DC Coupled Cathode Follower Questions

steeledriver

Member
2014-12-08 10:28 pm
After reading Merlin's post on the sound of DC coupled cathode followers when they're set up to steal current from the previous stage, I'd like to toy with them in my next amp, which uses a solid state tonestack. Does anyone have anecdotal evidence to support his praise for their effect on the sound in an amp?

I'm considering using them in two places in the amp - once right at the input before the tonestack, buffering the tonestack, and once right after, immediately before the PI (see attached). Is it a good idea to generate that kind of distortion BEFORE the tonestack? Or even a good idea to stack that kind of distortion in an amp twice? I'd like to hear it with an inverted signal (compressing both peaks) and a non-inverted signal (compressing positive peaks twice) for comparison.

Any experiences?
 

Attachments

  • New Amp - So Far.PNG
    New Amp - So Far.PNG
    48.2 KB · Views: 381
I can see this is an instrument amp, not a HIFI amp. Wrong forum section.

Anyway; what kind of tone results are you looking for? Those bootstrapping and buffering things make the amp better able to drive the tonestack. This is not necessarily a good thing, unless you're looking for a very clean, dryish sound.

On the other hand, a better drive capability could make an amp less sluggish and have more snap.
 
Is it a good idea to generate that kind of distortion BEFORE the tonestack?
Absolutely. Applying EQ to a pre-distorted signal has greater effect than doing it before distorting the signal.

Or even a good idea to stack that kind of distortion in an amp twice?
Yes, several high-octane commercial amps do this. However, with only one stage of gain before the stack, you won't get a huge amount of distortion unless you have active pickups.
 
The bootstrapped follower certainly does work. Pretty darn low distortion values for a simple all-tube circuit and the frequency response can be made very flat too. All of the sudden the low-reputation 12ax7 will do surprising things...

Tips:

1. D1 is usually not so necessary if you have a slow start-up... but if you use solid state rectification and are concerned about protecting the valve, a neon bulb works very well.
2. It's far better to use higher B+ and then use greater anode load (at least 2x 100k ohm) on the differential stage.
3. You will need a far larger C1 if you want to get frequency response super linear. These days I am using 2uF or sometimes a bit higher.
4. I think you might want to re-calculate your value for C2 as well...

Of course if you are looking for a distorting circuit then maybe its not a wise choice...

Ian
 
Last edited:

steeledriver

Member
2014-12-08 10:28 pm
Absolutely. Applying EQ to a pre-distorted signal has greater effect than doing it before distorting the signal.


Yes, several high-octane commercial amps do this. However, with only one stage of gain before the stack, you won't get a huge amount of distortion unless you have active pickups.

Hi Merlin, thanks for checking in. I own your book, and it was a godsend to me two years ago when I picked up this hobby. It's great to be able to clarify ideas with the person who engendered them.

If an overdrive channel were added on a switch after the first gain stage in the form of an extra gain stage and a second DC coupled cathode follower, the overdriven signal would be inverted and have the cathode follower distortion applied to the other side of the wave. With appropriate attenuation so that the op-amp tonestack doesn't clip, do you think that would make a desirable sound? I think that would lend itself to removing a lot of high harmonics from the signal.
 

steeledriver

Member
2014-12-08 10:28 pm
The bootstrapped follower certainly does work. Pretty darn low distortion values for a simple all-tube circuit and the frequency response can be made very flat too. All of the sudden the low-reputation 12ax7 will do surprising things...

Tips:

1. D1 is usually not so necessary if you have a slow start-up... but if you use solid state rectification and are concerned about protecting the valve, a neon bulb works very well.
2. It's far better to use higher B+ and then use greater anode load (at least 2x 100k ohm) on the differential stage.
3. You will need a far larger C1 if you want to get frequency response super linear. These days I am using 2uF or sometimes a bit higher.
4. I think you might want to re-calculate your value for C2 as well...

Of course if you are looking for a distorting circuit then maybe its not a wise choice...

Ian

Hi Ian - I posted this in the wrong forum. I'm building a guitar amplifier, so distortion is somewhat the name of the game.

Thanks for the great advice. Indeed, I would prefer a slow start up to avoid a transient spike that might damage the input of the op-amps. My LTSpice simulations show that the positive input only spikes to about 25V, but that's just a simulation and I'd prefer to play it safe. On that note, can you recommend a simple yet effective soft-start circuit? I was thinking of just using a varistor.

I chose 0.01uF to attenuate the bass around 60Hz in the amp. The tone controls I designed have +/- 8dB of control, so I'm worried about having too much booming bass in the amp. My previous amp is very bassy even with the Dumble-style bass control bottomed out.
 

steeledriver

Member
2014-12-08 10:28 pm
Thanks Mooly, my mistake.

Still unanswered, for those just joining:
- Soft start method - thermistor, varistor, or other?
- DC coupled cathode follower used twice with an inverting gain stage in between. Has it been done before? Does it work well to smooth a fizzy distortion? Or is it better to apply to a non-inverted signal?
 
Do you have any experience with how the positive peak compression affects the sound of an otherwise clean signal?
Well, this is a form of distortion, so it's not really a clean signal anymore! But initially it's just a fattening up of the tone.

If two gain stage/DCCF blocks were placed in series for overdrive, the overdriven signal would be inverted and have the compression applied to the other side of the wave, in addition to the initial compression on the clean signal in the first stage. With appropriate attenuation so that the op-amp tonestack doesn't clip, do you think that would make a desirable sound?
Yes. What matters is the biasing of the two gain stages, and the loading of the cathode followers. You will need to experiment with these for the best sound. Use 10k trim pots for the bias resistors to begin with, or something of that nature.

I think that would lend itself to removing a lot of high harmonics from the signal, and my aim is for a smooth overdrive with only 2 valves.
Leave yourself the option of bootstrapping the gain stages. This will allow you to get a more crunch if you decide you need it.
 
They say a picture is worth a 1000 words - so here is an example.
http://bmamps.com/Schematics/marshall/6100-6101_30th_Anniversary_CompleteSchematic.pdf
The Marshall 30th Aniv. Special.
Sheet 1 of the schematic shows the Clean Preamp (top), the Crunch Preamp (middle) and the Lead/Overdrive Preamp (bottom).
Note 3 off DC Cathode followers in the Lead/Overdrive Channel.

The V2B switchable operating point stage in the Crunch Preamp is another "iconic" circuit which shows up in many amps from many manufacturers (that is it has been copied many times).

Cheers,
Ian
 
Thanks Merlin and gingertube. That's one hell of a preamplifier on the Marshall! I looked it up on YouTube, and I daresay I didn't like the overdriven tone that much. It seemed too thin and fuzzy for my taste. Oh no! D:

Can I bounce one more idea off you? Inspired by DaGeezer's Fender-into-Skyliner mod which uses a 3-stage OD (https://soundcloud.com/dageezer/twin-ultraphonix-test-1), the medium gain topology in Merlin's book, and the original idea of DCCFs I was exploring here, I put together the preamp in attachment 1 in LTSpice. The drive channel should go smoothly from asymmetric to symmetric clipping from what my LTSpice simulations show, and both the clean and OD channels filter through a DCCF. The difference is that V2b switches in and out (oddly enough, I put this together a few days ago before ever seeing that Marshall schematic).

You'll notice that I have to attenuate the signal pretty heavily with an 820k/22k voltage divider, and that's because the active tonestack has a flat frequency response when centered, and +8dB when cranked to 10. Attachment 2 shows the Low-Mid, High-Mid, and Treble controls swept from 0 to 10. The attenuation makes sure that the largest signal that gets passed to the EQ is 4Vpp. The EQ has a +/-15V split rail power supply. I believe that after the tonestack, with a pre-PI volume and sufficient global negative feedback applied, this should be about the right amount to drive a 12AX7 long-tailed pair PI cleanly. It's easy enough to reduce the gain/attenuation of the tonestack.

Thoughts on the smoothness of a 3-stage OD (carefully attenuated between stages) vs 2-stage gain-CF gain-CF? Or anything else I've said?
 

Attachments

  • Three Stage OD Idea.png
    Three Stage OD Idea.png
    42.9 KB · Views: 139
  • Range of Control - Lo Mid, Hi Mid, Treble.PNG
    Range of Control - Lo Mid, Hi Mid, Treble.PNG
    105.3 KB · Views: 135
Last edited:
Hi Guys

It's not surprising that you did not like the OD tone of the 30th Ann Marshall. Those CFs are there simply to make the board layout easier. and otherwise they suck tone.

TUT6 (The Ultimate Tone vol.6) explains why every Marshall has a CF and why that CF is a tone killer. Passive EQs (real "tone stacks") tend to be very dull and ineffective when driven by a CF but then come to life when plate-driven - much more dynamic and useful. Your active EQ is not a tone stack in the normal nomenclature of guitar amps, rather it is an "active EQ". Most active EQs have to be driven by a low-z source to work properly.

Interstage attenuation is one of the two keys to having sustain for clean or distorted signals, as it eliminates grid rectification that can hang up the signal. The other key is biasing the tubes for equal plate swing. All of this is shown in the TUT-series, volumes 5 and 1 respectively.

High voltage supply values are preferred for clean tones. To get smoother distorted tones it is often easier to reduce the B+ to quite low values. Note that you can use traditional high rails for either and for whatever tone you wish to achieve - all roads lead to tone, as they say. The bootstrapped CF is great for low-THD drive at high signal levels but requires a floating heater supply (at extremes) for the highest voltage operation, all of which is completely unnecessary in MI unless it is what appeals to you.

If you incorporate active EQ, it is usually placed after the tube circuitry. In this position it will be working at a line level (0dB, or 1Vpk), as is an effects loop or buffer to a power amp. The input to this circuit block must be zener clamped to protect the ICs. In normal use, the zeners never come on. Of course, what is "usual" is not what you have to do since both DIY and MI are about tone creation and circuit exploration.

Late EQ is preferred for distorted sounds; late or early is fine for clean sounds.

Simulations are good for prototyping EQs and things like that, but generally in MI the real effective method is to build something and listen to it. Keep notes about the circuit iterations and changes made, dating them and keeping old versions of the schematic for later reference. Hands-on wins and is necessary for tweaking the tone of a guitar amp, besides being very satisfying.

Using traditional tube practices, three stages are needed to have a smooth distortion sound. With four stages you can create any tone you want. More stages usually add noise unless some or all are set for medium to low gain. All of this goes along with the fact that each tube gain stage adds a veil of tube character and a sweet clean tone can be built up that still has crisp definition.

Have fun
 
Hi Struth. Thanks for the valuable input. Does "MI" mean "musical instrument"?

I had the same thought about driving the active EQ from a low-z source. In my original post the load for the DC coupled cathode follower is an op-amp buffer, which then drives the EQ. I have a 330 piece set of 1/2W zener diodes from 2V to 47V. What value of clamp would you recommend from your experience?

The output of the EQ will feed into an active mixer with the recovered BTDR-3 reverb signal, which will then feed a 1Meg pre-PI volume pot, and then the PI itself. At least, that's what I've got right now. I'll design a global negative feedback loop so it's difficult to overdrive the PI, and the power tubes will be two 6V6's.

I would like to be able to test the circuit before assembling it, but I have no desktop setup for it at the moment. I do have a spare 300-0-300 transformer lying around, and I could get some spare tube sockets, terminal strips, and hook-up wire.
 
Last edited:
Hi Guys

The zeners are sized to keep the signal fed to the buffer opamp within its supply rails. For example, if you have +/-12V then use 10V zeners. There should be a high-ish value series R between the output of the tube block to the zener clamp.

The series R will be part of a voltage divider and presumably the resistances are chosen to keep the output signal at an appropriate amplitude for what follows. As I said, the zeners should never turn on for normal operation.

Those Belton reverb modules need well controlled input signal inasmuch as they should not be overdriven. The output needs boosting, too, and some colour added - this is where tubes or discrete jfet circuitry is handy. Opamps can certainly get the functioning circuit going for you.

Note that the output filtering on the regulated supply should be much larger than app notes indicate - those show the values needed for stability only. Distributed filtering and decoupling is best.

Have you built a guitar amp before? If not, then you should try building something very basic and traditional to gain experience. You will find that many of those techniques are very useful and indeed more effective than simulation particularly for MI. Sims are great for hifi.

Once you build a full amplifier, you will also find that the feedback loop around the PA is usefully played with. A fantastic guitar tone is had using a UL output stage and no global feedback. Open-loop tetrode-pentode is dark with flubby bass. You can make the loop variable and dial in whatever you wish.

Allow for more heater current than just for 6V6s so you can use alternate tube types and mix types. TUT3 shows all of this.

Have fun
 
Hi Struth,

I designed and built one guitar tube amp already, a very complicated 2 x 6L6GC amp with parallel effects loop, pentode-driven spring reverb, adjustable feedback loop with presence and resonance controls built in, and a number of other tone tweaking tricks gathered from Merlin's book and around the internet (scale control, bass/mid/bright boost, rock/jazz switch from Dumble amps). By some miracle of the tediousness it took to build, it works very well, and I'm having a really fun time tweaking the OD tone to my liking. Also sorting out a little buzz when the reverb or preamp gain are cranked - otherwise silent. Shielded wire hasn't helped with that so far.

After that amp, I moved onto building a solid state headphone amplifier for a university project, which is where I got the idea for the active EQ. That's also finished and working well.

This amp is meant to be a bit simpler. I'm pretty good at layouts, already successfully built many of the circuit blocks I want to use in this amp, and I have a signal generator and oscilloscope to help with testing. I've read that the BTDRs don't handle being overdriven, so I'll make sure the input signal is attenuated below spec and that low frequencies are filtered out.

I have some JFETs lying around from previous projects. I had planned on using a NE5532 op-amp recovery circuit. Indeed, those op-amps rarely work for me unless I strap a ceramic capacitor directly across their power pins when I solder them. You say the reverb recovery circuit could use additional coloring? I'll save a tube and try a JFET recovery circuit.
 
Hi Guys

"Also sorting out a little buzz when the reverb or preamp gain are cranked - otherwise silent."

The Galactic Ground method shown in TUT3 eliminates all this kind of noise and is universally applicable to all technologies. If the signal ground ties to the chassis ground at more than one point then it is incorrectly grounded. That point is NOT a star ground, which will generally increases noise in a randomly-grounded amp. That point should also not be at the guitar input.

You mentioned "scale control". Do you mean Power Scaling? I invented that and the term, but you'll find poor interpretations of older methdos on the web. My site has the real thing.

All opamps should have a local 100nF disc to bypass rail-to-rail. Some devices are just more sensitive to this, and again, normal decoupling and proper grounding are generally neglected with opamp circuits. In my builds each opmp has a minimum of 470uF per rail plus its decoupling Rs along with the 100nF bypass.

If you are laying out a PCB it is easy to end up with an antenna comprised of the opamp supplies and/or the heater wiring for the tubes. These should both be bypassed along their length to avoid RF ingress.

To keep the new amp simple I would leave out the active EQ and go passive.

Have fun
 
All components are grounded at their local capacitor, and then all capacitors are grounded together and to the chassis at a single point, which is indeed the input jack. Being new at the game, I gave each tube its own cap. The only other place a connection to chassis is made is at the safety earth, which is on one of the PT bolts.

The "scale" control I mentioned is not traditional power scaling, it's just the term Merlin used for a pot wired as a rheostat in series with the PI cathode resistor. Turning it up heavily cold-biases the tube to distort the signal and make it quieter. I actually enjoy that control when the preamp is already heavily distorted.

Maybe "simpler" was the wrong word to use. I think it would be fun to experiment with an active EQ. I just got pretty tired of wiring controls on my first amp, which has 13 pots, 3 jacks, and 5 switches on the front face. This second amp is going to have 8 pots, 1 jack, and 1 switch on the front face. I'll make sure the op-amps are adequately bypassed and decoupled - and I should only need 4-5 of them since they're dual packages.
 
Hi Guys

It sounds like the ground at the input is a star. Not good at all.

The individual stage stars should be on a buss in the order they appear in the signal path. This provides the correct isolation of quiet and noisy currents. The tie to the chassis should be from about the midpoint of the buss.

Do not confuse the brass stip in a Fender amp or the wire tacked to the pots in a Marshall as a true ground buss. A proper ground is entirely wired and separate from the chassis. Anyone who suggests that using the chassis as GROUND does not know what they are talking about. The chassis is a SHIELD and should not be used to conduct signal currents.

Intuitively one might think therefore that the standard guitar cable is wired incorrectly since signal current flows through the shield. There are better ways to connect a guitar to an amp but this economisation is nonproblematic. You should think of the wiring from the input jack to the first stage as an extension of that "tunnel" carrying the guitar signal, which tells you also that having the grid-leak on the jack is not a good idea.

Merlin's book came out after Power Scaling, so he borrowed the term and uses it rightly without caps. Skewing the bias that way does not achieve Power Scaling although it does allow access to distortion at lower loudness than the stock amp does.

A note about electrolytic filters: Radial-lead caps are far superior in performance to axial-lead types. For example, for a given value and voltage rating the radial will sustain twice as high a ripple current. Apart from that, axials are mechanically inferior with one lead welded to the can. This weld often breaks even if the cap is tied down to the board. The can caps of old and the modern drop-in replacements are all worth avoiding as they compromise grounding and are not of great quality otherwise.

Also note that when you are buying caps, you should look at the expected life. Cheap caps might last 1khr (one-thousand hours), where good caps are 10khr+.

Have fun
 
I thought that the idea of star grounding at the input jack was to prevent the chassis from conducting signal currents when a cable is connected - reducing the distance/resistance from the amp's own ground to the connected instrument's ground to minimize the size of loops through the cable. Intuitively though, it also brings current in the ground path closer to the input signal.

There is no ground bus in my amp, but I can try moving the star to a spare bolt further away from the input.
 
Hi Guys

What part of "don't use a star ground" is difficult to understand?

The star ground is the only organised ground that people grasp, so they think it must be good. In a few circuits a star ground will improve upon random grounding. In a guitar amp it won't. With a star ground you are assuring that noisy currents mingle with quiet currents and then you wonder why there is buzz.

Have fun