New build fine tuning

[cdemike]

Just fired up the plexi-style build that the generous members here helped me design (https://ampgarage.com/forum/viewtopic.php?f=6&t=36339). Still needs tweaking, but I'm overall very excited that most things seemed to work as-planned. I ended up making a 1968-9 lead spec 50W with some changes to help with versatility. Namely, it has a switch for 33k or 66k slope resistors, bright cap (100pf or 4700pf), and 0.68uf or 2.38uf cathode bypass resistor at V2. I almost never jumpered channels on any of the 4-holers I've had, so V1 is wired permanently in parallel with similar values as a typical lead-spec bright channel (56k plate resistor). One of the switches on the input jacks was wired to take a 1.5uf cathode bypass cap in and out depending on which input you select (sort of like having a "fat" channel and a "bright" channel). Getting rid of one of the channels let me put an Orange-style FAC knob so I can pick from 47nf, 24nf, 16nf, 9nf, 4.7nf or 2.2nf coupling capacitors at V1. It also has a 3-way switch for screens and reservoir cap values (16uf or 32uf and 50uf or 100uf, respectively). The last signal path changes are adjustable negative feedback using a 250KL pot in series with a 33k NFB resistor and a 470nf presence cap (typo when ordering parts led me to be short one 100nf cap, so it's what I had on-hand. Still going to give a try for kicks). I also used a Fender-style bias circuit since those tiny trimmers make me nervous (plus the chassis mount ones look cleaner to my eye), and the B+ rail has a 10K dropping resistor in place of the usual 20K (or series 10K's) to get preamp B+ higher per the advice in the other thread. Overall very happy with the outcome.

Played it some this morning and one of the KT88s red plated (thankfully caught it early, and it seems to be working fine and still biases closely to its counterpart). The bias circuit took a lot of fine-tuning to get the right values for a usable range compatible with KT88s, but all the tweaking led me to forget to solder one of the legs of the feed resistor. Thankfully I crimped it pretty well to the post so I had good bias while testing, but I'll need to actually solder that . There's not a lot of headroom, and while I was playing it after resetting the bias it made a crackling sound after I started testing the filter cap switch. It's now really low on gain (clean and quiet all the way up, whereas it had more gain than I wanted on 4/10 before the incident). So I'll need to track down what happened there and fix that. Thankfully no burning smell or burnt-looking components, so that's a start. Then I need to figure out how to get more headroom. Target number one is V1's plate resistor -- going to try 47k rather than 56k to see if that gets me where I'd like to be. If that doesn't do it, I currently have a 1k cathode resistor on the V2 gain stage feeding the cathode follower; I might try 1.5k to after that. If I need to keep going from there, I'll need to think through next steps, since I'm not sure how best to get more headroom.

The "fat" channel is very dark and quiet, so I suspect I'll need to revisit how I wired my input jacks to switch that cathode bypass cap in and out (might end up using maxkracht's "cloacal" input jack trick using a stereo jack. I suspect some weirdness is happening with signal getting bled to ground there. I am slightly frustrated that my B+ seems lower than I anticipated with 440V at the plates. I used a Mojotone-spec JTM45 power transformer, which I originally anticipated would put me closer to 490V, but I'm going to give this one a shot before I consider swapping the PT. The big OT in the picture is a 1202-132 style unit with adjusted secondary ratios to let me present a 3.4k load (basically made the 8 ohm setting 16 ohms, the 4 ohm setting 8 ohms, etc.). Phil at Heyboer was great at helping me get this transformer built.

[xtian]

Your lead dress is nice, Mike! But many of your solder connections look dodgy. Make sure you apply enough heat to the turret FIRST to get solder flowing (wicking) to it, then apply heat to the component lead(s) also, and fill those turrets with solder. Same for the pots, and other points of connection.

[cdemike]

Your lead dress is nice, Mike! But many of your solder connections look dodgy. Make sure you apply enough heat to the turret FIRST to get solder flowing (wicking) to it, then apply heat to the component lead(s) also, and fill those turrets with solder. Same for the pots, and other points of connection.

Thanks! And I appreciate the tip -- this was my first time working with split turrets (thought it'd be easier trying to thread my wires through the perfboard). I had a heck of a time finding what's an appropriate amount of solder for them -- glad I learned something today! I'll go back and fill those turrets up when I get back working on it. Here's hoping all that crackling was just a cold joint!

[Fischerman]

Cool amp! I've built/modded a few Marshalls and a Super Lead is my favorite amp. I agree with above about the solder joints. But be careful not to overheat the wires or the insulation will melt and look like crap. Put the tip of your iron on the turret (or whatever the most massive piece is) not the wire and let the turret heat the wire last if that makes sense.

I think building these 'All access' Marshalls with switches for everything is great to hear the difference all these tweeks can make. There is no substitute for hearing it for yourself. But having done all that, my last and favorite build was a 'mostly' stock '69-spec Super Lead with some EVH tweeks and that's it. I was going for the EVH 12-series thing and a one-trick pony. Push-pull Bright (4700pF) on Bright Volume and push-pull Boost on Normal Volume so it looks stock. The boost is just a 330uF fat cap on V2 820R cathode resistor. In other words, once you get it all up and running perfectly I think you'll find that many of the switches will have one favorite position and you won't need that switch anymore. But you'd never know for sure without hearing it first.

[Fischerman]

Also, I just noticed that it looks like you have the power tube grid stoppers mounted on the board. Is that correct? If so then I don't think they actually serve the purpose intended. They have to be mounted on the socket in order to work as intended. They should help prevent oscillation. Many Marshalls have as high as 5k6 there. I think I used 4k7 in mine because I was out of 5k6.

EDIT: Here’s one way to do the grid stoppers (red and brown wires to pin 5). These resistors have pretty fat leads so it’s quite rigid.

[cdemike]

Thanks so much! I'd love to hear some clips of that '69 build. What iron did you end up using? I think you're right about the "all access" build piece. I almost always run my amps to sound pretty similar, but you're exactly right that I wanted to hear what the different iterations of the circuit sound like. Really fun learning experience. Ironically, I think it's unlikely I'll end up using the push/pulls (bright switch, v2 cathode bypass) or the NFB knob very much after a brief use. I mainly play either rock or country and I do record from time to time, so having the mid cut and variable tightness does seem like it'll be useful.

I'm learning a bunch with my goofs on this build -- I had no idea the grid stoppers needed to be mounted at the socket. They are indeed mounted on the board -- good catch. Thanks! How does the different placement impact their function?

[Fischerman]

Thanks so much! I'd love to hear some clips of that '69 build. What iron did you end up using? I think you're right about the "all access" build piece. I almost always run my amps to sound pretty similar, but you're exactly right that I wanted to hear what the different iterations of the circuit sound like. Really fun learning experience. Ironically, I think it's unlikely I'll end up using the push/pulls (bright switch, v2 cathode bypass) or the NFB knob very much after a brief use. I mainly play either rock or country and I do record from time to time, so having the mid cut and variable tightness does seem like it'll be useful.

I'm learning a bunch with my goofs on this build -- I had no idea the grid stoppers needed to be mounted at the socket. They are indeed mounted on the board -- good catch. Thanks! How does the different placement impact their function?

It's a beast. Its the best Marshall or clone I've ever played. I've built others but this one is just special. I used Mercury iron with laydown PT. The B+ is just over 500vdc when plugged into 120vac but I always run it with a variac, I built it with that intention in mind. The OT is their fatter stack version of their typical plexi clone OT. I had to clock the OT as shown because it was too close to the power tubes the other (tenchnically 'correct') way.



IME, if you run a .68uF presence cap like many 70's Marshalls had stock then that's essentially a NFB control. I like that sometimes too but slightly prefer the 100nF cap.

[Fischerman]

Oh yeah, forgot to post this: https://www.aikenamps.com/index.php/gri ... -they-used

[cdemike]

Wanted to post an update after some fine tuning:

Life got busy, so I didn't have a chance to do much with it, but I have been playing it a fair bit. I could be fooling myself, or maybe it's not the factor responsible, but it does seem that there is some break-in period for the Sozos. The amp is sounding much more open and clearer, and I'm also feeling an improvement in headroom -- still totally possible there are a few factors at play (especially since I used all new preamp tubes). I did finally swap the 470nF presence cap for a 100nF and am liking the amp much better. I did that yesterday and it made a noticeable difference in headroom, as well as a major difference in the amp's overall sound, which I didn't expect. I also have found I prefer a 12BZ7 in the phase inverter position since it appears to give more headroom and touch responsiveness.

So I feel like I learned a few things, but wanted to get feedback to see if I'm taking away the right things:
-The presence circuit is more than a simple high-pass filter shunt to ground. When I built the amp, I assumed I'd be able to get the amp to sound the same as if I just had a had a smaller presence cap by setting the presence control lower. I found that the overall character of the amp is very different with the smaller presence cap, even with presence set to zero: more headroom, more balanced midrange response (almost sounded like Brian May's "Deacy" amp in some guitar settings). So the value of the cap makes a difference comparing the circuits with adjusted presence settings which would have compensated between two simple high-pass filters.

-In some applications the transconductance of a tube can stand-in for input headroom. In settings where I'd need to turn my guitar down to 2 or 3 to get a clean sound using a 12AX7 in the PI, I can get similarly clean sounds with my guitar's volume at 5 using the 12BZ7. I've had the opportunity to swap my 12AX7s between the 3 positions to see if it's just the tube, and there hasn't been a big difference with the 12AX7s shuffled around. I was surprised, since I expected the lower plate resistance combined with equal gain factor to produce more overdrive, but I'm probably still not driving the KT88s hard enough to produce significant overdrive downstream of the PI anyway. My original thought was that spreading the drive between stages would improve touch responsiveness and therefore allow me to get better touch responsiveness setting the volume lower, but I got what I wanted mostly by ostensibly reducing the amount of PI drive.

[martin manning]

Re the red plating incident, at what plate and screen voltages are you running your KT88's, and at what is the OT primary impedance?

[cdemike]

Re the red plating incident, at what plate and screen voltages are you running your KT88's, and at what is the OT primary impedance?

Plates are currently sitting at 435V and screens at 432V. Primary impedance is 3.4K. What are you thinking?

That reminded me about the still-ongoing B+ piece. I had originally hoped for about 480-490V at the output section plates, and I'm not exactly sure why my voltages are so low. I'm not finding there isn't any indication of a leak to ground anywhere along the B+ rail between there not being any weird voltage drops at any of the nodes, and that I'm getting expected voltage drop across the B+ dropping resistors, which I'm interpreting as getting the right current flow across them. The amp doesn't have any balancing resistors so the only paths to ground should be leakage through the filter capacitors and through the tubes themselves. But I also am not really certain how I'd end up with a weak PT since I figure it'd have to involve a significant manufacturing error in terms of winding ratios, which I consider pretty unlikely; this is not an exotic or experimental PT design and I trust Heyboer's quality control, so I'm thinking as time goes on there must be something amiss in terms of how the amp is presently configured.

I had considered switching it to a full bridge rectifier, but that'd involve a pretty extensive overhaul of the bias section.

[martin manning]

How much voltage drop at the OT CT do you see loaded vs. unloaded? How much idle current? At 435V Va I'd guess about 70 mA per tube at the cathodes. A parasitic oscillation could cause excessive voltage drop and red plating.

In general KT88 are designed to be operated with the screen voltage at one half to two thirds of the plate voltage. If the plate and screen voltages are equal, and low enough, you can get away with it. At 435V Va I would choose ~2k5 for the load. At 490V Va, I would increase the load to 3k6 and drop the screens to 325V.

[cdemike]

Thanks, Martin. The OT CT sees 449V with no tubes in the amp. I currently have the KT88s biased pretty cold at 50 and 51mA as a precaution following the red plating. That's measured using 1R resistors on the cathodes, and the specific tubes in question are EH KT88s I bought used, so it's possible I just have a tired set. No red plating since that initial incident, though, so I'm not really sure what would have caused that situation, sparing that bad joint in the first post, though I'd think that both tubes would have red plated if there were an intermittent connection on the bias supply. Maybe I just caught it early enough before both tubes got that hot?

Regarding the low B+, I remember when I first started the amp up and started testing, I tested the AC voltage across each of the poles of the standby switch and found 660V rather than the 700V I expected (PT datasheet states 350-0-350V at 200mV: https://www.mojotone.com/products/70203 ... sizew=1200). My outlet voltage was a strong 122V when I measured that, so it's probably not weak mains voltage.

[martin manning]

It doesn't seem like that PT was ever capable of 350-0-350 at 200 mA.

There's not a lot of headroom, and while I was playing it after resetting the bias it made a crackling sound after I started testing the filter cap switch. It's now really low on gain (clean and quiet all the way up, whereas it had more gain than I wanted on 4/10 before the incident).

Do you still have that problem?

[cdemike]

It doesn't seem like that PT was ever capable of 350-0-350 at 200 mA.

There's not a lot of headroom, and while I was playing it after resetting the bias it made a crackling sound after I started testing the filter cap switch. It's now really low on gain (clean and quiet all the way up, whereas it had more gain than I wanted on 4/10 before the incident).

Do you still have that problem?

Unfortunately, after rethinking the measurements at the standby switch, I think you're right about the PT. Good to know moving forward, and hopefully someone else can avoid a similar issue in the future after reading this.

Regarding the low volume issue, I reflowed all of the joints on the board per xtian's advice during that initial round of revisions, and the issue went away. So it seems like it may have been a cold joint somewhere. Glad I posted about it before I brought the amp out for a gig, jam, rehearsal, etc. I also wanted to see if the noise issue would be fixed if I installed shields on the preamp tubes, which did fix that issue. Specifically a shield on V1 made all the noise go away, and now it's quiet as a mouse. I live in apartment, and our living room is where we both keep the music stuff and where the Wifi router lives, so I suspect the router was the main noise culprit.

The headroom issue does remain, though. It produces more drive than most non-master Marshalls I've tried or owned. My concern is less that there's more overall drive available and more that I wish it would be more responsive to guitar volume and picking dynamic changes. Following the advice from the other thread, I set the B+ pretty high, and am currently getting about 380V at the PI node and 200V at the plate of V1. I measured the others while checking loaded vs unloaded B+ the other day but didn't write them down. If they're useful I'm happy to pass them on, but I do remember getting 185V at the cathode of the cathode follower, and I know 12AX7s are rated for a 200V maximum between the heaters and cathode. So while there probably is still some wiggle room in terms of preamp voltage, I'm thinking that's as close to maximum as I'd want to get if I ever gigged the amp and was unsure of wall voltages. I did buy some spare resistors to further lower resistance along the B+ rail in case I wanted to go down that road, but I'm hoping to get input about best ways to proceed in case someone else has a better solution.

I suspect that the low headroom may have something to do with the fact that the second gain stage is permanently bypassed. Given that I'm not huge on the 1.5uf cathode bypass cap in parallel with the always-on 0.68uf there, I may rework that to take the 0.68uf in and out of the circuit. I'd probably also change the cathode resistor to 1.5k to get more headroom from the stage, but I'm not completely convinced that's the source of the problem, given the big improvement in touch responsiveness after putting the 12BZ7 in the phase inverter. So I suspect I may be working the phase inverter and potentially the cathode follower too hard.

Filtering may also be a place I could focus. There is unsurprisingly a major improvement with the screens filters set higher, so I am also considering swapping the current 16/16uf can there for a 32/32uf can. My main hesitation is that I really like how saggy and vocal the amp is with 16uf on the screens. Currently this is how the B+ rail is set up: OT CT (switchable 50/100uf) > Choke > Screens (switchable 16/32uf) > 10k > PI (32uf) > 10k > V2 (16uf) > 10k > V1 (16uf)
So I also considered changing the preamp to 32uf on V1 and V2 64uf, but I'm not sure which node might be most effect -- maybe the PI? My main hesitation is that I really like the amp overall now except for the slight deficit in touch responsiveness, so I don't want to change too much around. I'm pretty embarrassed to admit that I'm pretty proud of how neat the amp's guts look (admittedly, a pretty distant second in priorities to how it sounds), so I don't want to end up with catawampus component leads.