Modifications on Hoffman Plexi 50, sanity check

[ac427v]

What an interesting journey. I think you are almost there. You started building a Marshall Plexi but many mods later you still do not like the sound. Been there...

Your ideal seems to be clean sound overall until at least half volume-especially clear bass, no mud allowed!

Sounds like you want a stock Fender Tweed Bassman:  Less preamp gain, less midrange, and cleaner feedback circuit. You can keep the EL34 tubes if you don't have a pair of 5881 tubes. You may prefer that sound anyway. The proper components should fit right on your current turret board. Way faster than doing Plexi mods one at a time

[stratomaster]

The 5F6A is voiced darker than his amp currently is, where he describes the normal channel as still too muddy.

It also uses a 12AY7 in V1, which is a huge reduction in voltage gain.  He described the 100k/22k split plate as too much reduction, so the AY would definitely be too much for him.

He could potentially adopt the NFB/presence circuit to a bit of success, but on all other counts I think a mod to Tweed Bassman specs is a few steps back from his goal.

Uki, I have a feeling you incorrectly installed the master volume after the treble. This is the solution, despite your hesitancy. Try this instead. Put a 220k resistor between the wire at the treble pot wiper and the lug.  Then put a 1M to ground at the input side of the 0.022μF cap at the PI. This will emulate a prePI master volume set back a notch or so with no holes drilled.  Alternatively you can fit a trim pot to the treble pot and just dial in the amount of signal attenuation you want.

[ac427v]

I stick by the Tweed Bassman circuit because it has great headroom and is not too bassy if the Bass control is properly used.

My other reason is that OP started with a vague schematic and has since made many changes but only documented them verbally. It is not clear which changes were later removed. Using a known circuit  such as the Hoffman 5F6A would allow the forum to communicate clearly what the current circuit is and which mods are needed

I whole-heartedly agree that a pre-pi master is needed regardless of his decision on what basic schematic to use. Both amps have plenty of gain available, even with a 12AY7. The master volume is needed to get some hair without blowing out windows.

[uki]

where he describes the normal channel as still too muddy

Indeed it is too dark, i though about adding a bright cap like in the bright channel but with a value that doesnt go as bright as the bright channel, would that work? what value?

He described the 100k/22k split plate as too much reduction, so the AY would definitely be too much for him.

I did set it to 18k, sounds good, punchy and clean until volume 6  (i dont have a AY)

He could potentially adopt the NFB/presence circuit to a bit of success

Reducing the resistor there will give more clean tone and headroom yes? 22k ?

Uki, I have a feeling you incorrectly installed the master volume after the treble.

I might have      here how i did it, output of trible to one side of the MV, wiper to cap before PI, the other lug to ground. It did increase and decrease volume.

Try this instead. Put a 220k resistor between the wire at the treble pot wiper and the lug.  Then put a 1M to ground at the input side of the 0.022μF cap at the PI. This will emulate a pre PI master volume set back a notch or so with no holes drilled.  Alternatively you can fit a trim pot to the treble pot and just dial in the amount of signal attenuation you want.

I'll try it as well.

I started yesterday making a schematic with all the changes, so we can keep track on whats going on.

[stratomaster]

where he describes the normal channel as still too muddy

Indeed it is too dark, i though about adding a bright cap like in the bright channel but with a value that doesnt go as bright as the bright channel, would that work? what value?

Because of channel interaction a bright cap across the mixer for one channel becomes a "dull" cap for the other channel.  This is part of why the normal channel is lacking top end. If you're going to experiment with a bright cap on the normal channel put it on the pot like on the 5F6A.

He described the 100k/22k split plate as too much reduction, so the AY would definitely be too much for him.

I did set it to 18k, sounds good, punchy and clean until volume 6  (i dont have a AY)

He could potentially adopt the NFB/presence circuit to a bit of success

Reducing the resistor there will give more clean tone and headroom yes? 22k ?

Yes, at the expense of some dynamic response and "feel". Try it, you might like it.

Uki, I have a feeling you incorrectly installed the master volume after the treble.

I might have      here how i did it, output of trible to one side of the MV, wiper to cap before PI, the other lug to ground. It did increase and decrease volume.

That sounds right.  I'm surprised you didn't notice the significant decrease of breakup by just turning it down slightly.

Try this instead. Put a 220k resistor between the wire at the treble pot wiper and the lug.  Then put a 1M to ground at the input side of the 0.022μF cap at the PI. This will emulate a pre PI master volume set back a notch or so with no holes drilled.  Alternatively you can fit a trim pot to the treble pot and just dial in the amount of signal attenuation you want.

I'll try it as well.

I started yesterday making a schematic with all the changes, so we can keep track on whats going on.

I hadn't realized you put C4b back in.  That will definitely increase the breakup.  If you want to keep the effect on the tone of that cap but reduce the boost, you can try adding a low value (2.2k or so) resistor in series with C4b.  Or you can reduce the 2.2M resistor from the grid to 1M.  Since you reduced mixers, this is needed anyway to get a similar effect as before the change.

[uki]

I hadn't realized you put C4b back in.

Ugh i forgot about that one, didnt remove it actually, ill check that out

Or you can reduce the 2.2M resistor from the grid to 1M.  Since you reduced mixers,
this is needed anyway to get a similar effect as before the change.

I was wondering about it, if could reduce it further, ill try it too 

Question about the PI , the 2 smaller resistors have smaller values than some Fender amps,
here we have 470 and 10k, some fender amps have 820 and 22k, is that related to the early breakup ?

 

[stratomaster]

Question about the PI , the 2 smaller resistors have smaller values than some Fender amps,
here we have 470 and 10k, some fender amps have 820 and 22k, is that related to the early breakup ?

Here's a good overview of the long tail pair.
https://www.valvewizard.co.uk/acltp.html

Those resistors set the bias and balance of the PI.  The 470 does bias hotter, but the designs you're comparing to employ a 12AT7 instead of the 12AX7 of Bassman/Plexi--so it's not apples to apples.

You can try a 12AT7 in that spot for experimentation. It won't hurt anything.

I'd recommend reading up on that page (and a deeper diver in the Aiken amp white papers) before you go changing values in the PI. Or if you insist on messing with it to pick a tried and true design like the BF or later SF-- but mind the voltages and tube types employed.

I think you're better off limiting gain into the PI than changing its design, though.

[uki]

The 2.2m resistor is now 1m, not much change, tone is clean ish, still grit, grainy, amp is somewhat clean until 4-5 depending on guitar pickups, the cleanest set on guitar still too much grit on amp, bridge humbucker is pure rock'n'roll !!    too much 

The C4b was out of the circuit already, I forgot to remove from the schematic, now updated in the post #53

NFB resistor down to 22k , not much change with it either. Try 10k ?

I'm thinking if there is some not so good component in the circuit, there are a number of super old resistors and caps.

I'm still thinking about to move the tone stack between V2 triodes to see if it will take out that grit.

[stratomaster]

I can assure you there are drastic changes in gain and tone with both the reduction of the 2.2M to 1M and the reduction of the NFB to 22k. 

Did you try removing the bright cap from the lead channel yet?  That will both reduce upper frequencies in the lead channel and add high end to the normal channel.

Did you try bypassing the cathode follower?  This will increase the tonestack related insertion losses and reduce gain into the PI.

[uki]

I can assure you there are drastic changes in gain and tone with both the reduction of the 2.2M to 1M and the reduction of the NFB to 22k. 

I did notice a good gain(drive) drop overall
When both volumes go up amp start going wild 

Did you try removing the bright cap from the lead channel yet?  That will both reduce upper frequencies in the lead channel and add high end to the normal channel.

Gonna try that next, bright is too sharp and i notice now normal can get better tone increasing treble control. Mid and Presence makes the amp go wild, specially together
 

Did you try bypassing the cathode follower?  This will increase the tonestack related insertion losses and reduce gain into the PI.

I'm not sure how to do it       jump from V2 grid to the input of tone stack ?

[stratomaster]

When you say "going wild" do you mean overdrive or is there a squeal/oscillations?

The pre phase inverter master will enable you to turn up the treble and turn down the gain into the PI. Again--it's the answer to get what you want. You should temporarily install one and play it for a week or two to get familiar with how it interacts with gain and EQ.

See the attachment for a quick way to bypass the cathode follower.

[uki]

When you say "going wild" do you mean overdrive or is there a squeal/oscillations?

Overdrive 

I did check again the NFB resistor, it was a bit too hight, one of the really old Rs, replaced for a 22k right on spot,
and tone did improve good amount

I'm getting some bad noise on the speaker, i was checking it and looks fine, checked for loose parts all looks good solid,
bad power tube maybe?     Gonna watch the PA tubes while playing

See the attachment for a quick way to bypass the cathode follower.

Is this jump bypassing only the sencond half of V2 ?

[stratomaster]

Is this jump bypassing only the sencond half of V2 ?

Yes. That's the cathode follower.

[uki]

Have done the cathode jump, no improvement, i totally disconnected the 2nd half of the tube.

the amp did improve a lot since we started thou and it is way louder !

I think one of the power tubes are going bad, when playing soft amp does ok, but if I play hard amp does some farting sound.
When tweaking bias some years back, went too far without knowing it and one of the tubes did red plate,
i didn't notice right the way, and that one tube had some sparks inside... so probably that...(i could be wrong thou)
maybe the screen voltage is too high for EH tubes ?

But now, I'm not sure which one is it thou 

I was thinking the spring coils holding the tubes were the cause of noise, but after putting it aside amp sound the same,
I got another pair of EL34 but totally unmatched. that is why the questions about the bias balance/adjust earlier.
(bough those two cos was very cheap, but not as pair)

[stratomaster]

You again should have had a significant drop in gain and an increase in clean headroom.  I think too much time is passing been comparison points. Perhaps record them to get a more accurate comparison. 

Plus it's hard to judge if the amp isn't healthy.  You need to get the power tube situation figured out before proceeding.

[uki]

You again should have had a significant drop in gain and an increase in clean headroom.

Yes indeed !

I think too much time is passing been comparison points. Perhaps record them to get a more accurate comparison. 

You got it !

As you can see the amp got good improvement yet still really agressive 

Plus it's hard to judge if the amp isn't healthy.  You need to get the power tube situation figured out before proceeding.

The fart sound is solved, the 22k NFB resistor was too low, i went back to the 30k ish one.

Is the screen voltage too high? Data sheet says 425v max, it is over the chart !



Side note: I use a preamp inside my guitar, it boost the signal, it isn't engaged in those videos, amp goes way more wild with it on.

[uki]

Back on it

I've check the amps on the other two PA tubes(Mullards) that i have, they are very unbalanced, 37mV and 28mV.
only now crossed my mind to see if it follow the tubes when swapping them, I think it probably will,
the tubes that were in the amp before are EH.
(I can't afford at the moment to buy another pair, tubes are way too expensive right now)

Then to use those two tubes amp need separated bias adjust per PA tube, or a balance/adjust.

Questions:
Which one would be better for setting the bias: adjust/adjust or balance/adjust ?

The voltage(478v) after the choke is the same as before the choke, is it normal ?

[stratomaster]

Questions:
Which one would be better for setting the bias: adjust/adjust or balance/adjust ?

The voltage(478v) after the choke is the same as before the choke, is it normal ?

Either one will get the job done. There are several examples of each on this forum.

Yes. The purpose of the choke is to create an LC filter to minimize noise and simultaneously minimize voltage drop vs plate.  If you had a benchtop meter you'd see the drop in the 2nd to 3rd decimal places.

[Willabe]

...

[uki]

I've made a small board for the bias circuit, balance/adjust.

I did follow the values like in the picture attached, or should i follow the values posted by Willabe?

[stratomaster]

They'll both work. Pick one. Adjust as needed if needed.

[uki]

The BIAS is in place, the range changed, now goes between -10v to -39v

Does it need more negative voltage ?

[stratomaster]

I personally don't like seeing anything closer to 0 than -20v unless the situation calls for it.

Please post a schematic of what you built.

[uki]

I personally don't like seeing anything closer to 0 than -20v unless the situation calls for it.

That is what i was thinking, before it was something around -20ish to -55ish


Here is the whole bias section.

[AlNewman]

Have you tried a 12AT7 in the PI position?

[stratomaster]

Here is the whole bias section.

I'd replace that 220k with a 100k and see where that gets you.

[uki]

Have you tried a 12AT7 in the PI position?

I don't have one 

I'd replace that 220k with a 100k and see where that gets you.

I'll try that !

[uki]

@stratomaster   The 100k resistor did the trick!!   

Now bias range is -20v to -70v

I did set the bias voltage to -46v and 27mV on each PA tube cathode , just to check if the circuit is working

what is the ideal setting for EL34 ? Tubes now are Mullard

Tomorrow ill be able to do a sound test

Schematic updated in the post #53

[stratomaster]

Ideal setting doesn't exist. There's a range. And that's not exclusive to EL34s.

For longevity I'd bias as cool as possible, probably 50-55% dissipation, before the crossover notch becomes too obtrusive. Hook to a scope to verify. If no scope, bias to 50% and listen/play for a while. If it sounds good, then leave it.

[uki]

probably 50-55% dissipation, before the crossover notch becomes too obtrusive.

    I'm not sure what you mean. I think i can figure out the watts output amp is generating, if that is what you mean.

Hook to a scope to verify. If no scope, bias to 50% and listen/play for a while. If it sounds good, then leave it.

I got a lil scope very simple but only can handle like 30v tops, never really learned how to use it thou,
except to check electrolytic caps 

[stratomaster]

https://robrobinette.com/How_to_Bias_a_Tube_Amp.htm

Read this.

[uki]

I got some time today to play for about 30mins, the amp bias still as posted above, it does sound much, much better with the Mullards! 
Now it is possible dial some clean tones, the amp does go on overdrive at about 5-6, depending on the controls setting, earlier.
I can't tell difference on volume after 6, feels like it stay the same, just adding drive.

I got another amp that is able to stay clean at volume 7(gain/master controls), it have 3 preamp tubes 12ax7s and  4 el84s,
someone on some forum said it is sort of a Peavey classic 50 or something like that.
Not sure if possible to achieve clean tone on that much volume with the Plexi.

Now something that is bugging me, the screen voltage, 470ish way above specification, which is max 425v,
would that shorten the tube life? I wonder if that is the cause of the aggressiveness of the amp as well.
good thread here

I don't think i have mentioned before, but the OT have UL taps, and it does have less voltage than the outer taps,
could that be one way to drop the screens voltage? Diagram of the PT attached.

[stratomaster]

Now it is possible dial some clean tones, the amp does go on overdrive at about 5-6, depending on the controls setting, earlier.
I can't tell difference on volume after 6, feels like it stay the same, just adding drive.

This is typical tube amp behavior and not exclusive to the plexi

If you have a scope you can check for output power at clipping onset. If it's in the 45w ballpark then there's no additional clean volume to be had.

[uki]

If you have a scope you can check for output power at clipping onset. If it's in the 45w ballpark then there's no additional clean volume to be had.

I do but not sure how to do it

I did this: cathode resistor mA times plate voltage. 478 * .027 = 12.9  , almost 13 watts per tube, is that right?
I didn't check the plate voltage after changing the bias circuit thou.

What about the screen voltage?

[stratomaster]

I do but not sure how to do it

I did this: cathode resistor mA times plate voltage. 478 * .027 = 12.9  , almost 13 watts per tube, is that right?
I didn't check the plate voltage after changing the bias circuit thou.

What about the screen voltage?

How to measure power:

You will need to recheck your plate voltage as it changes with the bias. But you're near enough to 50% that I wouldn't worry too much.

Reducing the screen voltage will reduce output power, not increase it.  It's not uncommon for tube guitar amps to run the tubes outside of their published limits.  If you want to have a heart attack look at how the AC30 treats the EL84s.

I think your amp is behaving normally unless there's an issue with the power output.

[ac427v]

The stock Marshall Plexi and Tweed bassman amps are capable of playing VERY LOUD AND CLEAN if the controls are set properly. Your amp is broken if it will not do that.
Turn the Bass control to 1. Is the mud gone? Does it play loud and clean? No? Then it needs repair, not tweaking. I'm sorry
Repair will be very challenging because you do not have a proven working schematic and proven good parts. You will need to use an oscilloscope to find the circuit locations that are adding distortion.
If that is not possible then I recommend rebuilding with a proven schematic--NO MODS YET! and new parts. This may include the transformers if you cannot prove that they are suitable for this project. OUCH!
Good luck

[uki]

Reducing the screen voltage will reduce output power, not increase it.  It's not uncommon for tube guitar amps to run the tubes outside of their published limits.  If you want to have a heart attack look at how the AC30 treats the EL84s.

Ok so that is normal on guitar amps.

I think your amp is behaving normally

It seen like so

I did move the bias a bit sounds even better now
tube base is getting hot, can't touch too long, no red plate, all seen normal,
played for about 20mins sounds really good!

here is the voltages
V4 plate to cathode 460 -  bias -45 - current .032
V5 plate to cathode 457 -  bias -42 - current .032



also i did the measurement like in the video, not sure if i did it right thou, i got a video 
in the video open the normal and bright volumes one at a time.

@stratomaster thanks so much for carrying me through this!





 

[stratomaster]

You have to turn it up until the peaks of the signal flattens slightly, just like that the video does.

Your traces show a bit of oscillation distortion, too.  The amp is still unhealthy.

If you are turning the volume all the way up on the video and you aren't seeing the signal flatten, turn up the input signal. I tend to use 100-120mV. As it stands right now you're demonstrating less than 1 watt output power.

If you're running into an 8 ohm load you should have about 19V at the dummy load.

[uki]

You have to turn it up until the peaks of the signal flattens slightly, just like that the video does.

Oh i thoutgh that was the limit, it was about half way of the volume, since drive start just after 5 i though it was the point,
now that you mention, the wave gets all crazy after that point, not flat but crazier than appear in the video.

Your traces show a bit of oscillation distortion, too.  The amp is still unhealthy.

oh dear 

If you're running into an 8 ohm load you should have about 19V at the dummy load.

Yes 8ohm

Ok here is the whole story of the amp: https://el34world.com/Forum/index.php?topic=23039.msg246988#msg246988
That may help

If you are turning the volume all the way up on the video and you aren't seeing the signal flatten, turn up the input signal.

I'll do it again and report back.

[uki]

The output reading got a bit different this time,
I've connect an 8ohms/45w dummy load(it does read 11ohms with the cable to the amp),

that crazy noise in the wave did not happen with the dummy load.
i did notice a tiny sound even w/o speaker, from PA tubes maybe?

Notice the painting of one PA tube V5 did decay more than V4, is it due to the unbalance and setting ?
I did feel the paint smell while playing, the decay seen rather quick although did not play that long, should I lower current in the tubes?

The wave start flatting at about 8v, round at about 7.5v with a small variation between clean and normal channels,
9.7v at max volume, on both channels isolated, and together it drops to about 8.9v

Here some pics:

[stratomaster]

If you're getting flattening at 8v, then you're only getting 8 watts out of the amp.

You've got a good bit of work to do before this amp is ready.  It should be deafeningly loud in the room.

Annotate the schematic with DC voltages especially the plate and screens of the EL34s.

[uki]

Done, I've notice a little drop on all voltages, probably cos the time of the day, more usage of electricity.

Schematic updated on post #53 just as before,  link here

470R is on the hot side for a 12AX7 LTP. You want more like 820R to 1k. If the LTP bias is too hot, if can muck up the LTP duty cycle, resulting in red-plating on one side of the output stage under large signal conditions

this was a post on another thread about this amp, only now I'm starting to understand what he said, yes one side runs hotter than the other.

[stratomaster]

The easy test if one side is significantly hotter than the other is to measure the voltage across the cathode resistor under load. You can even measure across the cathodes. The goal is for there to be a 0 measurement. If it's positive then the side you have the red lead on is running hotter. If it's negative then the other side is running hotter. This measurement will also tell you how much higher.

I think you have a different problem, though. But it's easy enough to check.

[uki]

You can even measure across the cathodes. The goal is for there to be a 0 measurement. If it's positive then the side you have the red lead on is running hotter. If it's negative then the other side is running hotter.

I did the reading across the cathodes, red lead on V4, with 1khz signal, it start at zero, and goes negative as the signal is injected about -2.3mV

[stratomaster]

That's nothing. A few mA difference is within matching parameters. You can tweak your balance pot if you really want it to be "matched" under load, but I think your problem is elsewhere.

What is your plate and screen voltage under load?

[uki]

Slowly turn bias from -53V toward -30V. Watch the milliVolts on the 1 Ohm resistors. For now, stop at 20mV (=20mA), which is on the cold side, but *safe*, and should play OK with slight hoarseness in note decays.

35mA may be a good final goal, but I would play it a long long time at lesser current to shake all the bugs out. Those vintage parts may resent being re-awaken after a decade of sleep.

I did follow his advice and turned down the bias

Here is the plate and screen voltages under load
end bias.

V4 plate 390v    screen 357     bias -43  .022mV
V5 plate 390v    screen 363     bias -47  .021mV

I did notice some variation while reading the screens, like going up and down then stable, i did use 2 meters at once, one on each tube.

[stratomaster]

You changed the bias so now we can't compare directly idle to load.

Please try not to add more variables during troubleshooting.

Leave the bias as is and remeasure the amp DC voltages, and please let me know what the load was that you used for the measurements you provided on this post. Thanks.

[uki]

You changed the bias so now we can't compare directly idle to load.

Please try not to add more variables during troubleshooting.

Sorry about that

i put the bias back to previous values, so now we are near to previous reading, all readings redone as follow

New readings for DC ,  load 1khz

volumes set to 50% on all readings

bias:
V4   -47  .029mV
V5   -44  .029mV

no load:
V4  plate 460v  screen 452v
V5  plate 460v  screen 454v

with load:
V4   plate 451v  -  screen with one volume either, 440v;    both volumes 405v
V5   plate 451v  -  screen with one volume either, 442v;    both volumes 407v

[stratomaster]

That screen is sagging more than I'd expected, but not enough for your output to be 20% of what it should be. I was hoping for an obvious voltage difference.

Without your amp on my bench I'm at about the limit of what I can do.

The only other suggestion I have is to signal trace with the scope at each grid and plate (except for the cathode follower, do grid and cathode). My only apprehension is I'm not sure that little scope can handle the DC voltages.

[uki]

The only other suggestion I have is to signal trace with the scope at each grid and plate (except for the cathode follower, do grid and cathode). My only apprehension is I'm not sure that little scope can handle the DC voltages.

Here is the scope manual, the probe I have is x1/x10 (gonna order a x100), maybe i can make a signal tracer got an schematic.

The parts of this amp, transformers, and some components, were salvaged from an old hi-fi amp, here, it came without power tubes and rectifier, maybe the OT is under rated, and it is the cause the amp isn't as loud as it should be? Follow a photo of the amp circuit where the parts came from
 
I'll check again all parts values, and connections, maybe i missed something.

I appreciate very much the time and effort        the amp is much better than ever!