Equinox reverb as an amp tank replacement

[Netbat]

Hi guys, this is not a new stompbox design, but more of an adaptation of an existing one...

So I have recently finished my latest project and, as I can't just control myself  , I started a new one: A Frontman 25R amp. Why this amp? Well, about 15 years ago I was playing in a garage band and the other guitarist had this amp. I loved the tone and specially the distortion and WTH ... Why not ?... So I got the schematics, made the layout, etched the PCB ... you know the steps ... and the amp is sounding great so far! ... Ok, so far I have just assembled and tested the board (no cabinet or reverb tank yet), but that's 50% of the way ...... (ok, maybe 40% max) ...
So, now I need a reverb tank ... The problem is that I don't like to spend too much on my hobbies and I live in a country where importation taxes are just violent  ... A small spring reverb unit would cost more than US$50 and would take 3 months to arrive. Being a cheap a**, I started thinking about other options and found out about the Equinox Reverb, based on the PT2399 delay chip. Although there are several other options, the referred one seems simpler cheaper (as I have some PT2399 available) and cooler  .

Now, I think it is just a matter of adapting the schematic so it can be used as a reverb tank replacement, but I'm not sure if I'm on the right track... See the changes I made on the original schematic:



And this is the buffer stage of the amplifier:



Reverb units have the option of grounded and ungrounded inputs and outputs. In this case, an ungrounded input/grounded output is used and I'd use P13 on the input of my unit and P15 on the output. Ground would come with the supply and P14 is not used. I wanted to make this reverb unit as generic as possible, so I put input and output trimpots to adjust the levels, aiming not to change anything on the amp.

See that the input pot is not REALLY necessary, as it would actually decrease the SNR, but I don't think it will make any harm here.

Before I prototype it, I'd like to hear some thoughts about the design... Another option would be to generalize the original pedal schematic/layout in order to create a PCB that can also be used as an internal reverb unit for amps.

At last, do I have to bias the input when using the PT2399? As I could see on the datasheet and on other schematics, there is always coupling capacitors on the inputs and outputs, so, I guess the PT2399 already does that internally.

Thanks in advance.

[Mark Hammer]

I can't see the pictures but there are a couple of aspects to consider.  The Frontman 25R normally parks the reverb tank after all the overdriving, then mixes the reverbed version of the overdriven signal in with the non-reverbed version just before feeding it to the power-amp chip.

This may well provide a signal level in excess of what the Equinox expects.  Additionally the delayed sound may actually sound better if tapped earlier in the signal path, such that the overdrive is straight, but the reverb is clean.  It may be better to tap your signal for feeding the Equinox after U1A or U1B.

[PRR]

Reverb tank has large loss.

Electronic reverbs typically have little or no loss.

You are going to have to cut a LOT of gain out of the reverb path. Do you know how?

[Netbat]

Hey guys, thanks for the response!

Mark, you are probably right, but I don't feel like changing the amp circuits righ now, as they are well and operational ... Besides, my goal is to come up with a "reverb tank replacement" sort of thing ...

PRR, yes, the spring reverb tank use magnetic transducers with rather low output, so amplification is needed on both ends. As far as I can see, the amp buffer line of FM25R has a gain of 7 on the input (feeding 630mV @ max volume to the reverb input) and a gain of about 67,66 on the output ... the expected signal amplitude to sum with the direct line should be around 10 or 20mV... that means that the reverb unit must have an output amplitude of around 0,15mV. That's 100 times lower than usual audio signals before amplification ... And I think I'm missing something ...

My strategy is to feed maximal amplitude possible to the reverb unit (hence the input trimpot) and adjust the output trimpot to cope with the expected buffer amplitude, as I guess the Equinox circuit has a gain of 1... This way I maximize SNR and have a versatile circuit ... but 0,15mV is too darn low ... I may not even be able to obtain this value by using a trimpot ...

My problem is that I can't find the actual max amplitude of PT2399 ... I assume it is 2.5V after the LPF, so 630mV is (supposedly) inside limits ...

I found another issue with the use of PT2399 here...

Can you guys see the pictures ? they show ok here ... And, as english is no my mother tongue (and I suck at writing  ), I'm not sure I'm being clear  ... If not please let me know ...

Thanks again for your thoughts ...

[merlinb]

My strategy is to feed maximal amplitude possible to the reverb unit (hence the input trimpot) and adjust the output trimpot to cope with the expected buffer amplitude, as I guess the Equinox circuit has a gain of 1...

Yes
You could remove R45 and/or R40 from the amp, which would convert those opamps to unity gain. Then you can probably feed the signal right through the equinox with no need to adjust the levels.

My problem is that I can't find the actual max amplitude of PT2399 ...

About 3.5Vpp, so about 2.3Vpp at the input to the Equinox. You might want to add a couple of diodes at the input of the equinox (one to ground, the other to 5V, reverse biased) to protect it from excessive signal levels, since the amp runs off higher voltages.

[YouAre]

...Additionally the delayed sound may actually sound better if tapped earlier in the signal path, such that the overdrive is straight, but the reverb is clean...

Sorry, just to clarify, are you suggesting that distortion be "un-reverbed" and then the reverb'ed portion of the signal be clear (signal tapped before gain/clipping stages)?

I'm probably misunderstanding what you were suggestion, but it could still be a cool accidental discovery.

[Netbat]

Thanks so much for this info Merlin...

I gave some thought about this and came up with a different approach ... I'm thinking, with Merlin's consent, to make an "Equinox reverb module" (I really like modular designs  ). This is the sketch schematic:



And the board should look something like this:



This way is can be used inside an amp or in a pedal in a similar way to the Belton digital reverb. I guess if I could get some SMT parts I could reduce this A LOT, but that's OK for a first idea ...

In my case, I'd use a "Mother board" (I don't know why but I always read this with a Russian accent...  ) such as this:



I included Merlin's point on input protection, but decided to restrain a bit more the input signal. I've put 2 leds to clip higher signals. As the input should be around 630mV, things should be clean and protected and I shouldn't need the input voltage divider.

I think I can't escape from changing, at least some resistors, on the amp board. Turning the output buffer (the one that receives the reverb signal) into unity gain should help a lot (by shorting R41 and removing R40) and I could have about 21mV on the output of the buffer without many problems...

The board should look like this:



They are meant to be placed side-by-side, connected with a flat cable (or something like this) and screwed on the amp's chassis...

Even the Equinox pedal can make use of this module, with a "L" shape assembly (the module should be mounted at 90º), having a "mother board" with this schematic:



That should look like this:



This should fit in a Hammond 1590B and the led is mounted at 90º, between the pots... What you guys think? Is this too much crazy rubbish?

[merlinb]

You need a 10k input resistor (in series with C33).

[Netbat]

You need a 10k input resistor (in series with C33).

Thanks for pointing that out, you are right. I corrected the layout and will be giving this dude a try over the weekend (hopefully) ... I'll come back with the results and the final versions afterwards ... Thanks so far guys! ...

[Netbat]

So I could finally assemble the board and connect it to the amp, but some strange results came out of this  ...

There is some sort of positive feedback that gradually increases the sound level, until it saturates (takes about 5 seconds) and I have to disconnect the supply to reset  the ICs. Even if there is no input at all ... the only thing that seems to avoid that is to constantly changing the "ambience" trimpot. With higher resistor values (about 200k) it also works, but the delay is too high, and unusable ... removing the ambience pot also solved the problem, but the sound is no good ...

The only difference from the original design is that I couldn't find non-poralized caps for C10 and C32, so I used tantalum polarized as represented in the original schematic ... As (i believe that) they participate in the feedback-forward process, I think they might be the ones to blame, but I can't see why ... And I also can't understand why they need to be strictly non-polarized ...

This is the reverb board considered:


And this is the "mother board". Input resistor is a jumper and input capacitor is 100nF. A 50K pot is connected in the output and the ambience resistor is a 20K trimpot.


Could be this beacuse resistor tolerances affecting the gain of the PT's (i.e gain > 1) ? Overall, this seems to work, and, behind the feedback problem, the reverb sounds is fine.

[Netbat]

I had a power electronics professor once that used to say that when you have a technical problem, you should keep looking at it until it starts to feel ashamed, slowly begins to be smaller and smaller, until the moment it disappears or runs away ... It seems to work every time  ...

So I decided to spend some time analyzing the matter and try some debugging... I went back to the original Equinox thread and found a Merlin's post about increasing R19 in order to reduce oscillations (increasing OP-AMP attenuation?). So I did (33k Ohms) and it worked!!! Hooray!! 

Now it was time to connect that on its actual place in the amp ... I initially thought about leaving the input buffer (on the amp) with its original gain, but that didn't work as signal levels would easily reach 2V (I'm thankful for Merlin's tip on the input protection). So what I did was leaving the input buffer with unity gain and adjust the gain of the output buffer. The reverb buffer stage is like this right now:



The reverb tank board documentation is as follows. As I said, I changed R19 to 33k. The schematic:



The board ready-to-transfer (board dimension is 45.974mm X 53.086mm):

http://i1306.photobucket.com/albums/s570/Netbat1984/Tank_PCB_pnp_zps61affb18.png

http://i1306.photobucket.com/albums/s570/Netbat1984/Tank_SILK_pnp_zps6412e055.png

The component placement (there is a bug on KiCAD that the track width don't show correct when printing, but this is just for reference and the PCB PnP above is correct though):

http://i1306.photobucket.com/albums/s570/Netbat1984/Tank_board_Placement_zpsafa2ced4.png

And it should look like this:



As for the "mother board", I designed it with a 7805 with a heatsink because of the voltage drop, as supply comes from the amps 27V bus and because I already had that heatsink. Eventually, one could also use a 78L12 and a 78l05 in series. This is the final version:



The board ready-to-transfer (board dimension is 36.576mm X 53.086mm):

http://i1306.photobucket.com/albums/s570/Netbat1984/MB_PCB_pnp_zps25cd09cc.png

http://i1306.photobucket.com/albums/s570/Netbat1984/MB_SILK_pnp_zps097b3d10.png

The component placement:

http://i1306.photobucket.com/albums/s570/Netbat1984/MB_board_Placement_zps1e83d860.png

And, finally, it should look like this:



Here is a pic of the implementation:



As long as finish adjusting the filters and get the tones right, I'll post an audio clip, but I can easily say that this Equinox  sounds great!

As I said before, this could be implemented inside an enclosure (I'm thinking on a 1590B) with a double board design for a pedal. I'll do the layout and post here later...

Thank you very much for the help and thanks Merlin for this AWESOME design! Cheers guys!

[Netbat]

As promissed, here goes the second mother board as for a pedal assembly... Here are the schematics:



The board ready-to-transfer (board dimension is 45.974mm X 53.086mm):

http://i1306.photobucket.com/albums/s570/Netbat1984/MB2_PCB_pnp_zps1dfb91d9.png

http://i1306.photobucket.com/albums/s570/Netbat1984/MB2_SILK_pnp_zps49942c9f.png

The component placement:

http://i1306.photobucket.com/albums/s570/Netbat1984/MB2_board_Placement_zps1145a9a3.png

And it should look like this:



I made the study and it should fit a 1590B with jacks, footswitch (mounted between the jacks) and a 9V battery (caps should be not higher than 10mm):



Thanks again...