Multiple secondaries… into two different loads?

[systemtruck]

Hey all,
Been searching for an answer to this. Let’s say we use a transformer with multiple “1” secondaries, and we send those to different loads? What gets reflected back to the primary? Simply the lowest loaded secondary’s impedance?

Case example.. in the context of a mixing console, I’m considering putting something like a Carnhill VTB 2280 / 2281 at the end of a line channel, before the passive summing resistor network. Then one secondary could continue to feed the mixing path, and the other extra secondary could provide a direct output of said channel/preamp. And if I were to discover an affordable transformer that offered even more, like a 1:1:1:1, the extra third secondary could feed an Aux bus.

Direct out would see 10k-15k of an audio interface, where as both the aux pathway and mix pathway would each see a 22k summing resistor amongst others that feed into a 12AY7 grid.

What happens to the primary impedance and to the transformer performance overall?

 

[metalockpick]

You'd get a complex load. Everything would be proportionally reflected back, less any losses.

The standard API output transformer (excellent versions are available from cinemag, UTM, and others, with Lndahl offering something similar) have four identical secondaries that can be combined or split, and will allow a 10: 1 Miss match before the sound starts getting screwy.

 

[systemtruck]

You'd get a complex load. Everything would be proportionally reflected back, less any losses.

The standard API output transformer (excellent versions are available from cinemag, UTM, and others, with Lndahl offering something similar) have four identical secondaries that can be combined or split, and will allow a 10: 1 Miss match before the sound starts getting screwy.

Not the API 2503 I assume, since that is 75 prim : 75, 300, 600 sec. Seems maybe usable though.
Any idea what the model is of the one you’re referring to?

 

[ccaudle]

What gets reflected back to the primary? Simply the lowest loaded secondary’s impedance?

The governing principle is conservation of energy, so whatever power (voltage x current) is supplied by all of the secondaries combined has to be provided by the driver of the primary winding.
In your example of 1:1:1:1 with 10k on one winding, and 22k/22k/22k on the other windings, it would be the same as if there were one secondary driving the parallel combination of those loads, or 4.2k. That would then be reflected back to the primary.

 

[systemtruck]

Understood. I think my problem may be that these 22k resistors are a passive mixing network that then enter directly into a tube grid. The summing amp design I’m copying doesn’t have a resistor to ground at the grid entry. It’s just right on into the grid. So wouldn’t these paths not have just a 22k input impedance but rather something like an impedance in the megaohms?
This being so drastically different than the 10k pathway makes me think that perhaps a transformer is not the answer for having line splits unless I can adjust the pathways’ impedances to be much closer. Or, use switches to only engage pathways that play together nicely.
But for what it’s worth, here’s a 1:1+1+1
https://edcorusa.com/products/xs440...ching-transformer?_pos=1&_fid=fde08e629&_ss=c

 

[systemtruck]

Understood. I think my problem may be that these 22k resistors are a passive mixing network that then enter directly into a tube grid. The summing amp design I’m copying doesn’t have a resistor to ground at the grid entry. It’s just right on into the grid. So wouldn’t these paths not have just a 22k input impedance but rather something like an impedance in the megaohms?
This being so drastically different than the 10k pathway makes me think that perhaps a transformer is not the answer for having line splits unless I can adjust the pathways’ impedances to be much closer. Or, use switches to only engage pathways that play together nicely.
But for what it’s worth, here’s a 1:1+1+1
https://edcorusa.com/products/xs440...ching-transformer?_pos=1&_fid=fde08e629&_ss=c

I suppose I could use that Edcor, or similar 1:1:1:1, and force the output of each winding down to a cumulative 600 with the appropriate resistor per winding. I wonder how that would impact the passive mixing resistor network that follows two of the winding outputs. All mixer inputs would have this shunt though, so at the very least the consequence would be even across all channels. Just not sure what a 600 ohm shunt prior to a 22K series resistor will do to the sound or level.

 

[Walrus]

Not the API 2503 I assume, since that is 75 prim : 75, 300, 600 sec. Seems maybe usable though.
Any idea what the model is of the one you’re referring to?

Incorrect, the 2503 has 3 identical secondary windings at 75 each.
It is only when you combine them is series do you get 300 or 600.

 

[systemtruck]

Incorrect, the 2503 has 3 identical secondary windings at 75 each.
It is only when you combine them is series do you get 300 or 600.

I see. Wouldn’t that amount to 75, 150, 225 though? I must have my transformer winding principals math off I guess.
Also, 75 seems like modern solid state output. Think this transformer behaves fine in the territory of 600:600 tube signal?

 

[CJ]

inductance goes up as a quadratic function of the turns count,

so if you double the turns, you get four times the inductance

if you triple the turns, you get 3^2 = 9 times the inductance

look into hybrid transformers if you want to get you brain blown up as to reflected signals,

 

[systemtruck]

inductance goes up as a quadratic function of the turns count,

so if you double the turns, you get four times the inductance

if you triple the turns, you get 3^2 = 9 times the inductance

look into hybrid transformers if you want to get you brain blown up as to reflected signals,

Ah yes, that was the missing factor. Makes sense thank you!

 

[ccaudle]

these 22k resistors are a passive mixing network that then enter directly into a tube grid.

Makes sense now, I originally misread how the 22k resistors were configured. In that case it is not a 22k load, it is actually a high impedance, so very little current would be supplied.

and force the output of each winding down to a cumulative 600 with the appropriate resistor per winding

There may be no reason to do that. Some transformers have a non-flat frequency response into a very light load, and some do not. Some are somewhere in between (see for example the Jensen transformer recommendations that do not have a minimum recommended load on the output transformers, but some of the line input transformers show a resistor/capacitor series connection across the secondary for cases where the load does not match the design load).

 

[moamps]

Makes sense now, I originally misread how the 22k resistors were configured. In that case it is not a 22k load, it is actually a high impedance, so very little current would be supplied.

Is it really so? Won't the other channels with their isolation resistors load that output? IMO, in this case I would calculate that the output transformer load is 22kOhm.

 

[systemtruck]

Time for a schematic because this is a bit complicated without it!
I didn’t mention a pan pot before, and didn’t clarify much about the Aux/Bus scenario.
The 22K resistor value is what is used for 9 channels of these preamps feeding a bus preamp, so that’s why I have 22K. This is from and RCA BC-6A console.
The mathing here is all labeled. Thinking that these values in pots and shunts creates 1800 load on each of the three secondaries, which amounts to a parallel load of 600 reflected to the primary. The output of the preamp circuit, I believe, is a 550 ohm cathode follower.
Any thoughts?
I’m figuring since i already need to choose pan pot values, which effectively is a shunt on two of the three secondaries, then I may as well shunt the third secondary that goes to the audio interface and then all secondaries have the same load.

 

[Admiral-dk]

Similar .... I have often connected speaker cabinets of two different Impedances to the same amp (if it is one of those with multiple Outputs).
The 16 Ohm speaker goes into the 8 Ohm Jack and the 8 Ohm Speaker goes into the 4 Ohm Jack - resulting in the Amp having the correct load
This is NOT a joke - but a perfect way to do i correct.

Must admit that I haven't got a clue about Line Level Gear - except that if you want to connect two 600 Ohm Inputs to the same Output - that Output should be 300 Ohm (if possible) ...!... but as most modern equipment has much higher Input Impedances - you are more likely to need to put a Dummy Load across the Output to get down to the 600 Ohm (if nessesarry) ....

Per

 

[systemtruck]

Edit:
Have updated the schematic above to include a forgotten AUX SEND pot, which the secondary C sees.
Also added values to the output of that Aux mixer. I put a 600 load there for consistency, since that’s what I’m loading down the 8 preamp circuit outputs with.
The big question at this point…
Do these RCA circuits work ok with a 600 ohm load on their 550 out cathode follower output?
The goal of this design was to provide a consistent load on the 1:1:1:1 transformer’s secondaries.
Is the better move to present a much much higher load? Formally the cathode followers went straight into the 22K resistors of the passive mixer stage. No pan pot. Just straight to the grid, essentially.
So can these circuits still function with load matching 600 to 600 at their outputs? Or are they meant to see a wall of high impedance?

 

[ccaudle]

Is it really so? Won't the other channels with their isolation resistors load that output?

I think you are right. I was thinking of a virtual ground summing node like often used in solid state designs, but I do not think that is how tube designs work.
With the additional schematic provided in post #13 I think it would be a complicated load to determine. If the other channels were 0V it would be 22k in series with all of the other channels in parallel, but with signals in the other channels the current supplied by one channel will depend on the voltage difference between the channels. The tube itself would be a high impedance load, but the other channels would have a changing effective load depending on how correlated the material between channels is.

 

[systemtruck]

I got in touch with Cinemag about 1:1:1:1 transformers and he suggested the CMOQ-2S, which is actually quite affordable, and not as bright sounding as the API 2503.

I mentioned this transformer stage structure, the one used in schematic in post #13, and he suggested not loading down the secondaries at all since the cathode follower’s overhead could be reduced. I mentioned that the two tube entry points could be high impedance, and that the third would be 15K for an audio interface input. He says the frequency response of this transformer should not be afffected much, maybe not at all, in this configuration.

So now i’m left trying to confirm… are those tube inputs actually high impedance in the megaohms like i think they are? Or are those input impedances determined wholly by the mix resistor networks and perhaps somewhere in the ballpark of 22-25K each?

That would actually be pretty good news if the resistor network is what determines the input impedance in those two spots. Because then all three input impedances from the secondaries would be kinda in the same ballpark. Although, now I’m not even sure any of this makes that much of a difference so long as I’m not loading down the cathode follower too much. Seems like maybe I’ll end up reflecting back to the cathode follower something around 5K-6K, considering a parallel 25K, 25K, 15K. Or a 22K, 22K, 10K (depends on interface) totaled amongst the secondaries. And if the two tubes are high impedance, then I’ll reflect back to the primary something like just under 15K.

As for material in other channels… they would be generally not related. The only time they’d be sorta related is when tracking a whole drum set. Otherwise, typically one or two channels would be active at moments when the rest of the other channels are completely inactive of any audio material.

 

[Admiral-dk]

As long as the Grid on your tube isn't in the area where it draws current (very dependent on voltages on Anode, Cathode + G2 {Penthode}) - you can consider it to be in the Giga-Ohms .... => the Grid Resistor is often the Input-Impedance (@ low frequencies) .... BUT even a small capacitance across can lower this in the extreme ....

So if you really what to know : a) Buld and measure - b) Simulate it - but please use very good Models for the Tubes ....

Per