AKG BX20E Compensation Circuitry

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Bo Deadly

Well-known member
Joined
Dec 22, 2015
Messages
3,266
Location
New Jersey, USA
Someone mentioned the AKG spring reverbs in another thread and how the dual driver/recovery coils at each end of the spring are used to cancel out correlated signals and thereby leave only the signals with no coherence whatsoever. That is fascinating. I found a document that describes this [1]. Here is a quote:

As can be seen from the block diagram (figure 6) the electronic portion, for each channel, consists of an input amplifier, an output amplifier, and two attenuation amplifiers. The spring is excited by feeding the original signal into one coil half of each of the moving coil systems arranged at either end of the spring.  o The "dry" signal is fed in phase, to each moving coil half on the ends of the spring, the reverberated signal is picked up by the two remaining coil halves, amplified, and connected in opposite phase. In this manner compensation of all interdependent and related Signals is achieved, TTL being a reciprocal and passive transmission system.

To illustrate this: If the circuit were to be applied to an ideal line the reflections would all be in phase at the output and would be cancelled by the out of phase connection of the output. However, since the transmission properties of the spring are statistically changed, it produces irregular, statistical, frequency independent reflections which appear at the ends of the spring with no coherence whatsoever. Therefore, these signals will not be cancelled at the out of phase output.


Note: There are no figures in the document but you can find pics in Google Images.

However I don't fully understand this because I would think that any signals would be convoluted by the spring by the time it reached the other end and so there would not be much correlation between the two ends anyway. Can anyone explain this cancellation a little more?

I wonder how effective it really is and if it could somehow be employed in a custom unit using conventional music wire springs. Of course one would have to craft the very small magnet wire coils that make up the transucers suspeded between the C magnets but there are folks around here that are good at making coils.

[1] http://lcweb2.loc.gov/master/mbrs/recording_preservation/manuals/AKG%20BX%2020E%20Reverberation%20Unit.pdf
 
Hi Square,
Interesting post , I did a bit of digging around and found a link to an old post in here .
https://groupdiy.com/index.php?topic=37106.0
theres a good photo in that link of the transducer coils.
looks like the coils are wound interlinked  a bit like a chain.
I did a bit of an experiment one time connecting an accutronics spring to the output of a 100w hiwatt,
then patching the spring output back to the second input on the amp,
It is definately possible to increase the reverb dwell time with the  feedback that occured ,
an attenuator on the spring input is essential in this situation! Of course push it too far and your transducer will go up in a puff of smoke .
I was wondering if it might be possible to mount both an input and output coil, each with their own set of laminations, on both ends of an acutronics tank ,seems like one of the people in the old post had a custom coil arrangement that acutronics were prepared to make for him ,unfortunately the link for this is now gone.

 
They're not interlinked. The document I linked to explains:

Each moving coil system consists of two coils which have a rigid mechanical connection between them and which vibrate in a strong magnetic field. The two coils are electrically and magnetically decoupled. The signal is fed into one half and picked up at the other.

They do look like links in a chain but I don't think they can be interlinked because that would make each the core of the other and they would be very much coupled magnetically. My guess would be that they are simply held together end-to-end using epoxy. Or maybe there's a bit of plastic or something to relieve the tension.

To make one of these driver/receiver transducer coil pairs I would just wind a small chain-link style coil with epoxy around an flattened chain-link shaped mandrel that the epoxy will not stick to. After it dries, it slides off.

Then they look like they are epoxied together end-to-end but rotated at 90 degrees. I guess this minimizes magnetic coupling? Now you have 4 wires sticking out that attach to what at first glance looks like an old TO-1 IC can:

normal_AKGbx20-DettagliFolli4.JPG


But my guess is that it's really just an insulator that ensures the 4 wires are held close to the axis of rotation and don't hinder the rotation of spring.

Now do this at both ends and you have a driver and recovery transducer at each end. With proper C magnets that would be enough to at least study the circuitry that does the out-of-phase cancellation of correlated signals.

To make a real solution the spring would have to be very long and change diameter and wire thickness and there would need to be mechanical dampening disks. Difficult but, with the right tools and skills, not impossible for an individual.
 
squarewave said:
"In this manner compensation of all interdependent and related Signals is achieved, TTL being a reciprocal and passive transmission system."

However I don't fully understand this because I would think that any signals would be convoluted by the spring by the time it reached the other end and so there would not be much correlation between the two ends anyway. Can anyone explain this cancellation a little more?
At low frequencies, the springs have similar transmission properties and in fact most of the initial transient response is very similar, because it's mainly related to the spring length and sound speed in the spring material. The output of the spring is a combination of correlated and uncorrelated signals.


I wonder how effective it really is and if it could somehow be employed in a custom unit using conventional music wire springs. Of course one would have to craft the very small magnet wire coils that make up the transucers suspeded between the C magnets but there are folks around here that are good at making coils.
This can be achieved using standard Accutronics tanks, using two tanks driven by identical signals and connecting the outputs out-of-phase. That won't be as efficient as in the BX20 because the manufacturing tolerances are much less stringent than they were at AKG.
 
Interesting that the patent was allowed to lapse ,I suppose the incoming digital technology made it uneconomical to build these things anymore. I wonder what the price tag would be if they still made them ,big money I'd imagine.
I have a bunch of old springs and parts from broken accutronics  tanks as well as several working tanks ,I was thinking of maybe converting the output transducer to a drive transducer and adding pickup coils and magnets at each end like the Akg. Not sure if its going to be possible to use the massive heavy magnets like in the akg but maybe smaller powerfull rare earth ones might do the trick,Im also not sure if this kind of set up would have the degree of magnetic seperation between inputs and outputs required to work properly .Any suggestions as to a source for suitable magnets would be great .
 
Although the BX-2 tube used in a number of AKG reverbs looks like it only has one transducer coil and not the two used in the feedback drive of the BX20E:

AKGBX2_4.jpg


and those units sound pretty good. There's a u tube video of a BX-15 that sounds pretty amazing.

I think it could be worthwhile exploring the coil suspended inside a magnet arrangement instead of the usual Accutronics transducers. But it is a very delicate arrangement. That and patents might be why Accutronics used the alternative mangnet suspended inside a coil method in which case doing it the AKG way might turn out to be better in some way.

Can you cut a neodymium magnet? More specifically, take a ring shaped neodymium, clamp very securely to some kind of metal cradle and then pass it into a metal blade on a table saw very slowly with lubricant that might make a good strong C magnet.

The dampening disks would be a problem though.
 
Im not sure what effect cutting magnets might have on the properties. I do know that mu metals properties degrade if its subject to stress .Like you mentioned a method of reducing the amount of heat generated is probably a good idea .
I was just looking at a long tank accutronics I have ,I might try mounting a coil in the middle ,where the two springs join on each line ,that might lessen any magnetic coupling between drive and recovery circuits .Maybe an old set of headphones or a small speaker might have an ndym ring magnet in them ,wouldnt really cost anything so would  be fine at least for test purposes.
One other thing that sprung to mind ,scuse the pun, was that if your driving the springs from both ends Im guessing you would need the direction of rotation to be opposite at each end
The quest continues.......
 
Actually thinking about this again one cannot use just any magnets. Most flat magnets (even the curved ones for motors) have polorized faces. A proper "horseshoe" or "c" magnet would be ideal but it doesn't look like a popular form factor for neodymium.  Three straight bar magnets or maybe wedges that have ends polorized and oriented as a simple U might work.
 
> cut a neodymium magnet? ... make a good strong C magnet.

Most flat face ring magnets are magnetized face to face. When we think C-magnet we usually want it magnetized end-to-end.

Keep the ring and fab two pole pieces to make a C.

EDIT - squarewave posted seconds faster.
 
Back
Top