7 VDC.

The first change included replacing the original Sziklai pair with a standard Darlington pair in the form of the TIP 142 and 147 complimentary power followers. These husky 10A TIP transistors prove tough and stable in my low power guitar amp. Mine have SOT-23 cases. 

To drive these transistors the BD 140 voltage amp or VAS runs 9.8 mA quiescent emitter current. Rather than going with an adjustable bias for the power followers, a simpler 2 diode scheme works fine and crossover distortion is acceptably low. The DC voltage difference across the TIP 142/147 bases = 1.44 V.  Adding a 47 µF capacitor across the 2 biasing diodes did not reduce crossover distortion.

I replaced the original 470 Ω collector on the input BJT differential input pair with a 1K potentiometer and tweaked the pot for the lowest possible distortion while viewing an FFT in my digital oscilloscope. After locating the sweet spot, I removed the pot and measured 699 Ω. This resistance was nearly achieved with a 4K7 in parallel with an 820 R. I stuck that into the circuit. With that amp set to maximum clean power into a dummy load changing resistor from 470 to 699 Ω lowered the amps distortion profile significantly.

The 100 pF capacitors arrangement in the Polytone was supplanted by that shown. A 100 pF collector cap shunt to ground (instead of to the 0 volt rail) plus another from collector to base stabilized this amp.

Although 1N4007 rectifier catch diodes were added to clamp any back EMF spikes caused by changes from the inductive load of a speaker, I added no PA current limiting circuitry.  For a 12 Watt experimenters amp, this seems OK.  With all power amps, overbuild them with parts that can stand some current; the only TO-92 transistors in the whole PA = the differential pair.

Above — The original Polytone PA that served as the basis for my power amp.

I liked the switchable "dry or wet" current feedback selector switch on old 1970's Polytones.
During listening tests, another version arose: a series 712 or 156 Ω which varies the midrange response heard in the speaker.
For my forthcoming experiments, I'm using 2 Jensen 8 Ω speakers:  a C12N or a C10R mounted in the same open back cabinet (using either 1 speaker or the other). I tested my switched feedback resistor values listening to these particular speakers while playing guitar and recommend doing this to anyone making a guitar amp. Find your current feedback resistor mojo to suit your guitar(s) and particular speaker(s).

Above — An FFT of my PA with a 1 KHz signal injected in the input and a dummy load serving as an output. Each vertical graticule is 10 dB. The third harmonic seems to be from crossover distortion and imbalances in the input pair — and also from the finals.

To get the 3rd harmonic down another perhaps 5-6 dB, I would have install an active current source for the input pair, add 68-100 Ω  degeneration to the input pair emitters, use separate emitter followers for the driver and final PA transistor, and change the 2 diode PA biasing to an adjustable "amplified diode" using a BJT plus a pot. Thus, the once simple Polytone amp now resembles a classic HiFi amplifier like I would build for a home audio PA.


Above — The nice looking sine wave in my DSO (time domain). I can not see any crossover distortion in this screen capture and while zooming in live on my DSO.

Still, though, the distortion in my original Model 102 version looked much worse. The modified version dropped the 2nd plus odd order harmonics down ~15 dB @ 12 Watts power compared to the original.  My current version proves a simple, easy-to-build, inexpensive, 5 transistor PA that sounds great during listening tests.

Beyond the tube versus transistor debate, you might argue that classic, beloved guitar amplifiers were highly non-linear from input to output.  The tendency of solid state amp builders to make their amplifiers ultra linear by using extensive local and wide area feedback circuitry and über perfect input stage balance perhaps makes them sound a bit sterile.  Warts and all, that's my PA.

Above — The final PA mounted in the chassis.

The board looks little messy from the previous 2 PAs built into it. On the right foreground, you may view the TIP42C BJT formerly used as a current source and now disconnected. I fashioned heat sinks from some bent aluminum sheet metal. The entire PA board is isolated from the chassis by cutting islands around the mounting bolts. At this point, the current feedback switch S1 (seen on the right panel) was not in use. Another panel 4 SPST multi-selector switch sits in the chassis in case that is needed in the experiments ahead.


Above — Front of the chassis with 2 input jacks and 4 switches mounted in situ. A third LED slot on the right is open in case an LED indicator is needed for future experiments with preamp and tone shaping circuitry.

Input Channel 1, Preamplifier 1

Above — I built this simple low noise preamp into channel 1.  The output was connected to the PA input with a piece of wire. This simple, low noise preamp allowed me to guitar test the PA and also set the current feedback resistors connected to S1 via listening tests. 

It's a great experience to listen to an amp with no tone controls.  I enjoyed playing through this amp for a few nights and even though preamp maximal gain ranks low, it's already a loud, small room practice amp.

Now I have a base platform for preamp and tone control circuitry experiments to blog about.

My guitar-related index is here