Class D chip for reverb tank driver?

[jubal81]

Been thinking about a design for an analog reverb and came upon this TI Class D chip that look very interesting. It's the TPA3137D2, which is 6W X 2 @ 8Ohms. It has differential inputs for noise cancelation and supposedly can drive 3-foot cables with only ferrite bead filters.

Never worked with Class D before and wondering if anyone has some insights before I get knee-deep.

Link to the data sheet:
http://www.ti.com/lit/ds/symlink/tpa3137d2.pdf

[Transmogrifox]

If you handle the switching noise on the power supply input well and use an inductor and/or ferrite to keep the high frequency stuff from going upstream I think the Class D will work as well as anything else.

I don't see a particular benefit from a Class D amp in this situation.  You're dealing with relatively low power, low current drive so the efficiency curve won't look particularly great when compared against a class A/B circuit.
Rod Elliot has several circuits for driving reverb tanks, one of which was just using a pair of the NE5532.

You do have to use a good layout and aggressive power supply filtering to keep EMI from going into your other audio circuits.  It would be best if you had an oscilloscope to look at the power supply ripple so you can best know how to filter it.  Sometimes the wrong types or combination of power supply filter capacitors will cause more problems than they might solve.  On a schematic the capacitor bank will look great, but in practice there may be some bad things happening between the non-ideal impedance characteristics and your PCB layout.

Summary is you need to give more attention to PCB layout than you would for an equivalent class A/B amplifier.

[PRR]

It is bridge-tie (floating) output. Reverb tanks are often grounded inputs.

You don't want anything like 6 Watts into a tank.

LM386 will smack a low-Z tank fine, with fewer pins and less supersonic trash.

[flanagan0718]

I wish I had more / something to offer but I really want to see how this develops.


Sent from my iPhone using Tapatalk

[jubal81]

Yeah, filtering is the challenge. Looks like they've tried to make it as painless as possible on the chip design.

I'm concerned about the power at 6W, but there are reverb drivers that use a 6V6 or el84 - with very good results.

[PRR]

> there are reverb drivers that use a 6V6 or el84

Most use the conditions for 6F6 giving 1.4W-2W max.

6F6 was minor overkill, but was going out of style in radios, so was cheap. When Leo/CBS finally ran low on cheap 6F6, they put 6V6 in the hole without trying to get the more-power of a 6V6. As EL84 got cheaper, it too was pushed into the job, still biased to the low-power 6F6 conditions (with different resistors for the 6F6 6V6 EL84).

Except for totally over-the-curl heavy SURF (and even then), there's much love for the 12AT7 stage used in combos. I can't see this making much over 0.1 Watts. For 4 to 50 Ohm tanks, the LM386 @ 9V will do that easy.

[jubal81]

My goal isn't the easiest or cheapest solution. Looking to challenge myself with something interesting and make something I can't buy.

Was looking at a headphone amp using mosfet + LM317 as a constant current source and thought up something like this for push-pull reverb with a phase-canceling dwell control. Just whipped it up, cap values aren't dialed in, but this allows for setting the exact power output with the resistors on the 317s.

[jubal81]

Also found an interesting dual mosfet for the job:

http://www.mouser.com/Search/ProductDetail.aspx?R=SQJQ960EL-T1_GE3virtualkey61360000virtualkey78-SQJQ960EL-T1_GE3

[Transmogrifox]

It is bridge-tie (floating) output. Reverb tanks are often grounded inputs.

It's not required to be bridge-tie, but you need more filtering to do it single-ended.  In either case you aren't required to ground your reverb tank so you could still use it bridged if you wanted.  This is just something you have to look out for. You may have to hack your reverb tank drive coil to isolate chassis connections from the input or output -- in the end, the driver and pickups are just coils.

You can also use capacitors or a transformer for DC-blocking.  I don't see how the Class D outputs are really any different than a single-supply linear amplifier driving a grounded load in that regard.

You don't want anything like 6 Watts into a tank.

Power rating of the chip is irrelevant.  The application defines how much power you pump into the load.  You can use a 100W capable amp chip and only drive the reverb tank with 0.1 Watts.  You may not be operating the amp in its optimal range, but it can be made to work without causing destruction to the reverb tank.

You can get class D chips that are optimized for <1Watt, so I would say Class D isn't off the table as an option.

As others have pointed out, Class D doesn't offer any clear advantage from a technical point of view.  As a fun project just to tinker with something and try something new the value is inestimable.

If your goal is to make the best spring reverb driver then existing designs probably give you the best options for a certain type of sound characteristic, but you have already stated your goal:

Looking to challenge myself with something interesting and make something I can't buy.

Some kind of Class D implementation is still high on the list of possibilities.

Your second option with the LM317 is another great one with this goal in mind.  I find it interesting.

Over the years I have played with all kinds of interesting impractical things -- ways of doing things that could be done more easily and cheaper but were really fun to explore the concept and prove to myself it could be done in some non-conventional way.

That basic skill of abusing IC's and circuit blocks to do things they weren't meant to do can lead you to some really clever (and cheap/simple) designs like the MXR envelope filter which implements a state-variable-filter with CMOS inverters and analog multiplexers.  The result is a respectable-sounding guitar effect that was also very inexpensive to manufacture.

Sometimes all the impractical things we do will inspire us to  discover something that turns out to be very practical and simple.  At the very least it is lots of fun, and in the context of guitar effects -- it usually makes some kind of sound