KT88 amp

[hylaphone]

Be sure the signal not roll off at the input due to internal capacitance.

Looks like 6SL7 and 6SN7 are 3 to 4pf input capacitance... my source Z is 150 ohms, no problem

[catalin gramada]

I was able to fix the phase shift by bypassing the 470K side of the divider with 150pF. This also made lower frequencies more symmetrical, but nowhere near even to 20kHz. I'll fiddle around with that more this week

Be sure the signal not roll off at the input due to internal capacitance.

[hylaphone]

I didn't found any difference in amplitude of both inverted/non inverted at least 20kcps. Anyhow found a slightly phase delay in between and I wonder myself where this problem come from and how I can align it ?

I was able to fix the phase shift by bypassing the 470K side of the divider with 150pF. This also made lower frequencies more symmetrical, but nowhere near even to 20kHz. I'll fiddle around with that more this week

[Helmholtz]

Anyhow found a slightly phase delay in between and I wonder myself where this problem come from and how I can align it ?

In a paraphase PI the second triode follows the first one and inverts the signal. A slight phase delay is caused by the RC coupling, probably increasing at lower frequencies.

[catalin gramada]

-- how far up the audio band do the PI outputs stay symmetrical?

I didn't found any difference in amplitude of both inverted/non inverted at least 20kcps. Anyhow found a slightly phase delay in between and I wonder myself where this problem come from and how I can align it ?

[catalin gramada]

Better pic than my broked English instead:
Be sure you use a good quality bypass capacitor. 100u will be better and you may put a film 100nF in parallel over it.

[catalin gramada]

That simple. In original circuit you have 240 ohm shunt nfb resistor in series with first cathode. Because the cathode is not full bypassed but partial you will loose some of you precious gain rising in same time the internal tube resistance. You need to make the shunt resistor as small as practical, but take a look at circuit: the series nfb resistor is also the bias resistor for those cathode: the closed path from cathode run also from: cathode- series nfb- OT secondary-signal ground which is in parallel with cathode- shunt nfb- bias resistor- signal ground. You don't need to bias the tube through series nfb resistor and OT secondary winding. This is a unwanted consequence of the circuit as time you want to keep common OT tied to the ground. But there is no real reason to do that. (..allright there are some but not subject of our analysis). Close the nfb loop over nfb series/shunt resistor and bias the tube just from bias resistor not through series nfb in parallel with bias resistor-. The common OT wire should not touch the ground, you need isolated speaker jacks.
And btw be careful when you do the measurements with your scope, sometimes may create a loop path through ground probe clip- earthing wire..and depend what you measure you may involuntary run some currents in circuit through this path....that's can happen...sometimes..
The "floating" OT secondary will not harm anything.

[hylaphone]

I just built a functional paraphase circuit using 12at7 you can easily drive you kt88. 6db global nfb applied. With 2.5 Vpk input sensitivity think is just enough to drive from tone control output. Slight tweaking may be necessary. You may share the bias cathode resistors linking together as in original circuit, it will not make any difference as time are full bypassed Please get inspired if you feel.


Late edit: series nfb resistor 1k was used from 16 ohm tap/ not 470 ohm. Common OT tap should be isolated from chassis.
I fast wired a B15 style preamp with -20db James tone control in between, using 12ax7 in front and definitely get some low end lovers may use.. Not too bad to push 300W rms with just two tubes in preamp...

Thank you for sharing that -- how far up the audio band do the PI outputs stay symmetrical?

And can you elaborate on that OT common arrangement?

I may try something similar with the 6SN7, 6dB feedback is cutting it close but certainly worth a try.

[catalin gramada]

I just built a functional paraphase circuit using 12at7 you can easily drive you kt88. 6db global nfb applied. With 2.5 Vpk input sensitivity think is just enough to drive from tone control output. Slight tweaking may be necessary. You may share the bias cathode resistors linking together as in original circuit, it will not make any difference as time are full bypassed Please get inspired if you feel.


Late edit: series nfb resistor 1k was used from 16 ohm tap/ not 470 ohm. Common OT tap should be isolated from chassis.
I fast wired a B15 style preamp with -20db James tone control in between, using 12ax7 in front and definitely get some low end lovers may use.. Not too bad to push 300W rms with just two tubes in preamp...

[Helmholtz]

So the measured sag seems to tell us it's not a PT issue.

A PT secondary can be seen as a stiff voltage source in series with some resistance.
This resistance (aka equivalent source resistance, needed for PSUD2) consists of the winding DCR plus the reflected primary DCR.
Sag is the voltage drop across this total source resistance and thus can be calculated from the resistance and the load current.
The core has no influence on sag.

BTW, a transfomer cannot saturate by load current.

[hylaphone]

But you must have specs for it?

Sure, the secondaries are 720VCT@.3A w/60V bias tap, 250V@.1A, and a few different filament windings

[g1]

It's custom from Hammond, with a separate winding for the screen supply. I couldn't find an off-the-shelf transformer that fit the bill.

But you must have specs for it?

[hylaphone]

Ok. So the measured sag seems to tell us it's not a PT issue.
But I've re-read the entire thread and one data point that is missing is the make & model of PT. It would be good to know.

It's custom from Hammond, with a separate winding for the screen supply. I couldn't find an off-the-shelf transformer that fit the bill. But now I have a bunch of these to try out in different designs..

[hylaphone]

I get that this might be an exercise in theory and a quest for knowledge, but I'd just like to point out that "squeezing out a few more watts" isn't something to be concerned about. A few more watts will be hardly noticeable with respect to SPL. You'd need to double the output power to get even a 3dB increase.

Yes, appreciated. I'm trying to have a more "holistic" understanding of all the factors discussed here so that I may design better output stages. This thread has been extremely helpful.

[g1]

A PT has a recommended thermal current limit. Doesn't mean it can't supply more current, but will overheat.

Ok. So the measured sag seems to tell us it's not a PT issue.
But I've re-read the entire thread and one data point that is missing is the make & model of PT. It would be good to know.