Marwatt build thread…

[Raoul Duke]

Above suggestions incorporated and documented in both layout and schematic as well as a few errors fixed. Hopefully I have this right.

After reading Mike's bias suggestions it jogged my memory to some Merlin reading I did a few months ago which led me to a thread on EL34.com where many ways to skin this cat were debated. Once I re-read everything - the light bulb went on and all the dots connected. In the end, I chose sluckey's method for implementation.

Any feedback is welcome and appreciated!

Marwatt board rev2.png

Marwatt rev2 schem.png

[cdemike]

Looking good! The new draft addresses most of what I had to say about the previous version. This is a true nit-pick since it's not reflected in your layout, but there appears to be an error with the bias reference resistors being connected to either end of the bias pot. Would hate for a future layout to reference the schematic and have that mistake though.

I'm not sure you need the series 220R + 470R B+ droppers between the reservoir and screens nodes, especially since they're spec'd at 5W and 10W, respectively, per the layout. A single 5W resistor equaling 680R or 750R across each side of the 50x50uF can cap would be fine.

Was there a specific reason for specifying 2W grid stoppers on the 6V6's? There shouldn't be too much current flowing through them, so 0.5W or 1W would be fine. Obviously won't hurt the amp to have the higher value, though; mostly just the wallet!

A few lead dress items:
- I would probably rework the input gain stage to look more like a 50W/100W Marshall in terms of component layout. Having the plate resistors straddling the cathode resistors/caps doesn't make a lot of sense in terms of keeping noise and potential for oscillation to a minimum. Here are layouts for a JTM50 and a Hiwatt signal board. Note the order of components as you move through the circuit will generally flow grid > cathode > plate, i.e., keeping the highest sensitivity components farther away from stronger sources of interference.

Blackflag 50 layout.pdf

- On a similar note, I'd recommend moving the tone stack components to be adjacent to the plate resistor for the Hiwatt side's input gain stage.
- It may work fine as is, but I'd reducing the number of leads feeding the Marshall side's volume/tone controls to one in order to minimize the total lead length the signal traverses through the circuit (less potential to pick up signals resulting in oscillation or noise).
- Relatedly, why not have the jumper running from the plate of the input gain stage on the Hiwatt side run directly to the slope resistor? Having it run to a turret and then to the slope resistor increases the lead lengths compares to having it just run diagonally.
- This might be silly voodoo, but I try to have the filter caps for given gain stages as physically close as possible to the plate resistors they supply. Same thing in terms of potential for noise/oscillation, so my thinking is that having the minimum distance possible to AC ground would keep things nice and quiet/oscillation resistant.
- Just wanted to confirm you're planning on laying the chassis out such that both input jacks are on the far side of the chassis rather than the typically Marshall 18w convention having one channel's inputs run into the center of the chassis. That'd be a big improvement over the typical 18w layout.

[Raoul Duke]

Thanks Mike, I appreciate the time you take to respond for sure.

- This is my first attempt at drawing a schematic, so please be nit-picky; I want to do it correctly. I read the bias sensing resistors going to ground and the 220ks to the wiper of the bias pot. Should the pot symbol be reversed or mirrored perhaps?

- Understood on the droppers, will adjust.

- Understood on grid stoppers as well. I think I went with 3w as a conservative estimate “just in case” based on the heat generated in that area. Still learning how and when to use the calculations correctly.

- Understood on the lead dress suggestions also. I’ll think on those a bit and come up with changes that follow your suggestions - which made total sense when you explained it the way you did and examples you used. I’ll likely have questions as I work through that - but I’ll get as far as I can first.

- I was planning to stick with the general 18w panel layout as my chassis was already drilled with matching TMB face plate. I’ll mull over ways to try and adapt it based on your suggestions but may not get there. I’ll try though.

Thanks again for the QC and advice. Very much appreciated!

[cdemike]

- I was planning to stick with the general 18w panel layout as my chassis was already drilled with matching TMB face plate. I’ll mull over ways to try and adapt it based on your suggestions but may not get there. I’ll try though.

Hmm if you already have a faceplate I get sticking with the original front panel organization. Do you ever use your low impedance inputs? One alternative (which admittedly lacks aesthetic appeal) would be to use one jack on the far right of the faceplate per channel by eliminating the low sensitivity inputs (one high sensitivity input per channel on the far right). That would leave the inputs in the center of the faceplate unused. You could then plug the holes in the front panel or repurpose those holes for switches or other controls. Variable cathode bypass cap options would probably be the first thing I'd think about, since the typical Hiwatt 47nF is super different than fully-bypassed per the typical 18w or the 680nF common to lead-spec big Marshalls. You could also try 680nF on one or both of the cathodes on the Marshall side. A master volume may also be nice, though the typical placement in one of the input holes is at odds with lead dress best practice: assuming it's a PPIMV (a bootstrap or type 4 master volume wouldn't make sense in these amps anyway), that would mean running the second-highest magnitude signal in the amp running immediately adjacent to the leads carrying the weakest signal in the amp. Although a twisted pair would minimize chances of interaction since the leads would be out of phase, I'm not really sure what advantage, if any, would be on the table arranging the TMB input that way.

Either way, I didn't realize you already had faceplates and I get not wanting to waste a perfectly good part. Same thing for the standby switch hole, by the way — you could always plug that unused hole, though it could be a reasonable spot to put in an additional control (lead dress-wise, this actually would be a reasonable place to put a switch controlling the value of the B+ dropping resistor between the reservoir and screens nodes). The standby switch hole would also be a good place to put a master volume.

[Raoul Duke]

Here's the modifications to the board layout. I think I got most of them - but please pick it apart. Had a few head scratchers trying to make things more direct. Also, I used actual part dimensions to get an idea where things are fitting.

Marwatt pre-amp change.png

Thanks!

[cdemike]

Here are my suggestions in red. I think I caught an error where the Hiwatt tone stack was being fed by the B+ rail rather than V2B's plate. The double-ended arrow is where I think you may have meant to place the jumper instead.

Reviewing the layout prompted me to look back over the thread, and I realized I provided you incomplete advice on the first layout draft regarding the Marshall side's tone and volume controls. I specifically should have clarified about alternatives regarding the parallel leads coming off the board. If you opt not to mount the components on the per that suggestion, I think there are 3 conventional solutions:
1. You could mount those parallel coupling capacitors with leads flying in a point-to-point kind of way in a v-shaped formation hanging off the pots;
2. Alternately, you could mount a terminal strip near the tone stack with the shared lead from the board connecting to a non-grounded terminal where the two coupling caps connect to the pots (basically the same as the point-to-point method in #1, but with the junction of the two capacitors and the lead from the board mounted to the terminal strip rather than left flying);
3. Or you could use the capacitors themselves as leads from the board to the volume and tone controls (sort of like how the 470K mixing resistors are mounted in the Hiwatt layout; this is what I drew into my suggestions on the layout).
I'd avoid number 1 since it's the flimsiest, which gives me pause in light of how much voltage would be on the plate side of the coupling caps. #2 would be fine, but involves buying and mounting a terminal strip when #3 provides a simpler option from an assembly perspective.

I also showed how I'd move the cathode resistor/capacitor turrets to minimize the possibility that the cathodes could serve as entry points for EM interference. Viewing the board as shown in the layout (with the power switch in the top left of the document), the side farthest to the right should be the quietest and least susceptible to interface/parasitic coupling, which is why I showed moving V1's cathode components farthest to the right. I swapped the order of the voltage divider going into V2A and V2A's cathode resistor/capacitor since the grid is a more sensitive input than the cathode and moving them to the right keeps components leading into that sensitive input farther away from the larger-magnitude signals associated with the phase inverter. I didn't show it, but I think I'd also run the 470K resistor to the same grounded turret as the V2A cathode resistor/capacitor (that cathode resistor and grid leak resistor are associated with the same B+ node, so it works from a grounding best practices standpoint by following a local star format. This also reduces the number of turrets you'll need to install while also reducing the possible points of failure in terms of ground connections).

marwatt layout suggestion.PNG

[Raoul Duke]

Thanks again Mike, really great suggestions that I will certainly implement. Trying to create the new layout using the original layout as reference - I’m pleasantly surprised that I only got the jumper wrong. My understanding of how things flow is still growing, so sometimes those types of things (tone stack on B+) aren’t always apparent to me. To my chagrin, I seem to think in a very linear fashion and forget that electricity is more like water in plumbing in that it flows to multiple places all at once vs. one point at a time sequentially. I’m readjusting though, lol.

- I’m sold on the 3rd option for the Marshall controls. I wasn’t sure how to go about that. Being a tremendously literal person - I was racking my brain trying to figure out how to get it down to just one wire - not even considering the “Hiwatt” method even though I’ve been staring at Hiwatt layouts for reference for days now. Forest for the trees I guess, lol.

- Understood and will do regarding everything else.

I’m fairly new to DIYLC as well as reworking an existing design, so this exercise has really been a great learning opportunity and I appreciate all your guidance. I’ll post the changes and any remaining questions on the few things I haven’t tried yet as soon as possible.

Thanks again!

[Raoul Duke]

After a few days away from this due to a freak micro-burst storm to include related clean-up and repairs and left us on generator power for just short of 48 hours - here's the latest revision to the Marwatt layout incorporating many great suggestions by cdemike. I included tubes this time as well. A couple of questions:

- Moved the 1M on V2 grid B to the jack to eliminate the turrets and jumper. Any downside to that I might be missing?

- Is there a better place for the 22uf/475v cap further downstream/after the PI? I was going to start consolidating things on the board and shrinking its' size - but thought about this move before I did and wanted a second opinion (or third, fourth etc).

- Looking at values - I want to check the math on that 2k2 dropper and think V=IxR is the way to do it - but I'm stuck figuring out "I". Am I just better off to wait and see what the actual voltages are once I get there?

- Last silly question: any benefit to using a 25k bias pot (like Marshall)?

Marwatt board rev3.png

Any feedback is appreciated!

[Raoul Duke]

Ok, here's my final board layout (I think). Caught a couple of errors on the previous one and fixed them, changed some values to match what I have on hand to use, consolidated some space. Feels like I'm ready. Going to start checking the chassis, x-formers, mechanical fit tomorrow. The chassis I got in this deal has a bunch of extra holes so I want to confirm where everything should sit. I'm sure I'll have a few adjustments to make as well as enlarging the power tube holes.

Marwatt board rev4.png

[cdemike]

New layout looks great! The order of components makes good sense. I did catch the jumper connecting connecting the plate of Marshall side's input gain to B+ appears to be mis-routed to the Hiwatt input gain stage's plate. The 47nF cap off the Marshall's input gain stage also appears to be on the wrong side of the plate resistor. The only other thing is that it seems like you may have omitted the filtering stage isolating the preamps from the PI. That may work, but I probably wouldn't chance it myself since that'd make for 6 triodes sharing a node, some of which would be in-phase with each other and could interact and cause instability.

- Moved the 1M on V2 grid B to the jack to eliminate the turrets and jumper. Any downside to that I might be missing?

I don't think so. Marshall and Hiwatt both tended to mount them at the jack, so it'd also make sense from that perspective. But I think the only resistor in the input network that should be mounted as close to the socket as possible is the grid stopper, which you've done.

- Is there a better place for the 22uf/475v cap further downstream/after the PI? I was going to start consolidating things on the board and shrinking its' size - but thought about this move before I did and wanted a second opinion (or third, fourth etc).

I'd probably try to see if I could mount it at the turret where the two Hiwatt preamp plate resistors are mounted. The other side of the Marshall side's plate resistor would also work. But I don't think it's super critical. I did some more reading after my last post about filter cap placement, and I think I'm eating crow for supper after seeing how Trainwreck amps laid out their B+ rails all together on one board on the PT side of the chassis. Ken Fisher definitely is a better source than I am in terms of best lead dress practice, so I think you'd also be fine to mount it over where the other board-mounted dropping resistors and filter caps are mounted.

- Looking at values - I want to check the math on that 2k2 dropper and think V=IxR is the way to do it - but I'm stuck figuring out "I". Am I just better off to wait and see what the actual voltages are once I get there?

Yup, Ohm's law would give you the answer. The current would be the idle current you'd get from the load line would be the best approximation. Short of that, go with previous Marwatt or similar builds' voltage readings. Tubes aren't linear, though, so changing the B+ dropping resistor won't be as simple as just determining a desired voltage drop and dividing by the current demand in the present configuration. I usually go with something either proven or conservatively large B+ dropping resistors and reduce them based on what I find I like best through incremental testing.

- Last silly question: any benefit to using a 25k bias pot (like Marshall)?

It could let you use lower power-rating resistors, if that's a concern, since the current through the bias circuit would be less with higher resistance. But no serious downside as long as you're not making an extremely high resistance bias circuit, which I wouldn't consider to be the case with 25k.

[Colossal]

On your new layout, caps C14 amd C15 are facing the (-) side the wrong direction.

[Raoul Duke]

On your new layout, caps C14 amd C15 are facing the (-) side the wrong direction.

Wow, I had that wrong on every iteration of this layout and never caught it (duh!).
Thanks Colossal!

[Raoul Duke]

New layout looks great! The order of components makes good sense. I did catch the jumper connecting connecting the plate of Marshall side's input gain to B+ appears to be mis-routed to the Hiwatt input gain stage's plate. The 47nF cap off the Marshall's input gain stage also appears to be on the wrong side of the plate resistor. The only other thing is that it seems like you may have omitted the filtering stage isolating the preamps from the PI. That may work, but I probably wouldn't chance it myself since that'd make for 6 triodes sharing a node, some of which would be in-phase with each other and could interact and cause instability.

Thanks Mike, the above is what I’m going to try and come up with solutions for.

- Should the B+ jumper go to the 100k/100k junction of the V2 plate resistors?
- Swap the 47nf cap over to share the turret with the 10nf?
- Understood on the preamp filter, I was wondering about that in the original design. Is it as simple as putting it between .022uf and .68uf caps, running the B+ to its positive end, ground the negative end, then continue the B+ to the previously mentioned 100k junction?

Thanks again for the helpful advice! I’m still learning how this all works and appreciate the coaching.

[Raoul Duke]

Like this:

Marwatt board rev5.png

[cdemike]

- Should the B+ jumper go to the 100k/100k junction of the V2 plate resistors?
- Swap the 47nf cap over to share the turret with the 10nf?
- Understood on the preamp filter, I was wondering about that in the original design. Is it as simple as putting it between .022uf and .68uf caps, running the B+ to its positive end, ground the negative end, then continue the B+ to the previously mentioned 100k junction?

New version looks right in terms of these issues.

- Understood on the preamp filter, I was wondering about that in the original design. Is it as simple as putting it between .022uf and .68uf caps, running the B+ to its positive end, ground the negative end, then continue the B+ to the previously mentioned 100k junction?

I don't think there's any issue in terms of physical placement in of the filter in the most recent draft, but you're missing the resistor separating the PI node from the preamp node. Best solution would be to move the all of the turrets between the 22nF PI input cap and the 100nF PI output coupler shown farthest to the left over one turret. That would free a column of turrets where you could mount the dropper. That said, I don't think it'd made a big difference if you just put the dropper along the empty column between the left-most PI coupling cap and the 22uF filter cap associated with the PI node.