Transformers - the how, why, when and where!

I’m hoping this thread will be a community effort on all aspects of transformers, power/output/reverb transformers (not forgetting chokes too) used in our favourite amps/creations.

There are suppliers of pretty much all we need in terms of transformers and chokes so why would anyone want to roll their own… this is the DIY channel! We could equally just buy a complete ready built amp from Fender et al.

We have quite a lot of info about the specs we need from transformers to build amps, but we don’t have so much at the ‘deep dive’ end to enable us build our own. The size/weight of a transformer is a good indicator of how much power it can supply at the same time as not getting too hot. The fundamental science/theory will be here but may deter folks from starting, I propose that in addition to the 'now the science bit' we get some ‘recipes’ together that will allow you to just build a working known good one.

I’ll start the ball rolling with a tear down and reverse engineer of an OT, the one used in a Jet City JCA22H. Along the way I hope to do a step by step build of a classic old school paper bobbin type build of a 5e3 PT and OT.


If anyone has tear down info on transformers/chokes, old busted transformers/chokes can be a useful source to reverse engineer too.

Let's get some of the basic stuff…

Transformers provide isolation from the wall outlet and provide a method to transform our mains/wall socket supplies into what our tube amps need and then match them to our loud speakers..

In 1831 Michael Faraday connected a dc source (battery) to a copper wire coil on an iron core and noted he could induce a current flow in a separate coil on the same core, it was transient though, but if he repeatedly switched the dc supply on/off an ac current could be sustained, electro-magnetic induction was discovered and the transformer was born.


Easy, we can all build a transformer, all we need is a 6” iron nail and 2 lengths of copper wire, some insulation and a power supply, all very well for a school science project.

Most tube amps will have a power transformer to provide a range of ac voltages/currents from a single source and an output transformer to match to our low voltage high current speaker load.


What do we need to build a useful transformer:

1) The iron core, we could use a big bunch of 6 inch nails.

E & I shaped steel plates are a better place to start, aka laminations, these are built up as layers like a pancake stack to form the core which allows the maple syrup/magnetic field to flow, the laminations help to reduce circulating eddy currents which helps power efficiency.

Bad news is these are not readily available for one-off builds, you can get them from old transformers or buy new ones that you can to rebuild to your own requirements. Bigger is always better here, makes fitting all the windings easier and helps bass frequencies for output transformers.




2) Enamelled copper wire, this is used to form at least 2 coils in a transformer, the primary (this carries the ac source eg wall socket) and the secondary (this carries the transformed voltage/current we want to use in our amps). For a choke, 1 coil is all we need.

The good news is that high quality enamelled copper winding/motor wire is readily available in all the gauges needed for tube amps on spools starting at $10.


3) Something to wind the wire onto, the bobbin, can be cardboard (old school vintage) or plastic. Paper sheets cut to the correct size to insulate between layers of wire. Hardware to hold it all together. Bell ends keep things tidy and high voltages out of harms way. Varnish.


4) A recipe, this would provide all the info needed to do the science bit, wire gauges specific to the role of the winding, the number of turns/turns ratios, the size of the core cross sectional area. The layout of the windings.


5) A winder/lathe to wind the wire onto the bobbin, these can be built from scrap wood or bought for a low cost.

TBC...

Doug

 

Here's a starter recipe for a 5e3 output transformer both the 8kΩ or 6.6kΩ into 8Ω speaker:

1"x1" core (25x25mm) (the 'tongue' of the E is 1" wide and the stack height is 1")

The turns ratio is 31.6:1 for an 8kΩ and 28.7:1

3 sections of windings with stiff cardboard insulation layers between each section.
1st half the primary (anode 1) 1420 turns of 34AWG (connect to 2nd half to make the CT)
Single secondary (single 8Ω tap) 90 turns of 19AWG
2nd half of the primary (anode 2) 1420 turns of 34AWG

The turns ratio is 28.7:1 for 6.6kΩ so use p1 = 1250 turns : 8Ω sec = 100 turns : p2 =1250 turns

Paper insulation layers can be used between each layer of wire. (this was done back in the day when the wire insulation was lower quality, with modern wire can be omitted and each primary section can be random wound)

Use 2 layers of stiff cardboard glued on top of each other to make the bobbin.
The bobbin is formed around a block of wood cut to the same size as the length E tongue and the 1"x1" core. This former also has a hole drilled through it to mount it on the winder.

Please chip in with any knowledge / improvements.

Doug

 

I'm reverse engineering the OT from my JCA22H head, lots of photos, but maybe too complicated for a starter OT so I'll do a step by step build with a 5e3 OT.

 

Lets's look at this a bit closer.

The 5e3 has 2 x 6V6gt tubes in the push pull output section. The 2 output tubes will give the best performance when they are matched to a suitable load impedance, this is the part often quoted as the load they 'like to see'.
The load matching is for the value of the 6V6gt "Effective Load Resistance, Plate to Plate" which we get from its data sheet:


This varies depending on the plate voltage, from the data sheet we are given 2 values 10000 or 8000 Ω.

This is where we get the 8kΩ or 6.6kΩ value for the output transformer, why the 2 values I hear you say. We don't want to get bogged down with the physics and mathematics behind impedance, one thing to note though is that it's frequency dependant. This ain't no hifi amp either, we want it to sound interesting. The 8kΩ or 6.6kΩ spec from the manufacturer will have been determined at a fixed frequency, eg 1kHz. If you really want to listen to a 1kHz sine wave then at least you'll be happy to know you have the correct transformer specification... but then your speaker impedance will also have to be 8Ω at 1kHz and then your speaker cab will have to be matched... you cleaned all the wax out of your ear canal 1st of course.
Bit of a moving target, we'll just have to settle for getting close and hand out the cigars anyway.



That 8kΩ primary impedance is only 8kΩ if you have attached an 8Ω load to the secondary winding (and only for that pesky 1kHz sine wave too). Also, with no load on the secondary side we have no load on the primary side back to the tubes, they don't like that. This can be useful when using the 'wrong' transformer, eg if you build an amp that 'wants to see' 4kΩ you could use your 8kΩ OT and put a 4Ω load on the 8Ω tap.

Let's drop the winding ratio for the OT into the mix, the transformer designer armed with a heavy pile of theory textbooks will eventually get the right numbers to spec an OT. The core size, core material, shape, wire gauge, insulation thickness, plate voltage, current, wind direction and magic pixie dust will be required to get the best results.

He will have defined a value for another often used term we need to breakdown, the 'reflected load', eg in our 5e3 amp, the 8Ω speaker is reflected back to the tubes via the transformer as 8kΩ.
The impedance ratio for our 8kΩ 5e3 OT with an 8Ω speaker is 8000Ω:8Ω = 1000:1
But we're not quite there, the turns ratio is the square root of the impedance ratio so now we have 31.6:1 (remember this ratio)
The recipe for the 5e3 OT gave us:
2 primaries of 1430 turns each (2 sections to give us the CT) total turns plate to plate = 2860 turns
1 secondary of 90 turns

The turns ratio is 2860:90 lets simplify the ratio, divide 2860 by 90 = 31.7 so we have a ratio of 31.7:1 (you did remember the ratio from above, didn't you? another cigar, it's close enough)

Next post will deal with measuring the turns ratio on an unknown transformer, PT or OT, same method.

 

Indeed, quite correct, when removing laminations from old transformers you have to be very careful not to damage them. Use varnish to repair any damage. The final stage of the reassembly is to use varnish to seal the laminations, this should be done once you have checked all the wires are correct and after any testing.

This is the one I use:

https://brocott.co.uk/electrical-insulating-air-drying-varnish-100ml-clear/

 

To find the turns ratio of an unknown transformer we will need another transformer and a multi-meter.

The extra transformer is simply a source of a low voltage ac, 12v would be good.

The next thing is to identify the primary on the unknown transformer, we don't want to attach 12v ac to the 6.3v filament winding as that might give 220v for those on 120v wall outlets or 440v for those on 220v (12v stepped up in the uk could give more like 480v). The same danger exists for output transformers if you connect the source to the secondaries.

1st thing to do is check for continuity between the various leads, if there are windings with no continuity we should give up and strip it down for parts.
The next step is to measure the dc resistance of each winding and checking the thickness of the leads or actual winding wire, grouping them together.

An example with 4 windings, it has 2 heavy gauge windings with 2 wires each, label them winding group (A) and winding group (B), there are 2 middle sized wires, label them winding group (C) and one group with smaller thickness wires with 3 wires, winding group (D) but label each wire D1, D2 and D3.
The measured dc resistances are:
(A) 3Ω
(B) 5Ω
(C) 70Ω
(D) 120Ω between D1 and D2
140Ω between D2 and D3
260Ω between D1 and D3

from this info:
(A) is a secondary filament supply, probably 5v (we'll find out after we supply the 12v ac to the primary)
(B) is a secondary filament supply, probably 6.3v
(C) is a primary mains supply, probably 5v
(D) is a secondary HT supply with a CT

Now we feel confident enough to connect the 12v ac source to winding group (C), also a good idea to use a light bulb limiter.

Once the 12v source is connected, measure the ac voltages on all the other winding groups., use these readings to work out the turns ratio.
Voltage readings:
(A) 0.5
(B) 0.66
(D) 30 and 60

Ratios
(A) 24:1
(B) 18.2:1
(D) 1:2.5

For an output transformer, the primary windings will have a much higher dc resistance than the secondaries.

Doug.

 

No questions are stupid here, all questions are good.

What post # do you want help with?

A ratio in its simplest form will be something:1. I admit 31.7:1 looks strange it's the same as 2853 primary turns:90 secondary turns, if we made it 31:1 the turns numbers would be 2790:90. The number of turns is limited by what can be squeezed into the size of the core window. Thinner wire can be used but that can cause its own issues, for the secondaries we need it to carry a decent amount of current or if too thin on the primary may cause a large voltage drop.

 

The numbers shown in the recipe are ones that just work as they have been used before with good results, this info might have originally come from someone taking a 5e3 OT apart and counting the turns (some vendors use this as a selling point to say their products are closest to the ones from the '50s).

I'll do a step by step vintage 5e3 OT build using the recipe in the coming days.

Further down the road we'll get into the theory/science/math behind it all, this kind of thing, don't want to deter folks just yet!

 

It's a lot more work to use paper. The more times the pri/sec sections are interleaved between each other (also means more paper layers) the better the high freq response, it's used in HiFi OTs. I think I read somewhere that the more paper layers (also thickness of paper) you add between the pri and sec sections means you are building a capacitor hence the freq response is impacted. The balance is between these.

The JCA22H amp OT I took apart has 4 sections, Pri scatter wound and bifilar, sec in 2 coils, pri scatter wound and bifilar then another sec in 2 coils . The sec wire is of 2 thicknesses, 2 coils are 16 awg and 2 are 18 awg. Very strange, this extra step must have been specified by Mike Soldano for the factory in China.