Hot rodding a Hiwatt

[cdemike]

I've been playing around with a Hiwatt DR504 clone which I modded to have a SAP preamp with an early 70s style phase inverter, i.e., with the cathode follower carrying the signal rather than just providing bias for the phase inverter (links to schematics below). I originally made the amp aiming for Jimmy Page's Royal Albert Hall sound following the schematic and layout from this build: https://www.marshallforum.com/threads/% ... ld.111713/

That build basically uses the SAP preamp but with the B+ rail shown in Mark Huss's interpretation of the Jimmy Page preamp (schematics linked at the bottom of the post). Shakti's build uses a choke like the original, but I originally opted to use the 100R resistor instead since I wasn't as confident in my metal working abilities at the time I built it and wanted to minimize the amount of work needed on the chassis. He originally built it with the early-style phase inverter like I have in it now, but he prefers the later style version. Based on that, I originally built it with the later style phase inverter, but switched it and found that I strongly prefer the earlier version.

The problem I'm facing is this: I like the amp overall but I'm not really bonding with it the way I do with Marshall-style amps. It's totally possible that I'm just barking up the wrong tree in terms of amp styles, but when I first built it I really loved how percussive and tight it is for rhythm. It's pretty easy to see how this style of amp fit Pete Townshend's playing, but the amp doesn't feel right for single note leads -- stiff and "pokey," for lack of a better word.

Since building it I made a few changes:
-I didn't really care for the boost switch, so it's hard-wired with a 100k resistor to ground in lieu of the balance control.
-I placed a 47nf cathode bypass cap in parallel with the first gain stage's 100uf and placed the 100uf's ground in series with a 25k pot wired as a varistor to work as a kind of "fat" control. That helped clear up the bass some, but it's missing that chirpy, almost "rubbery" quality that can be heard on the RAH recording (reference clip below) when the 100uf is out of the circuit.
-I've also departed from the SAP schematic by using the normally-unused V2 in parallel with the other half of V2 and correspondingly reduced the plate resistor to 100k and cathode resistor to 1k. That was mostly out of distaste for an unused triode, but I didn't find it made a lot of difference.
-Lastly, I changed the usual 100R smoothing resistor right after the standby switch a 500R for more sag. This made a big difference in how the amp feels, and I'm much happier with it, though I'm still not completely happy. Between the change in the phase inverter and the saggy smoothing resistor, single note leads have more depth and a vocal quality that I like.

I wanted to run some alternate approaches by the forum and solicit any other ideas for mods on the amp. Probably the most conservative approach would be to machine the chassis for a choke to see if that makes the different power supply bridges the gap in terms getting closer to the "original sound" (acknowledging the version of the audio available to the public has been heavily produced and probably isn't a realistic target for an amp builder).

Alternately, I was considering branching out to try more original design ideas. These were the ideas I was kicking around:
1. Paralleling V1 and running that into the normally-unused triode as a DC-coupled cathode follower to get a little overdrive from the first stage. That would then run into the volume control and keep the remainder of the circuit true to the SAP preamp.
2. Replacing V1 with a 6U8, using the triode section as the cathode follower. I'm dubious that the lower plate resistance of the triode section would produce the compression from the cathode follower I'm hoping for, so it's possible that idea 1 would be better in this regard.
3. Replacing V2 with a 6U8, also using the triode section as a cathode follower. This would avoid some issues with potential microphony in the 6U8, make an amp with no unused triodes, and maybe make for a cool, unique-sounding amp.
4. Replacing V2 with an EF86. Same potential advantages as #3, but I'd have concerns over the tone stack loading the EF86 down. Potential plus is that EF86s are still in production so potentially cheaper should 6U8 supplies dry out.
5. Replacing V3 with a 6U8, again using the triode as the cathode follower. Same potential plusses as #3, though I'm not sure how the triode will respond to being used as the AC cathode follower here, so it might require some tweaks (I'm guessing nothing too major as long as it keeps ~80V on the PI grids).

It's probably clear that I think a pentode would sound cool in terms of adding "bark" to the amp, but I'd really like to get feedback if anyone has tried anything similar or just has other ideas in terms of cool approaches to modifying these amps -- definitely open to other approaches and suggestions.

SAP preamp schematic: https://hiwatt.org/Schematics/DR_Pre2InputSAP.gif
Schematic with present phase inverter schematic: https://www.mojotone.com/Amp_Kits/Briti ... 50_SCH.pdf
Jimmy Page preamp schematic: https://hiwatt.org/Schematics/DR_Pre2InputJP.gif
Led Zeppelin, Moby Dick at RAH: https://www.youtube.com/watch?v=r9-42mu ... hE&index=5

[Stevem]

With the Hiwatt having about 50 more volts on the plates of the output tubes then the Marshall in my opinion it's going to be hard to get it to not sound as stiff.

[cdemike]

Thanks, Steve. My particular amp has about 470V on the plates after I put in the sag resistor. I was thinking with an output section that stiff, it might make more sense to approach it more like a Soldano and let the clean output section be itself.

I realized today that I could also ditch the DCCF in V1, switch the input stage to line up with a conventional normal channel, install an EF86 in V2, and use the unused triode in V1 as a DCCF after the EF86. That'd avoid loading down the pentode and give more preamp compression. As of now, I'm leaning toward option 5 since placing the pentode latest seems like it'd be best to get some pentode compression. That leaves figuring out how to configure the cathode follower, but thankfully there's a very similar circuit in the chapter on cathode followers in Merlin's book. Seems to me that the main difference I'd need to account for in having a triode with a lower Ra, though with the CF being fixed bias, maybe it wouldn't be an issue?

Anyway, it seems the big fork in the road between focusing on V2 and V3 is whether to get the compression from a cathode follower in V2 or from overdriving a pentode in V3. Again, I could be going down a weird road, so I'm all ears if there's another way of approaching it.

[cdemike]

Coming back to this project. The more I think about it, I think my design goals are to make an amp that is rewarding for both lead single-note playing and for rhythm. Since I'm primarily a "leave the amp at one setting and adjust with the guitar's volume control" kind of player, it seems like the key to threading both of those needles is maximizing touch responsiveness. I also really enjoy an amp that feels fast but without feeling stiff.

I'm leaning towards the pentode being the second gain stage rather than the third out of concern that the high sensitivity of the stage might kill touch responsiveness being late in the chain and out of logistical concerns in terms of getting the cathode follower position right. I'm also straying away from a compactron to get the pentode stage since I'm mostly reading lukewarm reviews of 6U8A and other triode/pentodes from those who have tried them as gain stages. The last update in terms of broad design ideas is that I've gotten pretty interested in the Dumble SSS style DCCF-driven output section. I've borrowed the design more or less verbatim from the SSS #2 schematic posted here with a tweak for adjustable bias. It seems to me that DC-coupling the output section helps achieve the fast-responsiveness that I'm looking for without necessarily compromising on the compression I like for single-note playing. To that end, I did adjust the reservoir filter to further add power supply sag, and I've compensated the NFB resistor to 7K from 10K to make it equivalent to a 100W amp's NFB level (read: more touch responsiveness).

I've posted my schematic to see if this makes any sense. I have a few points of concern:
-I'm not sure it makes sense to have V1 and V3 on the same B+ node since they're in-phase now that I added a gain stage as compared to the original Hiwatt preamp. Although a grand total of 320K resistance between them seems like it might be enough to prevent oscillation, I wanted to see if I'm wading into bad-general-practice territory. The reason I selected a 12DW7 for V1 was since it seems that the 12AU7 section would provide enough headroom that the third gain stage shouldn't add much overdrive, but I also wanted the pentode and the post-tone stack gain stage to be in phase, ruling out a cathode follower driven tone stack. The phase concern stems from wanting to maximize even-order harmonics and it seems to me that the pentode and the triode gain stage following the tone stack are the two most likely to clip in the preamp section.
-It seems like the B+ rail should work as-intended, since it's mostly stock as compared to the stock schematic, but the main difference is the added node (labeled B+3). I inserted a node to run the power section's cathode follower at a lower B+ to put less stress on the tube, and it seems like it shouldn't draw a ton of current, but nevertheless the series 220K resistor feeding the node does give me some pause.
-The value of the 3.3K bias feed resistor doesn't reflect anything specific other than its use in the Dumble schematic. It seems likely that I'll need to adjust values on a built version, but since the bias is so important I wanted to get an extra pair of eyes on this section before any expensive "learning moments" occur (hoping to avoid having to buy new transformers and EL34s).
-I didn't include any grid stoppers on the output section for maximum current capability there. The SSS uses 1.5K, so if that's bad practice to omit them, it seems like 1.5k might be a good starting value.

There is one minor error on the schematic that I just caught -- I plan on keeping the 12AT7 as the phase inverter for the added headroom.