I'm tired of anemic guitar amps!

[scruffy]

Still, I wonder if it might be worthwhile to cut the first stage to 2 or 3 volts? Reason? I rather like a device which utilizes the full range of its controls; rather than have half the range of the pot "dead room" where nothing good happens, being able to utilize most or all of the control makes the adjustment more sensitive by stretching the adjustment range out. And it might make some of the circuit happier not being driven so hard.

Yes. The player has to understand how the amp works. There's some complexity in it, it's not for everyone. But I do like the idea of simplifying. Also keeping things intuitive. For example cathode bypass caps and coupling caps => bass boost and bass cut.

But here's the problem: all the switching changes the volume around. For example, in changing a coupling cap from .02 to .002 you lose half your volume. It's impossible to compensate for every conceivable way the volume might change, so we provide the volume control and leave it up to the user. Some settings will require a different position than others.

Being a KT-88 amp, it's a lot about the bass. You don't want to just cut the bass out, if you're a bass player you need the bass and if you're a guitar player you need the area "just above" the bass, where the thump is maybe 150 hz (call it lower midrange). When you keep this area of the bass frequency spectrum, the voicing on the top end needs to change to compensate for the extra lows. In an EL34 amp you don't get a lot of low end so you don't have to worry about it, but in a KT-88 amp you can hear it.

The good news is, I got rid of 4 knobs today, by eliminating unneeded functionality. The bad news is, there are holes where the knobs used to be.

Do you mean value of capacitance or merely brand or type of capacitor (such as polystyrene vs. polypropylene)? Is there no way you can direct couple any of these stages or do none of the biases line up?

DC coupling is tricky. It would be the best, in theory, but in real life requires 1% tolerances and careful adjustment, and sometimes even additional (negative) supply voltages.

 

[scruffy]

The front end is down to six knobs.

Three for the up-front tone stack, bass middle treble.

Of course the obligatory gain control.

And two for bass boost and bass cut. These last two happen at different frequencies, they give you a bass "contour". They're essential for any kind of high gain work or if you're a bass player.

These six knobs will give you everything you need for clean, and everything you need to feed into the gain stack.

The gain stack has only 3 knobs: a "grunge" control, drive in, and volume out. Volume out is needed to match the clean level. Drive is self explanatory. Grunge is the cold clipper, it's very grungy sounding. The grunge control is a rotary switch with four settings "grunge level".

So that's a grand total of 9 knobs for the preamp, not including the reverb and not including the master section. (The reverb is only 2 additional knobs). If given extra room I would add a Dumble-style peak filter after the master tone stack. Master section is 3 more tone controls plus presence and resonance, and master volume. So that's 6 more knobs for master, bringing the baseline grand total to 9+2+6=17 knobs.

 

[toobfreak]

But here's the problem: all the switching changes the volume around.

I understand. In all prototyping, it really is an iterative process, almost like a diamond cutter--- each strike brings the stone a little bit closer to the desired end result. The amplifier is almost a blank canvass of possibilities and now you need to whittle it down and find the closest, simplest approximation to the amplifier you seek buried within its capabilities.

The good news is, I got rid of 4 knobs today, by eliminating unneeded functionality. The bad news is, there are holes where the knobs used to be.

Well, there is still the cover plate that can cover those holes. I would print the final stencil on a 2-5 mil thick sheet of aluminum then bond that over the whole front to give you the final look and layout you want.

 

[scruffy]

I ran into some weird problems debugging this thing. Weird stuff. For instance - I have a tone control, it's the world's simplest circuit, just a .005 cap in series with a 250k pot. This circuit is in millions of amps worldwide. The pot has maybe 300 degrees rotation, in this amp it cuts out entirely from 100 degrees to 200 degrees. The top part of it is fine, and the bottom is fine, but the entire middle of it has no signal. This time it's not the pot, I checked. And the other weird thing is it works fine with a .02 cap. It's just the .005 value that's causing it to fart out. Obviously it's not shorting out, that's the only thing that would kill the signal entirely is if it shorted to ground. I think I know what the problem is, it"a "phase cancellation" with something upstream. Finding that and fixing it is so much work I'm just going to clip the cap and call it a day. Replace it with a .02 so the knob will work.

Regarding the voltages - work backwards. A pair of KT-88's in ultralinear mode needs 100v of drive to get to full power. With a 12at7 PI you get a gain of around 12 or 13 which means you need 8 volts at the input of the PI, therefore 8 volts at the output of the master tone stack. To get 8 volts at the output, you need 80 volts at the input. So any stage that feeds the master section directly should be able to provide 80 volts of signal. In this amp that's stages 3, 4, and 5.

80 volts is a convenient number because it's one side of the swing of a 12ax7 at 350 volts with a 220k load. The maximum swing according to the AC load line is right around 160 volts.

 

[scruffy]

Okay well, both amps are tested and in the box. I'm going to stop with the build details for a while, and field test the amps in live situations.

That means they get played loud for extended periods of time. A rehearsal might be 3 hours, and the amp is near full bore the whole time. A studio session is usually all day, 8 hours.

This way I'll get a keen sense for what's missing and what's still needed. Also the amps will get exposed to real environments with jostling and bad power and beer spills and all that.

The only thing I still really need is rubber feet for the mother amp. Besides that both amps are gig ready if not yet perfectly cosmetic.

 

[toobfreak]

I ran into some weird problems debugging this thing. Weird stuff. For instance - I have a tone control, it's the world's simplest circuit, just a .005 cap in series with a 250k pot. This circuit is in millions of amps worldwide. The pot has maybe 300 degrees rotation, in this amp it cuts out entirely from 100 degrees to 200 degrees. The top part of it is fine, and the bottom is fine, but the entire middle of it has no signal. This time it's not the pot, I checked. And the other weird thing is it works fine with a .02 cap. It's just the .005 value that's causing it to fart out. Obviously it's not shorting out, that's the only thing that would kill the signal entirely is if it shorted to ground. I think I know what the problem is, it"a "phase cancellation" with something upstream. Finding that and fixing it is so much work I'm just going to clip the cap and call it a day. Replace it with a .02 so the knob will work.

Sounds like a possibility. I was going to wonder if somehow when you reach that part of the range of the dial, the circuit tunes itself or becomes reactive to or interactive with an adjacent circuit maybe through close coupling?

Perhaps it might help to relocate a part or move some wiring or add some shielding. Hopefully, you don't have to start chasing gremlins all around the pathway.

 

[scruffy]

Sounds like a possibility. I was going to wonder if somehow when you reach that part of the range of the dial, the circuit tunes itself or becomes reactive to or interactive with an adjacent circuit maybe through close coupling?

Perhaps it might help to relocate a part or move some wiring or add some shielding. Hopefully, you don't have to start chasing gremlins all around the pathway.

It was a lot of fun debugging those issues for a week. You get to the point where the amp kinda-sorta works but doesn't sound like it's supposed to. Also regarding the voltages again, it really is on the hairy edge of insufficiency. If you put a 1 meg resistor in the wrong place your level will drop and your cold clipper won't clip anymore and your cathode follower won't follow anymore and you'll have an anemic amp. (There's so many ways that can happen...) Just the simple act of changing a coupling cap can take your level below threshold. Can't sell the customer an amp where half the settings don't work!

I may put the extra tube in after all. It would be painful but the benefits outweigh the anxiety. I want to put a notch filter between stages 3 and 4, and that'll lose another 90% of signal, so makeup gain is required again. Just half a tube will gain back more dB than is lost.

I was working on the design last night, I can do the concept with 6 knobs. 4 levels plus bass and treble. Bass is a switch with three settings: "some", "more", and "lots". Treble is just a good old fashioned passive tone control. Four level controls are needed: a master (for obvious reasons), a crunch control, a main preamp gain control, and an input trim to match the instrument. With 6 knobs you only get the most minimal of tone shaping, but all your basic features are working, and you can remove bumblebees and icepicks.

I could build the full 200 watt version of this in a large Marshall case for about 4000 bucks. (Same price as a 100 watt SLO - same price as a Dumble in 1986 lol). And the 100 watt version in a Marshall small box for about 3000.

 

[toobfreak]

It was a lot of fun debugging those issues for a week. You get to the point where the amp kinda-sorta works but doesn't sound like it's supposed to. Also regarding the voltages again, it really is on the hairy edge of insufficiency. If you put a 1 meg resistor in the wrong place your level will drop and your cold clipper won't clip anymore and your cathode follower won't follow anymore and you'll have an anemic amp.

Been there, done that. I have a house filled with tube amps of various kinds both commercial and custom, some just modified while others made from scratch, and it is often amazing how two superficially similar circuits can sound different, good or bad, or be changed from sounding muddy, lifeless and wooden, to open, clear and alive by just making some small change.

 

[scruffy]

Here's the further changes from the schematic:

Coupling caps in order: .02, .01, .02, .005, .002, .001

Install a 500 pf plate bypass cap across the cold clipper's 12ax7. Insert the following in series with the .02 output coupling cap on the cold clipper: two caps in parallel, .005 and .002. The value is important, don't replace the whole mess with a .005 because it won't sound right. The switchable values for the cold clipper's cathode resistor are: 39k, 22k, 10k, 4.7k

I have the Gibbs reverb tank (Gibbs is Accutronics) so my reverb is totally lush and frankly a little muddy. Replacing the .02 coupling cap in the reverb amp with a .005 solves the problem. You don't have to do this if you have a modern tank, the modern ones are a lot less bassy. My tank does hum just a tad, it's because the "send" end is right next to the power transformer. There's still the option of using a new tank, a Twin-style long tank will fit upside down on top inside the case, and it's shielded and it'll take care of any residual hum.

That's it, easy peasy. This way I avoid putting in an extra tube and an extra filter. Never a good idea to drill a chassis with circuitry already mounted, or punch a 7/8" hole which requires a pipe wrench. During debugging I move the amp around a lot (it never "stays on the bench"), and eventually the loose connections get exposed and you get confidence in the reliability of the build.

Anything more I have to fix at this point is going to be a headache, further mods are a risk. The amp works great, for both bass and guitar. I might take the slope resistor back down to 33k in the master tone stack, I used 56k and it"a too much. But that's it, other than that the amp sounds and behaves like it's supposed to.

Next step is to get it back in the case for real, it'll take 2 additional screws in front and 2 in back. Word to the wise: use cage nuts if you can. Speed nuts suck.

 

[toobfreak]

Never a good idea to drill a chassis with circuitry already mounted, or punch a 7/8" hole which requires a pipe wrench.

Word to the wise: use cage nuts if you can. Speed nuts suck.

Speed nuts are cheap crap, minimal to their needs. One other option if you are using a steel chassis: weld nuts. But then, IMO, if you've gone full steel chassis, the ideal solution is to just use a heavy enough gauge steel that you can drill and tap it directly.

A nice touch is to mill the steel thinner where possible with a fly cutter to save weight, then fold and weld the sheet into the permanent chassis.

 

[scruffy]

I think they sold me some relabeled KT-88's.

Real ones bias up around -90 volts, to get 50 mA with 535 volts B+.

6550's though, bias around -65 or so, for 45 mA. The extra 5 mA is due to the different plate dissipation ratings of the two tubes.

These I got are brand new in original JJ boxes but they want -67 volts to get 45 mA. That tells me they're almost certainly relabeled 6550's.

Which is okay, they'll work, but they won't deliver the full 100 watts. Also if you're unaware and try to bias them to 50 mA they'll have a very short life and you'll keep buying new tubes.

 

[scruffy]

I'm going to post my reverb circuit. It's clever and unusual.

The requirement is it has to be switchable in and out. This means - in a normal Marshall, you have a cathode follower driving the tone stack, and this is because the CF has a low output impedance which results in minimal signal voltage loss. But now you're going to put a reverb between the cathode follower and the tone stack.

For reference, I'll direct you to the schematics for a Twin Reverb and an Orange. Here they are:

https://schematicheaven.net/fenderamps/twin_reverb_sf_135_schem.pdf

https://schematicheaven.net/newamps/orange_mkii1972.pdf

The way reverb is usually done, there is a series resistor in the signal chain, and the reverb input is tapped from the resistor input, and the reverb output goes back to the resistor output. The simplest example of this is the Orange. The series resistor is 100k, and you can see the points labeled Echo Send and Return.

In a Fender, the series resistor is 3.3 meg, and it has a small cap across it, usually 10 pf or so. It forms a voltage divider with the grid bias resistor of the reverb recovery amp. In the Fender schematic you can see the entire reverb circuitry. The input to the reverb tank is taken from the input of the 3.3 meg series resistor, goes through a 500 pf cap to cut lows and then gets boosted by a 12at7 and fed into the tank. The (weak) output of the tank is then boosted by another 12ax7 and fed back into the output of the 3.3 meg series resistor through the reverb control. The resulting mix of clean and reverb signals is then further boosted by yet another 12ax7, and fed into the power amp. But, in a Fender, there's no tone stack there, so no cathode follower.

An interesting thing of note here, is take a look at the schematic for Dumble's "Steel String Slinger", and look at the part he calls "filters" which are the big switches in the middle.

These filters are where the reverb series resistor would normally be, if this were a Fender.

So two things are going on here:

1. Level reduction of the clean signal, to mix it with the reverb signal

2. Tone shaping, as needed.

But I'd like to draw your attention to the strange way Fender accomplishes the mix. Leo was a cheap bastard, he didn't like spending money on parts. So instead of an actual mixer, he just fed the reverb output into the bottom of the grid resistor in the reverb recovery amp. That means, the mix will only work for a very narrow ratio of clean signal to reverb signal.

In my amp, that ratio is all over the place. In clean mode you're getting maybe 2 volts into the reverb, but in full overdrive you might be getting 80 volts. That's why there's a Dwell control, so you don't have to fry the springs. However you have the same problem on the output side, you have to mix an 80 volt clean signal with the volt or two coming out of the reverb. And then tone shape as needed.

In the next post I'll show you how I did it. Resulting in a much more flexible and powerful amp and a better experience for the player.

 

[scruffy]

Here's the strategy:

First, we need the reverb output to drive a tone stack, so there has to be a cathode follower there.

Second, we have to accommodate a wide range of signal levels.

Third, the reverb output is somewhat inadequate, a volt can't really compete with 80 volts.

And fourth, we would like a real mixer instead of Leo's discount approach. This gives us the ability to control both the level and the tone of the two signals being mixed, individually.

So right off the bat, we're going to put another gain stage after the reverb booster, and add a cathode follower to the recovery amp. But that means we also have to change the way we mix the signal. Because if we add a gain stage to the reverb path, the result will be out of phase from where it was.

To start with then, we focus on the bottom half of the voltage divider consisting of the series resistor and the grid leak of the recovery amp. In the Fender, there's a 470k resistor in series with the 100k reverb control, and those together form the grid leak. So that's 570k, so if we wanted to remove Fender's reverb circuit we would disconnect the 470k resistor and replace it with 560k to ground (nearest commercially available value), and this will retain the original raw signal level. And we note with a smile that this equates with a 500k pot in series with a 56k or 62k or 68k resistor.

Now we want add a gain stage after the reverb booster, and we already know the signal level at the output of the booster is in the 1 volt range, so we don't need a whole lot of gain - UNLESS we want to put a tone stack in that path somewhere, in which case we need more gain. More on this issue later. For now trust me that one stage is sufficient. And if we add a stage, we can take care of the phase issue by simply swapping the output leads of the reverb transformer.

We also know that we only need about a volt at the input of the power amp to get to full power. We don't need 80 volts, that's way overkill. So we can use a passive mixer, no problem. We can take that 100k reverb control and change it to 500k, and mix its output with the other 500k control with the clean signal. This way we can hear the reverb output without hearing any raw signal. However when changing that control we have to pay attention to the ,003 cap that feeds it, the purpose of which is to cut bass to remove mud from the reverb output. Mud is not something adjustable, you generally just want to get rid of it - so this cap value is a design decision, not a switch. However the treble portion of the reverb signal could be adjusted too, sometimes you want a little more or less brightness in the reverb. This part now becomes very easy, we just hang a passive treble cut on the output off the first reverb booster stage.

This is our first design iteration, schematic follows momentarily.

 

[scruffy]

 

[scruffy]

So how this works is:

In clean mode, your raw signal level is all the way up and so is your Dwell. You're basically operating like a Fender. You just adjust your reverb level to taste and you're done. But now you have an additional reverb tone control with which you can dial back the highs.

In overdrive mode, you turn the Dwell way down, and turn the raw level down to about 1/3, and then raise the reverb level till you can hear it. Now adjust the Dwell till you get exactly the right sound. You can tell when you get it, it's somewhat metallic and very distinctive. Now adjust the reverb level to taste. This is where that reverb tone control becomes useful. Usually you don't want a lot of bite in your reverb with a distorted signal, because distortion creates "fizz" in the springs, and if you drive them hard you'll get a lot of it. You can dial back the fizz and still retain the meat of the reverb. Also in overdrive mode, "mud matters" (and for bass guitar in overdrive mode it matters a lot). Different tubes sound different, so you can find the right reverb tube that's tight and not too bassy, and if you wanted to be a real Dumble you could tweak the value of the reverb coupling cap to match the tube lol.

 

[scruffy]

Oh - voltage wise:

The raw voltage divider cuts the raw signal by 80%, so 80 volts becomes maybe 15.

In clean mode you're getting about 8 volts into the reverb, meaning maybe 1-2 volts of raw signal. But you'll be getting about 20 volts out of the reverb, so if your Dwell is all the way up, your reverb level control will be way down.

In overdrive mode you'll get 10-15 volts of raw signal and maybe 20-30 volts out of the reverb, so the knob positions will end up being pretty close.

You can further overdrive the reverb recovery stage by cranking both the raw and reverb levels. This you can do independently of the master volume, so if you just want "that sound" you can get it. (It's a great sound, you'll want it).

 

[HaShev]

a way to get umphhhh in the amp would be with a built in compression/ limiter with guitar settings like they have in condenser mic pre amps.

 

[scruffy]

a way to get umphhhh in the amp would be with a built in compression/ limiter with guitar settings like they have in condenser mic pre amps.

You can get plenty of sustain from the cathode follower. I'm still working on sound clips, I'll post them eventually.

Something I forgot in the preamp portion of the schematic - the treble cap in the up-front tone stack has a switchable .0022 cap across it. It's activated by pulling out the midrange knob. This will give you an upper midrange boost, which you can then dial back with the passive tone controls if you wish.

It's a great way to make your reverb "quack". As a bass player I find this incredibly useful. Set your level so digging into the string activates the midrange boost. Now dial the reverb into the boost. When you play quietly finger style you get a good solid bass, and when you dig in a little you get grit plus highs. If you slap at this setting you'll get a nice little quack whenever you pop the G string.

 

[toobfreak]

If you slap at this setting you'll get a nice little quack whenever you pop the G string.

Last time I tried popping a G string, all it got me was my face slapped. But it would be great if someday, you could get some big rock guitarist to try your amp and do a video on it. This thing sounds like it has potential to outclass and outplay a lot of the amps out there in terms of power and versatility.

 

[scruffy]

The amp is back in the case! Boy, that was a pain and a half. Speed nuts suck, they lift the chassis off the case so you end up with the entire weight of the amp on 4 screws point-wise. Which is not a problem, it's well within the capability of the screws - however the amp doesn't sit flush against the case and the weight isn't distributed like it's supposed to be. In real life the amp gets jostled, so I used 1/4" bolts to minimize the chances of anything bending. However the thread on those speed nuts is not the best, the bigger danger is the amp lifts off the case vertically because a thread got stripped. Have to tell the roadies to be careful