Kay model 703 work over using a Triad N-68x isolation transformer

[SoulFetish]

A customer brought in a Kay model 703 for us. Only backstory and instruction on work order was to go through and make sure everything is a"as it should be". At first I thought it was an old widowmaker power transformer-less amp. While not completely isolated from the mains, there is a power transformer to isolate main power supply. Originally, all the heaters are run directly coupled to the mains supply.
This doesn't make any sense to me due to the potential fault from a heater shorting to any other node.
Anyways, the amp had had some service in the past. It had been fitted with a 3-conductor AC cord and the power transformer had been replaced with a triad N-68x isolation transformer. However, whoever did the wiring had no idea what they were really doing with really dangerous mains wiring grounding the neutral mains wire to the chassis/power supply ground making the chassis a mains current conductor and shorting the primary windings of the triad.

Here is a schematic for the Kay 703:
https://el34world.com/charts/Schemat...Kay/Kay703.pdf

here is the data sheet for the Triad N-68x:
http://catalog.triadmagnetics.com/Asset/N-68X.pdf

Triad's design uses 2 x 115 primary winding to be used with either 115V or 230V mains voltage with a single 115V secondary winding. Anyone see an issue with using connecting the mains to the single "secondary" 115V winding and using 2 separate isolated secondaries to supply the heaters and main power supply? I noticed that this transformer has a lead for a shield connection on the secondary.
I plan on leaving this unconnected if connected to the mains supply?

[loudthud]

I noticed on the Kay schematic that there is no fuse...

Is the existing Kay transformer a one to one ratio ?

It's reasonable to assume from the transformer spec sheet that each primary winding can handle roughly 200mA of current. The heater current for those tubes is 150mA so that's cutting it pretty close. The startup surge is probably twice nominal current. On the DC side, the rectifier tube is rated at 100mA DC, but it's only half wave. The transformer might not like the imbalanced current.

Is there a Triad model that is perhaps twice the VA rating ?

Grounding that shield might help reduce noise.

[35L6]

I'm glad you asked that question . I've often wondered what you could do with that split primary running it "backwards" . For a little background , my AA5 conversion uses an N51-X which is a 35VA transformer has working fine for years . The amp has 35Z5 , 35L6 , 12SQ7 , 12SG7 , 12J5 , and 12SK7 . The transformer barely get warm , so if you wired up the N68-X in normal configuration and connected the secondary where the power cord is connected it would work , even with the unbalanced load from the half wave rectifier . Now we get to the tricky part . If you reverse the transformer and split the windings will that half winding act like a fuse or does it need to heat up to fail and will be fine with the brief heater warmup time ? Are you spliting the winding to reduce hum ? Were you thinking about putting a bridge before the 35Z5 ?

[Helmholtz]

One more remark: If the secondary voltage (115V) is specified for nominal load, reversing primary and secondary will result in lower secondary voltage (maybe 10%-15%).
Isolating transformers often have a few more secondary turns to compensate for sag under load.
(Might not matter here.)

I also wonder if the isolation test from the transformer spec also applies between both primary windings or only between primary and secondary.

[SoulFetish]

One more remark: If the secondary voltage (115V) is specified for nominal load, reversing primary and secondary will result in lower secondary voltage (maybe 10%-15%).
Isolating transformers often have a few more secondary turns to compensate for sag under load.
(Might not matter here.)

I also wonder if the isolation test from the transformer spec also applies between both primary windings or only between primary and secondary.

You are 100% correct! The unloaded secondary voltage was around 106V at each secondary with 121V at the wall, so I had to adjust the value of the series resistor in the heater circuit.
this was basically a complete disassembly and rebuild.
There was some excessive power supply noise running the amp as designed. I altered the power supply and redrew the schematic to show the changes labeled in blue. Objectively this was a total success in filtering out any audible power supply noise. The amp is really, really quiet at idle at all volume settings (with the input grounded).

[SoulFetish]

There was something interesting I found when working on this amp. I experimented with removing the low input network and grid leak biasing, and cathode biasing the pentode. This made it necessary to alter the load impedance and screen resistor value. I did some experimenting with bias points and load resistances to get the gain I was looking for.
My goal was to ultimately redesign what appeared to poorly designed input circuit! The resistor divider at the input attenuates the guitar signal by half, creating a 44k input impedance. The divider network is then capacitively coupled grid leak resistor of 3.9M to set the bias.
The whole thing appeared to be a noise machine....
I was completely wrong about this. My redesigned input stage was considerably more noisy, making the volume control highly sensitive to circuit noise. After I changed the input back to the original design, the amp was so quiet even with you ear up to the speaker, you could barely hear that it was on. Who knew – grid leak pentode?...

[loudthud]

Is the 22K just shunting the noise to ground ? Or, is the change in noise just a change in gain ?

[SoulFetish]

I didn’t look at the signal coming off the plate to compare, but the voltage at the output transformer secondary with the grid leak input was as much if not more than my modified input.
With the 22k shunt, the amp was still easy to overdrive.
I wonder if the impedance network at the input may be the key to filtering out the 120V heater supply noise

[Helmholtz]

AFAIR, input stage noise is dominated by Johnson (thermal) noise of the resistors in the grid circuit.

The larger the total AC resistance from grid to ground - the more noise.
As the coupling cap can be seen as an AC short, effective AC resistance of the original circuit is only 22k with open input (and maybe 13k with guitar connected).
So they get away without shorting input jacks.

With cathode bias and traditional input wiring the grid only sees the large value grid leak R as long as the input is open.