LTPI little help, please.

[catalin gramada]

BTW, how did you measure current source impedance?

I didn't measured, I just compared with what I get using a fixed 22k tail. The AC balance was worst for identical current with 12au7 used as ccs as I did. So, I started with a fixed resistor and then replaced with a ccs hoping for the best.Bias resistor was tailored to get same current to the sections as before (no quite engineering way, more monkey around).It was not. Now a fixed 27k do the job (datasheet states till 150k can be used but already get in HK voltage limit for this current) for 10% imbalance, as I stated.

[Helmholtz]

What is surprising for me,with 12au7 clip relative hard . Why ? The outputs are quad of el34 triode strapped into 4 ohm load resistor.
Put some pics with output and PI output scooped together. No nfb was used.
Is something you don't want to use for guitar, I guess...

[ATTACH=CONFIG]54673[/ATTACH]

[ATTACH=CONFIG]54674[/ATTACH]

I suspect EL34 grid conduction, meaning that grid drive voltage is too high for the circuit. This would show in clipping of the positive peak of each PI output.

What is your bias voltage?
Can you show a schematic of the complete power amp?
How does the PI output look with power tubes removed?

[catalin gramada]

Thank You.
The first pic show the signals at onset clipping. Yellow-OT output on 4 ohm load, blue-output of inverted PI section (PI outputs are reversed to can acomodate eventually nfb in cathode resistor of gain stage before pi) Bias voltage is 43V and was arbitrary choosed for cca 70% plate dissipation (not sure if quite right in triode mode)

[Helmholtz]

Bias voltage is 43V

This means you will get clipping by grid conduction as soon as the grid signal amplitude exceeds +43V.

[catalin gramada]

This means you will get clipping by grid conduction as soon as the grid signal amplitude exceeds +43V.

Thanks.
Two question please: why is happen that, and how can be solved? A big value grid stopper help, please?

[Helmholtz]

Thanks.
Two question please: why is happen that, and how can be solved? A big value grid stopper help, please?

It happens because the grid-cathode path acts like a diode. As soon as the grid gets more positive than the cathode, the diode is forward biased and starts conducting current. This in turns loads down the positive going PI signal.

Large grid stoppers may somewhat smooth out the hard clipping.

[catalin gramada]

It happens because the grid-cathode path acts like a diode. As soon as the grid gets more positive than the cathode, the diode is forward biased and starts conducting current. This in turns loads down the positive going PI signal.

Large grid stoppers may somewhat smooth out the hard clipping.

Yes,yes this is clear but asked why is happen to have this hard clipping characteristic? It looks almost as SS one.Never seen to clip so hard identical tube in almost identical conditions in pentode mode. It is somehow related to triode strapped connection?

[catalin gramada]

Remember I read somwhere in this mode screen grids induce a lot of internal feedback. Can be this a consequence or I.m wrong?

[Helmholtz]

Yes,yes this is clear but asked why is happen to have this hard clipping characteristic? It looks almost as SS one.Never seen to clip so hard identical tube in almost identical conditions in pentode mode. It is somehow related to triode strapped connection?

Not sure but maybe you didn't encounter grid conduction with pentodes. Pentodes have higher gain and thus require less grid drive to be driven into (smoother) plate saturation.
Also pentodes have higher internal plate resistance than triodes which also makes clipping smoother.

[pdf64]

Given the direction this thread has taken, to avoid a whole lot of possibly incorrect assumptions being made about the amp in question, a full power amp schematic is essential

[catalin gramada]

Yes,You,re right pdf64. Till now no schematic, just a plan. I still try to understand how to do EL34 usable in triode mode, after some times ago had a very deceptive experiment with kt88, or 90...don,t remember...
Anyhow el34 seems to be a good candidate. My goal is to provide a very good damping factor with minimal amount of nfb or not at all for a bass geetar aplication. I don't have to many options around here ,so I used what I have on hand : hefty 480v supply , with 200uF filter cap. 1.9k OT(not quite optimal load) and a quad of Russian el34. Keeping toped at 480v - no problems, screen grids was tied to the plate with a 100ohm resistor to prevent osscilation ( no really think it need it) . Power dissipation over screen was max 5w in hard clipping condition. ( datasheets states 8w max). I found it sound way better with g3 disconnected from cathode, so I keep that. The power at onset clipping with 1.9k into 4 ohm load was 70w, 24Vpk.not bad.. Just for those hard clipping I cannot decide. Maybe it sound still usable, which I doubt, but had not possibility ti crank it in my basement yet