Pentode Reverb Driver Part II: subminiaturization of the stand-alone reverb and preamp

[SoulFetish]

This is inspired by the discussion we had here:
Question about Pentode reverb drive circuit
The impetus for that topic was the born out of trying to retrofit my buddy's Silvertone 1474 with a quality spring reverb; using the existing power supply, and trying to maintain the original preamp topology.
That only left me one tube to drive the transmission line, recover the gain, and mix the wet and dry signals. Since that was my first attempt at designing and building any reverb circuit, getting the noise rejection, gain performance, and overall stability proved to be a much bigger challenge than I anticipated.
How did it all turn out?..... pretty good But there are some things I would've done differently in hindsight in order to make it a better fit with the silvertone. The reverb I chose for the project was the long delay 9 series because, go big or go home I thought. However, the only realistic location to mount the tank was in the cabinet below the speakers. Because of the material and way the Silvertones were built (cheaply!), the cabinets can suffer from heavy acoustic vibration during output. That, combined with the long decay of the tank made it very susceptible to acoustic feedback and instability. But, man, there was a lot there to love and I've been thinking about it ever since.
I've been thinking about taking a different approach to designing a stand alone tube reverb/preamp than what I've seen commercially available or in DIY builds. Because of the size of traditional spring tanks, transformers (power and reverb driver), and tubes themselves, these often end up being the size of small head cabs. We don't have to use any of those things.
The design concept is this: Develop a modern, miniature, all tube stand-alone spring reverb unit using Acutronics miniature Blue Reverb; SMPS boost converter with an input of 12V DC for the Heaters and HT power supply; 6943 Subminiature Pentodes for the reverb current drive and virtual earth mixer stage, and 6111 Subminiature twin triodes for the preamp circuitry.

Acutronics Blue Reverb:







One of five 6943s I just picked up: (but I also have a bunch of 6111s)



6943 curves for a screen voltage of 125V:



6111 twin triode curves:



Open Source HV boost converter designed by Jan Rychter:
https://jan.rychter.com/high-voltage...-tube-projects
HV PSU 2.3-2014-05-03.zip
HV PSU 2.3 Schematic.pdf

More to follow. Now, I sleep....

[Old Tele man]

Interesting, but I wonder about lack of EMI shielding with that plastic case since the old (original) models had metal case which served as shield against EMI.

[SoulFetish]

I've been thinking about the power supply requirements, assuming a 6943(x2) and 6111(x2) topology, I will need a supply of 250V able to supply a max current of 50mA. I figure this will be worst case scenario because some of the stages will be inverting, but I want to give myself plenty of headroom.
After thinking about it, I think the supply should be isolated. The boost converter I listed about is not, so it probably isn't a good fit for an instrument amplifier.
It seems that the best way to approach this is using an isolated flyback power supply.
Texas Instruments and Linear Systems both have some good options for me, but I like the simplicity of the LT3748. The chip has the capability to measure the output voltage without the need for an isolated secondary winding or optoisolator, so this simplifies another aspect of the design. I found this circuit example in the data and application notes, which I was going to use to base my supply on. My application will have a higher load current demand, and I was looking for some guidance in finding an appropriate flyback transformer, and to a slightly lesser degree a suitable switching mosfet. I will have to design a board anyway, so I will probably have a few populated for use with other effects projects. Here is the reference circuit:

and a LT3748 datasheet:
LT3748fb.pdf
I know some of you cats cut your teeth designing switching supplies, so I'm all ears for anything you want to recommend.

[nickb]

The snag about switchers is that of speed and custom windings. That means a really good layout.

I think you'll have a lot less pain, at least for a small quantity, if you simply use a power transformer back to front i.e. say 6.3V in 120VAC out (==168VDC). Maybe use a 9V in 240V out to give you 240 x 6.3/9V = 168VDC ( ==235DC).

[SoulFetish]

The snag about switchers is that of speed and custom windings. That means a really good layout.

I think you'll have a lot less pain, at least for a small quantity, if you simply use a power transformer back to front i.e. say 6.3V in 120VAC out (==168VDC). Maybe use a 9V in 240V out to give you 240 x 6.3/9V = 168VDC ( ==235DC).

I have nothing against linear power supplies, and I'm much more familiar with them. I believe Seymour Duncan uses a linear power supply in their mini tube overdrive pedals.
I've not ruled that out, but there are some disadvantages in the size, cost, and low frequency noise. I could probably achieve better efficiency with a switching power supply, but the main attraction is the relatively small physical footprint AND I can series/parallel the 12V in for a theoretically quiet filament supply.
Maybe a linear supply is the way to go for now. I would still like to have a flyback design ultimately, but the process of specifying an appropriate transformer, reducing EMI, and trying to eliminate/reduce all the modes of failure is probably a big enough project on its own.

Okay, I just did a little research, using a 9VAC input voltage is a good place to start. After rectification, I can still use a series/parallel combination for the heaters, and supply a board mount toroid to step up back up the AC voltage. Talema manufactures a 70051K transformer which can be had at a relatively inexpensive price.
I may be able to use the size down, but I want to test the circuit under full load first, and then see what kind of regulation the transformers actually have.

https://www.digikey.com/product-deta...026-ND/3881341
The external supply needs to supply minimum 525-600mA for the DC heaters plus the primary side load current.
What do you think?

[SoulFetish]

Double check design of Mu-Follower input stage

I'm thinking about this as the basic circuit architecture (see schematic).

Many reverb units are unity gain devices, and I remember there was some affinity for this type of reverb circuit in one of the older threads. The Fender 6G15 and Premier 90 stand alone reverb units use cathode follower input stages, for instance. Perhaps, the idea was to maintain as much clean headroom as possible, and were concerned with overdriving an amplifier's input. Or they wanted to buffer the input for the benefit of a low impedance dry signal.
The issue I saw in those input stages was the potential for increasing noise. So I decided to use a Mu Follower input stage to provide some voltage gain, amplify the signal above the noise floor, drive the pentode with the necessary grid voltage, and provide a nice low output impedance.

I'm planning on using a cascode reverb recovery for max gain. This is fortuitous because I have an extra triode, and with this setup, I get proper phase alignment of the wet and dry signals (although, some have suggested that this is not so important with the nature of reverb signals).
Lastly, the final stage is a pentode virtual earth mixer.

Before I go on, I just wanted to get some feedback on the Mu Follower input to double check that I didn't make any mistakes or over look anything.
Thanks.



I have to work out the design details of the last few stages, so forgive the incomplete component values.

[SoulFetish]

-
I'm nearly finished with the circuit design. All the stages are done except the mixer:



I wasn't trying to turn this into parts soup, or make it unnecessarily complicated. One of the main objectives was to design the circuit to make as efficient use of the power supply (9VAC/3.33A wallwart) as I could think of. Since this supplies the heater circuit as well as the HT, the the biggest challenge was/is not overloading the supply. The problem is that these particular tubes have high transconductance, and draw about as much current as you're going to find in preamp tubes. By stacking the 6111 tubes into compound stages, I'm able to cut their current demand in half. Plus, it give's me a chance to do some experimenting on circuits I've never used before, and see how they sound.
The last bit, I would like some help with. I'm not sure how to calculate and design the feedback network for a proper virtual earth in the final pentode stage.

edit: oh, the red dots are so I could keep track of phase

[nickb]

You'll need to add a DC blocking cap in series with the feedback resistor on the mixer stage. Since you want heavy local feedback you might as well make the stage have as much gain as possible by bypassing the cathode. Make the feedback resistor Rf high enough so as not to load that stage, say 1 meg. Then roughly, the gain will be Rf/Rin where Rin is resistance in the input path. The problem is that is you want low gains the input resistors will have to be big. You might want to forget the VE , lose the cathode bypass cap and use a passive attenuating mixer.

I expect the final three stages will have a pretty good HF response as they are either pentodes or cascode ( sort of the equivalent of a pentode) which minimize the plate to grid one feedback capacitance. Therefore you might need to be liberal with grid stoppers, especially on the reverb recovery and input. The last two resistors especially will add noise so it's a balancing act.

[tubeswell]

-
I'm nearly finished with the circuit design. All the stages are done except the mixer:



I wasn't trying to turn this into parts soup, or make it unnecessarily complicated. One of the main objectives was to design the circuit to make as efficient use of the power supply (9VAC/3.33A wallwart) as I could think of. Since this supplies the heater circuit as well as the HT, the the biggest challenge was/is not overloading the supply. The problem is that these particular tubes have high transconductance, and draw about as much current as you're going to find in preamp tubes. By stacking the 6111 tubes into compound stages, I'm able to cut their current demand in half. Plus, it give's me a chance to do some experimenting on circuits I've never used before, and see how they sound.
The last bit, I would like some help with. I'm not sure how to calculate and design the feedback network for a proper virtual earth in the final pentode stage.

edit: oh, the red dots are so I could keep track of phase

That pan input impedance might work better driven straight from a transformerless SRPP driver (using the 6111s - which are comparable to a 12AU7) - seeing as how you expressed concern about power supply'n'all. YMMV

(or another idea - use all the 6111s on the driver side as in: 2 cascaded inverting stages driving a SRPP, and use both 6943s on the recovery side - with one being the VE mixing stage)

[SoulFetish]

Tubes' and nickb, thanks for your feedback.

tubeswell, I actually think the pentode is a better driver for the reverb. I gotta run in a few minutes to go do sound for a show, but I can go into it a little more later on.

nickb, I thought about your suggestions and comments about the cost of resistor noise. Perhaps you and tubeswell are right, and as good a mixer as a VE might be, this isn't the best implementation. I was thinking about some ideas and came up with this:



gotta' go!

[SoulFetish]

would like to finish this project at some point

I have all the parts, minus enclosure, and I started putting together some physical layout ideas. So far I'll be using some kind of aluminum enclosure with around the following dimensions: 6.5"L x 4"W x 4.5"H.
My biggest concern is making sure to keep the sensitive input and reverb recovery stages away from the power entry and rectifier modules. I want to get some feedback on a couple of layout ideas for mounting the reverb tank, and toroid/power supply PCB.
All the components are to scale, and I've attached a schematic of the power supply. I've run several simulations in order to keep everything as efficient as possible, particularly for the 12V heater supply. This is my second biggest concern, actually. I want to make sure that the layout is as compact as possible, while allowing proper heat dissipation and prevent overheating on the PCB. Anyway, here are the files:





Any suggestions or preference?

[SoulFetish]

Oh, one other thing.. these are the components I'm using.
With a toroidal transformer like this one I'm using, where are the points of flux leakage I need to worry about?

[Helmholtz]

Generally toroidals have very low leakage, because of the way they are wound and because there is no airgap.

A little leakage flux emerges from the wire turns and rapidly drops with distance.from core.