Help me decipher an Acoustic tonestack?

[Derringer]

I picked up an Acoustic 320 a little while ago. Great amp. I've been slowly recapping it to hopefully extend some trouble free useable life.
Anyway, this is what the tonestack in each preamp, it has two, looks like.

The bass/mid/treble knobs seem to do a cut/boost and I'm assuming that's accomplished by how they seem to be blended between the collector and emitter of the feeding transistor.

Can anyone tell me what frequencies these controls are centered on?

This amp was primarily designed for bass so I'm also toying with the idea of modding one of the channel tonestacks to be more suited for guitar frequencies.

thanks!

[DougH]

The easiest way to analyze one of these ladder tone stacks is to simulate it in something like Spice, run an AC analysis and look at the frequency response (Bode plot). By adjusting the controls and looking at the plots, you can work out what each control is doing.

[Derringer]

yeah, I hear you Doug.

Most of my questions could be answered if I knew how to use a simulator. I got a hold of the Tina TI one but I couldn't figure out how to make it go. Granted I didn't put a lot of time into it but time's a limited thing when you've got 3 kids, a job, and a 55 year old house

Now math I can handle just fine ... was hoping for an equation of sorts that factored in the standard LC equation.

anyone care to sim this? I'd greatly appreciate it

[PRR]

Really, this is matchbook work. (A calculator is helpful.)

> what frequencies these controls are centered on?

T and B are slope or shelf filters. No "center". These yield to RL or RC equations, use 6.8K for start of slope, 1K or 2K for level-out.

M is a bandpass and "the standard LC equation" applies.

[Derringer]

awesome, thanks Paul!

Last night, before I pulled the preamp boards for recapping, I ran my signal gen through the amp and scoped the output. Assuming my signal generator is accurate, it looks like the bass starts at around 100Hz and goes down, the middle peaks around 500Hz and the treble starts a bit before 2K and goes up

Knowing which resistor to use in the calculations was definitely holding me up. When I calculate the middle, I get 580 Hz.
I'm being little lazy and using this: http://www.calctool.org/CALC/eng/electronics/RLC_circuit
The bandwidth changes for 6.8k and 1k or 2K, that's how wide the affected frequency range is, yes?

Now for the bass, L1, what capacitor do I factor in? Would it just be C9? If I do, that comes up to about 120 Hz, which jives with my scope readings.

And if that's the case, should I be using the series value of C13 and C9 for the mid calculation?

[DougH]

Really, this is matchbook work. (A calculator is helpful.)

> what frequencies these controls are centered on?

T and B are slope or shelf filters. No "center". These yield to RL or RC equations, use 6.8K for start of slope, 1K or 2K for level-out.

M is a bandpass and "the standard LC equation" applies.

Where are 6.8k and 1k or 2k coming from?

[PRR]

> Where are 6.8k and 1k or 2k coming from?

[Derringer]

I should probably post the whole thing so you can see what's before and after the tonestack


I dove back into Tina-TI. I did the tutorial, was able to sim a BJT gain stage and  an opamp bass boost and used the signal analyzer to look at how different frequencies were affected on the output.

I'm not having luck with this circuit though. The gain is WAY down in the sim which maybe is to be expected but the frequencies affected are just little blips, nothing like a shelf filter.

[PRR]

> was able to sim a BJT gain stage.... I'm not having luck with this circuit though. The gain is WAY down in the sim

Do you have proper DC voltages?

What is "way down"? This stage should have gain near 1. Put in 1V, get out 0.95V or so.

[Derringer]

found the problem .. just had couple connections around the 680R switched up

seems to be sim-ing ok



The treble seems real touchy though. Seems like it's either off or blasting up through the roof. Might be a log/lin thing that I haven't delved into yet in the program.


totally got yelled at by the wife for "spending all that time with your circuits" today too 

[PRR]

T at 5% should slope-off to infinity loss.

100K pot on presumed 6.8K resistances should not be "touchy".

Sorry you got in trouble.

[jatalahd]

If you are simulating the circuit only around Q3, there should not be that much gain I see in your plots (unless the LC resonance gives some). If I see correctly that R14 and R15 both are 6.8k, then Q3 is a unity-gain phase splitter, where you only get 180 phase difference for the signals from collector and emitter. Could you break your simulation in parts so that for example simulate the circuit only with the low-pass filter section (other components removed)? Then you should clearly see a basic 1st order filter curve with the shelving effect at the high frequencies. The pot should have a range to move from high frequency cut to a very small boost in the high frequency side.

I did a lazy simulation with this by excluding the Q3 all together and replacing it by two voltage sources, first with 6.8k internal resistance and the second in opposite phase with 1k internal resistance. However, I don't yet know if this approximation is valid in this case. If I have more time I can try to simulate it with the transistor included. From the full schematic document I got that C12 and C13 are 0.1uF (couldn't see them values in the schem).

I just don't get it why have they done it like this, currently I see that the boost side of the tone control is very very small compared to the cut so the practical significance is a bit doubtful.

[PRR]

> replacing it by two voltage sources

Poor simplification.

Cathodyne (emitter-dyne?) is a gain stage with two dependent outputs, not two independent voltage sources.

The cathodyne gain is Rp/Rk. If values are equal, gain is unity. If Rp is small, gain drops. If Rk is small, gain to plate rises. So they are not independent voltage sources.

The phase difference actually has no effect here. We are not taking output from both ends. The pot impedance is high enough that there is little mixing.

> the boost side of the tone control is very very small compared to the cut

The boost "could" go to "approximately infinite".

Take C12 turned all the way to Boost, omit the 1K. At 16KHz, 0.1uFd is 100 Ohms. Parallel with 6.8K is 99 Ohms. Transistor internal emitter resistance here is maybe 20 Ohms. Re is 119 Ohms. For Rc we have 6.8K, and effective ~~16K for the tone pots, and say 50K for the Volume control: 4,300 Ohms. 4300/119 is gain of 36, or 31dB boost!

(If you keep going to 100KHz, boost appears to be 140, 43dB. Good for very-long-wave radio pickup.)

At this point we must wonder about "minor" effects: input of Q3 with 119r under it may be near 10K, which is heavy load on R9 network, so we suspect it will fall short of 36. But Acoustic knew we don't want 31dB boost, so threw-in some 1K to limit boost to more like 4300/1119, 3.8, 11dB. Q3 input will be over 100K, so not a big load on R9.

[jatalahd]

Thanks Paul for taking the time to explain the circuit. I did a simulation with the transistor included and indeed for the top RC branch I got a max boost of +12dB and cut -12dB, quite symmetrical.

Elegant circuit, I want to use it somewhere else now. But I need to do the math homework still to fully understand it

[Derringer]

If you are simulating the circuit only around Q3, there should not be that much gain I see in your plots (unless the LC resonance gives some). If I see correctly that R14 and R15 both are 6.8k, then Q3 is a unity-gain phase splitter, where you only get 180 phase difference for the signals from collector and emitter. Could you break your simulation in parts so that for example simulate the circuit only with the low-pass filter section (other components removed)? Then you should clearly see a basic 1st order filter curve with the shelving effect at the high frequencies. The pot should have a range to move from high frequency cut to a very small boost in the high frequency side.

I did a lazy simulation with this by excluding the Q3 all together and replacing it by two voltage sources, first with 6.8k internal resistance and the second in opposite phase with 1k internal resistance. However, I don't yet know if this approximation is valid in this case. If I have more time I can try to simulate it with the transistor included. From the full schematic document I got that C12 and C13 are 0.1uF (couldn't see them values in the schem).

I just don't get it why have they done it like this, currently I see that the boost side of the tone control is very very small compared to the cut so the practical significance is a bit doubtful.

I started with just Q3 and did realize that there would be gain loss, and I was getting something ridiculous like -150 dB. So I simmed the Q1 connection to Q1/Q2 circuit without the pad and bright switches and then the Q4 buffer. After still failing I started trimming things and found that I messed up the 6.8K/680r/6.8uf node on the collector of Q3. Then I was able to get the results I posted ... which seem pretty close to what I saw on my scope and what I calculated. The treble is still odd but I haven't jumped back in yet.

It is a nice circuit, you just need to source inductors somehow that aren't huge chokes. The Xicon 4TL models get you in the ballpark and seem to have a proportional relationship between the DCR on the windings vs. their inductance. I already subbed the 10k:3.75H one for an open inductor that was on the amp's 5-band graphic eq.

I'm going by internet word though of course ... found this:

42TL022 1.5K Ohms .56 Henry   
42TL021 4K Ohms 1.5 Henry
42TL018 7K Ohms 2.6 Henry
42TL019 10K Ohms 3.75 Henry
42TL025 17K Ohms 6.4 Henry
42TL017 20K Ohms 7.5 Henry

here http://www.soundonsound.com/forum/showflat.php?Cat=19&Number=932504&Main=932105

Thanks Paul. All storms have died down and been forgotten here   

.. though she did try to kick the mouse out of my hand a few times while we were couch potato-ing the superbowl

[Derringer]

and I just found a response to my original question posted on the UnofficialAcousticControlForum:

Here are the published specs for the 320 & 330 amplifier tone controls:


CHANNEL A TREBLE CONTROL..........A continuously variable high frequency control with a range of 36 db at 6 K Hz.

CHANNEL B TREBLE CONTROL..........A continuously variable high frequency control with a range of 36 db at 6 K Hz.

CHANNEL A MIDRANGE CONTROL......A continuously variable mid frequency control with a range of 36 db at 600 Hz.

CHANNEL B MIDRANGE CONTROL......A continuously variable mid frequency control with a range of 36 db at 600 Hz.

CHANNEL A BASS CONTROL............A continuously variable low frequency control with a range of 32 db at 150 Hz.

CHANNEL B BASS CONTROL............A continuously variable low frequency control with a range of 32 db at 150 Hz.

[PRR]

> Elegant circuit

Many-many Boost/Cut EQs reduce to two basic schemes. Steve Dove's console essay calls them Swinging Inputs and Swinging Outputs.



Fixed resistors may be 10K, pots 100K, "X" is a reactance which typically is 10K-1K over the frequency range to be bent. Alternate values are common: 6.8K instead of 10K to reduce pot-loading, 2K fix 10K pots for slightly lower hiss at the cost of much increased drive current and less maximum action, etc.

Swinging Inputs is used in many "graphic EQs".
http://www.geofex.com/article_folders/eqs/paramet.htm

Swinging Outputs is used in Ampeg VT-40 (sub-model with 3-band EQ) and some others.

Swing In causes signal loss before the gainstage and hiss can be an issue. (These will rarely be used in low-level stages.)

Swing Out loses signal after the gainstage and headroom can be an issue. (VT40 used a 300V stage driving a 0.2V sensitivity stage so the ~~10:1 loss was moot.)

The Acoustic 360 looks different because the Cathodyne is 1-in 2-out, not 2-in 1-out like the opamp implementations. But the "cathode output" is not used as an output, it is used as a gain control. It reduces to Swinging Outputs.

Baxandall is a modified Swinging Inputs (gain stage worked inverting not differential).

In all cases, you can imagine the pot wiper left center right, and the curve shape reduces to a simple R-X problem. Pot center almost always reduces to a hair less than unity-gain. The more complicated question is: what happens at part-boost/cut? Some part of Pot is series resistance which reduces the effect of X. Low pot resistance decreases the interaction but hurts hiss/headroom compromises.

There are other schemes which are neither S-In or S-Out, but they use many more gain-stages to invert and mix the various signals. They may avoid signal loss and hiss/headroom compromise, but often the complexity (cost) over-takes the performance gain. RANE probably has a paper on most of the variants.

[disorder]

I have nothing to contribute other than a great appreciation for those old acoustic amps. Here is me sweeping the MID pot from 1 to 99% 10 times logarithmically.  Other pots set to 50%

[Derringer]

hahah

makes for a purty picture

yeah, these amps are tough to beat

Is that sim from the tonestack for the ACC 320 I posted or from another Acoustic?

[disorder]

That's the 320 you posted but I can't read all the values so mine may be off a bit.