PFM Model Train Spring Reverb Unit

[mth5044]

I found a PFM Sound Reverb Unit all boxed up among my grandfathers belongings. He was, as well as my father, really into model trains - having built multiple huge platforms in all different scales. I've always been fascinated by how much time and effort he put into making his layout look so real. It was really an incredible work of art thinking back on it now.

Anyway, I found this box labeled Reverb Unit, so I had to check it out. It is part of a Pacific Fast Mail (PFM) system from Japan for train sounds. From what I understand about this system (I never saw my grandpa use it, it apparently was old technology even decades ago) a series of train sounds were recorded on different tape loops (mellotron?) and sent through the rails of the layout. The trains were fixed with speakers and some circuitry that picked up the sounds through the wheels (and must have blocked out the power driving the train) and played them from the train itself. While I'm not sure if the system ever worked on my grandfathers layout, he must have discarded the main console but kept the reverb. It appears to have come in a kit as he has a little envelope of the remaining solder and some screws to affix the reverb tank to somewhere.

Anyway, I took it home and traced the circuit and took some pics. Doesn't come with any instructions, but I thought it would be interesting to run a guitar through it.. you know, normal behavior. From what I understand about reverb circuits is that they are primarily made up of a driver stage, recovery stage and an impedance matcher. In the schematic below, I believe each of these stages are set up. Two 2SB187's make up the recovery and drive stages and a transformer (ST-32) which is 1.2k:8 if google is correct.

I traced the circuit, it is clearly not set up in normal schematic fashion, which is why I come to you all with a few questions. But first, some initial observations. I think that the Red RCA from the transformer would be sent to the reverb input and the black from the output. That leads me to believe that the grey wire is an input, black is the output and blue is the power? I'm pretty confused about how the circuit is set up and I think that is mostly due to how the schematic is drawn. What kind of power would be needed to drive this thing? B5k pot is a gain control? I'm guessing to not sent too much signal to the pan.

Any ideas are welcome, thanks!

[analogguru]

red wire: +9V - 12V (= gnd)
black wire: - 9V  - 12V
grey screened wire: input (you will need a [pre-]amp, guitar signal is not enough)
blue screened wire: output

R2 +  R3 mix the original signal with the reverb signal

analogguru

[mth5044]

Thanks Analogguru! I'm a bit confused though as there is no red wire, unless you mean the short clipped one on the transformer. The red wire in the picture goes to potentiometer. The only unconnected ones are blue and gray covered ones and the black one.

[analogguru]

You are right... I mixed it up, simply because the potentiometer is not visible on the pictures.
Anyway, the screen of the red screened cable would be a place to connect  V+.

[PRR]

It's only part of a full spring reverb.

You need a BIG drive signal. Like 7 Volts in 500 ohms. A small power amplifier, but not at our popular 4-8 ohm impedance.

The dry signal comes out about half that level, at about 2K impedance.

I am not sure how this fit into the model-railroad system. The 7V 500 ohm might be the feed to the locomotives; however the 2K output is so much weaker I would think they'd need a second power amplifier to feed the locos.

Personally I think this board is hardly worth fooling with. Except it IS Germanium (ooo!), and it was Grandpaw's.

While it takes large voltage at ~~500 ohms, the actual spring-tank is 8 ohms. (It is an Accutronics or clone, you can look-up the part number.)

An LM386 will drive the tank without that transformer. Put a volume (Drive) control in front of the LM386. You can use the 5K pot if you like. In any case it will need a buffer, and maybe gain, to get from guitar to pot/LM386.

The tank output goes to the two-transistor amplifier. (I re-drew it for functional clarity.) The output of this amp, and the dry signal, are mixed together with two resistors; the 3K+1K network is fine. This output may be a bit too hot to go into a guitar amp, try 1K to 330 ohms to ground on the mix output.



NOTE: This board wants to be positive-ground, an LM386 really wants to be negative ground. The 2-transistor amp only pulls 3mA, so a voltage converter is a real possibility; personally I'd use a 9V battery for this board until it worked. A 9V batt will run it for hundreds of hours. (The LM386 can eat your Spot-One or whatever POSitive 9V supply you use for general pedalling.)

[mth5044]

Thanks for the explanation and redraw! Makes sense. I'll take your advice and keep the board as my grandpa had made it. Germanium never seem to turn me on anyway.

I'll figure out an LM386 circuit for it. Thanks both of you for your help.!

[mth5044]

This is a slightly modified variation of Rod Elliot's Project #34 Spring Reverb circuit. http://sound.westhost.com/project34.htm

I added in a buffer before the circuit, cap blend for bass at the output of the spring tank and a trim for setting the resistance pre-tank. I'm thinking it might benefit from having C3 be selectable between 22n, 47n and 100n for cutting or boosting the bass going into the tank. There is a relay on the mute section for remote switching - that circuitry hasn't been added onto the schematic yet. Comments welcome.

[PRR]

> C3 be selectable between 22n, 47n and 100n

You surely mean C2? (C3 in that drawing is the '386 output snubber.)

That was my first thought. "Usually" we cut bass going into the tank below 500Hz so it doesn't mud-up. A good place is _C2_ more like 0.003uFd (3nFd in new-speak). Ah, the '386's 50K input shunts the 100K pot, so maybe more like 0.01uFd. Or maybe Drive pot 10K so the '386's extra 50K is not-much effect, and C2 on the order of 0.03uFd (as you suggest).

If you need to frill-up, I would add treble boost/cut, for a mellow-room tone versus a zzzzinggg tone.

I do not understand the power voltages. If there is no negative supply, IC1A IC1B inputs have to be up at half-voltage, not GND.

[mth5044]

I was actually just about to post that I missread the schematic - I had read R5 and C3 as a highpass, but I was pretty wrong. It seems there isn't much deliberate bass roll off, so I suppose a 22k / 0.02u highpass filter with switchable cap would achieve what I was shooting for. As far as placement, I suppose right after C4 would be a good place to start.

For the treble control: would just ditching the bass cap blend and putting in a passive baxandall tonestack do the trick? I'd hate to think that the signal coming out of the tank is so weak that any kind of passive tonestack would just run it into the ground and/or amplify and unwanted hiss. Perhaps placing it after C5 and putting something like 10nF in the place of C5/6.

I have looked at the Project 34 again and realize he does as a +/- 15V powersupply is needed. I will have to think about that some more. I suppose all the GND would now be -15V besides the RCA shields.

Thank you for your comments.

[PRR]

High-pass in front of the LM386. After it, power is large and too precious to waste, also the impedance into the tank may be fussy. (I'd lean to a "high" impedance drive, like 100 ohms into a ~~8 Ohm tank; that's what Hammond suggested.)

R5 C3 is just something the LM386 *needs*. Especially with loudspeakers. Or with coils, which is what you are doing. The exact reason is complicated: at high frequencies, when a speaker/coil impedance rises, the '386's emitter followers gyrate that into an internal capacitance, which with another internal cap, cause the '386 to break-out in radio waves. Which is bad for nearby radios, but also the '386 can't scream 900KHz and also do a sweet job with your audio. So you "always" add the 10R+0.1uFd to the output.

> all the GND would now be -15V besides the RCA shields.

You are mis-understanding or mis-stating something, I'm not sure what. Find a clear description of single-supply versus dual-supply operation. Yes, this can certainly be done with a single 12V-15V supply but the opamps will need a 6V-8V "Vref" bias source.

[mth5044]

Misunderstanding is most likely.

For the power - are the two options either 1) running the 5532 at +/-15V (I was thinking much more needed to be done to the circuit if this were the case) or 2) tacking on a 100k resistor on top of R6 that connects to VA to bias the signal to ~7.5V. The signal is set to 7.5 at the input, but I suppose that gets lost somewhere.

[mth5044]

Things got a bit crazy, and ended up with another schematic. Seems this has gone pretty far from the original intent of this thread. Perhaps I should make a new one...

Anyway,



Runs on a 16V AC adapter. Power section on the top converts to +/- 15V DC and also 5VDC for the ucontroller. The 386 has remained to drive the reverb tank. There are two inputs, line level and instrument level each with an input volume. Then a summing amp with reverb drive and dry level afterwards. The output of the reverb tank is met with a baxandall tonestack and a reverb level then a mute relay for on/off. Then the instrument out and line level out.

Thanks for any advice!

[PRR]

I think your relay is permanently on, the LED(?) permanently off.

The path through C20 gives "infinite gain" when D_LV is full up.

The tank recovery amp may need a LOT more voltage gain than a unity-gain tone stage. Remember the original used TWO transistors just for this function.

You surely want an output cap on IN_O.

I don't understand why LL in has a gain of 10X, then a loss of 1/77 relative to IN in. Or a bass-boost.

[mth5044]

I think your relay is permanently on, the LED(?) permanently off.

You are correct, silly mistake. Pin 5 of the relay should only connect to VR and pin 2 to the collector. That solves both of those problems.

The path through C20 gives "infinite gain" when D_LV is full up.

I assume infinite gain is in quotes because it's only possible in ideal situations, but I guess I don't understand why it has infinite gain. The IC4B opamp has a gain of -1ish of gain, but why only the dry level and not the reverb drive. Hmm

The tank recovery amp may need a LOT more voltage gain than a unity-gain tone stage. Remember the original used TWO transistors just for this function.

I will insert the volume control into an opamp stage as schematic before this.

You surely want an output cap on IN_O.

Surely! I will tack that on.

I don't understand why LL in has a gain of 10X, then a loss of 1/77 relative to IN in. Or a bass-boost.

To be honest, the LL input is a mystery to me. I got it from another reverb schematic. I guess the goal is to get both the instrument level and line level to similar levels. If pading a line level signal 1:10 is something that is used, I guess multiplying the gain by 10 then cutting it by 1/77 is.. close to that. Perhaps since the line level is already coming from a low impedance source, I could just use a voltage divider with a potentiometer to get the same effect. Perhaps it be better to use a FET buffer on the line with a 1:10 pad and use the now free opamp stage for boosting the volume after the tonestack.

Also, if one were to use this schematic for another 8 ohm reverb tank, how can it be adapted to an isolated input of a tank? The majority of the 8ohm tanks have isolated inputs and grounded outputs. I see examples for it when using opamps, but nothing with the 386.

[mth5044]

Hmm

[PRR]

> The IC4B opamp has a gain of -1ish of gain

No; the ratio of the feedback resistor to the input resistor.

D_LV pot full-up, there is no input resistor. Rfb/0 is infinity. (In Pure math, "undefined"; in real life >100 at 10KHz and >10,000 at 100Hz, slightly tempered by a couple 1.0uFd caps..... IAC, I would not even try D_LV full up.)

> an isolated input of a tank

That just means the input jack is strapped to the tank case. That can be useful with lower-power drivers using current-feedback to get a high output impedance without large loss. '386 has heaps of output, tis simpler to just throw ~~100 Ohms in series with the tank.

> I guess multiplying the gain by 10 then cutting it by 1/77 is.. close to that.

Except that 2V input, times 10, is 20V signal, but +/-15V rails only support 10V signal. Hi-Fi CD players output 2V; so does much other stuff, especially when pushed. True, another 1.77 gets you into a reasonable match for instrument level, except you now have a signal with arbitrary and un-adjustable fuzz-tone on it.

Generally a Line input should go right to a pot. Especially if running lower supply voltages in your box than in whatever box is sending this line level.

[PRR]

[mth5044]

You have a magical way of making things much more clear. Thanks so much for the schematic.

[PRR]

> making things much more clear

I'm lazy. I look at your plan and shudder. I'm never going to build all THAT!! Decide what you need to do, figure how simple you can do it. Whether mass-production or DIY, pencil is cheap, parts+solder costly.

[mth5044]

Hand drawn, in PRR fashion. Added to PRR's previous schematic: instrument level input volume, dry cut switch and an output control.