18 Watt Plexi Build

[xtian]

The heat sinks are probably unnecessary. Presumably you have the resistors firmly mounted to the chassis with thermal paste, right? Then the box itself is the heat sink.

[dbharris]

Thanks Phil and Xtian. I really do appreciate the feedback. I'm new to this amp stuff but have a half dozen or so amps I'd like to build. The glue is not under the heat sinks, just around the perimeter base. The thermal paste is underneath. It runs much cooler now than before I added the thermal paste and heatsinks, but certainly won't win any beauty pageants. When I get up to a 100 watter I'll reconsider whether to rebuild this or not.

I got my light bulb limiter built and tested with a working amp. About to fire up the plexi....

Thanks!
-Dan

[Paul G.]

I've built 2 of the Trinity Plexis. One was my gigging amplifer for a number of years. The second was the spare ready to go. I don't gig 3x/week anymore, so I sold the spare, but I'll never get rid of the original. It is a wonderful amplifier.

These are photos of the spare, that I built into a standard 18W (18") chassis with Trinity guts and Classictone transformers.

[dbharris]

That looks great Paul!

I went slow with the start up procedure and brought up voltage slowly on the variac with a light bulb current limiter and dummy load. No problems. Removed light bulb limiter, all good. Installed tubes, no problem.

Hooked it up to the speaker cab and plugged in a guitar, all good. Everything works as it should. Sounds good and is pretty loud.

I'm going to burn it in this week on the dummy load and afterwards roll some tubes. I'll prob play it for 2 or 3 weeks before installing the VRM. Right now I have the SS rectifier in, Mullard reissue long plates in v1 and v3, a Mullard reissue shortplate in v2, and JJ 6V6s in v4 and v5.

Voltages:
Mains 121.4
B+ 386
B+1 370
B+2 281.4
B+3 242.3
B+4 227.5

V1
1- 172.8
3- 1.408
6- 131.9
8- 0.775

V2
1- 141.8
3- 0.982
6- 240.8
8- 142.9

V3
1- 181
2- 19.34
3- 31.17
6- 184.8
7- 18.75
8- 30.95

V4
3- 376
4- 370
8- 25.13

V5
3- 373.3
4- 376 (maybe 370, can't read my handwriting)
8- 25.13

-Dan

[dbharris]

I wanted to provide a quick update and ask for some thoughts/suggestions. I burned in the amp on a dummy load for about 25 hours total (over a few days) by playing music from an iPad into the input with the channels jumpered.

When I initially tested the amp, I was playing through just the normal or just the plexi channel, not both. When I play either channel there are no issues, I can dime the gain and volume and it reacts like I would expect. I have never owned an amp before with channels that you can jumper, so I am not sure what I should be expecting. Overall the amp does seem quite bright (even with a Les Paul and very bright with a strat). It is manageable when playing either channel. When jumpered, it seems much more bright and I can only run the channels with moderate gain/vol otherwise it runs off into feedback like crazy. At levels where I can control the feedback notes will sustain indefinitely, but there is very nasty distortion happening that is very present/resonant on lower notes.

I am plugging straight into the amp with a 20ft cable, no pedals. Maybe I need to reverse the leads of the OT primary? I don't otherwise have any "classic" symptoms of the leads being reversed like squealing, but I understand it can manifest more subtle than that. Is this a lead dress issue? I can chopstick around if needed.

I tried to at least eliminate the speakers being an issue. I tightened all the screws on the back of my cab, did not really help. I removed the middle piece to make it semi-open back. Helped a little, but I think that is because the bass is less dominant. Loosened the screws mounting the speaker a 1/4 turn each, no change. I also tested through an egnater tweaker 1x12 cab with a broken in G12H30. The highs were much more in line with what I expected, also a lot less bass than I would have imagined (it is a vented cab). But the nasty distortion is still present, though not as prominent.

I swapped tubes around too in the preamp and do not think that is the issue. I don't have an extra set of power tubes, but can order if that is advisable to eliminate it as a possibility.

Thanks in advance!

-Dan

[Paul G.]

There is a forum for Trinity, and a subforum for the 18W Plexi. You can sign up (https://www.trinityamps.com/phpbb/viewforum.php?f=7) and post questions there. There are some good people on there, including Stephen, so...

[dbharris]

Hi Paul, thanks for the response. I do have a thread going over at Trinity but there doesn't seem to be much activity over there. I may poke around and try some things to see if anything changes the issue.

-Dan

[Phil_S]

You might try the 18watt forum. I haven't been active there for years, so I don't know what's happening. My login still works. I see a few threads there. Activity level looks rather low though. https://www.18watt.com

[dbharris]

Thanks Phil, I will make a post over at 18 Watt too.

To report my progress since my last post, I have moved wires around with a chopstick quite a bit with no significant changes. Sometimes by moving wires I could induce more background hum, etc. But it did not impact the amp having nasty distortion riding under bass notes or it wanting to feedback with channels jumped.

I swapped OT leads for the primary. When I did this the amp did not squeal/howl, etc. It did seem like it wanted to feedback easier and more brittle/high pitch sounding when channels are jumped. Honestly it was very hard to perceive any difference. So, I swapped them back as they were.

Just realized I had not posted the schematic or layout here, they are attached.

I took another set of voltage readings, a couple notes...I have swapped preamp tubes a bunch so this is a totally different lineup than when I took the initial voltages on fire up. Also, when I first took voltages, I was on a dummy load. This time I was on the 1x12 G12H30 cab. When I took the voltage for pin 3 of V4, it induced squealing and the voltage was moving quickly around in the range of 393-399VDC. No behavior like that on pin 3 of V5. I did not take voltages with OT leads swapped.

I am going to visit my parents for the weekend. When I get back, I'll make a short recording of the sound. I am thinking it may be crossover or blocking distortion and I may need to try the Paul Ruby zener diode clamps.

V1
1 - 170.6
3 - 1.478
6 - 129.1
8 - 0.799

V2
1 - 143.3
3 - 0.973
6 - 241.3
8 - 144.1

V3
1 - 173.1
2 - 19.17
3 - 32.14
6 - 181.6
7 - 18.94
8 - 32.14

V4
3 - 393-399 (squeals when tube pin touched by DMM probe)
6 - 366
8 - 24.42

V5
3 - 369
6 - 366
8 - 24.42

Thanks!

-Dan

[dbharris]

Alright yall, I'm back and hoping this post may help shed some light on where to look to fix this issue. After much digging on several forums it seemed like I was going to need a scope to properly diagnose the problem, so I took the dive. Here are some screenshots from the scope. It appears to me, I am dealing with crossover distortion. But, not sure what to do about it.

I ran a 1K sine wave into the input of the amp, from an app on my tablet. I'm commenting the amp settings in the caption below each picture. Scope was on on the output jack. I'm attaching pictures at clean settings and then the setting where crossover distortion is most apparent. Past that it goes very square wave.

On the Plexi channel and with them jumped, I can control/effect that hump in the trough with the treble control.

After this I did probe a few different parts of the signal, and I believe this problem first shows up pins 3 of the power tubes.

Let me know if there is anywhere else you would like to see scoped out or anything different, etc. I have a lot more screenshots at volume settings 2, 4, 6, 8, and 10 for a few configurations.

Thanks for any insight!

-Dan

[dbharris]

Normal channel scoped out:

[pompeiisneaks]

Those notched waveforms scream crossover distortion, meaning your bias is set too low.

A good article on this with a photo of that type of distortion can be seen here:

https://sound-au.com/valves/analysis.html

Scroll down to the image just before '10 - Power Supply'

also to read up more on blocking distortion read here:

https://www.aikenamps.com/index.php/wha ... distortion

the easiest fix for that is to increase the grid stoppers on the power tubes, plexi's often didn't have them, but more common standard builds have between 1.5 and up to 5.6k grid stoppers to help block this.

The schematic you seem to have from trinity has 5.6k there so that should handle/manage any blocking distortion.

Other than that I'm not 100% sure what else could be going on with the other issues. have you calculated the bias on the tubes to see that it seems reasonable?

Looking at your data, you show a 300 ohm resistor per the schematic. You show 370VDC on the plates 25.13 before the cathode resistor, and from that calculation for 6V6's I show you at 105% dissipation. This is going to be pushing the tubes a bit hard.

That is at least one area you may want to improve to avoid the tubes being eaten up fast.

~Phil

[dbharris]

Hi Phil,

Thanks for pointing me to those resources. I'll give them a read in the next few days. Somebody on the 18Watt forum suggested trying 1K screen resistors, so I may give that a go.

I had used Rob Robinette's tube bias calculator and it seemed like each tub was dissipating 14W. I know the JJ 6V6s can handle more abuse, but maybe this is pushing this particular pair too far. I will look at running the tubes cooler.

Measured voltages again on the power tubes and came up with:

Mains 120.7
B+ 388

V4
3 - 375 (the voltage still moves around a bit when I have a DMM probe on this pin but now just between 375-378 and slower than before)
4 - 372
8 - 24.66

V5
3 - 374
4 - 371
8 - 24.66

-Dan

[dbharris]

I spent the weekend substituting parts, inspecting with the scope, listening to the amp, rinse and repeat. I also touched up the solder joints, but there was no change, so I think they were fine before. I took voltages after every step, no unexpected readings. I tried a different set of power tubes, no change.

I changed the cathode resistor for the power tubes to 330 Ohms. This had very little to no impact on the cross-over distortion. It is biasing to about 95% now. I think I could go back to 300 Ohms and it would be fine.

I added a smallish CDE Mica 1KV cap between the power tube plates. This had a noticeable effect on crossover distortion and taming the brightness somewhat. It did not minimize crossover distortion, but delayed the onset a bit. Still not a huge change. This was more apparent in the normal channel. I started with 220pf, then went up to 330pf. There was not much difference between the two, so I kept the 330pf in. I wonder if trying a smaller cap may be advisable?

Next, I swapped the screen resistors to Vishay metal film 3 watt at 1K. This created about a 5 volt delta between the plates and screens. I could not perceive much difference in sound or feel after this change, but kept it.

Next, I swapped the cathode bypass cap for the power tubes to a Vishay 1000uf 63V cap. The amp already felt quite stiff, but this seemed to tighten the low end response. I also have a 1500uf that I may swap in.

I then proceeded to chopstick the amp again while hooked to a dummy load and 1K sine wav monitoring through the scope. I was not able to perceive any differences in lead dress around the tube sockets, except with getting the grid wires too close to the speaker jacks causing extra hum. When I touched the shielded wire between the plexi gain pot and V2, I could sometimes change the character. I resoldered that joint and it sounds the same as before. I think eventually that connection would've gone bad and it was not a strong enough mechanical connection. I may run a new section of shielded cable here anyways for peace of mind.

In summary, the brightness is more in check but still requires non-typical tonestack settings to dial in a good sound. The plexi channel still runs off into uncontrollable feedback at settings on the Gain and Volume of about 3/4 and up. Crossover distortion is still present. But I wonder if this is perhaps typical of the power amp design anyways?

I'd like to be "done" with the amp next weekend. I plan to install the VRM and I might try dumping some signal after the gain pot. I am a little hesitant to do that b/c the gain control is warmer sounding than the volume control, so I am typically running that up higher. But I would like the controls to be usable over the full range.

I was referred to two methods of signal attenuation? After the wiper of the gain pot insert a 470K with treble peaker cap-470K to ground vs. just a100K to ground? For some reason, I ordered 390K resistors from Mouser instead of 470K. I know that will be the same 50% signal dump, but with a lower impedance I think? The 100K to ground dumps more signal to ground as the gain pot is run higher, right? So, at full on it would only pass about 10% of signal. Or am I mixed up here?

Open to any other suggestions.

Thanks
-Dan

[dbharris]

I implemented the Ruby mod, which nearly removed all of the crossover distortion. Because the highest cathode voltage I had recorded was about 25.5 VDC, I went with a 3 diode array: 20V and 6.5V zener in series (both 5 Watt) and back to back with a 1n4007. That was connected in parallel with each 220K grid leak resistor with the 1n4007 side going to ground. Some pics of both channels on my scope. This was the most prevalent I could make the crossover distortion with a SS rectifier. It is even less apparent with the 5AR4. Sounds much better now.

NRubyMod.jpg

PRubyMod.jpg

I then installed the VRM and tested it. All good. But I did not have time to rewire the input jacks and V1 socket. So, I'll be tackling that next weekend. And then tidying things up in the amp with zip ties and thread lock, etc. I plan to play the amp for a week at least after that before doing any other mods.

I am still considering reworking the amp to incorporate GNFB similar to what a real plexi would have. That requires several changes: channel mixing resistors and moving both channels to 1 input on the phase inverter, changing the tail resistor, adding the NFB resistor and feedback tail resistor, and phase inverter plates. Or if I don't do that, I may try to make the gain range more usable on the plexi channel. Maybe I'll be happy with it and leave it as is. Who knows....

I'll try to grab a short audio clip next week.

-Dan