World Designs KT88 amplifier

[ppppenguin]

Trevor, job well done.

Funnily enough I've never built a valve amplifer, unless you count mucking about with an ECL80. Built a number of transistor amps out of a Mullard book which worked OK. Did a major rebuild on a pair of Quad II which had been badly mucked about with. That was back in the late 1970s.

[valves4amps]

(7 hours ago)Murphyv310 Wrote:
I think I made my first amp when around 13, in 1967. It used a 6SN7, 6V6 & 5Z4.

My history also around the same time and the same valves in the amp. My circuit came from a Gilbert Davey book Fun With Radio.
I used it for several years with my guitar, built many since over the years, mostly virtual copies of Marshalls requested by my customers when having repairs done.

Dave.

[Murphyv310]

Hi.
Here is the amp in action!!!
https://youtu.be/noYNsaLDOBs

[Murphyv310]

Hi.
I've been told the clip sounds shrill!!! I'd recommend listening on a decent pair of headphones on your PC.

[Geordie McBoyne]

Hi Trevor I like the symmetrical appearance of the amplifier, above and below the chassis.
My experiences of power amplifiers is not all that great. We can take it as read that the output transformer is the most important component in an amplifier, also the phase splitter and output drivers of a push-pull amplifier must receive attention. The RGD 1046 amplifier I'm presently servicing has a simple phase splitter consisting of a low impedance triode which has equal value anode and cathode load resistors. Grid drive to the PX4 triodes is via 0.1mfd capacitors. Is this a satisfactory arrangement? The Williamson amplifier employs a similar phase splitter but signals developed across the two load resistors do not drive the output valves directly, but instead triode buffer amplifiers are interposed between the phase splitter and the grids of the output triodes.
The Mullard 5-20 amplifier employs an ECC83 as the phase splitter which has equal value anode load resistors. The Mullard description of the amplifier informs us that by using a high mu triodes the high value anode load resistors can be the same value, the balanced outputs across the load resistor should be within 3%.
There are other phase splitter and driver circuits which we can discuss later.

Geordie McBoyne.

[Murphyv310]

Hi.
Thanks for your input David.
Valve amp design of course can be very basic, conversely complex. Some of the circuitry seen today that appears on the net is a load of tosh to be honest.
The Mullard designs take a lot of beating, the fact that the 3-3, 5-10 & 5-20 are so good is the use of DC coupling to the phase splitter (in the P-P versions and output in the 3-3) this reduces phase errors and improves frequency response.
The 5-20 also had 0.5uf coupling caps to the output valves, this too reduces phase errors. The NFB can also be "adjusted" with an RC network to "Doctor" the response curve at the expense of slight phase errors, these are just about impossible to detect with the ear though.
What I do though is to try to move most "Push-Pull" amps out of class A, this can be achieved fairly well if there is enough HT headroom. The World Design amp had low cathode resistor values of 200 ohms, the HT was approaching 500v and the 6550 anodes were glowing red, valve life was horrendously short and after the previous "engineer " had a go with fixed bias the scenario was worse.
When I rebuilt the amp all the fixed bias stuff, zeners, pots all over the place and the low DC heater feed to the outputs were binned, veroboard and other nasties were chucked out.
Now the cathodes sport 500 ohm bolt on gold resistors, decoupled by 330uf 63v. There is 55v across the 25 watt cathode resistors that are attached by heat conductive epoxy, running temperature is around 50c on the resistors, the amp is now not running in class A, P-P but more like AB.
With the higher value of cathode resistor the standing current dropped but I was unsure if distortion would be a problem. Fortunately a thorough test showed that there was virtually untraceable distortion till near maximum output. The owner of the amp rarely uses the amp at anywhere near its maximum so the reduction in anode current will improve reliability, don't get me wrong the amp is still an excellent central heating system.

This bring me to another amp. I got this amp from Mike (Radiomike) a few years ago, it was a home constructed amp (nicely made too). It's line up was EF39, ECC35, 2X 6L6, GZ32.
The output was low and severely distorted, the first thing to do was to swap out the EF39 with a EF37A, This made a marked improvement but the amp lacked punch and bottom end. The circuit was loosely based on the Mullard 5-20 but the EF37A was biased wrongly and was upsetting the DC conditions of the ECC35. The anode load and screen of the EF37A were too low, a 270k on the screen and another 270k decoupled with a 10uf on the HT feed sorted out the valves voltage, the HT on the EF37A anode is now 44v and it biasses the ECC35 correctly, mind this is DC coupled, I replaced the 0.1uf couplers to the 6L6 grids from 0.1 to 1uf, upped the cathode resistor values on the 6L6 and did a couple of other small mods. This amp now does a genuine 12 watts and sounds amazing.

I think if you know valve theory, have time and are happy to deviate just a little from the norm then you can make a seriously good amp, far cheaper and with a great satisfaction it is home built.
My next project is to have a go at a Williamson amp, mind you I have two orders for stereo Mullard 3-3 amps and another repair has come in!

[valves4amps]

Hi Trevor, at low listening levels it probably will stay in class A. You could afford to reduce the anode current further to maybe 75 or even 50mA, cathode resistor of 680 - 1k and still not encounter crossover distortion. This would reduce the heating bills further? . With an output of maybe 20 watts max the 6550's are under used and as you say the fixed bias was unnecessary at this low power.

The Wialliamson amp seems to be based around correctly wound transformers where the layers of primary and secondary are interleaved to give equalised coupling over the whole winding. I've seen some interpretations where the circuit topology can change from long tailed pair phase splitters to concertinas. I personally prefer the long tailed pair for linearity. The other point of the design was distributed negative feedback around each stage with a smaller amount of overall feedback applied. Too much overall feedback will invoke phase errors at some frequency and should be there only to account for speaker damping.

Dave.

[Murphyv310]

Hi Dave.
Yes I would agree that at low levels the amp is still in class A. The problem with an even higher value of cathode resistor is its dissipation, increased heat from the resistors will offset to some extent the reduced heat in the bottles.
The other problem is of course the effective output will drop, to compensate the HT needs to rise then of course we are talking about different mains transformers or going to silicon rectifiers and in a way going back to the mods that were in the amp before to its detriment.
The customer has had the amp back for nearly a week now, so far no calls to say there is any problem and when demonstrated on collection he had a big smile on his face............. that was good enough for me

[valves4amps]

Yes I agree with that, if it ain't broke don't fix it. Also customer satisfied win win, they don't come much better than that?

Dave.

[tiv48]

Just for interest here is a PDF of the WAD kel34 circuit based on the mullard 5-20
Rob T
ps. i have found the KT 88  article but can not get my scanner to work.