Jose Plexi mods

[Matador]

Referencing the secondary of the output transformer (regardless of NFB) also simplifies debugging, in that you can always just attach a scope probe to the 'non-grounded' side of the output transformer and see what is happening referenced to the common ground potential: in other words, you can see something in the preamp and the output signal on a common scope shot (provided the vertical scales are appropriate, etc).

Utilizing chassis points as circuit return paths is a poor solution, as one loses the ability to predict and control the return currents in the circuit(s).

It does then bring me to think where is the best noise option for gnd for the speaker jack?

There is no (appreciable) current flowing in the ground reference for the speaker jack: the current circulates through the secondary and the speaker only. Since most of my designs use NFB, I always return the feedback signal as a twisted pair from the speaker jack back to the point where it's injected in the phase inverter, so that's where the speaker's ground reference is attached.

 

[Spencerleehorton]

Ah ok so do not put the speaker gnd to the filter caps gnd.

 

[Matador]

Ah ok so do not put the speaker gnd to the filter caps gnd.

In a typical Marshall LTP with a presence control, there is a 5K pot to ground right at the tail of the LTP. There is also the 47K resistor that is fed from the secondary 8 ohm tap. I typically feed the presence ground and 47K attachment point as a twisted pair out to the speaker jack, and attach the two sides to the two sides of the secondary (keeping in mind the phasing is important). So the 'ground' of the speaker jack is actually the same ground as the bottom of the presence pot. This Almost Always (TM) works fine.

https://groupdiy.com/threads/pcb-ground-planes.83954/post-1089789

 

[john12ax7]

It might be helpful to consider what is actually going on. For a NFB amp you essentially have a transformer balanced differential signal that gets fed back and inserted single ended. How would you handle a transformer coupled mic pre feeding a single end unbalanced eq? It's the same thing. Looking at it that way logically gets you to what Matador is describing.

 

[Spencerleehorton]

Just wanted to check I’ve got this right before I move forward

 

[Spencerleehorton]

On the OD channel I want to adjust the bottom end a little and have gone with 1.5k cathode and 0.68uf bypass caps but there it’s still a little muddy if I change the plate 22nf to 2.2nf will that cut a bit of bottom end as well?

The cathode cap calculator on amp books seems to look like it should sort it!!

 

[Alva Goldbook]

On the OD channel I want to adjust the bottom end a little and have gone with 1.5k cathode and 0.68uf bypass caps but there it’s still a little muddy if I change the plate 22nf to 2.2nf will that cut a bit of bottom end as well?

The cathode cap calculator on amp books seems to look like it should sort it!!

Take note of which frequency is each guitar string.

https://homes.luddy.indiana.edu/donbyrd/Teach/MusicalPitchesTable.htm

Amp books has a good coupling cap calculator. I think you gotta figure out your output impedance I can’t remember. For a 12ax7 with 100k plate it’s 38k. But there’s an amp book calculator for that too in case you need it.

At 22nF you basically pass the full frequency range of the guitar. So run a simulation of 10nF or 4n7 before you use crazy low values of 2n2, 1nF or 500pF like Vox did.

I don’t recall what Soldano did off the top of my head, but Marshall also boosted treble using a RC filter of 470k/470pF.

Marshall traditionally used cathode bypass caps and RC filters to boost treble, Vox boosted treble using coupling caps and Orange did both. I think Marshall’s approach gives a more aggressive sound.

 

[Spencerleehorton]

yep thats done the trick, 22nf to 2.2nf on the last OD ch preamp, sounds great now.
still a few bits to tidy up but looking and sounding so much better now.

 

[Alva Goldbook]

Suppose you have 3 triode gain stages. Any adjustment you make to the first stage will have the least impact on the over all tone, and any adjustment to the last gain stage will have the most impact on tone. Another way of thinking about it is the closer the gain stage is to the power tubes the more the power tubes will hear it. Many amp designs from Marshall to Fender use larger coupling caps on the later part of the amp and smaller ones closer to the input. This helps balance things out if what you’re hearing is too harsh. Just something to keep in mind.

 

[Spencerleehorton]

I looked at the B+ voltages today as wanted to try and tighten up the bottom end and get a bit more clarity.
Had 15k as first resistor, thought it was 10k so changed it to 5k, then changed the next 10k to 5k and left the last 10k.
Everything sounds so much better now, really made a difference to the feel and tightness of the bottom end.
B+1 is 430v
choke
B+2 is 427v
5k
B+3 is 375v
5k
B+4 is 345v
10k
B+5 is 300v

 

[Alva Goldbook]

I looked at the B+ voltages today as wanted to try and tighten up the bottom end and get a bit more clarity.
Had 15k as first resistor, thought it was 10k so changed it to 5k, then changed the next 10k to 5k and left the last 10k.
Everything sounds so much better now, really made a difference to the feel and tightness of the bottom end.
B+1 is 430v
choke
B+2 is 427v
5k
B+3 is 375v
5k
B+4 is 345v
10k
B+5 is 300v

Tubes are very forgiving with their B+ voltages. I don't think anyone would notice a difference between a 12ax7 run at 150v vs 230v. You could tell a difference between say 80v and 300v, but not in tone, but fidelity. As a triode is run at a lower plate voltage, the grid voltages has less separation, and minor changes to a bias resistor will have greater impact on how the triode clips and distorts. At higher voltages, the grid voltages have more separation. As the voltage gets higher and higher, for the same cathode resistor values, the triode becomes cleaner and cleaner. You get more clarity and fidelity, but it shouldn't have any real impact on bass or treble response.

 

[Spencerleehorton]

Well it’s definitely made a difference that I can clearly hear and feel, much better, voltage difference of about 50v more per preamp tube

 

[Alva Goldbook]

Well it’s definitely made a difference that I can clearly hear and feel, much better, voltage difference of about 50v more per preamp tube

Here's the Average Plate Characteristics curve of a 12ax7. The X axis is the plate voltage, the Y axis is the plate current. The black lines are the grid curves. Notice each line goes from 0 volts to -5 volts. This is what you're measuring if you measure the voltage on the cathode. I drew a blue and red line. One at 300v and the other at 150v. Each line is drawn assuming a 100k plate load resistor.

Now if you wanted to center bias for max headroom and volume, you'd shoot for the middle. But notice with the blue line this spans 0 to -4 volts. At 150v this only spans 0 to -2 volts. So to center bias at 300v you'd go for -2v and at 150v you'd bias to -1v. So if you biased to -2v at 150v you'd be approaching cut off, and clipping quite a bit, while at 300v you're center biased.

Now suppose we drew a line horizontally at our center bias point for each plate voltage. Orange for the 150v and green for the 300v. What we find is that at 150v we're at 0.5mA and at 300v we're at 0.8mA. So what do we have here? We have a grid voltage and a current. With these two things we can know a third thing, resistance. And a handy Ohm's Law calculator will tell us what the resistance is. Resistance for what? For the cathode bias resistor. This is how we calculate it.

Let's start at 150v. For this to work, we have to convert mA to Amps.

1/0.0005=2000

So at 150v and a 100k plate resistor, to center bias a 12ax7 we'd use a 2k resistor.

Now at 300v.

2/0.0008=2500

So at 300v and a 100k plate resistor, we'd need a 2.5k resistor to center bias.

So notice this relationship. As the voltage increases by a LOT, we're talking double the plate voltage, it will shift the center bias point by 500k. So if you center biased at the lower voltage, at a higher voltage you'd not be reach your full head room. If you biased intentionally for distortion at the lower voltage, with the higher voltage you'd get less clipping and a cleaner sound. But we're talking 500k over 150 volts difference.

 

[Spencerleehorton]

I’ll take some time to digest that, thank you.

 

[Alva Goldbook]

I’ll take some time to digest that, thank you.

It’s worth practicing. It’s like high school math homework. You do it over and over again until it becomes easy. Just like practicing scales or whatever.

You pick a plate voltage using a reasonable guess of what it will be for a given circuit.

To find the slope, which determines the plate resistor, you just draw it. A high slope will give you less gain but more harmonics. A flatter slope will give you more gain but less harmonics. This is why 100k is used so often on 12ax7. It’s the best of both worlds.

The plate voltage and the current on the Y axis gives you the Ohms Law calculation for the plate resistor.

Remember we gotta translate mA to A.

300v / 0.003 = 100,000
Plate resistor is 100k.

If you warm bias you draw the horizontal line at the top half of the slope. This gets you clean tones.

If you cold bias you draw a horizontal line at the bottom half of the slope. This will get you dirty tones.

Center bias gets you the most volume.

 

[CJ]

Fender had a pile of chassis' out in the back yard, these were layouts that did not work well, oscillation, hum etc. In other words, they built amps by experimentation.

Musicman and Ampeg are the only tube amp I can think of with isolated speaker jacks. Ampeg amps have the most hum out of all of them.

Some amps ground the speaker frame to the chassis to kill oscillation but they are rare.

 

[Matador]

To find the slope, which determines the plate resistor, you just draw it. A high slope will give you less gain but more harmonics. A flatter slope will give you more gain but less harmonics. This is why 100k is used so often on 12ax7. It’s the best of both worlds.

I seem to remember that the 100K was picked because RDH says that the "optimum" plate resistor is 2xRp, which for a center-biased 12AX7 is roughly 60K, so 100K for Ra is pretty close. 2xRp balances gain with output voltage headroom, being the operating point that is capable of delivering the most power into a plate resistor, and thus "most" capable of driving follow-on loads (e.g. optimum output impedance).

This TubeCAD article goes over this in depth:

https://www.tubecad.com/articles_2003/Grounded_Cathode_Amplifier/page11.html

It's also equally likely that Leo got a good deal on a truck-load of 100K resistors, and then just went with that.