Transformer Coloring Box Again

Help Support GroupDIY:

apzx

Active member
Joined
Apr 1, 2017
Messages
25
I've done a bit of searching, but did not really find too much on how to really go about this beyond it is pretty easy. I have not really played much with transformers, and this is really my first attempt at trying to work with them. I could very much have messed something up in that aspect. So, please try and forgive my ignorance regarding transformers :)

Below is my initial thought process for this. Fine tuning to come at a later date.

1kFh6wa.png


I'm looking to use Edcor transformers in this because they are relatively easy to obtain and very affordable. The intention is for some tasteful distortion and for the lulz distorting/clipping the transformer.

For the input I'm looking to use the PCW10K/10K because I want it to be handle most any reasonable signal that is thrown at it which for a +28dBu rated device I think should be okay. I also want to use this transformer to debalance the input signal, and I think that is how it is supposed to be wired up...? The primary & secondary side of the Edcor PC & PCW transformers is center tapped and they do say either unbalanced or balanced. Anyway, following that is a bridged-T attenuator so I can keep the load impedance of the transformer roughly constant (in this case I'm aiming for about 11K-9K) so that it does not adversely affect the load presented to the primary side. I'm also not sure if the transformer will require a Zobel network, so for the moment I've just got the parts there, but no values or anything. I'm just trying to get the basics sorted out right now.

Following this is the gain stage, which I have decided is going to be a DOA (have not decided which yet though), and right now it is driving a PC10K/10K because if I want to clip the transformer (got to leave your options open, you know?) it is a much easier ask to get 5Vrms than it is 20Vrms. Right now, if I have done my math correctly-ish then with a gain of about 21dB I only need about 650mVpk to get about 7.15Vpk on the output, which starts exceed the ratings on the transformer. This would equate to an attenuation in the attenuator of roughly -34.6dB. I also want to use this transformer to balance the output, and again I think I have it wired correctly for the purpose...? Again a zobel network is drawn because I am not sure if I am going to need it or not. I am aware that the output of the DOA should be AC coupled into the transformer to remove any DC bias that may be present (primarily because I just forgot to add it in this drawing...). I am also aware that for a bridged-T attenuator if I want to change the attenuation settings things get a bit complicated. I've settled on the idea of using a rotary switch to alter attenuation settings, but not shown here for the sake of simplicity. Besides that I am still debating on the best way to fine tune the level coming out of the DOA.

With this all said though I am left with some questions. Do I need to AC couple the input transformer, or is it more of a situation of where I should kind of thing? The other big question that I am having has to do with the impedance presented to the transformers. I get that they reflect their impedances from the secondary to the primary. What I do not fully understand is how say 10K/600 transformer would work on the output if it is presented with say 10K input as its load? My gut tells me that the signal level would decrease, but if I opt for say something like the PC600/10K then signal level should increase? At least that is my newbie transformer understandings. The last thing I am bit concerned with is driving the output transformer with too much level. This would very likely run on +/-15V maybe +/-18V and that means the signal can be well in excess of the 5Vrms rating of the output transformer. But I'm getting a sneaky suspicion that the main thing I really need to be concerned with here is how much the transformer is going to be loaded more so than the input voltages...
 

ruffrecords

Well-known member
Joined
Nov 10, 2006
Messages
13,528
Location
Norfolk - UK
Looks like a good start. I would replace the bridged T with a 10K pot and series capacitor to the op amp and fit the latter with it own bias resistors. On the other hand you could be adding 20dB of gain to the input signal so it might be an idea to add a T attenuator on the output to restore the level to sensible levels. Your basic setup is almost exactly how I build my 'tube tone boxes' but for them the idea is to drive the tube harder but it does also drive the output transformer harder (but I use a VTB2291 which can take it).

Rupert Neve's 'Silk' circuit I believe adds a little dc current to the output transformer. This shifts the magnetic bias so the transformer clips unevenly which tend to create odd harmonics.

Cheers

Ian
 

apzx

Active member
Joined
Apr 1, 2017
Messages
25
Very interesting! I got to thinking a bit more and as I said my transformer theory is not the strongest. I am trying to figure it out a bit more as I stare thinking about this.

The reason I opted for the Bridged-T attenuator is because I incorrectly assumed that I would want to keep the load impedance about the same. But then I broke out the basic transformer maths and for a 1:1 input transformer it really does not matter much as long as the load is high enough in the first place. The high gain on the opamp was there more for amplifying lower level signals to the point where it could more easily saturate/clip the transformer. I also selected the lower rated output transformer on purpose specifically for the possibility of clipping it if desired. The PCW series as I said before is rated for 20Vrms whereas the PC is rated for 5Vrms. It is much easier to get 5Vrms instead of 20Vrms and that is the reason for my selection there.

I have revised my idea a bit based on your suggestions and a bit more thought on my part.

V04WYGO.png


The input is still the same. After that is a rotary switch that selects a resistor to go in series with a 10K pot for adjusting the output level of the opamp. The values are kind of arbitrary-ish, but mainly there to extend the range of input levels. For the opamp itself I added a hi/lo gain switch. Finally, I really don't know the best way or even the right amount of voltage to add a small DC bias to the transformer. I wonder if it is a better idea to apply that through the center tap rather than the high side. I get the feeling that if I apply a voltage to the center tap it would essentially cause the lower tap of the transformer to be shut out or cancelled out in some fashion (I don't know the appropriate verbiage here). I've got +2.5V right now through a 1K resistor feeding into the primary input, but that is easily adjusted and will probably need to be done experimentally to find a voltage and resistor value.
 

ruffrecords

Well-known member
Joined
Nov 10, 2006
Messages
13,528
Location
Norfolk - UK
I have never tried adding dc to a transformer either but I have seen the Neve silk schematic on line somewhere so that might be a good starting point.

Cheers

Ian
 

apzx

Active member
Joined
Apr 1, 2017
Messages
25
I have never tried adding dc to a transformer either but I have seen the Neve silk schematic on line somewhere so that might be a good starting point.

Cheers

Ian
Thinking on this some more, it really doesn't inherently make sense to apply a DC bias voltage to the transformer. At least to my mind because an inductor should just pass that to ground. Would it not make more sense to actually bias the primary of the transformer with a small DC biasing current from a simple constant current source?

Running a constant current through the transformer should saturate the core more readily, and be easier to control. Unless I am way out in left-field with my thinking here.

Edit - Actually, having done a bit more reading on things (granted in the context of power transformers and such), it seems that applying a DC offset would be the correct action here rather than a current source. :unsure:
 
Last edited:

ruffrecords

Well-known member
Joined
Nov 10, 2006
Messages
13,528
Location
Norfolk - UK
Thinking on this some more, it really doesn't inherently make sense to apply a DC bias voltage to the transformer. At least to my mind because an inductor should just pass that to ground. Would it not make more sense to actually bias the primary of the transformer with a small DC biasing current from a simple constant current source?

Running a constant current through the transformer should saturate the core more readily, and be easier to control. Unless I am way out in left-field with my thinking here.

Edit - Actually, having done a bit more reading on things (granted in the context of power transformers and such), it seems that applying a DC offset would be the correct action here rather than a current source. :unsure:
This is an area of transformer theory I am still not 100% clear about but a dc current flowing in the winding will cause a static flux in the core. Because the flux has direction, just like current has direction, this static flux acts like a dc offset so one half of the waveform the core saturates at a lower level than the other half. CJ knows all about this so maybe he will chime in. So I think the current source idea is the way to go. As I said, I am sure I have seen it in the Neve schematics so I would suggest you have a look at how they did it.

Cheers

ian
 

apzx

Active member
Joined
Apr 1, 2017
Messages
25
All right, so I've had a quick gander at this silk circuit and it is way more involved than I thought it'd be. I will have to ponder on this one because it has some feedback through the transformer. But the DC bias part of it is pretty simple, just applying a small amount of voltage to the non-inverting input of the opamp. Not 100% sure, but I think the other aspect of simply a tuned circuit to help shape the feedback a bit for well, I guess more silk.

Thank you Ian, you've been incredibly helpful with demystifying some of this transformer stuff for me. I've got lots more to think on now. :unsure:
 

apzx

Active member
Joined
Apr 1, 2017
Messages
25
Well I placed the order to Edcor yesterday. So, in a couple of months I ought to have the transformers on hand to test the ideas for biasing the transformer. Gives me some time at least to contemplate some ideas. Right now I have AC coupled the transformer and from what I've been able to surmise adding a DC bias is not as easy as I thought it'd be. So, I'll just have to head scratch and try some different things. I figure as long as I am a bit careful about the amount of current I let flow I shouldn't blow up the transformer.
 

ruffrecords

Well-known member
Joined
Nov 10, 2006
Messages
13,528
Location
Norfolk - UK
The formula for calculating the flux in a transformer from the number of turns, the core size, its material and the applied voltage has a 1/f term which is why transformers saturate more easily at low frequencies. But this means that at zero frequency i.e. dc, the flux would be infinite. The confusing thing for me is that the flux is related to the magnetizing for by the BH curve (B is the flus and H the magnetising force) and H is measured in ampler turns so maybe this is the formula you use for dc. I don't know. Some aspects of transformers are still a mystery to me.

Cheers

Ian
 

abbey road d enfer

Well-known member
Moderator
Joined
Jan 22, 2008
Messages
12,644
Location
Marcelland
The formula for calculating the flux in a transformer from the number of turns, the core size, its material and the applied voltage has a 1/f term which is why transformers saturate more easily at low frequencies. But this means that at zero frequency i.e. dc, the flux would be infinite.
That would be if the total DCr of the circuit was zero. The magnetizing current at DC is the simple result of U/R.
 

apzx

Active member
Joined
Apr 1, 2017
Messages
25
Actually, the magnetizing current isn't that simple (I've done a lot of reading today haha). But it really does boil down to Ohm's Law with some additions namely in the form of non-ideal properties of the inductors in use, and also looks really familiar to the equation for finding the current in a LR series circuit. I did find a derivation for the saturation current that requires knowing the saturation limit of the core material, how long the winding is, the magnetic permeability of the core material, and the number of turns.

Isat = (Bsat * Lm) / (u*n)

I'm sure it'd be possible to estimate the length by guessing the gauge of the wire and the DCR of the primary. The number of turns can also probably be estimated. The other two values are available from Edcor. Knowing this would allow for the initial back of the napkin estimating values for the DC biasing circuit.
 

abbey road d enfer

Well-known member
Moderator
Joined
Jan 22, 2008
Messages
12,644
Location
Marcelland
Actually, the magnetizing current isn't that simple (I've done a lot of reading today haha). But it really does boil down to Ohm's Law with some additions namely in the form of non-ideal properties of the inductors in use, and also looks really familiar to the equation for finding the current in a LR series circuit. I did find a derivation for the saturation current that requires knowing the saturation limit of the core material, how long the winding is, the magnetic permeability of the core material, and the number of turns.
You're making things more complicated than need to.
We're talking about DC current. Inductors just do not existe at DC (as well as capacitors.)
 

skipwave

Well-known member
Joined
Jul 8, 2005
Messages
1,378
Location
Los Angeles
To calculate the amount of current that will flow through the primary winding, as you say it’s simply Ohm’s Law. The DCR of the primary winding as the divisor of the volts. V/R = A

The impact of any change exact amount of current flowing on the flux and therefore asymmetry of clipping does not seem useful to calculate. Try it with a variable voltage source, simple pot as voltage divider will do, and twist the knob until it a sounds rad.

I’ve been wanting to do this for awhile, particularly because all the vintage Triad transformers I have specify a DC current limit in the primary, which I just don’t know whether it was because builders were making small signal Class-A stages to maybe reduce part count or something else.
 

Michael Tibes

Well-known member
Joined
Jun 5, 2004
Messages
776
Location
Berlin, Germany
I have revised my idea a bit based on your suggestions and a bit more thought on my part.
A small detail: I'd reconsider S2. The way you designed might cause a short moment of 'no connection' when you flip the switch, leaving the opamp without feedback which might result in a big pop or whatever noise. Maybe always keep R10 and only switch R7 (needs t be recalculated) parallel to it?

Are there any practical experiences with these concepts yet or has it only been a theoretical discussion so far?

Michael
 

SWAN808

Well-known member
Joined
Sep 13, 2009
Messages
85
Which discrete op amp are you thinking about using? That will impact the sound to a degree. I have the DIY Colour CTX Mk1 and Mk2 and the LTL Mass Driver (which uses an Edcor)...they are similar but the difference is in the way they start to clip...the discrete clip sooner and nicer...before that they are very similar...
 

apzx

Active member
Joined
Apr 1, 2017
Messages
25
To calculate the amount of current that will flow through the primary winding, as you say it’s simply Ohm’s Law. The DCR of the primary winding as the divisor of the volts. V/R = A

The impact of any change exact amount of current flowing on the flux and therefore asymmetry of clipping does not seem useful to calculate. Try it with a variable voltage source, simple pot as voltage divider will do, and twist the knob until it a sounds rad.

I’ve been wanting to do this for awhile, particularly because all the vintage Triad transformers I have specify a DC current limit in the primary, which I just don’t know whether it was because builders were making small signal Class-A stages to maybe reduce part count or something else.
It is more of a curious question on the amount of flux that is actually biasing the core of the transformer. What will end up happening more than likely is tweaking it by ear to come up with the values.

I did change out to a voltage source (LM4040 feeding into a pot) and I ended up opting for DC coupling the DOA to the transformer, and adding a trimmer pot so I can adjust the offsets. All of this gets added to the non-inverting input of the DOA. I may also do a DC servo. I am a little undecided right now. But then again I am still waiting on the transformers so I still have some of the finer points to think on, but overall the general idea is there.

A small detail: I'd reconsider S2. The way you designed might cause a short moment of 'no connection' when you flip the switch, leaving the opamp without feedback which might result in a big pop or whatever noise. Maybe always keep R10 and only switch R7 (needs t be recalculated) parallel to it?

Are there any practical experiences with these concepts yet or has it only been a theoretical discussion so far?

Michael

Yeah I forgot about that tidbit. Last time I did something with variable gain like this was through a rotary switch that was MBB. It is an easy thing to tweak though. Not a hard fix, but thank you for the reminder!

Which discrete op amp are you thinking about using? That will impact the sound to a degree. I have the DIY Colour CTX Mk1 and Mk2 and the LTL Mass Driver (which uses an Edcor)...they are similar but the difference is in the way they start to clip...the discrete clip sooner and nicer...before that they are very similar...

Right now I am looking at the NTP M100 actually. But I know that the interaction between the DOA and the circuit it is in can influence the behavior a fair bit. Though I am also thinking that maybe an AM10 might be fun in this. But I am not 100% set on which just yet again because the transformers have not yet arrived. So, in the thinking phase still.
 

Latest posts

Top