1
\$\begingroup\$

I'm making a simple SMPS to create multiple isolated outputs to audio effects circuits.

I need to make a toroid pulse transformer which has one primary and at least 10 secondaries. These secondaries will be used in pairs to make 5 positive and negative outputs.

I've never wound my own transformer before so I have some basic questions :

  1. Can I wind each 'pair' together. That is physically push two wires through the core together for each turn? Then use those pairs to create my positive and negative outputs? Will that have any effect on noise or galvanic isolation etc?

  2. Can I safely wind my secondaries with thinner wire? The intention is that the primary will pass 3 amps, and each secondary will only be expected to supply 100mA or so. Is it safe to do that?

  3. Will the transformer still work if the iron powder compound is not exactly right. So in an ideal world I'd be using a T130-3 (switching at 100KHz) But what if it's a T130-6 etc? Will it work, but just much less efficiently?

  4. what would happen if I wound secondaries, on top of other secondaries? Would that work? So if I ran out of space on the toroid, could I just wind more secondaries on top of other ones? I appreciate that there'd be some difference in output voltage, but would those coils still work?

  5. Sort of related to question 1. Would my negative secondary coils need to be wound in the opposite direction? And also, should my secondaries be wound in the same direction as the primary?

PS :

  • T130 core size : OD-1.3" ID-0.7" height:0.25" (roughly)
  • Type 3 iron powder (grey clear) : Carbonyl HP Permiability 35 80Khz-500Khz
  • Type 6 iron powder (white clear) : Carbonyl SF Permiability 8.5 (10Mhz up)
\$\endgroup\$
  • 2
    \$\begingroup\$ If this is a one-off, just get 10 separate transformers. These would either need a center tapped secondary or two secondaries. If this is not a one off, then stop and have a detailed conversation with a transformer manufacturer about a custom part. After that, you may be back to the 10 separate transformers anyway. \$\endgroup\$ – Olin Lathrop Jan 5 '18 at 13:44
  • \$\begingroup\$ This is a one off... But one critical factor is space. I already have a huge enclosure full of mains transformers. I'm trying to replace that with something that will fit into a 100x200mm space. Also, I'm not sure how multiple transformers would work in a switch mode power supply? \$\endgroup\$ – Richard Jan 5 '18 at 13:50
  • 1
    \$\begingroup\$ Then there is the question of why you think you need 10 isolated supplies in the first place. This smells of X-Y problem. \$\endgroup\$ – Olin Lathrop Jan 5 '18 at 13:58
  • \$\begingroup\$ @Olin X-Y problem? I want to have my instrument inputs and outputs galvanically isolated using 10K 1:1 audio coupling transformers. The circuits are +-9V, I'll need a supply on the left of each transformer, and a supply on the right.. that's at least 4. They'll also need to supply enough current to drive a 600k coil, if I can't properly source 10k ones. \$\endgroup\$ – Richard Jan 5 '18 at 14:16
  • 1
    \$\begingroup\$ So draw a diagram already! We have tools for that. It still isn't clear why your box needs so many stages of galvanic isolation. \$\endgroup\$ – Dave Tweed Jan 5 '18 at 14:52
3
\$\begingroup\$
  1. Yes - it's called bifilar winding.
  2. Yes. Work out copper loss (I^2*R) for each winding, choose wire such that the loss on the primary is about the same as all the secondary losses added together (and as small as possible, i.e. use most of the space). Copper loss is simply the power loss due to current passing through the wire's resistance. Work out the length of wire, look up the resistance per metre of your chosen wire diameter in a table (online) and plug in your current. If losses don't match, choose more appropriate wire. Doesn't have to be precise - see the difference between adjacent wire sizes.
  3. Probably not, but it depends on your definition of "work". You need to work out the flux density in the core, and compare it with where each iron saturates. Add datasheets for both cores to the question if you're in any doubt what to do. This is basic to transformer design, easy to get wronf if you don't understand.
  4. It'll work, but the coupling between secondaries will be affected - e.g. noise on one sec may be coupled more strongly to another.
  5. You can wind everything in the same direction - just pick the correct end of each winding when wiring it up. (Easiest done if you have an oscillator and a scope so you can monitor the phase of each winding).
\$\endgroup\$
  • \$\begingroup\$ What a great answer.. thank you. 1) Thanks.. Can google that now. 2) I'm not sure what this means? So far consideration has simply been to choose wire gauge that can pass the expected current. Where can learn about this copper loss issue? 3) I'll do that 4) I'll aim not to even if I have to buy a larger core. I'll need 30-40 windings. Primary 0.7mm diam wire, secondary 0.2mm, so I think I'll be ok for space atm. 5) Thanks! I have a cheap osc which can manage 100 Khz (MC34063 switching regulator). Lower switch freq is a worry as this is audio. But the larger toroid makes work easier. \$\endgroup\$ – Richard Jan 5 '18 at 14:13
  • \$\begingroup\$ Added some info on copper loss. \$\endgroup\$ – Brian Drummond Jan 5 '18 at 14:19
0
\$\begingroup\$

I'm going to try to address the underlying problem, rather than the actual question.

It appears that you have some sort of audio processing "box", for which you want both the input and the output to be galvanically isolated from the power source as well as from each other. There are two general ways to acoomplish that, as illustrated below.

The first requires two audio transformers and a single power supply. If the transformers require compensation in order to get adequate performance, then post-compensating the input transformer and pre-compensating the output transformer are done inside the isolation barrier. This scheme is easiest to expand to multiple inputs and/or multiple outputs — all you need is more audio transformers. All of the active circuitry still runs from the same power supply.

schematic

simulate this circuit – Schematic created using CircuitLab

Sometimes it makes sense to use just one transformer in the audio path, in which case, there might be active circuitry on both sides of it, requiring two power transformers or a transformer with dual outputs.

schematic

simulate this circuit

\$\endgroup\$
  • \$\begingroup\$ Wow.. thanks for your work! Ok.. so scheme one.. If my guitar (for example) is going straight into the input coil of the input transformer, then I'll have potential impedance issues surely? Likewise with the output transformer, though that's probably easier to deal with. What form does "input comp" take? Some sort of gain stage? Seen next comment also \$\endgroup\$ – Richard Jan 5 '18 at 17:41
  • \$\begingroup\$ Directly connecting to the input transformer also probably requires an Jensen level £80 per unit transformer.. Whereas if I use an input buffer stage I can drive a much cheaper transformer? \$\endgroup\$ – Richard Jan 5 '18 at 17:43
  • \$\begingroup\$ I can't answer these questions because you've told us nothing about the actual signal processing that you're trying to do. Transformers transform impedances, too, so any "matching" that's required can be done on either side of the transformer. Or use a transformer designed for the impedance you need in the first place. \$\endgroup\$ – Dave Tweed Jan 5 '18 at 17:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.