diyAudio (
-   Tubes / Valves (
-   -   Claus Byrith mods to Mullard 5-20 Need update (

coylum 25th June 2019 02:40 AM

Claus Byrith mods to Mullard 5-20 Need update

I have recently built the first of two Mullard 5-20 monoblocs using Transcendar UL OPT's. I am having issues with 30dB NFB and want to make the front end mods suggested in the Claus Byrith 4-30.

There are many references to this design but I am confused by certain conflicting statement regarding the need for a cathode bypass capacitor and step filter on the EF86. Initially these were discarded when the EF86 was wired as a triode with a 47K ohm anode load but I read on a thread (that I can longer find) that these elements were reintroduced at a later stage.

Can anyone provide me a link to the final schematic for the mods to the EF86 input stage.

Cheers, Malcolm

Diabolical Artificer 25th June 2019 05:32 AM

"I am having issues with 30dB NFB" Oscillation? 30dB is a bit heavy...Not familiar with the Claus Byrith 4-30 but if your having oscillation use 1/(2 * pi * f * C or R) and use what caps and R's you have, something like a 4k7 and 1n. But first you could try reducing the FB a tad to get things stable enough to proceed.

Is this tested with a dummy load or speaker? A 0.1u and 10r across the OPT sec can help or similar for spkr loads..

Good luck, Andy.

coylum 25th June 2019 08:47 AM

Basically the Claus Byrith mod to the front end is to wire the EF86 in triode mode, dramatically reduce the NFB, do way with cathode bypass capacitor and step filter and end up with a stable, linear result with a sensitivity of 500mV! Hopefully it will work for me!

DF96 25th June 2019 10:01 AM

Whether you need a step filter in the first stage anode circuit depends on the OPT and details of circuit wiring, as the HF stability will be affected by stray capacitance. If you want to use someone else's HF compensation then you have to use their OPT and their circuit layout in detail.

coylum 25th June 2019 01:35 PM

I am well aware of the points you make - I was basically asking if anyone knew if Claus Byrith had made further modifications to his initial design as there seemed to be conflicting opinions. I now have my answer in that he did issue an update in 2008 confirming that neither a step network nor cathode bypass were required in his design. Obviously he used a specific transformer but I am interested to try his principle in my Mullard 20 watt and see how my transformer performs.

20to20 25th June 2019 03:02 PM


Originally Posted by coylum (
Obviously he used a specific transformer but I am interested to try his principle in my Mullard 20 watt and see how my transformer performs.

If you really want a 5-20 and feel the only problem is the OPT, then why not just use a shunt filter for everything over 40KHz and keep the rest of it as designed?

zigzagflux 25th June 2019 05:58 PM

I have made three 5-20 variants over the years, with different OPTs (Lundahl, ST70). I have found best sound obtained by straight resistive feedback (would have to look up the amount, something between 10 and 20 dB) in combination with the step network on the first triode's anode load. It is adjusted by 1 and 10 kHz square wave, somewhat subjectively, but gets pretty clear what works well. Each OPT gets its own adjustment.

gregt 28th June 2019 02:38 AM

I built this design quite some time back. Worked well but I ended up using a constant current source rather that the 82K in the tail of the long tail pair. I found that gave better balanced output from the phase splitter. Stability was fine , although I used locally wound Xformers wound to my spec.

Johan Potgieter 30th June 2019 01:34 AM


As so often happens, discussion of a certain amplifier circuit widens out to other alternatives and quickly becoming OT. In the following I admit guilt of the same, but since other aspects have already been touched on:

As said you did acquire the alternatives you were looking for. But regarding the whole Byrith alternative:

I read that the original purpose was to have an amplifier with much lower gain, and that vis--vis 5-20s. To that end, the simplest solution for existing amplifiers was to change the input stage to triode plus cathode feedback (unbiased cathode resistor). Naturally NFB-affecting networks had to be altered correspondingly.

But I think one must not lose sight of basics. If knowledgeable members will kindly have patience: The purpose of the step network usually included at the anode of the first tube is to lower the loop gain so that at frequencies when other capacitances/OPT leakage reactance etc. started playing a role, loop gain was reduced to <1 for NFB stability. In that sense the first tube step function should not be governed by other reactive elements coming in later. (BUT: Unfortunately the latter were often such that said step function had to cut down into the audible range and lower NFB where it was most needed.) Further stability was (hopefully) then taken care of by a feedback resistor bypass capacitor.

Back to topic: As said the initial purpose was stated to make existing amplifiers more acceptable with as few changes as possible. Regarding subsequent NFB stability, one obstacle remained: The use of the ECC83 driver with its (also) high gain, relatively high internal impedance and subsequent Miller contribution to phase shift.

Not to teach basics where already grasped: For new 5-20s my suggestion was to keep the EF86 as pentode (with a number of advantages) and rather replace the 12AX7 with a lower gain tube (12AU7 or 12BH7 or similar, then with some resistor changes). I found such an alternative yielding a more acceptable open loop/phase response which subsequently eased getting acceptable NFB stability etc., even at 26 dB NFB.

Again now rather OT (apology Coylum), but the above to me more acceptable before faffing around with stability-affecting capacitors or significantly lowering the NFB to give poorer performance than the original 5-20.

And of course: All still naturally dependant on a suitable OPT!

All times are GMT. The time now is 05:33 AM.

Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 17.65%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2019 DragonByte Technologies Ltd.
Copyright ©1999-2019 diyAudio