Ampeg Gemini 2 OPT 7591A Push Pull

here is another Ampeg OPT,

this time the Gemini 2, which sports a 15 inch Jensen and a fat stack of 100EI,

p/n OT-215  682-529  from the town of Linden.

did you know that the Ampeg guy was an accordion player?

don't know if he ever served time, 

ok cut the chit chat and serve it up,

this thing has a bigger core than Dylan's Gemini 1, probably to match up with the 15 inch speak,

Henries and freqs,

here is some background information>

usually do not do both pri and sec inductance, as to get sec inductance you simply divide by the Z ratio,

in this case we have a turns ratio of about 18 to 1, so Z is 324,

324 x 4.25 H = 1,377 H,  which seems to match the pri Henries number for peak inductance.

one thing can be seen by the sec Henries is that as freq goes up, Henries approach the same value for both 8 and 16 ohm sec taps,

this would seem to indicate that speaker matching would be more important for bass than guitar, anybody find this to be true in the real world? 

this HP generator will not saturate the pri because there are too many turns and the voltage will not go high enough,

however, since the sec is low turns, probably around 100 compared to 1,800,
we have the Power.

we need a low freq so we can really saturate the core with max flux, so use 15 Hz.

we see that max Henries occurs at about 5 volts AC on the sec, since our turns ratio is 18:1, this means that max Henries will occur when pri volts equals

5 x 18 = 90 V-RMS = 255 V-peak to peak,  400 volt pwr supply would handle this with room to spare for max overdrive,

so here is a saturation curve for the sec- Henries vs Level>

here is the pri leakage inductance for the Gemini 2 OPT,

very low compared to most guitar outputs checked,

this is because due to the hi-fi winding structure, P-S-P-S-P-S-P,

balanced DCR also, so we either have 40/60 pri 1 turns and 60/40 pri 2 turns with sections alternating 1-3 pri 1 and 2-4 pri 2.   

or equal turns per section with sections 1-4 for pri 1 and sections 2-3 for pri 2.

pri 1 and 2 form the center tapped push-pull sections,

here is leakage by itself and then compared with pri inductance vs frequency,

leakage is checked by shorting sec leads and measuring pri inductance,

> as freq goes up, Henries approach the same value for both 8 and 16 ohm sec taps,

I see a fairly constant 1.5-1.6 ratio from 20hz to 400Hz.

To judge above 400Hz I'd want an expanded graph or raw numbers.

The expected ratio is 2, so something is odd. But we often see ideal turn-counts rounded to give a a full-width winding.

good catch PRR!

attached is a graph that is more aligned with the 8 and 16 ohm ratings,

exciting the sec is tricky, it is more sensitive to voltage level, and the core seems to wake up and go to sleep if it sits a while, at least at the really low freqs,

some people saturate outputs with 60 cycles and get the core real hot in order to break in the silicon steel before shipping as this seems to loosen them up a bit,

also, i do not know if testing a 16 ohm winding with the HP generator (600 ohm out?) is the best way to go, 

> i do not know if testing a 16 ohm winding with the HP generator (600 ohm out?)

My inclination would be to get a good Power Amp, put a few ohms in series to prevent smoke-down, and sock it good.

A tube-amp has to be much better than the OT being tested, which suggests a Mac.

Cap-output transistor amps can be screwy driving inductance. Fortunately many are direct-coupled. These can get sore driving a low resistance, hence a few ohms added so they never see much less than rated impedance/resistance.

Most power-amps, even direct-coupled, have a bass roll-off at the input or in the NFB loop. To a point, you just allow for this by cranking the signal generator below ~~~20 Hz to maintain desired level. If you switch to a Heath tranny sig-gen (goes to 1hz) you may need to shunt some caps with bigger ones.

I had my Heath rigged with a 1 Watt 8 Ohm power amp (LM383?) off the 6V lamp-winding, 4,700uFd output cap, so I could excite speakers directly down to a few Hz.

pic 1 - just some hardware, note paper covers for screws which keeps the screws from shorting out the core, not a big issue but something you will see in older high quality transformers,

pics 2 & 3  - not enough room between the coil and core to check turns by PRR method, (winding about 10 turns in the window and checking the voltage ratios) so we grab an equal stack of 100 EI and do test coils, checking current in the original transformer and the test coil at a fixed frequency and voltage, if the lams are semi identical as to perm, then the turns count in the test coil should match up pretty well with turns inside the transformer, coil structure might affect the readings a bit, but we should be in the ballpark with this test, we did this test twice, once for pri turns and once for the sec, unwrapping the turns on the test coil until the excitation current matched up with the orig. xfmr.

this coil gets used in a similar version of the Gemini 2 which runs 6L6GC tubes instead of the Westinghouse 7591a tubes,

by the looks of the splices seen on the ends of the layered paper coil, there are at least 4 pri winds, maybe 5, so the structure is either

P-S-P-S-P-S-P  or P-S-P-S-P-S-P-S-P,

here is a print of the best we can do without taking this thing apart>

finally got some JJ 7591S tubes for this Ampeg Gemini 2,

they sound great,

also replace the orange multicap shown here, it had a bad hum when you got your hand close to it,

replacement caps had a reserve parking place on the board, so we stuck them there,

no more hum from the the proximity effect,

this amp has not the best grounding scheme, everything is daisey chained to the filter cap can ground lug, and the reverb pot is really close to the mains,

fixed this with a simple ground wire from the reverb pot to the tone stack ground,

this amp has the weird tone circuit PCB which can be seen mounted to the pot,

we also replaced the Wima .1/400 main coupling caps with .22 /1000 volt Erocaps,

this cuts down on  bass a little as the Jensen 15 inch is kind of tired,

plate voltage was 490, s/b 430 for the 7591, not 6L6 voltage,

so we dropped the screen voltage with a 27K resistor in the pwr supply,

the grids were pulsating red with the amp was on 10 which is probably why the amp came in with no power tubes, they probably blew up,

schemo on the back panel,

reverb sounds good for jazz type stuff, Fender is more surf,

foot switch sits in a spring mounted to the inside of the amp,

word has it that Stevie Ray liked these amps, i believe it,