Another newbie output transformer question

[shakeshuck]

I've been playing a bit with LTSPICE as I'm still trying to work out what matters and what doesn't (at least from a measurements point of view).

The schematic I was experimenting on is using 6L6GC output tubes in PP UL Class A. According to the specs (if I'm reading them right) the impedance should be about 5k. Out of curiosity I tried a 10k/8 tranny instead of the 5k/8 I had been using and saw a big improvement in the low frequency noise floor.

Are there any reasons why I might not want to use the higher impedance tranny if the simulated results are better? I presume the tube specs are given as they are for a reason...
The difference in simulated output power is less than 10%.

 

[DF96]

Were the two OPTs identical in every respect apart from turns ratio?

 

[shakeshuck]

As far as I can tell.
I'm using the Rev4 spreadsheet given elsewhere on the forum. The tranny types were Hammond 1615 and 1609. The only differences in the numbers given for each prior to calculation are the impedances.

 

[45]

Never tried with 6L6 but with EL34 in triode mode running at 400V/50mA, about 70V grid-to-grid swing I could get 15-16W in pure class A with 10K and about 22 W with 5K. Real amp, no simulation!

 

[artosalo]

I think that your results are mainly caused by how the LT Spice "behaves" with different input data.
Can you post here the .asc files you have used ?

 

[shakeshuck]

Here we go...

I reduced the bias negativity slightly since the first post, which has altered the noise floor further.

You can give me a list of 1000 things I've done wrong.

 

[artosalo]

Here we go...

I can't simulate. Need to have transformer symbols (ULPPTran5000 and ...10000)

 

[shakeshuck]

I've just noticed there are some ridiculously low currents on this - I've done the typical newb thing and watched the voltages but nothing else...

 

[shakeshuck]

I can't simulate. Need to have transformer symbols (ULPPTran5000 and ...10000)

Sorry, I thought they'd been pulled into the .asc.

 

[shakeshuck]

I take back what I said yesterday about the ultra low current, I must have been having a bad day.

I can now see there are issues with the splitter, and I'm not sure how to fix it. Now that I'm more current aware (!) it is obvious that altering the anode resistors between the two sides can help balance either the voltage or the current, but of course that means that the other is quite a bit out of whack and I don't know which is more important or what to do about it. I'm sure they're supposed to be better balanced than they are in this case.

I'm re-posting the files as well, hopefully there aren't the help-unfriendly issues I had with those from yesterday. Sorry about that.

 

[hpeter]

I've been playing a bit with LTSPICE as I'm still trying to work out what matters and what doesn't (at least from a measurements point of view).

The schematic I was experimenting on is using 6L6GC output tubes in PP UL Class A. According to the specs (if I'm reading them right) the impedance should be about 5k. Out of curiosity I tried a 10k/8 tranny instead of the 5k/8 I had been using and saw a big improvement in the low frequency noise floor.

Are there any reasons why I might not want to use the higher impedance tranny if the simulated results are better? I presume the tube specs are given as they are for a reason...
The difference in simulated output power is less than 10%.

In general, higher Z OPT gives stronger self-feedback, better DF, lower Pout. It makes sense with (noFB) triodes.

But with tetra-penta you are aiming for higher Pout+efficiency anyway, and likely you will have strong global FB´s somewhere..
Therefore higher Z is against you goals (squeezing most W possible, beat the crap out of it)