Help designing a more versatile Marshall style circuit

[cdemike]

I'm taking on a challenge of trying to downsize the amount of equipment I have and have a "capsule collection" of versatile amps (floor space issue in a small-ish apartment). I love my Marshall SV20, and I'm selling a Fargen Mighty Plex that I really enjoy (especially since it's cathode biased, so tube rolling in the output section is a lot of fun for me), so I'm planning a build that checks both boxes. I generally play Marshall style amps aiming to get in the ballpark of Jimmy Page's live sound from around 1973 (The Song Remains the Same). The SV20 does that pretty well, but it has its things that I'm hoping to improve on. I've read that the Met exhibition Page did said his main amp was most likely actually a Super Bass (not Lead), and I've used that as a starting point. I know its bordering on tin foil hat speculation territory, but I'm in the camp that thinks he had KT88s in it by that point as well.

So this is what I'm planning on:
1986 (Bass)-spec 50w build
Toroidal transformer (weight savings): I was considering Antek's AS-1T300, which is rated for 300V, which I'm estimating will give a B+ of around 420V using the multiply-by-1.414 method
Built-in EP3 preamp using a 20V zener and filtering to supply power
A pot to control input sensitivity by putting it in series with the V1 cathode bypass cap (borrowed idea from Merlin)
Variable slope resistor for more tone stack versatility (see below)
100w output transformer for "big amp feel" -- I was considering using a 1202-132 (2") style output transformer with the 4- and 8-ohm taps moved to 8- and 16-ohm settings, respectively (no 4-ohm option) to adjust for primary impedance (idea borrowed from the Mojave Scorpion)

I've attached schematics for the plan 1986-spec (with 6550s to adjust for KT88s) and the SV20. I think I have most of it worked out at this point, but I'd really like to get feedback early on in the design stage before proceeding to layouts. I'm also trying to figure out what's the best way to approach the tone stack modification (if it's an idea worth pursuing in the first place). I've attached 3 different versions: one with the pot working parallel to halve the 56k slope resistor to ~33k, one with it in series to double 33k to 58k, and one similar to the parallel version but with a big capacitor to avoid having any DC at the control panel. The big cap's value wasn't chosen around much more than what film cap values I could find easily available, so I'm not committed to 680nF. I did think that the bigger value would be more "transparent," but I'm hoping to get feedback about that and generally which approach might be best.

I'm also on the fence about adjustable vs cathode bias. My thought on cathode bias is that it might cut down on costs in the future to use independent cathode resistors to avoid having to buy matched pairs, and I really like my SV20. I do have some concerns about bass looseness, but I understand that 100uF bypass caps should offset some of the difference between cathode and fixed bias in terms of feel?

[dorrisant]

I like where you're headed... pulling up a chair.

Food for thought, check this out: https://thesubjectmatter.com/calcptcurrent.html

That site will help with your predictions. Also, +1 on the Antek PT. I have used them before and will again. The voltages tend to run a bit high, but there are things that can be done to regulate it down as far as you want.

Not telling you to change your plan, but you need a cap to block DC from the slope pot. I attached a pic of the way I have done it in the past. the 0.1µ cap seems to not affect the bandwidth. Without that cap, the pot is scratchy sounding and dangerous.

[cdemike]

These are great! Thanks so much. The calculator seems to indicate that 325V may be better for an on-spec B+ of around 440V. I'm somewhat hesitant to bump the B+ too far, but that's primarily a function of uncertainty over what would happen if the OT primary impedance is too far below the "rule of thumb" estimations I've been relying on. The equation I've used to estimate primary impedance is:
Z(primary)=B+/([# of output tubes]*[idle bias current])

That said, 440V would definitely be closer to original spec per the 100W and 50W schematics (definitely not the the SV20, but that's probably a decision I need to make in terms of design direction... That decision probably revolves around whether cathode bias is a good idea given the overall design goals).


Feeling a little silly after reviewing my "Version 3" schematic aimed at keeping DC off the pot and finding the wiper is still connected to DC... I'm glad to hear 100nF worked for you. I know Fender designed the circuit with the cathode follower aiming to buffer the losses in the tone stack circuit, but I also have played with Bassman/Marshall style tonestacks enough to know they're not exactly the most responsive... So in other words, if the DC cathode follower didn't work well as designed in the first place it definitely seems possible that converting it to an AC circuit would involve minimal audible changes. Any chance you were able to compare it vs the "plain" DC cathode-driven tonestack? Thanks again -- keep the suggestions coming!

Edit: forgot to ask about working with toroidal transformers! I haven't worked with toroidal transformers before and have read about the big inrush current for the first few cycles after power up. I planned on using slow-blow fuses on-spec with the Marshall schematics, but wasn't sure if those would blow regularly given the different power supply and become an inconvenience. Did you use a NTC thermistor and/or adjust your fuse values? I also have seen warnings about adequately insulating the mounting bolts to avoid the transformer welding itself to the chassis. Did the mounting hardware Antek provided work for you?

[dorrisant]

I've never had to adjust the mains fuse value really... didn't seem to be an issue.

As far as isolating the toroidal PT from the chassis, I have designed a 3D printed a cup and lid that worked out great. Some of the first few I installed, I used the metal enclosures that Antek sells. They really didn't look safe to me, with the enclosure about 1/16" from the chassis. I ended up putting a plastic disk between it and the chassis to prevent from contact. You can see in the pic that the disk is painted the same color as the chassis. I used the two enclosures that I had bought and switched to the 3D printed type from then on. I still use the fender washer type mounting hardware that Antek send with the TXs, it installs under the cap. there are holes in the bottom of the cup for mounting and anti-rotation as well as bigger holes for all of the wires to pass through.

Another good thing about the toroidals... you can wind you own secondary taps to get control voltages etc..

[cdemike]

Very relieved to hear how painless working through mounting and current surges was for you. Really cool artwork on the top of the chassis, by the way!

The next hurdle I'm facing is "options paralysis" in terms of the output transformer specs.

Question area 1: I'm seeing differences between manufacturers' interpretations of these output transformers in terms of choice of steel choice. This kind of metallurgy is beyond my knowledge, so forgive me if this is overly-simplistic, but I understand "higher grade" steel like M-6 to be more hifi sounding versus lower grade like M-19 to roll off highs and lows. Is that the case? s there any difference in terms of attack/compression?

Question area 2: another difference I'm finding is wire gauge (e.g. thinner wire in the 4K output transformer for the JTM45/100 [Drake 1202-84] versus the Drake 1202-132). Does this make an audible difference?

Regarding the tone stack: Dorrisant, I'm curious if you tried the slope pot with and without the coupling cap ahead of the tone stack. I've read enough threads to understand that there is some DC leak across most common capacitor types used in tone stacks, hence the need to isolate DC using the coupling cap, but shouldn't this be minimal? I understand keeping high voltage DC away from the control panel is a safety issue as well, but I'm not very clear on how much of an issue this might be given the existing DC buffering in the stock tone stack. Maybe I'm misunderstanding this?

I'm doing my best to avoid "cork sniffing" during the design process, so I apologize if this is getting in the weeds.

As an update, I'm leaning toward adjustable bias with independent bias controls so I can be a cheapo and use mismatched KT88s. I've attached my schematic which was based on Mark Huss' adjustable Hiwatt circuits and should, as far as I can tell, provide the same bias range as the stock setup.

[Roe]

adjusting the slope resistor works best if you also adjust the 500p/250p treble cap.

a 1202-132 here seems overkill, expensive and heavy. the -84 sounds different from the -132 due to different Z. 440v seems ok. The jtm/bass circuit normally works best at 450+v, whereas the 1987 circuit works well at lower voltages. Filtering and NFB matter greatly on these amps

[Reeltarded]

I do much different things, but Roe's point on the treble cap is a key, as are filtering options. I like 50u mains/screens and decreasing values toward input for the more relaxed period response.

I don't believe in 33k slope. I start at 47k as a general rule.

Jimmy Page's Marshall sound is a relaxed preamp with early value treble and slope modified PT selection with KT88 and around 180w RMS. HIGH internal voltages. I could believe .047 output coupling.

The soul of it is moderate volume setting with again early brite setup. Think JTM with Bass model output.

I do not use a blocking cap on variable slope. It makes no noise for me. I started building my tonestacks off the board on the controls. Minimizes stray fields between V2 and PI.

[martin manning]

I do not use a blocking cap on variable slope. It makes no noise for me.

I don't think there is an issue there, no DC path to ground.

[cdemike]

a 1202-132 here seems overkill, expensive and heavy. the -84 sounds different from the -132 due to different Z.

That's a good point -- certainly at odds with going with a toroidal PT for the weight savings... Do you have a suggestion for an alternate solution to get the amp to sounds more like a 100w equivalent? Based on the clips I've heard (e.g. this clip from Johan Segeborn: https://www.youtube.com/watch?v=58_dr4WcnEE), it does seem like there are some intrinsic differences between the 100w and 50w, and I'm inferring that is mostly a function of OT and, to a lesser degree, B+. That line of thought led me to 100w output transformers broadly, but I narrowed on 1202-132 for the bass response. About 15 years ago I had a late 4-hole metal panel 1959 that burned its OT which I replaced with a Metro 1202-132, which I really liked, but that amp is obviously pretty different than a bass-spec 50w (not to mention nearly heavy enough to need a crane to put on top of a full stack). I considered a C1998, but I didn't really like OT that came out of my old Super Lead (most likely a C2668, if I remember correctly, but I can't remember). Again, though, very different circuits, so I'm totally open to suggestions on OT choice and approaching the "big amp sound" question.

440v seems ok. The jtm/bass circuit normally works best at 450+v, whereas the 1987 circuit works well at lower voltages. Filtering and NFB matter greatly on these amps

Good deal -- I'll probably change course and plan for a 350v PT. The calculator dorrisant shared estimates 474V B+, which is on-spec with the Mojotone schematic and other's I've seen. Regardi0ng NFB, most of the bass-spec schematics and descriptions I've come across (mostly on the Metroamp forum) had NFB sourced from the speaker jacks and spec'd 47K. I could put a parallel 47K on a switched presence pot to halve that to closer to the 27K JTM spec, which I suspect would also achieve some of the cleanness I associate with the bass-spec amps.

I do much different things, but Roe's point on the treble cap is a key, as are filtering options. I like 50u mains/screens and decreasing values toward input for the more relaxed period response.

I don't believe in 33k slope. I start at 47k as a general rule.

Jimmy Page's Marshall sound is a relaxed preamp with early value treble and slope modified PT selection with KT88 and around 180w RMS. HIGH internal voltages. I could believe .047 output coupling.

The soul of it is moderate volume setting with again early brite setup. Think JTM with Bass model output.

I do not use a blocking cap on variable slope. It makes no noise for me. I started building my tonestacks off the board on the controls. Minimizes stray fields between V2 and PI.

I'm just recently paying more attention to filtering, so I'm still learning here. Just to clarify, you're describing 50uF rather than the 100uF referenced in the schematic I posted? I think I understand what you mean by Page's sound stemming from a relaxed preamp -- I'm reading that as also a function of filtering (again, tell me if that's an over-simplification). Would you suggest lowering values to something like 32uF filtering, or something more radical like JTM style filtering with fewer filtering stages overall?

Aside from filtering, which seems to be a major factor on the sound and feel of the amp, the differences I'm seeing in the JTM preamp are 270K mixing resistors (which I'm thinking would probably be pretty subtle), 27K at the head of the NFB feed, changes in the arrangement and values of the presence control. I don't know how I missed the big difference in the cap value and the addition of the 4.7k resistor in the 6550-loaded Bass spec vs presence controls in JTM and EL34-loaded JMP circuits until now. I imagine the additional resistor is subtle (limiting effective sweep of the presence pot from 0-5k to 2.5k-5k), but the 6550-loaded Bass-spec cap being nearly 7x larger at 680nF seems like it would result in a pronounced difference. It seems to me that the larger cap would widen the frequency range impacted by the presence control. Is that accurate? If so, the JTM/Bassman spec 100nF presence cap would seem to restrict action chiefly to high frequencies. Interestingly, the SV20 schematic I have uses the 100nF cap, which I think can sound too bright for a lot of the pot's sweep, but that might not be the case when applied to a Bass-spec amp rather than the Lead spec in the SV20.

So to recap between Reeltarded and Roe's posts, it seems like the big NFB circuit differences to keep an eye on during design and tuning would be:
-Placement of the NFB source tap (16 ohm tap [JTM] vs 8 ohm tap [JMP lead] vs 4 ohm tap [SV20] vs speaker jack [JMP bass])
-27k vs 47k feed resistors
-Value of the presence cap (100nF [JTM, SV20, JMP with factory-installed EL34s] vs 680nF [JMP with factory-installed 6550s])

I may need to experiment with values once the build gets underway, but do y'all have any thoughts in terms of which might be the best starting point? It probably won't make a big difference from a planning perspective, but I'm trying to think through a clean layout (an addition I picked up after a Hiwatt build). Seems to me that, since Page's amp started life with EL34s, it'd make sense to run with the 100nF cap on a full-sweep 5k pot (i.e. no resistor) as a starting point?

I'm glad I consulted the forum on this, especially since you and Roe agree on the high B+. I'll experiment with the 47nF coupling once the build gets underway -- seems like a reasonable way to filter the amount of bass produced in the preamp.

I do not use a blocking cap on variable slope. It makes no noise for me.

I don't think there is an issue there, no DC path to ground.

Thanks for the update, Martin and Reeltarded. It seems like I'll just need to ensure I get good quality caps to avoid DC leakage like in this build (https://ampgarage.com/forum/viewtopic.php?t=18077)from xtian a few years ago. Here's a revised tone stack schematic to blend between 250/500pF treble cap values in addition to variable slope. I'm thinking I'll need to mount these in on concentric pots (again a weird visual thing -- I don't like busy control panels).

Edit: typo, added proposed tone stack schematic

[Roe]

For NFB, use a 220k pot as a variable resistor in series with 27k @16 ohms.

A 50w can never give the sound or feel of a 100w. that's why I got 10 plexis at home. The best compromise, if you really must, is perhaps a -84 slightly lower voltages (490 or so), somthing that is still more than 50w. Alternatively, consider 4x6v6s or 4x5881s running somewhat inefficiently

[Reeltarded]

I think looking at JTM45-50 gets close but that (100w) +3db makes a reactive difference in the guitar hearing the amp along with the other small things.

I have a 50w amp with 50/50 mains and screens, 16/16 PI and V2 with 10u (Solen film) input stages that plays a little squash like a pro... a broken-in baseball glove. Smooth transient. No tricks to get great attack envelope.

I like Heyboer 3.4k 50w OTs as much or more as I did ClassicTone. Marstran 50w replacement is fantastic! (Heyboer)

Merren is always excellent.

I get my sets from Amp Parts Direct. APD tags on generally straight up Dagnall pattern.

[wpaulvogel]

Use a JTM45 350-0-350 power transformer and a 784-139 clone output transformer. This will get you the correct voltage and lamination steel for the feel. You need a squishy power transformer to sag the voltage to mimic the 100 watt feel. The KT88’s are going to make about 65 watts clean rms which is perfect and you’re going to love the feel without the vibration of your eyeballs compared to the 100 watt output. Stick with fixed adjustable bias because output tubes vary wildly and cathode bias will require changing resistors for consistency when you swap tubes. I’ve built several amps with this method and they are great.

[Reeltarded]

I endorse all aspects of Paul's post.

[cdemike]

Coming back to this project finally. I really like the idea of using the JTM45 transformer, especially on the feel basis, Paul. Did you use full wave rectification or a bridge rectifier on those builds? (Does it matter in terms of feel? I keep reading bridge rectifiers are more efficient, so I'm reading that as a stiffer response?)

I drew up a layout so I can run it by the forum, especially after experiencing some lead dress issues with my last build. I agree about the fixed bias piece, so this one has 2x bias pots mounted to the board, and decided to work the adjustable features into push-pull pots to cut down on chassis drilling. I'm thinking I'll try to build this in an aluminum JTM45 chassis with an extra hole for the "thickness" pot drilled between the bright channel's inputs (explanation below). I also planned to use the Mojotone JTM45 PT, since it has the highest current capacity rating of the JTM45 transformers I could find. If I understand correctly, this should give the right B+ at around 480V at the plates, since the tube-rectified JTM45 kit with the same Mojotone transformer has 465V at the plates. I also worked in the lower filtering suggestion from Reeltarded. Even though this won't be as loud as a 100-watter, I still expect it to be very loud, so I drew up a Lar/Mar style master volume as well. The inputs are wired to make the "high" jack feed both V1 grids, whereas the "low" jack will feed only V1A. The "thickness" control is a 1M pot in series with a 100uF cathode bypass cap. If I understood Merlin's book right, that should be the correct value to completely negate the 100uF cap when all the way down. There's also a coupling cap switch on the thickness control, a bright cap switch on the gain control, and, per Roe's suggestion, a NFB switch on the presence pot. If I did the math right, the lower NFB resistor should approximate the amount of NFB an equivalent 100W amp would have.

Hoping to get feedback before getting officially started with this. Thanks again for all the help so far!

[cdemike]

I think looking at JTM45-50 gets close but that (100w) +3db makes a reactive difference in the guitar hearing the amp along with the other small things.

I have a 50w amp with 50/50 mains and screens, 16/16 PI and V2 with 10u (Solen film) input stages that plays a little squash like a pro... a broken-in baseball glove. Smooth transient. No tricks to get great attack envelope.

I like Heyboer 3.4k 50w OTs as much or more as I did ClassicTone. Marstran 50w replacement is fantastic! (Heyboer)

Merren is always excellent.

I get my sets from Amp Parts Direct. APD tags on generally straight up Dagnall pattern.

No idea how I missed this the first time I read it, but I just was looking up Solen FAST capacitors as a potential option for this build since I'll be mounting the filtering for V1 and V2 to the board anyway. Am I reading right that this has JTM45-style filtering with a single shared section between V1 and V2? Did you use a sag resistor on any part of it or was the adjustment down to 10uF enough to compensate for the ESR difference? I did a little digging, by the way, and found the folks at Royal seem to agree regarding lower filtering. They used 64uF on the PI, 32uf on V2, and 32uf on V1 (pictures showing caps here: https://reverb.com/item/6871165-royal-m ... tyle-plexi), and I'm considering a similar filtering setup to yours to get the squish and to potentially avoid recapping down the road.

Paul, you mentioned the squishy transformer -- I'm curious if you have any thoughts or experience in terms of tuning that. It sounds like the JTM45 PT is a good match, so I'm trying to learn about design moving forward. The calculator dorrisant posted predicts 294mA of current demand with a 3.5K primary impedance OT, so with this using about 50% more power than rated for the PT it seems like we're in good shape for some sag. Would a PT rated at, say, 250mA have any appreciable sag if the current demand is only 40mA higher? I know a lot of this will likely come down to the specific PTs in question, but trying to learn what I can. The line of thought started when weighing the EI JTM45 replacement PT vs a toroidal. I was leaning toward this one (https://www.antekinc.com/as-2t350-200va ... ansformer/), which if wired with the HT secondaries in series should make for 350-0-350. Based on the data sheet, it appears each side of the secondary should fall by 28.75V per amp of current draw. Since those figures show the secondaries wired in parallel, I'm estimating 58.5V per amp. Using the 1.414 rectification coefficient, I'd estimate about 24.4V of sag at a full 0.3A DC load. Jason at Headfirst measured 60V of sag in a 1983 2204 (https://www.youtube.com/watch?v=x5feoCREGnY). That said, that was a 50W, and I know 100W are stiffer, so I imagine the voltage drop would be less in the 100W amp. I may be off base, though, so trying to figure out how to match numbers to the way the amp feels.