Handbook Amp Project – Part 4 – Input Section

Print Friendly, PDF & Email

NOTE on terminology – I have learned to NOT use the terms “Earth” and “Ground” interchangeably. They are functionally different electrically:
http://www.valvewizard.co.uk/Grounding.pdf
An Earth is a voltage reference point (Planet Earth and 0V usually) and a connection point for equipment electrical safety. The connection to Planet Earth at some point is made by driving a metal spike into the ground and attaching an “Earth” cable to it. The ocean is probably the most consistent relative voltage point worldwide, as salt water is a very good conductor of electricity.
It is important to realise that different points on Planet Earth can have voltage differences between them that can cause currents to flow between them if connected suitably. They may only be metres apart to have significant effect – such as to cause current to flow from one end of an “Earth” cable to the other. It has happened with telecoms cable links where Earth voltage differences between points, kilometres apart, has caused damage to equipment upon connection due to this potential difference:
http://www.cablinginstall.com/articles/print/volume-4/issue-9/contents/special-report/ground-potentials-and-damage-to-lan-equipment.html
This has become more of a problem with the extensions of LAN telecoms equipment between buildings, cities etc. as the Earth is no longer used solely as a safety connection, but as an extension of interconnected network circuits – a Ground on a large scale.
A Ground is also a measurement reference point in a circuit path (usually 0V also) but is the return path to minimum voltage difference that current takes, to flow to or from 0V, relative to the circuit’s higher energy state point – this may be a positive or negative Electro-Motive-Force (E.M.F.).
Current will take the path of least resistance to return to a point of minimum energy difference.
This principle is also a fundamental law of nature – the propensity of all energy to dissipate to a minimum possible state.
Higher things fall down if allowed (potential gravitational energy), heat flows to cooler areas whenever possible (2nd Law of Thermodynamics), and electrons flow between materials where there is a magnetic or electrical imbalance (electrical pressure difference) if a conductive path exists between them. If there isn’t one then one will be made – if the difference is great enough – lightning for example – by breaking down the molecules of the matter that separates the difference. Lightning breaks the air molecules apart when the voltage difference between cloud layers or Earth becomes large enough.
http://simple.wikipedia.org/wiki/Second_law_of_thermodynamics
OK – Physics lesson over. I hope you appreciate the power of electricity more now…
————————————————————————————————————————————

The Pre – Amp Section

A general rule in electronics is that low impedance outputs connect to high impedance inputs – when signal voltage is transferred. It is different for power transfer, as then it is better if impedances are matched, not different. Both concepts are utilised in a valve amp – voltage transfer between gain stages, and power transfer at the speaker output stage.
The signal input for the amp comes from the guitar pickups, over (not through!) the guitar volume and tone controls (capacitor + resistor), along the guitar lead to the input jack of the amp, across its own output impedance of usually 10k ohms or so:
http://www.soundonsound.com/sos/jan03/articles/impedanceworkshop.asp
The signal level at the amp varies greatly depending on pickup types and strength (Humbuckers or single coils, standard or custom, windings and magnet flux density), volume/tone control values, guitar lead length and impedance (cable quality), strength of playing (chords or single note) and distance of strings above pickups…Phew..!
A major factor in signal quality (e.g. how noisy it is to start – a single coil Strat under a fluorescent light…?) is how well the guitar is screened and grounded also.
The signal’s first amp component it encounters is R1, the 1M ohm, 1/2W input impedance resistor. This input impedance is a higher resistance value than the output impedance of the guitar to ensure the bulk of the signal transfers to the amp, and is not lost across the guitar output in preference.


The 12AX7 pre amp valve is two amplifiers in one tube. The “12” means it was designed for 12V DC across the heaters (no mains hum), or 6.3V AC wired in parallel (pins 4+5 linked) across pin 9:


The amp Triode sections in the schematic are:
1 – Anode, Gate, Cathode pins 1, 2 and 3
2 – Anode, Gate, Cathode pins 6, 7 and 8
I want to measure the Gain of each triode roughly myself, to get an idea of how much amplification occurs at each stage, and so the pre amp section overall. This is basically a ratio of the output signal/input signal in amplitude (voltage) terms.
https://en.wikipedia.org/wiki/Gain
For a quick general overview of the basic functions of the resistors and capacitors used in this design, see here:
http://www.diystompboxes.com/pedals/tubedummy.html
The guitar signal passes through a 68k resistor to the gate/grid (pin 2) of the first triode for initial amplification. The screened lead is prepped for the braid being grounded at one end only, to prevent an ground loop in A below – from the Grounding PDF (have you read it yet?!):




Above – After soldering the lead to the 68k resistor and removing the braid at that end, I check for a good joint with the DDM. The input signal is the most important to keep clean of noise, at the start of the amplification chain.
This is where the build process gets harder, as I have to think of board component placement to ground lug positions, and future connection wires to other sections and to the board, if I am going to try to keep each section’s grounds apart electrically, as an experiment, rather than just joining any ground to the nearest convenient chassis lug. Also, I have to remember to think what other components are going to share the lug so I don’t solder too early, so I have to view this section overall, as well as component by component, so I don’t miss one and have to de-solder a shared connection – this can get messy, and is frustrating going backwards. It is inevitable to a degree, building in this experimental way of course.
I drew these shared section grounds onto the Maggie schematic to try and keep better track of the separate sections – it helps modularise sections in your head also:



Also, at this point, it’s worth noting that C+ feeds the pre-amp section, and it is designed to be furthest from the noisiest rectifier section, with B+ in between, on the board. From the Grounding PDF (I will buy this book!) the ripple currents are shown left to right in the same way:

So, obviously, the guitar input jack and screened lead path to pin 2 of the 12AX7, ideally should be placed as far away a possible from other AC sources – mains wires, transformer’s magnetic fields, or output tube HV anode and speaker wires – as possible, as this is the lowest voltage the signal will be, so more susceptible to noise distortion as a proportion of overall signal strength. A high signal to noise ratio is required here – SNR. This can be thought of as wanted signal divided by unwanted signal, so the lower the noise, the higher the ratio.
To simplify the cathode side of the pre amp pair, ignore the EQ “mods” and just see both cathodes at pins 3 and 8 as being grounded via a 1.5k each:

The R14, 68k on the second cathode goes to the top of the OT secondary, and is a negative feedback line which I will cover later (if I understand it myself!). It is left disconnected for now to see the raw behaviour of this stage.
http://www.aikenamps.com/NegativeFeedback.htm
A note from the Grounding PDF regarding “noisy” switch circuits such as this EQ/boost section:
“Any non-audio s (e.g., for channel switching) should be considered noisy and should not return directly to an audio star but to the reservoir capacitor.”
I will look at this should noise be an issue when the amp is complete. For now, my cathodes grounds terminate at the input lug. As an experiment at this stage, I am going to terminate each section ground at the same point to see what happens – and it is physically more convenient due to the board component layout, even though the Grounding PDF states that an ideal grounding scheme has only 1 circuit connection to the chassis in an ideal build and the rest of the circuit should “float” disconnected from the chassis completely where possible.
This input section is the most difficult to build, as it has the most components with many off board, like the tone and volume pots. I messed up in a few places like the switchable input socket – soldered to the wrong side pins – and had to de-solder, so it’s worth going really slow and checking as you go. This type of socket shorts the tip to ground when the guitar lead is unplugged so no noise is amplified unplugged, so you need to know how these work. The socket tip pin gets soldered to the ground pin on one side, the signal lead braid and core to the other.


http://www.ampmaker.com/store/JCM-type-jack-socket.html

In a test build like this, it will be messy as there is no board layout to follow exactly, and I found a couple of mistakes in the Handbook wiring diagram anyway – most importantly, a ground at pin 2 of the 12AX7 – it should not be there, as this is the guitar signal pin. It was easier to just follow the Maggie schematic in general, but with some component value changes, to use what I have that were bought from the Handbook parts list. I don’t anticipate this causing major issues as these two circuits are so similar.
First Checks
After connecting all the pre amp section parts, and double checking wiring continuity, resistor values, and valve pin connections, I think I am ready to do the first test, which is fire up the circuit without the 12AX7 in, first.
If all is well with no bangs or funny voltages, I will then add the valve to check the heater wires work. If that’s ok, I will start voltage level checks, and add a 1V test tone from FLS, and see what happens…
Ok – it all went well!
The first tests with the 12AX7 present since the Rectifier stage measurements show voltages:
A+ = 277V (down 0V)
B+ = 264V (down 13V)
C+ = 174V (down 90V) – now I know why an 8 or 10microF, 250V rated capacitor is sufficient here! For higher output amps, 450V rated caps should still be used though possibly, but manufacturers seem to disagree – bloody duck’s teeth to find it seems…? Ampmaker nor Maplins stock these.
Triode 1 Volts
Pin 1 = 111V
Pin 2 = 0V
Pin 3 = 0.96V – NOTE this negative voltage compared to the grid for later understanding.
Triode 2 Volts
Pin 6 = 106V
Pin 7 = 0V
Pin 8 = 1.00V – NOTE this positive voltage compared to the grid for later understanding

    Triode 1 – FLS Test Tone – 1V Input

Using the test Probe on X 10 setting and 1V/div for the first triode, with a 1V peak to peak signal voltage from FLS at 1kHz, at the input jack, I get 44V inverted peak to peak signal:
Gain44.jpg
This cannot be altered with the volume or tone controls as they don’t connect to this triode, as DC is blocked by C1.
As a break, I thought it would be fun to find out what the pre amps sound like at this stage, and to get an idea of any hum or noise present, if not, it can be eliminated as a source, if there is noise in the power section later.
As I already have jacks mounted in the chassis, I wired from the triode 2 output cap to the tip of the jack socket, and from ground to the sleeve. I can now connect (very quietly!) to another amp input set at low volumes. I now have the most expensive pre amp in the world!
All I am doing really is adding another 2 triodes to the 2 that are in pre amp stage of the other amp, as this (Aston) uses a 12AX7 also. Some high gain amp designs use 2 pre amp valves before the power section.
It just shows that all sounds ok so far. I took the output from the 2nd triode (gently!) also to check the volume and tone controls work too – but at very low (position 1 only) volumes to not damage the other amp.
I found this page to get an idea of the output impedance for this stage, for these resistor values of 100k anode and 1.5k cathode:
http://www.ampbooks.com/home/amplifier-calculators/output-impedance/
About 68.15k ohms unbypassed, and 38.46k bypassed.
Frequency Response – Triode 1
I took readings of the voltage levels at set frequencies using the FLS Tone Gen circuit with the intent to generate a Frequency Response curve – a purely academic exercise and not definitive for the 12AX7 in any way. It is just to get an idea of the workings of a valve, and give an idea what is involved in plotting a Frequency Response curve should you wish to try it, and it will give you a better understanding of why your amps sound the way they do – their tonal response.
A calibrated signal generator with a very smooth and flat frequency output curve would be used in industry for accurate results.
http://en.wikipedia.org/wiki/Frequency_response
This Frequency Response curve gives an idea of the Bandwidth of an amplifier, which is an indication of how linearly an amplifier responds to a defined range of frequencies, within a particular maximum and/or minimum value, say.
These values are set as a general industry standard (for audio) which are usually from 20Hz-20kHz within +/- 3dB from the value at 1kHz – the common Test Tone frequency. Hi spec audio circuits may demand only a 0.1dB change in certain bands – a really flat response for “top” quality kit. (Yeah, like most people can hear the difference anyway!).
NOTE: This graph does not take into account the frequency response of the OUTPUT of the M Audio unit that is feeding the input to Triode 1 from Fruity Loops in the first place. It is a bit of fun and a learning exercise only.
The FLS circuit below works correctly as far as accurate frequency generation now, as it was not set up right before. It looks like this now:

Remember, the human ear does not respond to sound linearly – it is closer to a logarithmic scale, which is why dB – a ratio reference – not voltage, an absolute linear value – is used to graph these results. It also means the results fit on the paper scale – 1 to 20, 000 lines on a graph would be a problem.

T2BW.jpg

This gives a very rough Bandwidth result from the graph, if the reference level is taken at 1kHz. It is about 1-11kHz +/- 5dB, to include the strange trough and peak in the 500Hz region! It was really odd here – with big differences in amplification from 100-500Hz.
Again, how much is down to the headphone socket out I don’t know – but you get the idea…
For a complete guitar amp, average Bandwidth it is about: ” typical guitar amp bandwidth = 5-10kHz”
http://www.aikenamps.com/ResistorNoise.htm
OK so far!
Triode 2 – FLS Test Tone 1V Input
The first thing to notice for stage 2 is the phase reversal of 180 degrees from Triode 1 and the input signal, at 20V/div, 1kHz, Probe X 1.
pic here
With the Triode 1 signal feeding Triode 2 via C1 and the volume/tone section I get range of outputs that fall nicely on the Volume knob – as a rough guide:
20V at position 1
40V at position 2
60V at position 3
80V at position 4 – start of distortion:
At volume 10 I get full distortion and 88V Gain:


From the WikiP statement:
“If the cathode resistor is unbypassed, negative feedback is introduced and each half of a 12AX7 provides a typical voltage gain of about 30”
2 triodes of 30 times amplification – undistorted – gives 60 times which is less than what I see here for both triodes – a 80V clean gain at anode 2. Particular component values influence results though. I’m most interested in the volume position that distortion starts with this amp – volume 4 onwards – as I want an amp that distorts a lot this time – I already have a very clean amp up to high volumes in the Aston.
Now I have logged the “normal” operation of the amp with no cathode bypass capacitors, you may understand how the Boost/Voicing section will work when I add it.


This will change negative feedback to positive for the frequencies determined by C5 or C6 (as R9 and R10 are the same) so will boost these more than the rest that are still being negatively feedback, as the capacitors allow these frequencies to bypass R8, so won’t increase its voltage, causing NFB at these frequencies. I will look at positive and negative feedback in more detail later in the Power section, hopefully.
As R14 brings overall circuit negative feedback from the OT to Triode 2 cathode when connected, I looked at the voltage variation here also so I can compare how feedback changes this signal later.
Triode 2 Pin 8
Probe x 1, 2V/div – input 1V, at 1 kHz, voltage Gain = 2V:
Frequency Response – Triode 2

Vpp

5

12

32

42

44

40

38

18

40

44

40

40

38

35

24

22

20

20

14

Vpp
dB

14.0

21.6

30.1

32.5

32.9

32.0

31.6

25.1

32.0

32.9

32.0

32.0

31.6

30.9

27.6

26.8

26.0

26.0

22.9

dB
Hz

1

2

5

10

20

50

100

300

400

500

600

1000

2000

4000

10000

12000

14000

15000

20000

Hz
Ref @ 1kHz

32.0

32.0

32.0

32.0

32.0

32.0

32.0

32.0

32.0

32.0

32.0

32.0

32.0

32.0

32.0

32.0

32.0

32.0

32.0

 
Ref -3dB

29.0

29.0

29.0

29.0

29.0

29.0

29.0

29.0

29.0

29.0

29.0

29.0

29.0

29.0

29.0

29.0

29.0

29.0

29.0

 
Ref +3dB

35.0

35.0

35.0

35.0

35.0

35.0

35.0

35.0

35.0

35.0

35.0

35.0

35.0

35.0

35.0

35.0

35.0

35.0

35.0

 

T2BW.jpg
To know how I did this, I have included the Excel doc here:
Triode2FreqResp.xls
Obviously, Triode 2 will only mirror – or exaggerate at worst – the values of T1. There is that same big kink again between 100 and 600Hz.
NB: If you do FR curves on output stages where you have to have a speaker load connected to not damage the valve or OT, WEAR EAR PROTECTION! 1kHz sine is annoying anyway, continually.
The EQ/Boost Switch


I used a .22microF cap that I found in a dead home theatre unit to substitute the .47micro I don’t have. Along with a 22microF, 25V rated cap that boosts the bass frequencies more, used with two 1M resistors and a toggle switch, that’s all there is as a mod addition for this part.
From doing all these base measurements I know that there is only 1V max at pin 3, so no problem with any cap rating here.
Fresh from doing the frequency curves for the triodes, I did another for the two switch positions to find where they have the most boost/cut. Again, it was interesting and worth doing. After this I played the pre amps through the Aston again to hear the sound – it works REALLY nicely. The thinner sound makes my guitar more Strat funky thin, and the boosted sound is a warmer, full and pleasant sound.
Taking voltages at the Triode 1 anode again, I noted the values when switched in and out for each frequency:

Vpp

8

18

40

40

50

48

40

15

36

41

33

40

40

40

34

33

31

30

26

Vcut

8

16

22

26

28

26

22

9

21

26

21

32

36

39

34

33

31

30

26

dBcut

18.1

24.1

26.8

28.3

28.9

28.3

26.8

19.1

26.4

28.3

26.4

30.1

31.1

31.8

30.6

30.4

29.8

29.5

28.3

dB

18.1

25.1

32.0

32.0

34.0

33.6

32.0

23.5

31.1

32.3

30.4

32.0

32.0

32.0

30.6

30.4

29.8

29.5

28.3

Hz

1

2

5

10

20

50

100

300

400

500

600

1000

2000

4000

10000

12000

14000

15000

20000

T1BoostBW.jpg
This is a nice big boost of up to 5 dB from about 4-600Hz, which adds that bass warmth.
I would like to compare these boost values to the Fallen Angel now, for future reference. I should be able to do that just by hooking the scope probe to the FA speaker wire, and taking 2 sets of measurements as above – boost and no boost. I’ll do that sometime. It will also be good to compare other amps frequency responses with my FLS setup, as a yardstick for this – see if the extreme bumps at 100-600Hz are there also depending what pre amp tubes are used in what model. The Vox and Marshall are transistors, so worth a go. The Aston, Fallen Angel and Champ use 12AX7s anyway, so no point doing them all.
As a quick compromise I swapped 12AX7s from the Aston which has a JJ tube in it, as the one I was testing with so far turned out to be a Marshall badged one. When I re-tested, adding another reading at 200Hz, the same kinky profile was apparent as for the JJ.
Ok, I’m happy it is a characteristic of the 12AX7 series, and not the amp, as the input components on the Aston are the same as this amp also – 1M input resistor, and a 68k grid resistor.
This EQ section is a nice and quite simple mod for any amp that uses this triode “Resistance Capacity Coupling” or “Common Cathode” configuration – 5 components and a switch drill hole!
This type of dual triode amplifier configuration is called a “Phase Splitter” and can be found along with many other types of valve circuits in the Briggs PDF. The link is here:
Briggs_amplifiers.pdf
I will try and work out how to incorporate a footswitch function for this also – it will have to use a switching type jack to bypass one side of the toggle switch, I think, so it takes its place when kicked in…a project addition maybe, at the end?
Looking at the graph, I wonder if the small cap is needed at all. An on/off switch would do for the 22microF cap, so there is either PFB with the cap, or NFB with just the 1.5k resistor.
This would give the default amp sound instead of the thinner option, or with a boost – which is all I want.
Just to finish this off to put my mind at rest, I graphed the Aston at very low volume through the 60W Marquee speaker at 1V (still annoyingly loud to test with!) at 1kHz reference, to see if the addition of a speaker and the whole amp chain does anything to iron out the kink but it doesn’t.
Here is confirmation that my results aren’t that far off regarding this kink, as here is a 12AX7 plot from:
http://www.amplifiedparts.com/tech_corner/12ax7_comparison_of_current_made_tubes


The same big dip from 100Hz to 1kHz. I’m quite pleased with myself now!
If you really want to understand pre amp valves in detail, download the ValveWizard’s PDF here:
http://www.valvewizard.co.uk/gainstage.html
—————————————————————————–
Next section – Power Tube components and testing….
Unresolved Questions
What resistor values cause the top of the signal only to clip?
It is the bias point on the Load Line that determines whether the amp clips on the positive or negative going signal. If centrally biased and with correct components it can swing swing equally and cleanly either side of the bias point. See page 9 of Blencowe’s book.
What values would keep the amp distortion free, even for less gain?
Values that prevent saturation of the valve – see above.
How will this be distortion be changed by NFB from the power stage?
NFB reduces gain across all frequencies generally if not equally, and if gain is sufficiently reduced, no clipping will occur for a given input strength. The max NFB that can be applied globally is 6-12 dB before oscillation and instability may recur. 
Think how and where the stomp foot switch would go in the boost section to enable what options.
Further Reading
http://www.aikenamps.com/NegativeFeedback.htm
http://lenardaudio.com/education/14_valve_amps_7.html
http://www.hans-egebo.dk/Tutorial/amplifiers.htm
http://www.diystompboxes.com/pedals/tubedummy.html
http://www.muzique.com/schem/gain.htm
http://www.ampbooks.com/home/amplifier-calculators/output-impedance/
http://www.diyguitarist.com/GuitarAmps/SD-Convertible/patgf1.htm
http://www.ampbooks.com/home/amplifier-calculators/output-impedance/
http://www.geofex.com/ampdbug/ampdebug.htm
http://www.geofex.com/tubeampfaq/taffram.htm