Marwatt build thread…

[Raoul Duke]

Hi All,
Quite a few months ago I got some Weber 18w iron and a Mojo 18w chassis on CL for super cheap. Bench tested everything and it's all good to go. Haven't done anything with it because I'm not really pining for an "M-type" 18w; but I think I've found something that I'm interested in trying - a "Marwatt". Found plans on the 18w forum and a build thread here but want to make a couple of changes to the design such as SS Diode rectification (seems more Hiwatt to me) and the distributed ground scheme similar to the HAD/Larry combo I used on my 2204. Looks to be a nice companion to my Sluckey Dual-Lite (which is the most versatile amp I currently own BTW) with sort of similar tube compliment (2xEL84, 1x12AU7 PI, 2x12AX7 pre-amp).

My first step was to try and create a board in DIYLC that suits the rectification change. I am hoping to get some feedback on the design and component values. For instance:

- Is 16uf enough for B+3,4,5?
- Is the main res/B+ filter (dual 32uf 500v) a good starting point?
- Are UF5408s preferable to UN4007 for the rectification?
- The PT has 2 sets of HV taps (340-0-340 or 270-0-270) I chose 270 because of the SS rectifier based on my reading. Good choice?
- Are there any errors in the design?
- Any "if it were me, I'd do it this way" suggestions?

Any advice is most certainly appreciated! Thank you!

Marwatt board.png

[sluckey]

- Any "if it were me, I'd do it this way" suggestions?

If it were me , and I wanted a critique on a layout, I'd also post a schematic.

[Raoul Duke]

Understood Steve. I’ll try my best to mark one up and post. You have to promise not to laugh too hard - my schematic skills are still evolving, lol.

[Raoul Duke]

Here's the original schematic. I'm going to try and make my proposed changes on another document and I'll post that when I think it's correct(ish), lol.

MarWatt_version2.1.png

[mhuss]

The Marwatt "HIWATT" channel has a HIWATT tone stack but still a normal unregulated PI. The regulated PI is part of the sound IMO. Also, you're not going to get a lot of headroom (another HIWATT trademark) using EL84s.

[cdemike]

If it were me... I think I'd probably try to get the closest to the circuits the "Marshall" and the "Hiwatt" sides are aiming to capture. I read the Marwatt as being primarily a Marshall 18w/Watkins Dominator with the capability of doing something Hiwatt-adjacent by including the tone stack. Hiwatts are so unique that I'd probably go the opposite way and make it primarily a scaled-down Hiwatt and have the ability to make it capable of doing something 18w Marshall/Watkins Dominator adjacent via switching. It's probably not too surprising that I completely agree with:

The Marwatt "HIWATT" channel has a HIWATT tone stack but still a normal unregulated PI. The regulated PI is part of the sound IMO. Also, you're not going to get a lot of headroom (another HIWATT trademark) using EL84s.

That said, part of my reasoning behind errring toward the Hiwatt side is that I think it'd be possible to avoid some of the big sacrifices that come with the standard Marwatt design.

I wouldn't bother having the channels operated by separate inputs and just change the signal chain via switches. So rather than mixing a parallel channel into other end of the PI, I'd probably start using a 3-position center-off 4PDT power switch that uses two poles for the usual power switch function and the remaining poles to switch the PI format from the 18w self-bias style to the Hiwatt-style regulated, directly-coupled setup. This how I'd set up the power switch for that setup (shown only as DPDT for clarity's sake). Admittedly, this makes some sacrifices erring toward the Hiwatt side in terms of tail resistance, and I also opted to balance the PI and use 12AT7. But this would work fine with 12AX7.

IMG_2032.jpg

From there, you could direct the signal out the 18W's input gain stage either to the Hiwatt tone stack or the regular 18W tone control. That wouldn't require more than a DPDT switch which you could operate via a push-pull pot.

The last piece of it would be including the additional gain stage Hiwatts have ahead of the stage driving the tone stack, which I'd just set up like a JCM800-style input circuit using a switched jack to optionally add an additional gain stage cascading into the 18W input gain stage. I think this is also a key part of the sound since the second gain stage is not center-biased and additionally can be overdriven (albeit lightly and much later than both EL84's and tone stack recovery stage).

Optionally you could also have variable NFB control. Part of what makes the Hiwatt PI do its thing is very strong NFB signal compared to Marshalls of similar wattage. 4K7 or 6K8 would be the most appropriate NFB resistor values to scale the NFB to 100w or 50w levels, respectively. I'd probably pick the 100W level and use an audio taper 250K pot in series to basically eliminate the NFB if I wanted full 18w-mode. Use of a dual-ganged pot here with a parallel resistor to reduce the pot's value to ~30K could also simultaneously increase the tail resistance as NFB decreases. If you did that, you'd be able to basically have a regular stock 18W on one side and a fairly complete representation of a Hiwatt on the other (albeit with much lower headroom owing to the EL84s).

Edit: forgot to mention that another solution would be to make all the above switchable via a single SPDT operating several relays or optocouplers that have the same functions above but makes for a much more ergonomically approachable amp than the kind of all-access build I described.

[Raoul Duke]

Wow, thanks for the input Mike! Those are a lot of really good/practical/useful suggestions and improvements. Honestly, for this design I’d rather concentrate on getting the Hiwatt side right - i.e. scaled down Hiwatt circuit as you suggest. Your schematic is really appreciated and I will be studying it. Very much appreciate you taking the time to comment and draw that up!

Which brings me to Mr. Huss’ post:
Firstly, thanks for your awesome websites! Your Hiwatt site is the source I go to for everything I’m interested in learning about Hiwatts; and I used your JCM 800 site for help with my 2204. Great stuff, but I’m sure you’re used to hearing that, lol. I agree that the PI really is important to the sound and about the EL84s.

So considering both comments I’m now wondering if I might be better off just building a straight Hiwatt 6V6 type of circuit - which actually appeals to me more at the moment. I think there’s a Trinity build thread floating around here somewhere that might show such a beast. I think the iron I have will work for that as well - but I’m going to research it a bit more to be sure.

Thank you both! Really appreciated!

[cdemike]

Of course! Interested to see where this project goes.

I did catch an error in my schematic with respect to the 470K bias reference resistor's placement. Here's a corrected version:

IMG_2033.jpg

I totally agree about the 6V6's being a better match. In keeping with the theme of the thread, if it were again me approaching this build, I would explore an output section based around 6K6 since they're true pentodes. Both 6K6 and 6V6 datasheets describe operation with screens and plates at 285V, and something else that favors the 6K6 ahead of 6V6 is that it requires a greater magnitude bias voltage than 6V6 (-25.5V vs -19V) under those conditions, which would help also by creating a greater margin before the output section gets into blocking distortion. I don't know that I'd actually describe them as a higher-headroom tube given the slightly lower plate dissipation limit, but they at least should tolerate a larger input signal prior to drawing grid current (so I think it'd be fair to describe them as having higher input headroom).

On the other hand, there is a big limitation in that 6K6's don't have a modern 6V6S-type equivalent that would tolerate the kinds of high voltages I'd expect for a Hiwatt-type of build. I don't want to put words in anyone's mouths, so I'll also just add the counterpoint that Hi-Tone builds something along the lines of what you're describing using 6V6's that I've read only good things about. So for me the real question is what would be more important to you in this build between true pentodes in the output section and high operating voltages. The 6K6's have the true pentode thing going for them, but you probably can get more headroom out of 6V6S's since they can reliably operate at higher voltages, which also would make more sense for getting the right operating points for all of the triodes in the amp.

Of course, a third option would be to try running EL34's inefficiently by generally lowering output section voltages and/or cathode biasing the output section (thinking about the Marshall "Studio" series). With EL34's you could alternately lower only the screens' voltages.

Lowering any or all of the output section voltages (or using 6K6 at or near the limits) wouldn't necessarily mean lowering the preamp voltages if you had a PT with multiple secondary windings.

[Raoul Duke]

Again, lots of great perspectives and advice here Mike; thank you.
The up-side: lots of fantastic ideas here.
The down-side: lots of fantastic ideas here, lol.

I need to really work through these ideas and figure out what I want to accomplish. I’d likely stick with the 6V6 idea but the EL34 idea is interesting as well. The PT has secondary taps for 540v and 680v as well as primary 120v and 125v so I reckon that gives me some options.

Thanks again, fun stuff!

[Raoul Duke]

Ok, after sorting through all CDEMike’s great ideas (thanks again Mike!) and taking into account what I already have on hand - I think my move will be to go ahead with a modified Marwatt and plan a super versatile Hiwatt build for the future (and do it right: EL34s, Heyboer iron etc.). Treat this one more like a proof of concept type of build.

I did some calculations for what I have here and here’s what I found (please correct me if I’m wrong):

PT with full-wave/two phase CT rectification: 270*1.414 = 381.7*.95 = 362.6 estimated loaded plate voltage. Sounds ok for 6V6.

OT was more convoluted to find answers on, but from what I think I found - the 4k primary impedance is a bit low, but it being a 2,4,8R secondary - this might get me closer to 4,8,16R.

Still would like to make this fixed adjustable bias if possible. The PT has a ~45v bias tap - so hopefully that makes it easier.

Now I just have figure out how to link all this together correctly.

[martin manning]

OT was more convoluted to find answers on, but from what I think I found - the 4k primary impedance is a bit low, but it being a 2,4,8R secondary - this might get me closer to 4,8,16R.

A 4k to 2-4-8 nominal OT running with 4-8-16 secondary loads will certainly work for 2x 6V6 or 2x EL84, as that will result in an 8k primary impedance.

[Raoul Duke]

Thanks for the confirmation Martin! I’m getting better at figuring out how to calculate stuff thanks to you and Steve - but it’s still so new to me that the double check is certainly appreciated!

Next exercise is double checking all the Marwatt values for 6V6 compatibility…

[Raoul Duke]

I did some research regarding the aforementioned 6V6, SS rectified, adjustable fixed bias changes to this design and here's what I've come up with. I'm presently drawing a schematic that (hopefully) captures this layout as well, but had a few questions:

1. Is it worth making the PI more Hiwatt (value-wise) or is this PI a compromise to get both channels in the ballpark?
2. Couldn't find a formula for the "choke" resistor, so I used what seemed to be in the middle of the designs I looked at for reference. Too small?
3. Worth trying 22k grid stoppers (Hiwatt value) or is that trivial/silly?
4. Would stiffer filtering give it more "Hiwatt" punch or is that trivial/silly?

I understand some of this will be "try it and see"; but I'm just trying to start in a reasonable place as my experience and predictive analysis skills are still pretty low when it comes to understanding these types of changes. Any comments or suggestions are welcome and appreciated! Schematic to follow shortly.

Marwatt rev1.png

[cdemike]

I did some research regarding the aforementioned 6V6, SS rectified, adjustable fixed bias changes to this design and here's what I've come up with.

The values for this look good to me, though as-shown the bias control would have an almost negligible range (approx -55 to -60V) since the 470R resistor is so small that the bias pot doesn't significantly change the ratios in the voltage divider. Increasing that resistor to 4K7 would make change that to approx -42V to -50V. You'd have a very broad and usable bias adjustment range by using the bias control as a potentiometer rather than a variable resistor, though, which would only require using a safety resistor in lieu of the jumper across the pot's terminals.

1. Is it worth making the PI more Hiwatt (value-wise) or is this PI a compromise to get both channels in the ballpark?

I think values tweaks can adjust things like headroom and balance, but at the end of the day an AC coupled LTPI based around a 12AX7 will have pretty different performance characteristics than a LTPI directly coupling a 12AT7 to a cathode follower, especially if you're driving the PI. Additionally, I think one of the keys to what makes a Marshall sound like a Marshall is overdriving an AC coupled LTPI, so you might end up sitting between two stools, so to speak, by trying to get the phase inverter to have middle-ground characteristics between a Marshall and Hiwatt phase inverter. So I'd probably stick with either the stock 18w values and knock the tail resistor back down to 47K. FWIW, increasing the tail resistance from 47K to 56K would lower the PI's headroom, which seems like it'd be the opposite of what you'd want if you wanted to make the PI more Hiwatt-like?

2. Couldn't find a formula for the "choke" resistor, so I used what seemed to be in the middle of the designs I looked at for reference. Too small?

I'm not really aware of a specific calculation other than to determine values for a desired amount of ripple attenuation. What parameter were you trying to determine values for?

1650R is way higher than the100R native to most Hiwatt power sections. 500R or 750R would probably be where I'd start to get a limited amount of screens node sag over the standard 100R value while being high enough to the 1K5/2K value that lends some sag to the Marshall/Watkins' power supply. I don't think you need both the 150R resistor feeding both of the flying 1K screen grid stoppers, by the way, unless you wanted to lower the potential at the screens while keeping B+ at the phase inverter high but even then I'd probably swap the positions of the 1K5 and 150R resistors.

Having this on a switch could be a good place have both, which you could use in lieu of a standby switch. Related I think Merlin's points about DC switching on the standby switch are pretty convincing, so I probably would avoid using a standby switch that handles DC altogether. This amp wouldn't have a cathode follower, so I can't think of a reason why it'd need to have a standby switch in the first place in any event. If you were committed to the standby switch, you could riff on the "least bad" topology Merlin suggests (i.e., having the switch short large-ish resistor on the B+ supply to limit current flow) and use a 3-position switch placing different values in parallel with this specific B+ dropping resistor to get a lower resistance to do the tighter Hiwatt thing or a higher resistance for the saggier Marshall thing.

3. Worth trying 22k grid stoppers (Hiwatt value) or is that trivial/silly?

That probably wouldn't matter in this specific case, since you're using a 12AX7. It does matter in a Hiwatt since those big grid stoppers help keep blocking distortion at a minimum despite lower output impedance from a 12AT7 phase inverter.

4. Would stiffer filtering give it more "Hiwatt" punch or is that trivial/silly?

This is a good idea and likely would make the amp punchier/tighter like a Hiwatt. I made an amp that had switchable filtering values on the reservoir and screens node, which I didn't actually end up using very often but was very helpful for finding what values I eventually did permanently set at those nodes.

[Raoul Duke]

Thanks again Mike, I appreciate the second set of experienced eyes for sure!

- I used Martin's design for a fail-safe bias control; so what you're saying is that there's a variation on that just by moving the 470R to the wiper in place of the jumper and changing it to 4k7?

- Agreed on the PI tail resistor, will do.

- Understood on the choke resistor - that was guess work on my part, so I appreciate the advice. Probably start at 750R and try to split them at both locations (I think?). I'm good with omitting the standby switch as well. I left it because it was in the original design.

- Understood on the PI grid stoppers and filtering, thanks for that info as well.

As promised, here's my first attempt at drawing a schematic - this one being only for the proposed changes to the power section of the layout above with one correction: 220k bias splitting resistors in place of the 470k pictured. Any feedback is appreciated. Still trying to learn the right way to use the documentation together.

Marwatt rev1 schem.png

Thanks again! I appreciate all the help!