Pete Millett Uniamp

[sam14]

Hi, I'm planning to build this amp with 6550s. I have the pcbs & Sowter PS tx's too. My question is about negative feedback. There isn't much info on Pete's site but I want to lower the O/P Z with about 10dB of global loop neg feedback from the O/p Tx's (Hammond 1650Rs). I was wondering what calcs I need to do to get suitable cap/resistor values. Any help would be much appreciated.

[Eli Duttman]

The phase compensation cap. value has to be empirically determined, at the bench. Use a 2 KHz. square wave and tweak the phase comp. cap. for best results, using your o'scope.

[Mickeystan]

Hello Eli, What is the bases for 2khz being the target frequency to optimise at say versus 5khz of some other frequency such as 10khz? Thanks for any insights Mickeystan

[Yvesm]

The goal is to tweak the cap for the smalest rise/fall time with lowest ringing.
You may use any frequency and corresponding sweep time that give readable display on the 'scope.

Yves.

[Mickeystan]

Yvesm, your reponse is in line with my own thinking and I have typically set the compensation cap value based upon the most opitmum squarewave response I could obtain at 10 Khz usiing my Wavetek square wave generator and my Tek 465b scope when driving a 8 ohm dummy load. My question to Eli was in interest of why he specifically stated 2 khz. Maybe it is only his frequency of choice or was merely stated as an example. Eli is a smart guy and I was curious if he might have futher reasons for his recommendation. Mickeystan

[HollowState]

Well formed square waves are made up of many upper harmonics. Because of this, a square wave can be reliably used to convey information by a factor of ten times. Therefore 2KHz will equate out to 20KHz. A 2KHz fundamental is the accepted standard for square wave testing when an output transformer is involved. To use a much higher frequency would require the OPT to properly pass ten times that frequency. Generally speaking, the response of most OPTs are unreliable at elevated high frequencies and could produce distorted and abnormal square waves. One would not want to compensate for an overly distorded square wave.

[sam14]

Apologies for my ignorance, but does it matter wrt the component values that the neg feedback will be fed back to the Sowter tx rather than the traditional I/P valve/tube?

[pmillett]

FYI, I never actually tried applying loop feedback on this amp this way myself - you may have some phase margin issues! Applying feedback around two transformers can be a difficult thing...

Pete

[Deco-Doctor]

Originally Posted by pmillett

FYI, I never actually tried applying loop feedback on this amp this way myself - you may have some phase margin issues! Applying feedback around two transformers can be a difficult thing...

Pete

Bump

Has anyone applied global NFB on these Uniamps?

I built a pair of mono blocks using these boards with EL43's in the output stage with Hammond 1650P transfromers. I built the amps to drive a pair of Quad ESL 57's, but they dont sound good into the ESL's as they are a difficult load dropping down to below 2 ohms at high frequencies. I feel they would sound better with some NFB. I would like some help on how to go about it.

Thanks.

Martin