KT88 amp

[catalin gramada]

Sorry have not a photo editor but did a sketch similar with you.

[loudthud]

It should be mentioned that there is significant capacitance between windings inside the output transformer. Some of it may cancel out, but it's hard to know without knowing how the OT is interleaved. In any event, a circuit that does not ground the Common of the secondary is a invitation to all sorts of ciaos from frequency response anomalies to parasitic oscillation. In other words, it's a really bad idea.

[catalin gramada]

I don't think so.. the secondary it was supposed to"float" in the top of 100 uF cap as ac shunt to ground.

[loudthud]

I don't think so.. the secondary it was supposed to"float" in the top of 100 uF cap as ac shunt to ground.

Ever wonder why an amp will hum like crazy if there is not virtual or real center tap grounding on the 6.3V winding ? Even if you just ground one side of the 6.3V, it kills 90% of the hum. Same reason, inter-winding capacitance impresses a common mode signal on the heater winding that can be hundreds of Volts peak to peak.

[catalin gramada]

Ever wonder why an amp will hum like crazy if there is not virtual or real center tap grounding on the 6.3V winding ? Even if you just ground one side of the 6.3V, it kills 90% of the hum. Same reason, inter-winding capacitance impresses a common mode signal on the heater winding that can be hundreds of Volts peak to peak.

And again, the capacitor- (which happen to be cathode bypass cap in the circuit we're talking)-serve as good as well giving the possibility for winding to "float" at some DC level. Many elevated heaters circuits shows no issues-reffering at you example. I.m very sorry -It simply works.
Anyhow op claimed his intention to keep nfb circuit into bottom side-as it is in the corrected layout sketch, but very easy to move into the top to see the difference, if any...

[hylaphone]

Sorry have not a photo editor but did a sketch similar with you.

Silly mistake, thanks for the sketch. It is working good now!

[catalin gramada]

By comparison: That.s the picture...

[catalin gramada]

Silly mistake, thanks for the sketch. It is working good now!

No worries. Get rid of 100k grid leaks and put 200k. You.ll get better excursion and better linearity, better swing, less distortion...100k looks horrible in the circuit
Make series nfb 1k to 10 ohm shunt and com output jack to ground tight twisted wires to minimise possible risk of interferences with sensitive parts of you circuit...

[catalin gramada]

When you're done a quick run very small signal square wave test to trace any sign of severe high freq osscilation who can endanger you power stage and may try to make it stable if necessary.

[hylaphone]

When you're done a quick run very small signal square wave test to trace any sign of severe high freq osscilation who can endanger you power stage and may try to make it stable if necessary.

Will do. Is this the reason for bypassing the 1K?

[catalin gramada]

Don't bypass nothing yet. You will see later if necessary, but I don't think is need it.Sometimes may see some sign of burst oscillations on a sine signal if you push the power stage up. It is happen sometimes when a fair amount of global nfb is applied. Sometimes is happen to be caused by reversing high freq phase to an amount when becomes positive- due multiple freq poles over nfb is applied and here a small bypass cap to compensate may help. But most probable you will find there high freq issues caused by undampened ringing in primary side. You don't know if you amp need it any compensations , and probably not as time you use a decent quality OT.

[catalin gramada]

Disconect the nfb and apply sine wave signal to the PI input till start to clip. Note the Vpk 1 value at the input. Do the same thing with nfb applied. and note what Vpk 2 is necessary till start to clip. Determine by ratio how much nfb was applied with 1k/10 ohm feedback network. Usually 6-10 db may be used.
You may put the values into a volt/db calculator to be easy, like that :
http://www.sengpielaudio.com/calculator-gainloss.htm
If want to apply more nfb just reduce 1k resistor to something less, say 470 ohm, but you have to calculate the power dissipation om resistor in respect with output voltage of you 16 ohm tap - maybe you.ll need a 3-5w resistor or so...