tube circuit layout

hej!

i have set up both channels in my passive eq with point to point tube gain stages.

looking at the outputs of the OTs i get quite different square waveforms and more resonance in one of them. one have bandwidth to about 55Khz and the other about 85Khz...

ive read comments about bad layouts can cause funny stuff but i dont know what to suspect, where are the sensitive parts. its a two stage circuit with global NFB.

is there some literature on the subject anywhere?

Treat everything like a high impedance so look out for crosstalk and stray capacitance.

JR

Swap the transformers and see if the difference follows the transformer.

Cheers

Ian

Swap valves too, variations in layout  and lead dress may play a role.

thanks!

id like to wait with swapping transfomers since they are tricky to install with screws...
low end is basicaly the same, -3db @ 6hz. also DCR.

i have swapped tubes already with no result.

when i pulled some interstage and feedback caps around i got little difference, both bad and good difference...
so i am suspecting that maybe components that are close to each other are messing with things...

5v333 said:

thanks!

id like to wait with swapping transfomers since they are tricky to install with screws...
low end is basicaly the same, -3db @ 6hz. also DCR.

Click to expand...

Low end is primarily determined by primary inductance  which in turn is mainly determined by the core type and the number of turns. Not too hard to make many transformers with very similar bass performance.

Top end performance is very much affected by leakage inductance and this in turn is strongly affected by the way the transformer is wound and constructed. It is not hard to imagine two supposedly identical transformers having very different leakage inductances. If you have access to the leads of each transformer you can measure the leakage of the primary by shorting the secondary and vice versa.

Cheers

Ian

Again, check transformers first.  Most likely thing. 

ruffrecords said:

Low end is primarily determined by primary inductance  which in turn is mainly determined by the core type and the number of turns. Not too hard to make many transformers with very similar bass performance.

Top end performance is very much affected by leakage inductance and this in turn is strongly affected by the way the transformer is wound and constructed. It is not hard to imagine two supposedly identical transformers having very different leakage inductances. If you have access to the leads of each transformer you can measure the leakage of the primary by shorting the secondary and vice versa.

Cheers

Ian

Click to expand...

perfect advise! would it be enough to measure leakage inductance one way? shorting the primary for instance?



actually the OT with higher bandwidth and more resonance lost connection betweein 3 pri windings and their pins some time ago. either from over heating the pins with my new solder iron which turnd out to be much hotter than the former one. or from fiddeling with the screw when mounting the OTs to the inside of the casing.
i had to save it by cutting a bit in the plate that holds the pins so i could reach and  resolder the wires...

5v333 said:

perfect advise! would it be enough to measure leakage inductance one way? shorting the primary for instance?

Click to expand...

I think so.... in theory it should be the same measured either way but you never know with transformers.

Cheers

Ian

ok i had some sweep test with both OTs.

i shorted the primaries, left the centertap connected to HT.

desoldered load components on the secondary, leaving it open circuit.

i put a 10nf across the secondary and also the tone generator with a 10k in series.

resonance peaks are  62.5khz for left channel (the one with higher bandwidth) and 61.2khz for channel right.

online calculator gives 0.648mH and 0.676mH.

this seems trivial...


agree to dismiss the OTs as the problem?

Sounds like it. 

something else seems strange with the channel with the smaller bandwidth. i have been sitting and trying to optimize a zobel network for the input transformer of the eq that preceeds the tube amp.
and so i wanted to boost the whole system to check the head room of the whole system once more.
putting a 40hz sinewave through out i dont see an clipping at the output until about 30Vp-p!!!!!!

watta frack

OT stepdown ratio is 9
a load resistor of 470R sits on the secondary
output 6sn7 tubes are biased at -6Vgrid.

is that 30V class B im seeing... or does it smells like OT is not doing whats expected from it??

maybe i still havnt understodd gain figures in balanced systems..

5v333 said:

is that 30V class B im seeing... or does it smells like OT is not doing whats expected from it??

Click to expand...

Without a schematic it is hard to tell.

Cheers

Ian

just checked headroom in both channels, they both have this high threshold of clipping...

there was one difference between the channels.
c12 c13 between the first plates is ceramic in one channel and mica in the other.

i just unconnected theese caps. now bandwidth is up around 120khzwith in both channels. need to recheck stability margins inside the amp again.

could this have been the problem? ceramic vs mica???

i have been ripping up zobel networks on transformers, both input and interstage, and made them better. things looks a bit better. i might have had a loose end somewhere before.

i still have something strange going on. attached is a pic of the output.

theres a little overshoot on the end of the squarewaves (5khz) upper and lower flat side. i have been chasing this some what and it prob comes from the interstage transformer. if the transformers casing and internal shields is not grounded, the same funny overshoot is present on the secondary of the interstage. as soon as i ground it the overshoot dissapears. but stillpresent on the output..!
i cant see it on the output plates either...

gonna do some more crazy experiments but just wanted to know if anybody recognize it and can tell what its about.

the ll7903 has a note in the datasheet:
Note! Isolation between shields is not guaranteed!

is it possible that this transformer is not very good as a phase splitter?

5v333 said:

just checked headroom in both channels, they both have this high threshold of clipping...

there was one difference between the channels.
c12 c13 between the first plates is ceramic in one channel and mica in the other.

i just unconnected theese caps. now bandwidth is up around 120khzwith in both channels. need to recheck stability margins inside the amp again.

could this have been the problem? ceramic vs mica???

Click to expand...

404 error on the attachment....fwiw

heres the schematic  again.

a little bit updated than the last one.

i asked lundahl if they recognized the funny shape, maybe from their transformers or something.

they answered that the waveform looked great! the funny figure could be counter forces in leaking inductance.

so i think im on the right path.