• The move to the new server is done. There are some software and database maintenance updates in process. This has us passing the hat around to help out. We appreciate any donations. Seriously, even a dollar helps. The payment page may be found here - https://www.audiokarma.org/support.html

Mullard vs. Williamson

tekuhn

tekuhn

AK Subscriber
Subscriber
I think I understand the general design differences - Mullard has pentode input stage with LTP inverter. Williamson used triode input stage and split load inverter. My question is what are the sonic differences and which is your preference? From an experimental standpoint, is it fairly straightforward to swap between the two front-end circuit styles (prior to the output stage coupling caps) with the understanding that there could be gain differences and NFB adjustments needed?
 
Register to hide this ad
original Mullard is pentode input, but lots of variants exist using a triode, or triode-wiring the pentode to get less overall gain.

so long as you have enough drive from both circuits you could swap the whole front end easy enough. If you did an amp with a removable board or sub-chassis it wouldn't be a massive effort to swap it all out. Just need to provide B+, ground, feedback signal, input and output connections. All tuning could be done on the removable bit, and the output stage would stay in place. Think Dynaco amps and different circuit boards. Those exist in a Mullard-style, a cross-coupled design from ARC, and if someone hasn't done a Williamson board I'd be amazed. Plus of course the original Laurent driver. The stuff on the chassis stays the same, unless the specific board has individual bias adjustments that replace the stock per-channel arrangement. Thats not really a topology-specific thing though.
 
The two different front ends stabilize very differently however when using a true pentode stage in the Mullard configuration, as while the original Williamson front end displays very good supersonic response, the original Mullard configuration displays a rather poor supersonic response due to the very high output impedance of the pentode stage, and Miller in the input stage of the differential phase inverter stage. As a result, both topologies use greatly different stabilizing component values all else being equal to achieve otherwise identical performance and absolute stability.

Dave
 
Thank you both. I remember learning from Dave’s Fisher 400 thread about Miller killing high frequency response on a triode input stage when the input impedance was in the middle range of the volume control, so thought perhaps a Williamson (triode input) might suffer a similar issue with certain input sources.
 
the Williamson amps are lower gain triodes, so its probably not as much of an issue. Miller capacity is plate to grid capacitence * (gain + 1), and most Williamsons use some sort of medium mu tube there. 12AU7, 6CG7, 6SN7, etc. Gain would typically be under 15. The Fisher uses a 12ax7, gain there is probably more like 50.
 
I wouldn't prefer either of those (stock Mullard vs. stock Williamson). I'd prefer a hybrid of each.

On the side of the Mullard 5-20, I'd prefer a triode input stage with much less than the stock ~30 dB global feedback. On the Williamson side I'd prefer a KT66 output stage run in pentode mode and keep the approximate 20 dB global feedback. Having built/refurbished both types of amps, both are decent performers when properly designed and tuned. But I hope I never need to live with just one single amp. It's like eating the same cut of steak every day. You get bored after a while even though the steak is tender, juicy, and great tasting. You just need something different once in a while. So I'm working towards good examples (either commercial or self-designed) of each type that I rotate in every now and again to keep my sonic palate refreshed. :)

I guess I'm primarily thinking of the output tube types EL34, KT66, KT88 (or their American siblings). And since this is a wish list, we might as well throw in good examples of single ended vs. push pull, too.
 
Both of the Mullard-ish 6L6 family amps I own are all-triode for the driver stuff and run ~20dB of feedback. Both pentode mode output. Middle tube is a 6CG7 which is less gain than the 12ax7/ECC83 originally specified, and one I went with a 12AY7 in front while the other is a 12ax7. I found that the 12AX7 gave more gain than I really needed. The 12AY7 came out very nicely though.

My 5-10 is a triode-wired EF86 and the inverter is also a 6CG7.

kward said:
You get bored after a while even
Click to expand...

Its a good formula, but the next two amps I'm brewing are not that front end for the same reason. Part way through construction of a Williamson that will most likely be EL34 triode mode. The other will be pentode mode but probably with a floating paraphase driver.
 
actually I'm remembering I own two commercial Mullard-ish designs too. A Bogen DS-225 and a DS-265. Same basic design for both, 12AU7 voltage amp, 12AX7 inverter. The 225 uses 7355 tubes, the 265 runs 7027A, both in pentode mode.
 
The Williamson has two sets of time constants (coupling caps) as opposed to one, and a generally longer signal path. Mullard designs like the Citation V are simpler and IMO easier to get right.

Jack
 
dcgillespie said:
The two different front ends stabilize very differently however when using a true pentode stage in the Mullard configuration, as while the original Williamson front end displays very good supersonic response, the original Mullard configuration displays a rather poor supersonic response due to the very high output impedance of the pentode stage, and Miller in the input stage of the differential phase inverter stage. As a result, both topologies use greatly different stabilizing component values all else being equal to achieve otherwise identical performance and absolute stability.

Dave
Click to expand...
Would adding a cathode follower after the ef86 solve the impedance and miller issues in the Mullard design?
 
would be ideal if you could direct couple that, otherwise its going to add two more phase shift points which will probably turn it into a real mess to stabilize. EF86 into the grid of whatever the follower tube is, cathode of the follower direct coupled to the grid of the inverter. Depending on the follower tube the cathode ends up a few volts lower than the grid so if its already direct coupled it wouldn't be a huge issue to just insert a buffer. Maybe tweak the screen resistor for the EF86 to shift the voltage slightly if needed.
 
It certainly should, in that now the driving impedance presented to the cathode coupled inverter is way lower now, but now, much lower than even a triode stage would present if directly coupled to it. And, the preceding stage (triode or pentode) would have better response not operating into the input capacitance presented by the inverter stage. So now, both the inverter and AF Amplifier stage have much better HF performance than it before, likely even better than the Williamson front end displays. Bottom line then, is that inserting a cathode follower would greatly improve the supersonic performance of the Mullard design, but would still likely require its own unique values of stabilizing components for optimum performance -- although they would no doubt take on values much more in keeping with those that would be used if a Williamson front end was employed. Direct coupling would be necessary to avoid any increase in LF phase shift however.

I've never tried such a topology, butt would be an interesting exercise to try out.

Dave
 
You must log in or register to reply here.
Back
Top Bottom