18-watt TMB Build

BenTobith

Member
Joined
Jan 6, 2021
Messages
59
Reaction score
23
HI all,

I recently completed my first amp build, an 18-watt TMB style from Tube Amp Doctor (TAD). The main reason I selected TAD was they had it in stock, and they offered a version without master volume. It seems most TMB kits elsewhere use MV, and I wanted to start without it.

I'm in the phase of making tweaks and trying different things to hone in on the sound and characteristics I'm looking for. For example:
- Lowering the noise floor a bit (some hiss between channels and when volume/TMB knobs are turned up on TMB channel), although I'm mostly using metal film resistors
- Likely reducing the gain a bit on the TMB channel
- Perhaps reducing a bit of the dark or "woofy" nature of the normal channel

Ideally, I'd have a balance between some good Marshall cleans on the normal channel (especially through Low input) and higher gain Plexi tones from the TMB channel when the volume is turned up. Right now, the TMB channel breaks up VERY early, and seems a bit "hot."

Some of the changes I've made vs. the stock layout/schematic:
- Added 47pF cap across PI plate resistors
- Replaced 100R screen resistors on output tubes with 470R
- Replaced 120R cathode bias resistor on output tubes with 150R (initially, tubes were running way too hot; they are still at 125% dissipation at idle, so might need to go higher on bias resistor?)
- Used a ground buss bar on the turret board instead of grounding buss on the back of the pots
- Have a single ground point (near input) for the amp, and connected north side of cathode bypass cap (output tubes) to ground buss
- Elevated heater CT to south side of cathode bypass cap

Other things I'm considering:
- Replacing 1/2 watt carbon film resistors for cathodes on preamps with 1 watt metal film
- Reducing 4.7K cold clipper resistor on V3 with 2.7K (or lower?)
- Reducing 100K cathode follower resistor on V3 with 56K
- Adding 470K resistor to middle lug of treble pot, to help regulate gain sent to PI (read about this elsewhere in the context of a Weber version of this circuit)

Any other thoughts on ways to tweak this amp, both to attenuate the gain a bit on the TMB channel, "clear up" the normal channel tone, and lower the noise floor?

Here's a link to the schematic and layout (stock): https://www.tubeampdoctor.com/media/pdf/f5/41/8e/18w-tmb_full_documentation_1114.pdf
 
Last edited:

neikeel

Well-Known Member
Joined
Feb 12, 2014
Messages
7,192
Reaction score
6,732
I will not claim to be an expert on 18w but have built four:
-first a Ceriatone based TMB18EF86 in a 1x12 combo
-second a NOS 18w Trem in a 2x12 combo
-third a NOS 18w Trem head (now in Italy)
-fourth an 18w Superlite IIb (ptp conversion of an RS equipped Elpico)

I have not found heater mods or output stage mods necessary. The best were/are the two stock Marshall builds with RS trannies and NOS parts (Mustards and Pihers)
The best way of tweaking the normal channel is to reduce the size of the first stage cathode bypass cap down to 4n7 or similar.
I tend to prefer the Trem channel as it is brighter and sharper but has no hiss or any adverse features. I have been using it as my gigging amp for smaller venues for a few years now.
The Ceriatone was my first and only really works well on the TMB channel with the EF86 turned right down as v noisy and most EF86 seem to become microphonic. When I get time I will redo that channel.

I will be interested to see what suggestions crop up to solve your problems.
 

BenTobith

Member
Joined
Jan 6, 2021
Messages
59
Reaction score
23
The best way of tweaking the normal channel is to reduce the size of the first stage cathode bypass cap down to 4n7 or similar.
Interesting. Mine is using a 25uF/25V electrolytic currently, with a 1.5K resistor.
 

Max Gahne

Well-Known Member
Joined
Dec 1, 2020
Messages
271
Reaction score
400
I'm rebuilding my 18 watter for the umpteenth time right now. This time it will be like a DSL 1-20, 4 gain stages into the tone stack. I like to use 1W metal film resistors for plate load resistors rather than 1/2W to help with the hiss. If you're going 1W with the cathode resistors I'd go 1W with the plate load resistors also.
 

BenTobith

Member
Joined
Jan 6, 2021
Messages
59
Reaction score
23
I'm rebuilding my 18 watter for the umpteenth time right now. This time it will be like a DSL 1-20, 4 gain stages into the tone stack. I like to use 1W metal film resistors for plate load resistors rather than 1/2W to help with the hiss. If you're going 1W with the cathode resistors I'd go 1W with the plate load resistors also.
To clarify, most of the resistors are 1W metal film. The carbon films are 1/2W, though I plan to replace a few of those with 1W metal film.
 

_Steve

Well-Known Member
Joined
Apr 29, 2020
Messages
566
Reaction score
478
Location
Los Angeles
EL84s are really hard to dial in. The problem with the TMB channel is likely its just blasting way too much signal into the PI then the EL84s. They only need around 10V peak to clip! If you really dont want a master volume you could add a fixed voltage divider between the TMB channel and the PI to drop the signal around 50-60%. That channel is basically a standard plexi preamp designed to drive EL34s!

The EL84s are also really prone to nasty fuzzy blocking distortion. In a recent build I did a lot of experimentation with this and found that biasing cooler, closer to 100% INCREASED the blocking distortion, while running them hot with 120-150R Rk mitigated/eliminated it. I wish I understood why, perhaps someone can enlighten me. Maybe thats why all the popular EL84 amps run them really hot!?
 

Pete Farrington

Well-Known Member
Joined
Jan 18, 2021
Messages
1,443
Reaction score
1,591
Location
Staffordshire UK
found that biasing cooler … INCREASED the blocking distortion
For class AB cathode bias to accommodate being overdriven nicely, the full power conduction angle needs to be close to 360° ie hot class AB. For a given set of anode voltage, screen grid voltage and anode load, the anode dissipation resulting from that is what it is really.
The issue everyone’s having being that the PTs available put out too much HT voltage, and then that’s often compounded by people’s mains voltage running high.
If we try to cool off idle dissipation by increasing the value of cathode resistors, that moves the bias towards cool AB. That results in excessive bias shift at high signal levels, which can easily be assessed by monitoring cathode voltage. An Increase of much over 50% from the idle level when cranked will tend to sound bad.
The real answer to get idle dissipation more reasonable is to reduce the HT voltage. Anything else will be a bodge that messes with the tone the design is supposed to provide.
Simple bodges are to increase the HT dropper resistor feeding the screen grid node, use a higher value cathode resistor with a zener diode in parallel to limit the bias shift, use a mains bucker transformer to reduce the primary voltage at the PT (most useful if the heater voltage is also high, otherwise be careful that it won’t drop below 5.7V).
 
Last edited:

BenTobith

Member
Joined
Jan 6, 2021
Messages
59
Reaction score
23
For class AB cathode bias to accommodate being overdriven nicely, the full power conduction angle needs to be close to 360° ie hot class AB. For a given set of anode voltage, screen grid voltage and anode load, the anode dissipation resulting from that is what it is really.
The issue everyone’s having being that the PTs available put out too much HT voltage, and then that’s often compounded by people’s mains voltage running high.
If we try to cool off idle dissipation by increasing the value of cathode resistors, that moves the bias towards cool AB. That results in excessive bias shift at high signal levels, which can easily be assessed by monitoring cathode voltage. An Increase of much over 50% from the idle level when cranked will tend to sound bad.
The real answer to get idle dissipation more reasonable is to reduce the HT voltage. Anything else will be a bodge that messes with the tone the design is supposed to provide.
Simple bodges are to increase the HT dropper resistor feeding the screen grid node, use a higher value cathode resistor with a zener diode in parallel to limit the bias shift, use a mains bucker transformer to reduce the primary voltage at the PT (most useful if the heater voltage is also high, otherwise be careful that it won’t drop below 5.7V).
Helpful.

There is a 6.8K dropping resistor before the power screen grids currently. Any thoughts on a different value to try? EDIT: Or maybe that’s actually the 1.5k 5W..

What would the zener diode approach look like? Literally wiring in a diode in parallel with the output bias resistor?
 
Last edited:

Pete Farrington

Well-Known Member
Joined
Jan 18, 2021
Messages
1,443
Reaction score
1,591
Location
Staffordshire UK
Helpful.

There is a 6.8K dropping resistor before the power screen grids currently. Any thoughts on a different value to try? EDIT: Or maybe that’s actually the 1.5k 5W..
1k5 is a typical value for these amps.
How about if it’s doubled, eg 2k7 10W; does that bring the g2 voltage down enough to help with the idle dissipation?
Literally wiring in a diode in parallel with the output bias resistor?
Yes, banded end to the cathodes. Needs to be beefy. The bypass cap is still needed.
 

BenTobith

Member
Joined
Jan 6, 2021
Messages
59
Reaction score
23
1k5 is a typical value for these amps.
How about if it’s doubled, eg 2k7 10W; does that bring the g2 voltage down enough to help with the idle dissipation?

Yes, banded end to the cathodes. Needs to be beefy. The bypass cap is still needed.
I noticed a few schematics of this circuit use 2K 5W instead, that might be worth a try to start (it is across the lugs of the 32/32 filter can).

If I just increase the value of the cathode bias resistor, does that not also raise the plate voltage? Feels like whack a mole.
 

Pete Farrington

Well-Known Member
Joined
Jan 18, 2021
Messages
1,443
Reaction score
1,591
Location
Staffordshire UK
If I just increase the value of the cathode bias resistor, does that not also raise the plate voltage? Feels like whack a mole.
Yes, ‘fraid so, likewise if the HT dropper under discussion is increased.
Root cause will be that the voltage out of the damn PT is too high.

I suppose another option is to insert a resistor between the rectifier and the reservoir cap, eg try 470ohm 10W. That’ll bring down anode voltage AND current.
 

BenTobith

Member
Joined
Jan 6, 2021
Messages
59
Reaction score
23
I suppose another option is to insert a resistor between the rectifier and the reservoir cap, eg try 470ohm 10W. That’ll bring down anode voltage AND current.
So I think in my layout, that would be from the red wire off the standby switch to the first lug of the can cap?
 

BenTobith

Member
Joined
Jan 6, 2021
Messages
59
Reaction score
23
I got inspired after reading some Merlin on phase inverters and avoiding blocking distortion. Also from referencing other schematics of TMB builds. Made these changes:
- Lowered cathode/cold clipper resistor on V3 (pin 3) from 4.7k to 2.7k
- Lowered cathode resistor on V3 (pin 8) from 100k to 56k
- Added a 470k resistor between middle lug of Treble pot and grid of phase inverter (pin 7)
- Lowered 22nF coupling cap to grid of phase inverter to 2.2nF
- Added a 10k grid stopper at pin 2 of phase inverter
- Lowered 22nF caps after plates of phase inverter to 10nF
- Raised 8.2k power grid stoppers to 100k
- Raised 150R cathode bias resistor to 180R

Overall, I think the noise floor improved a bit, and the amp feels pretty good. TMB channel has less gain, and feels more stable. However, I think it’s a bit more ice pick-y than it was before. Might’ve been one of the cap value reductions I did. I might try walking one of those back to see if it helps.
 

ibmorjamn

Well-Known Member
Joined
Nov 3, 2012
Messages
11,373
Reaction score
11,810
Location
North
Is it possible to use a variac to drop the wall voltage a little ?
 

ibmorjamn

Well-Known Member
Joined
Nov 3, 2012
Messages
11,373
Reaction score
11,810
Location
North
Yeah, I definitely could. Although the amp seems to be running great in that regard right now. I used to have a Brown Box..
I just thought of that due to Pete Farrington’s post
“The issue everyone’s having being that the PTs available put out too much HT voltage, and then that’s often compounded by people’s mains voltage running high.”
 
Last edited:

ibmorjamn

Well-Known Member
Joined
Nov 3, 2012
Messages
11,373
Reaction score
11,810
Location
North
Mains voltage running 120 , not sure if dropping it 5 volts would compensate enough.
 

Latest posts



Top