Partridge TG 6556/2 Output Transformer

starting a new thread pending imminent tear down...

this transformer is used in the HIWATT DR 504,  3.8K P-P to 8/16 Ohms

two EL-34 tubes drive the output transformer in Push Pull configuration for about 50 watts of power.

frequency response is flat from about 10 Hz to 23 K Hz at which time there is a slight drop in amplitude.

sq wave test at 1 K,  damped sinusoidal are amplified in bottom pic,

both secondaries connected,

10 K and 100 K sq wave

Is it just me, or is that a REALLY flat frequency response for an output transformer?

I wonder if one of Edcors HiFi style units would be a good substitube for this in an amp build, versus a 40/50w "Fender type" 4k output transformer.

you do not want a transformer to be too good, Hiwatt tried the Partridge C Core transformers and the amps kept blowing up. leakage was so low that the freq response was off the map,


here is the Pri Inductance vs Voltage at 60 Hz>

Pri Ind vs Freq

kink at 200 Henries is from switching range sw on GP 200, the voltage goes up a bit at the beginning of the dial.

expanded version of previous graph

was wondering about the steep induction curve and the perm calculation that says 19,000 instead of 10 to 12K (normal 29gaM6 specs)

certainly not Nickel Alloy that would cost a fortune, turns out to be 0.006" lams instead of 14 mil,

38gaM6 instead of 29gaM6,

gonna be hard to find that,  and this varnish must have had a high solvent content as it hardened pretty good,

we got about half way done tonight, have to re-heat and get the rest tomorrow so we can hack the coil,

never seen 6 mil lams in an output, even though they are not nickel, they still cost about twice as much as 14,  but you have less eddy currents and better inductance, K1 stacking factor goes from 0.9 tom 0.85 for a 1 x 1 which is what this is, so B-max goes up about  5.8%  as there is more air between lams which means less steel to take the flux,

both the Fender Twin Reverb and the Hiwatt OPT  have the same size core, laced 1 x 1,

but the Hiwatt (Partridge) OPT has thinner lams,

even though the partridge has fewer turns than the Twin, it still has more Henries,

if we ran a B-H curve test on the lams, you would probably see less waste (area inside the B-H loop) with the Partridge due to less Hysterisis .

so the Partridge will have less phase shift and distortion at the bottm end than the Fender due to the lam thickness being thinner.

alright then,

de-lam complete,

take some ref pics just in case we have to cross check something left out of our notes,

all pruned up and ready to go,

10 PB lead pad with another one that goes over the top, sometimes 2 over the top,

everything was baked pretty good over the last 45 years so it was a slow one,

outer-most layer is a primary of 3 layers and 373 turns,

0.0025" KP, was the layer insl use throughout, for pri and sec,    since this is from the UK, they probably used metric for everything,

so this was probably 0.06 mm  paper.

i bet this was hand wound,  nice and slow as to get even tension for a tight coil,

wire diameter was right in between gauges, so was DCR/1000 ft, so probably metric wire,

measured 0.012" = 0.30 mm wire

or maybe not, there are differences among different wire tables out there, throw in double coated or triple coated, heavy build, 155 vs 205 degree, nominal, min and max,  plus the wire you but today will ohm out different so we do not stress exact wire parameters, the right wire size is the one that fits on the bobbin without making it tough to get the lams in.

layer insl was impossible to get off, even with the heat gun, lots of solvent in the varnish ,  and probably Pete Townsend dimed all the controls for full bakeage, although i hear he liked the 100 watt models,

so you just rip the wire thru the paper.  outer layer was very brittle, the wire snapped a few times, just broke apart in my hands so good thing we did a re-wind,

secondaries were two strands of 0.015" which sits right between #26 and #27 AWG wire,

97 turns total with a tap at 68 for the 8 ohm wind,  2 layers,

caked paper still a problem, got easier as we moved onside,

every transformer we unwind has something new that we have never seen before,

this one was no exception, the 8 ohm tap was soldered right to the sec wind, just about every power output transformer you see will have the individual sec winds brought out of the bobbin and then spliced,

one of the twin wires broke off at a splice, so this is not the most dependable way to do it,

bi-fi sec wind splice sandwiched between the 10 PB.

life getting easier as the varnish does not always reach all the paper near the center of the coil,

interesting that no polyester insl was used, only 2.5 mil KP.

pri-sec insl was 6 layers of 2.5 so 15 mil total insl  between hi and low voltage,

another primary being counted,  we have four primaries, all 3 layers at 373 turns so about 124 turns per layer,

no varnish in the middle, the wire was probably moving around a bit and singing a song at 1 K Hz,