Experimenting with 6BM8 reverb - questions and demos!

[chrisd]

Hi!

In a bid to keep myself busy/sane during the UK's second lockdown last year, I built my first amp, based on the "Double Six" kit from ampmaker.com (thanks Barry!). However... I really like reverb. It makes everything magically sound good, and having it on the amp means one less thing to set up and stop me practising.

So I ended up removing the mid control, and redoing the layout to cram a transformerless 1-tube 6BM8 reverb driver in - and I got it working last week :)

I realise there are already a bunch of threads on 1-tube reverbs on here (I'm pretty sure I've read them all!), but I thought it might be interesting to add some recordings of how different tweaks to the circuit affect the sound, as I couldn't find anything like that in other threads - instead I basically just chose a design at random, then had to commit to building the whole thing to see how it would turn out.

To this end I've attached the current version of the schematic to this post, and recorded a few chords here:

Note that the "mid" knob is actually bass, and the "bass" control is the reverb mix.

I've got a few issues still though:
- Having the reverb/"bass" turned all the way off doesn't kill the reverb completely... any idea why?
- The reverbed tone isn't great - I get lots of harsh twangy dissonance, and not much sparkly lush goodness... :(
- I suspect the reverb mix stage is losing a decent amount of volume from the amp. I can only just get crunch from the power stage, by I use the bridge pickup + boost switch + volume all the way up. The amp's designed for clean tones, but I don't want to risk losing potential volume.

So I'm on a quest to improve the reverb tone, increase the mix, and drop less volume... easy right? :D

Does anyone have any suggestions? And if there's interest I'm happy to record more demos as I make changes.

Thanks!
Chris

[PRR]

Short R13. What changes?

[tubenit]

Hey, I really appreciate your sharing the schematic and 6BM8 reverb design.  I think it sounds excellent!  Nice guitar playing
also. You've come up with a pretty nice reverb tone, I think.

You can try a small smoothing cap from plate to cathode on the 6BM8 triode.  Maybe 120p?  See if that helps at all?
Other experiment would be a .001 or 680p or 820p going into the reverb instead of the 330p. 

Regarding the drop in volume/gain.  Maybe try 470k for R77 with R16 being 330k?

With respect, Tubenit

[chrisd]

Short R13. What changes?

I will try it tomorrow :)
That's keeping the NFB away from ground... presumably when shorted I would basically end up with no NFB and a fractionally hotter bias on V1b?

I think it sounds excellent! Nice guitar playing also. You've come up with a pretty nice reverb tone, I think.

Thanks!

I haven't tried a smoothing cap, will give that a go.

Originally I had a 470p going to the reverb, which sounded about the same... bigger is interesting - would that increase the drive power as well, or just add more bass? 1/(2pi*R*C) = 1/(2*pi*1M*C) is 482Hz at the moment, increasing to say 1nF would give me 159Hz, assuming that's how the maths is supposed to work.

Regarding the drop in volume/gain.  Maybe try 470k for R77 with R16 being 330k?

Oh yeah - I forgot to say why I made the mix resistors so large! Originally I had 33k and 220k as R77 and R16, similar to dchang0's "'59 5F1 Champ w/ 6BM8 Reverb" schematic.

But that meant that turning the reverb up also seemed to decrease the guitar signal at the same time, around the last quarter of the range. I tried reverb pots from 100k to 1M and the same thing happened on all of them - but increasing the reverb mix resistor seemed to alleviate it.

My guess is that at higher reverb settings, some of the signal could be disappearing back down the reverb coupling capacitor, through the relatively small plate resistor and up to B+... is that a thing that can happen?

[tubeswell]

Going by your schematic:


R17 is in parallel with your reverb level (including the pot and the 260k R77). This makes the grid load for the output tubes Rl = 470k||1260k = 342k with the reverb level dimed, but only 470k||260k = 167k with the reverb level pot cut.

Accordingly, at the dry-wet mixing stage, the voltage output for the dry signal is attenuated by 342k/(1000k+342k) = 0.25 of Vin when the reverb pot is dimed, and 167k/(1000k+167k) 0.14 of Vin when the reverb level pot is cut.

So that's where all your dry signal is going.


You might get some more gain by increasing R11 to 150k or 220k, and increasing C12 to 22uF, and decreasing R12 to 1k or 820R. You may or may not like the end result, so I suggest 1 tweak at a time.


Also, R9 is quite lossy, being in parallel with the 1M volume pot. As C3 provides DC blocking from V1a, then C10 is not necessary, and so you could eliminate both C10 and R9, and get more dry signal gain. (The Vol pot would provide the necessary grid leak for V1b).

[PRR]

> keeping the NFB

Your reverb loop is INSIDE the NFB loop. Simplistically the NFB should "cancel" the reverb. It can't due to time differences. It may just sound bad. Shorting the 47r looks like the quickest way to try this.

[chrisd]

Thanks for the analysis @tubeswell! That's a lot more load than I thought (I just kind of eyeballed it and made some resistors bigger at random...).

And good spot on R9||Volume - that seems like an easy way to get more from the first stage. I wonder if I could move the reverb mix there? I could replace C10 with a mix resistor and run some wires over the board to make it work with the current layout.

Also - I started experimenting, but got as far as the reverb tank before I noticed something... The tank has red and white jacks on it, and on my Blues Junior (the other reverb amp I've worked on), the red cable is send, white is return. That makes sense, right? Red is "hot", which the reverb driver is... so I've been matching up the colours like that.

But no! Turns out every other reverb tank ever has white for the input, red for output. I never noticed because the Accutronics tank on my Blues Junior doesn't have coloured jacks

Swapping the wires doesn't change the level massively, but the tone is a lot better - it's a lot "smoother", there are more high frequencies and less mud, and a bit less of the unpleasant twanging noises.

Re. the reverb mix interacting with the volume - has anyone ever put it _before_ the recovery stage? I.e. wired like a guitar volume pot, or maybe a variable resistor across the 6BM8 triode's grid?

[sluckey]

Re. the reverb mix interacting with the volume - has anyone ever put it _before_ the recovery stage? I.e. wired like a guitar volume pot, or maybe a variable resistor across the 6BM8 triode's grid?

You can put the pot before the recovery stage. But, the signal is very small coming out of the tank. I think it's important to get that low level signal into the grid with the least amount of wire necessary. This gives less opportunity for noise and hum to get into the signal.

You may consider re-wiring your reverb pot so C75 feeds the top of the pot and R77 connects to the wiper. Probably wont make much if any difference, but easy to try.

[chrisd]

This gives less opportunity for noise and hum to get into the signal.

Hmm good point.

I finally got the amp disassembled and set up on a workbench again yesterday.

You may consider re-wiring your reverb pot so C75 feeds the top of the pot and R77 connects to the wiper.

Tried this one first - it did have some effect, I think it let me get more reverb while cutting more "proper" signal. I was losing more signal towards the end of the reverb travel than before though so I've put it back for now.

Either way round I'm still getting quite a significant amount of reverb coming through with the pot all the way down though :( My multimeter claims the pot's OK though - I get continuity from the wiper to ground when it's off.

Short R13. What changes?

Not much unfortunately :( I put it on a switch so I could turn NFB on and off while playing, but there weren't any massive differences - I could only barely hear the change in the main guitar tone, let alone the reverb :(

Next discovery though: I hooked up a DI box between the reverb send and tank and recorded what I was getting out of the 6BM8 pentode. I'm pretty sure I'm overdriving it, which would explain some of the ugly trails on louder notes/chords. Cutting the reverb driver input will lose some reverb level though :(

If I "un-triode-strap" the pentode, will that allow it to handle more input swing without clipping?

[Williamblake]

Yeah, nice sound so far, keep it coming. What reverb tank are you using btw?

[chrisd]

Yeah, nice sound so far, keep it coming. What reverb tank are you using btw?

Thanks, will do!

It's a 8EB2C1B tank, so the second-highest input impedance.

Originally I was using a "Mox" tank I found on eBay, I think that's a rebrand of the generic Aliexpress ones. I've just swapped it with the Accutronics one in my Blues junior (also 8EB2C1B) which seems to emphasize the highs a tiny bit more.

[chrisd]

 No real updates so far. However, my shielded wire has arrived from Aliexpress today so I'll be trying some more things over the next couple of days.

In the meantime, here's what it sounds like with the tank the right way around!

(note: recorded on laptop instead of phone for simplicity, so it can't be directly compared to the previous recording. I'll try and do subsequent demos the same way for comparison).

EDIT: Original video deleted because the audio was clipping. Link updated with one through an e609 and audio interface that sounds much better.

[tubenit]

That first reverb setting on this YouTube sounded absolutely lovely to me!  I would be very happy with that.

PLEASE keep us updated on this!  I'm very interested in seeing what tweaks you ultimately settle on.  I've been interested in
the 6BM8 reverb for some time and have never gotten around to trying it out.

Would love an updated schematic when you get her all done.

Thanks,  Tubenit

[Williamblake]

I second your reverb sounded better in the first clip and think the chchch or tinny boxiness results from too much drive into the reverb or too much bass going into the reverb drive signal. If you dont want tank artifacts.

[chrisd]

Thanks for the feedback, really appreciate the interest!

I reckon ignore that second clip - I've just noticed the clipping about 45 seconds in, it didn't sound like that in person so I think the laptop mic must have clipped.

I was kind of hoping that expensive laptop > cheap phone, but I guess I was wrong :( I'll redo the "tank correctly wired" video before and after this weekend's mods.

First I think I'm gonna try putting the reverb control _before_ the recovery gain stage, with lots of shielded wire. If it's still too noisy I'll put it back, but I think if it does work it'll give me way more flexibility with the mixer stage.

Then I'd like to try taking the reverb drive signal from somewhere with lower signal level to avoid clipping the drive - maybe just after the tone stack.

[pdf64]

I suggest to try lifting the cathode bypass off the reverb driver before rearranging things.

[chrisd]

I suggest to try lifting the cathode bypass off the reverb driver before rearranging things.

Good suggestion, I've just tried it.

It does decrease the reverb level a lot, but as a result of having less driver gain it seems to sound a lot better. Obviously the problem now is that the level's too low :D

I connected up the reverb DI box again, and the signal going to the tank is a lot cleaner with the bypass cap removed. When I put the cap back and record the reverb driver output, it's _really_ crunchy though - this is definitely something to try and improve.

[sluckey]

I would replace R71 with a 1M-A  "Dwell" pot. Then you can dial in any amount of drive that you want.

[kkillebrew]

Hi - I think your reverb sounds pretty darn good already - just like your playing!
 
I use a similar circuit/concept when building reverb into small amps, and I really like the  sound.  As good as traditional Fender style transformer drive? Maybe - maybe not, but it is very good.  On mine I eliminated the ringing you describe with the circuit below.  I am using a 6GH8 - different pinout but very similar tube - it also works with a 6U8 but with slightly less drive.

 A separate screen supply/resistor with cap to ground in the 6bm8 pentode section might give you some improvement.  Biased at 6 -7 ma and using a 12K to 15K plate resistor should match your tube better to the pan input impedance.  I have included a drawing of the latest implementation I used on a 5881 SE amp similar to yours.  You might have to adjust the 33k/15k combination for the 6BM8 but probably not.   I also moved the send and recovery points to the output of the first preamp triode to eliminate the amp volume control changing the reverb mix.  Good luck and have fun!

[chrisd]

Hi - I think your reverb sounds pretty darn good already - just like your playing!

Thanks!

That's a neat schematic... I briefly tried a separate screen supply but at this point I'm suspending things that don't fit the layout above the turret board, so my extra power supply node was a bit unreliable :D

I'll definitely give your schematic a go!

At the moment, I've moved the reverb mix to the preamp section. That's helped the mix a _lot_. I've got a 2-band EQ in that stage too, so it means the recovered reverb signal is competing with the much lower post-EQ signal, so I can use much smaller mix resistors.

Annoyingly now that it's louder I'm getting a 120Hz buzz when I turn the reverb up :( I'm hoping that'll go away when it put it back in the chassis but if not I'm going to be chasing that down. Could also be related to how I've butchered my previously beautiful layout to try stuff out...

Next step is to write up the schematic as I've got it currently and do another quick recording for comparison later.

[tubenit]

do another quick recording for comparison later

Continue to follow this with interest and look forward to your next recording.  Thanks for sharing!

With respect, Tubenit

[chrisd]

Here's the latest!

Made the following changes:
- Instead of putting the reverb around a 1M resistor in the second gain stage, the send is taken straight from the output of V1a, and the recovery is mixed in after the EQ section and a 200k resistor. This makes the reverb mix go _way_ higher, as well as giving the amp a load more volume.
- Volume control is moved to the second stage (between preamp and the EL34) to keep things simple in the first stage.
- Reverb driver pentode bias is now a bit hotter, with a 200r cathode resistor. It's dissipating about 2.7W at idle.

There's a tiny bit of gain at times - since the volume is now in the second stage and I've removed some attenuation from the first stage I think I might be pushing V1b a bit too hard.