Inductors and Gyrators (Acoustic 360 Project)

[Rob Strand]

The info here is applicable to the Acoustic 360 project
http://www.lynx.net/~jc/pedals360.html
-------------------------
The other day in this thread by rankot,
http://www.diystompboxes.com/smfforum/index.php?topic=118943.0

I realized the Acoustic corp amp circuits rely to some degree on the inductor's DC resistance.
The DC resistance is not specified.

When you see an inductor it looks like an ideal inductor in series with the coil resistance.  The coils
resistance is often not on the schematic but in reality it is there and affects the circuit's behaviour.
No one bothers measuring it because it is a "hidden" parameter.  So when you need to know what it
is there is little info out there.

The same problem exists for the Mesa Boogie equalizer.  It is a common problem with circuits containing inductors.   Just look at the number of Wah threads discussing inductors.

After much detective work  (which I'm not going to clutter the post with) there is very little concrete info
indicating measured values of DC resistance for the inductors.
After combining the info from all the sources, including correlating manufactures specs with spice simulations.  I've managed to reduce the window of uncertainty.

From photos the inductors appear to be wound on P14/8 pot-cores which have a screw in the center.

1) Estimated inductors and DC resistance (DCR)

I made these reasonable assumptions:
Potcore:    P14/8
Material:   N48
ui:     ~ 2300   (not ue)
type:   gapless core, screw hole in centre leg
AL    2100
All coils to be wound with same wire.

All wound with 42AWG wire (**** The 6H inductor will not fit on bobbin.)
L [H]       6.0   3.0   1.75   1.5   0.75   0.375   0.187   0.093
DCR [ohms]    280   184   134   123   83   57   34   27
Turns      1690   1195   913   845   598   423   298   210

All wound with 44AWG wire
L [H]       6.0    3.0   1.750   1.500   0.750   0.375   0.187   0.093
DCR [ohms]   399   268   198   182   125   87   60   42
Turns      1690   1195   913   845   598   423   298   210

The 44AWG wire seems to be the overall best match to the info I gathered.  The 6.0H inductor
fits on the bobbin with 44AWG but not 42AWG. 
The aim here is not to wind an optimum inductor. 
The aim is to second guess what the manufacturer has done.

As a side note, gapless core inductors are likely to have +/- 20% to +/- 30% tolerance.  The DC resistance  is probably going to be +/- 10%.

2)  Gyrators

Obviously winding such inductors is a pain so it is desirable to replace them with a Gyrators.

Once trick with LCR based equalizers is the ordering of the L, C and R can be changed from what
is on the schematic.  The first step to replacing an inductor with a gyrator is to order them as,
Rs + C + DCR + L where Rs is a real resistor in the circuit and DCR is the inductor's DC resistance.

The reason for doing  this is to make one end of the gyrator grounded.  This simplifies the type of gyrator circuits and avoids using more complicated floating gyrators circuits such as that used here, in
JC Maillet's  Acoustic 360 project,

http://www.lynx.net/~jc/pedals360.html

3)  Gyrator Example



As an example, take the 1.5H inductor,

R2  = DCR = 180 ohms
R1  = 100k   (high enough not to affect circuit but low enough not to cause noise)
C2  = 82nF
L  = R1 * R2 * C2 = 180 * 100k * 82nF  = 1.48H

C1  = capacitor in series with the inductor in the circuit. 
If there isn't one then, without knowing the specifics of the circuit, it is probably a good idea to put in say 10uF.


4)  FYI: Acoustic schematic bugs

I noticed there's bugs on the Acoustic schematics.   

Some people mentioned there were parts on their boards which weren't on the schematics.

eg.
the Acoustic 220 and 320 Equalizer schematics have missing resistors.
Most schematics contain "Rev B, Feb 1978, uses Equalizer 170084".
which shows

Band   Resistors
2000Hz   390R
820Hz   270R
350Hz   220R
125Hz   s/c      }   bug on this schematic.
70Hz   s/c      }

However the Rev, E, July 1978, 170084 EQ Schematic
specifically says "R27 to R29 Corrected" and has the parts
people see on their PCB's.

2000Hz   390R
820Hz   270R
350Hz   220R
125Hz   390R (R27)   }   the bug is fixed
70Hz   560R (R28)   }   schematic represents actual board

[PRR]

The coil series resistance is knowable by tank Q and L-C values.

In the example circuit, the Q goes up with frequency.

Winding coils is not a major pain. But the gauge-math may be.

With modern TL072 opamps, the gyrator series resistance can be a little lower than the calculated "coil resistance". This leaves room for the additional loss of Q due to the large value resistor.

[Rob Strand]

The coil series resistance is knowable by tank Q and L-C values.

For a circuit given as a schematic the Q is not documented and cannot be derived.

In the example circuit, the Q goes up with frequency.

Depends what is kept constant and what is allowed to vary.

With modern TL072 opamps, the gyrator series resistance can be a little lower than the calculated "coil resistance". This leaves room for the additional loss of Q due to the large value resistor.

Sure, but usually the series value is so much smaller than the parallel one you don't have to worry.  The only time you would worry is if you dropped the parallel resistance to reduce noise and the series resistance is relatively high.   Even if you compensate for the Q you can't get rid of the parallel resistance,  which can affect the circuit at high frequencies.

One thing about those equalizers, if we adjust the output to be a specific dB boost or cut, and we keep L and C the same, then the response is essentially independent of the DC resistance.  The boost/cut adjustment effectively hides the difference in DC resistance.   The only time you see a difference is when you set the pot to one extreme then the one with the higher DC resistance runs out of adjustment range!

[Transmogrifox]

There is also an element of parasitic parallel capacitance to be considered. You can estimate this using a similar method used to estimate coil resistance.  You have both coupling from windings-core-windings along with winding-winding capacitance.  A more advanced model includes series impedance to account for coupling paths.

In practice you can get a good idea for whether the parasitic capacitance makes a material difference at the selected center frequencies of the resonant tank. 

[PRR]

> For a circuit given as a schematic the Q is not documented and cannot be derived.

Of course.

For a "nice broad" bump assume Q~~1.

There is little use of Q>10 in audio.

Musician's EQs may fall anywhere in there, just like all other musical effects. Some guesswork needed.

> Depends what is kept constant and what is allowed to vary.

If they don't give you a tapped inductor, then the only "vary" is the C; L and R are constant. (Unless they switch an R with each C.) This gives the increasingly narrow curves shown in JC's article.

[Rob Strand]

Musician's EQs may fall anywhere in there, just like all other musical effects. Some guesswork needed.

In this case I used the +/- dB specs to narrow down the R somewhat.   I don't trust the specs 100% but they are a good guide to squash down the region of uncertainty.  Especially if you take enough samples over the different models.

It wasn't as easy as it sounds as there were cases where different bands on the same equalizer had the same valued inductor but different added series resistors.  So there's no way the equalizer could produce the spec'd +/- dB on both of those bands.

I noted the narrow bandwidth bands tended to have the lower resistors, and hence more boost.   So I suspect they tried to make the perceived level volume increase in each band the same as opposed to the same maximum +/- dB in each band (as the spec. implied).

Using that and the fact the wire had to fit onto the bobbin, I'm fairly confident the estimates narrowed below a 2 AWG step in wire.   If for some reason they used an odd number AWG that would be within +/- 20% for resistance.     Not enough to worry about, especially on that type of equalizer.

[rankot]

Rob, I happened to buy one not working Acoustic 220, and AFAIK, it has 1.5H inductors inside. I will soon have time to take care of it, and I promise to measure that inductor and post resistance here (in case it's working)!

[Rob Strand]

Rob, I happened to buy one not working Acoustic 220, and AFAIK, it has 1.5H inductors inside. I will soon have time to take care of it, and I promise to measure that inductor and post resistance here (in case it's working)!

Good to know.   The other thing that would be nice to know is the DC resistances of the inductors.  I struggled to back-engineer those, it's pretty much impossible to do with any confidence.

[rankot]

Glad to see you back online!

[Eb7+9]

The other thing that would be nice to know is the DC resistances of the inductors.  I struggled to back-engineer those, it's pretty much impossible to do with any confidence.

Rob, those Acoustic 1.5Hy coils have about 240 ohms of parasitic resistance

[Rob Strand]

Rob, those Acoustic 1.5Hy coils have about 240 ohms of parasitic resistance

Thanks!   That's a lot higher than I expected.   The tables above only get to 182 ohms with reasonable assumptions of a P14/8 core, 44AWG wire, N27 material.  The 3.0H hits 268ohms.  So maybe the cores have very small gaps?  Philips offered small gaps but perhaps not in that era.

[PRR]

> very small gaps?

Cheap wire.

Most coil uses want low resistance. But audio chokes are usually low-Q. You can use a high-Q choke and add a resistor. If you custom order them, it makes sense to go thinner on the wire, save a penny there and another 11 pennies on a resistor.

You can do the math and add a resistor.

[Rob Strand]

Cheap wire.

Perhaps.  That's a far easier explanation.

I was assuming 44AWG was about as thin as you would like to go before it is a pain to wind (which can indirectly push up the costs).  Also 44AWG is near the edge of basic wire gauges.    The problem is it's an assumption.   Those miniature transformers could very well use smaller wire than 44AWG and they make those all the time, and cheap too.

EDIT: Actually, 46AWG will give higher resistance than 240ohm, so we don't need much thinner than 44AWG.  That might explain it.

[rankot]

Hi Rob, did you build a preamp clone, or just tried to fix an original one? I am struggling with my clone, it's working really fine, but it's noisy. Used BC550B transistors instead of unobtanium originals, but they're low noise. Tried both inductor and gyrator versions and they're almost the same, although inductor based seems to be a little bit trebly.

[Rob Strand]

Hi Rob, did you build a preamp clone, or just tried to fix an original one? I am struggling with my clone, it's working really fine, but it's noisy. Used BC550B transistors instead of unobtanium originals, but they're low noise. Tried both inductor and gyrator versions and they're almost the same, although inductor based seems to be a little bit trebly.

I didn't build a clone I was just trying to work out what these things *really* are.

Gyrators can be a little noisier (hiss-wise) than real inductors but since you aren't seeing a big difference the problem is elsewhere.  The EQ is at the end of the signal chain so while the EQ can *affect* the noise it's probably not the thing *causing* the noise.  The noise could originate from previous stages.   In many circuit changing the transistors won't change the noise because the noise is determined by the circuit design itself and the physics of the transistor.   Noise often comes from first stage.

One way to get idea of a noisy stage is to short the input of the stage.   Then, disconnect that stage from next and short the input cap of that stage to ground.   Keep moving up stages until you find where the noise stops.

h inductor based seems to be a little bit trebly.

I would only expect that to be the case if the inductor and the gyrator aren't really the same value, or have the different equivalent  series resistances.   Imagine if the real inductor version had more boost and/or the peak frequency was higher, it would seem more trebly.  You should be able to see the differences by measuring the frequency response.

[rankot]

Thanks for advice, I'll try to isolate noise in stages.

[jonny.reckless]

I typically use a slightly different topology to simulate series LCR. This does away with the parallel resistance, and allows you to have a fairly high Q pretty easily. The minimum resistance is now the sum of the two resistors. Here's an example (log-log) plot of impedance vs frequency. An op amp can be used as the unity gain buffer. You do however get a phase shift of pi (rather than zero) at resonance, but this doesn't really matter too much in most of the circuits I use it for. It works just fine for a frequency boost (or cut) in a graphic EQ circuit, for example.

[Rob Strand]

I typically use a slightly different topology to simulate series LCR. This does away with the parallel resistance, and allows you to have a fairly high Q pretty easily. The minimum resistance is now the sum of the two resistors.

Those work best when the series resistance of the inductor is high-ish (say over 1k to 2k).

IIRC with transistor buffers and series resistances on the low side the parallel resistance starts to re-appear.

At this point the problem looks elsewhere.

[rankot]

Gyrators can be a little noisier (hiss-wise) than real inductors but since you aren't seeing a big difference the problem is elsewhere.  The EQ is at the end of the signal chain so while the EQ can *affect* the noise it's probably not the thing *causing* the noise.  The noise could originate from previous stages.   In many circuit changing the transistors won't change the noise because the noise is determined by the circuit design itself and the physics of the transistor.   Noise often comes from first stage.

One way to get idea of a noisy stage is to short the input of the stage.   Then, disconnect that stage from next and short the input cap of that stage to ground.   Keep moving up stages until you find where the noise stops.

The first stage is a simple buffer, so I don't think it adds noise. I will start from the second stage, because my o-scope is not sensitive enough to see noise that is less than 2mV, and, according to that, there's no noise in the first stage.

This is my PCB (for gyrator version, inductor based one is very similar). Maybe the layout is a problem?

[Rob Strand]

The best plan is to narrow down the cause.

The first stage is a simple buffer, so I don't think it adds noise. I will start from the second stage, because my o-scope is not sensitive enough to see noise that is less than 2mV, and, according to that, there's no noise in the first stage.

Tracking down noise (hiss) with a CRO can be a bit unreliable.  Some CROs produce lot of noise themselves.  If you have a low bandwidth or filter setting on your CRO enabling that can reduce the CRO noise. Also you can pick-up a lot of noise above audio which makes the audio noise hard to see.  Ideally you need to have a x10 or more preamp and a 10kHz to 20kHz 2nd order filter between the probe and the CRO.    To see noise, turn off averaging if you have that setting.

Power supply noise could affect the buffer.  Especially with the bias divider on the input.
JC Maillet's  page mentions he added an active filter on the power rails in his later version.

With the input shorting trick the noise is probably best judged by ear.   Narrowing down the noise to a preamp stage (or stages) is a very helpful moving forward.

This is my PCB (for gyrator version, inductor based one is very similar). Maybe the layout is a problem?

What stands out to me is the switcher on the supply.

I doubt the layout is causing noise in the audio circuits themselves.   However, with the switcher present, the layout could have an impact.     The layout of the switcher part could also have an impact. The switcher can create noise on the power rails and also couple noise into the audio circuits.

To completely remove any causes related the switcher it's probably best to:
- remove the switcher inductor; so when the switcher is disabled it doesn't short across the supply
- disable the 555 from oscillating
- power the unit from a clean external supply.

If you get a significant change doing this them the cause of the noise is related to the switcher.

If the noise is still present it's probably a wise move to debug the noise problem with the clean external supply in place.