TL Reverb driver idea - from triode cathode circuit??

[jimboyogi]

I am trying to design a transformer-less (TL) reverb into an existing amp design. The driver is 1/2 of a 12AU7 triode, recovery is very conventional 1/2 12AX7. I am having 2 main troubles getting it to where I like it.

1 - Phase issues - I have read on this forum, and elsewhere, that there is no real phase to the reverb return signal, so there should not be any phase issues. Well, I'm seeing phase issues. When I take the reverb drive signal from the anode of the 12AU7, I get a strong comb filtering effect when I mix a lot of reverb signal back into the dry signal. Gets very hollow and phase cancellationey (I hope I spelt that made-up word correctly).

2 - I am using a "F" style tank, high input impedance, 1.5k - 1.9k (depends on whether type 4 or 8, I am trying both). I have both Accutronics and MOD tanks, both medium and long decay. When driving from cap coupled anode of the 12AU7, I just can't get a reverb tone that I'm happy with. Does not sound like the Fender style that is in my mind as a reference, lacks 'lushness' and clarity. Seems to sound like the springs are vibrating at their resonant frequencies, and the frequency of this is not closely related to the frequency of the note/s played, the guitar signal just sets them in motion.


To solve problem #1, I am trying to drive the tank from the 12AU7 cathode, instead of the anode. I was hoping that it may solve problem #2 also, but that's just wishful thinking.

It works, actually quite well, but I wanted to discuss the various concepts and issues. I have not heard of this being done before, but of course it has, so if there are previous threads, or other info out there, please point me towards it.

Reverb tank coils are current driven, not voltage driven, this is why I thought the cathode circuit could work. I have very limited theory, but I think that all of the current of the tube flows through the cathode circuit, and the current varies with the grid signal. So here we have a potential driver circuit.

The input coil of the type 8 has a listed impedance of 1.925K ohms, and I measured the DC resistance at ~200 ohms. The 12AU7 has a 22K anode resistor, and in normal operation the Rk was 1.5K, giving a Vk of ~7V. I have tried putting the input coil in series with the Rk.

Also another potential issue is the tank coils and laminated steel cores. I assume that these are made like a PP OT, designed to not have a DC offset applied. But putting them into the cathode circuit creates a DC offset?

[jazbo8]

John Polstra presented his current drive tank design years ago on AX84, you may want to Google it.

[nickb]

You might find the Fisher K-10 Spacexander interesting too.

Whether it's a current drive or voltage drive depends on the driver impedance. An ideal voltage source has zero impedance whilst an ideal current source has an infinite impedance.

[g1]

I'm sure this is not all of your issue, but I would not ever try to compare a cap. coupled "F" type tank to a Fender transformer driven tank. Compare to a similar Ampeg or Traynor circuit, fine, but I've never heard anything that matches the *iconic surf type vintage Fender transfomer driven reverb sound (tm ) *.

[Enzo]

Agree, when you make a circuit nothing like the Fender circuit, don't expect it to sound like the Fender circuit.

[J M Fahey]

I am trying to design a transformer-less (TL) reverb into an existing amp design. The driver is 1/2 of a 12AU7 triode, recovery is very conventional 1/2 12AX7. I am having 2 main troubles getting it to where I like it.

1 - Phase issues - I have read on this forum, and elsewhere, that there is no real phase to the reverb return signal, so there should not be any phase issues. Well, I'm seeing phase issues. When I take the reverb drive signal from the anode of the 12AU7, I get a strong comb filtering effect when I mix a lot of reverb signal back into the dry signal. Gets very hollow and phase cancellationey (I hope I spelt that made-up word correctly).

2 - I am using a "F" style tank, high input impedance, 1.5k - 1.9k (depends on whether type 4 or 8, I am trying both). I have both Accutronics and MOD tanks, both medium and long decay. When driving from cap coupled anode of the 12AU7, I just can't get a reverb tone that I'm happy with. Does not sound like the Fender style that is in my mind as a reference, lacks 'lushness' and clarity. Seems to sound like the springs are vibrating at their resonant frequencies, and the frequency of this is not closely related to the frequency of the note/s played, the guitar signal just sets them in motion.


To solve problem #1, I am trying to drive the tank from the 12AU7 cathode, instead of the anode. I was hoping that it may solve problem #2 also, but that's just wishful thinking.

It works, actually quite well, but I wanted to discuss the various concepts and issues. I have not heard of this being done before, but of course it has, so if there are previous threads, or other info out there, please point me towards it.

Reverb tank coils are current driven, not voltage driven, this is why I thought the cathode circuit could work. I have very limited theory, but I think that all of the current of the tube flows through the cathode circuit, and the current varies with the grid signal. So here we have a potential driver circuit.

The input coil of the type 8 has a listed impedance of 1.925K ohms, and I measured the DC resistance at ~200 ohms. The 12AU7 has a 22K anode resistor, and in normal operation the Rk was 1.5K, giving a Vk of ~7V. I have tried putting the input coil in series with the Rk.

Also another potential issue is the tank coils and laminated steel cores. I assume that these are made like a PP OT, designed to not have a DC offset applied. But putting them into the cathode circuit creates a DC offset?

One accepted way to drive reverb tanks is to put them in series with (usually) a transistor, so passing DC current, which gets modulated to vibrate the driver end.

FWIW that was one of my first succesful products, around 1969, the schematic straight from Popular Electronics,

It's based on a factory suggested schematic; old reverb specs stated acceptable idle DC current through the coil, so it was an "official" and "accepted" rating.

You are doing basically the same, only a 12A*7 is incredibly inefficient "transistor" , which can do with, say, 9 to 25V supply what requires 120 to 250V with a tube.

Don't worry about saturating the laminated cores, because gap is *huge* , way more than what would be used in, say, a Class A`rated OT, so you are safe in that regard.
Only problem with excessive DC is that it will twist the driving magnet and might have it rest in a non linear or insensitive position, but as long as you respect the datasheet value, no problem.
Not sure now, but it might have been as high as 3.5 mA or thereabouts.
All this through a 45 years time tunnel, of course.

FWIW here's the schematic of what I did, 45 years ago, mounted on PCBs "printed" with my Mom's nail enamel:

worked very well way back then, will work just as well today.

For the full article, including Space Age Technology (transistors, a FET and a PCB):
Popular Electronics January 1968

You'll have to provide your own nail enamel.
Inferno Red suggested for a hotter sound

The article includes tips to interface it with an existing Tube amp, and feed it from its existing PSU

[Mick Bailey]

Are the phasing issues really phasing?

The output of a reverb tank compared to the input shows very little relationship when scoped. The springs used are also often dissimilar (opposite-wound, different tension, different number of coils) and this also has a bearing on the output. The time delay alone and reflections along the springs means that there's no fixed relationship between the input and output.

I've experimented with cathode-driven reverb but could never get enough current off a single triode to give a really lush-sounding reverb in the territory of the 6G15. A Fletcher-Cooke configuration (two cathode-followers in P-P) works, but means using both triode halves and adding more complexity. The problem was I had a 'tubes or bust' determination and after rethinking my approach ended up with a transformerless reverb that used a MOSFET + BJT as the driver in a constant-current configuration. Everything else is tube.

[jimboyogi]

Thanks so much for all the responses!

jazbo8 - Thanks for the info. I looked up the thread, and some interesting stuff there. Along very similar lines to the reverb driver circuits on Merlin's website. I have actually made a couple of amps using this style of driver, plate driven with series resistor to simulate constant current. They sound good, but I'm trying for better!

nickb - that Spacexander is an interesting circuit, and does use the cathode circuit for drive, although via a matching transformer.

g1 and Enzo - good points, with a circuit nothing like a Fender, it's unlikely to sound exactly like one!

Mick Bailey - I appreciate what you are saying about phase. Using my modern scope, when I scope the reverb output, it actually traces as a sine wave (with sine wave input). I know that this isn't exactly the true waveform, but maybe it's the scope's line of best fit for the complex reverb signal? I have found though that when mixing the wet and dry, it does respond as if there is some type of phase issue. So when taking the reverb drive from the driver plate, there is strong cancelation when mixed back. Taking the drive from the driver cathode (opposite phase), there is far less cancelation when mixed back.

J M Fahey - Thankyou for your post, absolute gold You really addressed the question, and gave me some piece of mind about the DC current through the transducers. Love the nail polish for pcb printing!

ATM I am getting 0.8V developed across the reverb input transducer (200 ohms) at idle, so 4mA static current. At 1kHz, when the driver triode is fully driven, just at onset of clipping, I am getting 15V p-p signal across the input transducer.
Can anyone please help me with the method to calculate the signal current I am supplying to the tank?

[jazbo8]

ATM I am getting 0.8V developed across the reverb input transducer (200 ohms) at idle, so 4mA static current. At 1kHz, when the driver triode is fully driven, just at onset of clipping, I am getting 15V p-p signal across the input transducer.
Can anyone please help me with the method to calculate the signal current I am supplying to the tank?

Quick back of the envolope calculation, assuming very little distortion:

Vo = 15Vp-p = 5.3V rms
Io = 5.3/200 = 0.0265A rms
less idle current, so
Isignal = 26.5 - 4 = 22.5mA rms or ~64mA p-p

[jimboyogi]

Wow, that's heaps of current! Thanks for doing this calculation jazbo8.

I'm wondering if , for Io, we should use the impedance at 1kHz = 1925 ohms say 2k, instead of the DC resistance? This would divide the answer by 10, giving 2.25mA RMS.

I just checked the Accutronics website. They specify the nominal drive current for a type F tank as 2.0mA (@ 1 kHz). So if the above is true, the tank is being driven fully.

I have also just changed the Ra on the 12AU7 from 22K to 12K. This has increased the max clean voltage across the tank input transducer to 20Vp-p = 7.07 Vrms.
If the transducer impedance is 2K, then 7.07Vrms/2K impedance = 3.5mArms at 1kHz?

Please correct me if I'm wrong, but I think that this is telling me that I am driving the tank hard enough from the tube cathode circuit.

[jazbo8]

I'm wondering if , for Io, we should use the impedance at 1kHz = 1925 ohms say 2k, instead of the DC resistance? This would divide the answer by 10, giving 2.25mA RMS.

Yup, that makes more sense, I wasn't really thinking... of course it is the impedance that matters not the DCR, then that begs the question, why is the idle current set at 4mA?

[jimboyogi]

Yup, that makes more sense, I wasn't really thinking... of couse it is the impedance that matters not the DCR, then that begs the question, why is the idle current set at 4mA?

It isn't necessarily, that's just the idle current through the transducer in the bias set-up I am running. I mentioned this in relation to J M Fahey's comments about acceptable DC current in the transducer coil.

[J M Fahey]

The above calculation is fine, using the impedance.

Only small correction would be that impedance, being inductive, varies with frequency, and guitars provide all kinds of frequencies in a random way, depending on whatb you are playing, from 80 to 5000 Hz.

But you can see and visually measure that current: put a small (think 10 or 20 ohms) resistor from tank ground terminal and actual ground and clip your scope across it.

Then just strum your guitar.

[g1]

I believe the drive level is recommended to be 2 to 5mA, so 2 may be more "minimum" than "nominal".
And as JM alluded to, you should use a variable frequency to get any idea of what is going on when either looking at scope, or listening. There is an incredible amount of variation between tanks of even the same model when using a fixed frequency. The newer Fender schematics make note of this for any signal voltages related to the tank. The spring action is all about the change in dynamics of the signal.

[jimboyogi]

J M Fahey, thanks for the calculation check. Yes, as you say, that calculation is only relevant for 1kHz, where the specified coil impedance is ~ 2K ohm. Because of the coil inductance, the impedance would be 4K ohm at 2kHz, and only 1K ohm at 500Hz. So the drive to the tank is very frequency dependent. In practice I am finding the tone of driving the coil from the cathode circuit "dull", lacking in higher frequencies.

g1 - Absolutely! I have a bunch of different "F" tanks here, and have tried numerous in various circuits. From measurement, the AC voltage generated across the input coil using the same drive circuit varies considerably, which indicates to me that the actual impedance at a fixed frequency varies significantly between different makes (MOD and Accutronics), as well as between different "F" tanks, type 4, 8, and 9.

Not to mention the completely different reverb sounds available from plugging into each of these different styles and makes of tank. The ideal is to develop a circuit that sounds great with all of them, but in practice this is not so simple. For instance, my MOD type 8 long decay tanks have an extremely long decay, and a huge mid-range emphasis. So for the best ability to dial in smaller, more useful amounts of reverb, I have subbed in a log pot for the reverb return control. But for all of the Accutronics tanks, particularly the type 4, the reverb has a shorter tail, and less mid-range. So these work best with a linear reverb return pot! And in the driver circuit, the Accutronics tanks benefit from being fed more low frequencies, whereas the Mod tanks sound better being fed less low freq. to avoid mud.