Hi everyone,
I am working through my design on the classic Williamson high power amp ala Acrosound with a few modifications.
Here is the specific schematic:
I first plan to do the modifications described in the article "Improving the Williamson amplifier" by Wright to get the 6SN7 tubes biased right. I have heard of others doing this with good results.
Later in the artical he talkes about improving the power supply of the amp to improve the transient response. The schematic I am working with above uses a different power supply than the one Wright works with so I am wondering what would be a good way to improve the design I am working with?
I know you have to be careful with capacitance directly connected to the cathode of rectifier tubes so I presume that changing the 10uf cap before the choke is not advisable to keep inrush current down. However, the rest of the caps seem wimpy so im not sure what an advisable amount is to increase them.
Also I am not sure if this effects possible improvements to the power supply, but I plan to loose the ultraliner design and go with some 6146W output tubes and use a regulator to drop the 450v supply to 220-230v for the screen grids. The 6146W does not like screen grids above 250v.
Thank for the help
Matt
I am working through my design on the classic Williamson high power amp ala Acrosound with a few modifications.
Here is the specific schematic:
An externally hosted image should be here but it was not working when we last tested it.
I first plan to do the modifications described in the article "Improving the Williamson amplifier" by Wright to get the 6SN7 tubes biased right. I have heard of others doing this with good results.
Later in the artical he talkes about improving the power supply of the amp to improve the transient response. The schematic I am working with above uses a different power supply than the one Wright works with so I am wondering what would be a good way to improve the design I am working with?
I know you have to be careful with capacitance directly connected to the cathode of rectifier tubes so I presume that changing the 10uf cap before the choke is not advisable to keep inrush current down. However, the rest of the caps seem wimpy so im not sure what an advisable amount is to increase them.
Also I am not sure if this effects possible improvements to the power supply, but I plan to loose the ultraliner design and go with some 6146W output tubes and use a regulator to drop the 450v supply to 220-230v for the screen grids. The 6146W does not like screen grids above 250v.
Thank for the help
Matt
Last edited:
I agree with Eli, and further I suspect that pentode mode operation will make the problem even worse unless the very best iron is used.
I think the Mullard 5-20 design or the Citation II approach might be a little bit more likely to work and be stable.
The charm of the Williamson was that properly implemented it had sufficient open loop gain for a rather prodigious amount of negative feedback to be applied, the price paid for this was the need for an exceptionally good output transformer. The original Partridge transformer used in this design was rather special.
I think the Mullard 5-20 design or the Citation II approach might be a little bit more likely to work and be stable.
The charm of the Williamson was that properly implemented it had sufficient open loop gain for a rather prodigious amount of negative feedback to be applied, the price paid for this was the need for an exceptionally good output transformer. The original Partridge transformer used in this design was rather special.
Williamson provides some basic OT inductance measurements to make - did they at least compare with the benchmark levels - your winder would have checked them I guess.
You should measure your PT winding resistances, and check the design limits for capacitor input level with the rectifier tube you have chosen. You may be able to increase the first cap a little. You can increase all cap levels downstream of the choke - back in the late 1940's, the range of cap values was limited and cost was a big concern - but not so nowadays.
You can full wave rectify the bias supply, with a soft series resistor, and extra downstream capacitance, so as to lower any residual ripple. Increasing the cap bypass on the heater elevation may also be worthwhile, as its impedance to ground is a hum loop path.
You should measure your PT winding resistances, and check the design limits for capacitor input level with the rectifier tube you have chosen. You may be able to increase the first cap a little. You can increase all cap levels downstream of the choke - back in the late 1940's, the range of cap values was limited and cost was a big concern - but not so nowadays.
You can full wave rectify the bias supply, with a soft series resistor, and extra downstream capacitance, so as to lower any residual ripple. Increasing the cap bypass on the heater elevation may also be worthwhile, as its impedance to ground is a hum loop path.
True orig Williamson uses cathode autobias...as for transient response in near class A, power supply demands is different and softer.
I still beg to disagree to others who find the o/p tranny criteria to be so critical. The LF pole becomes critical due to the absurd values of original concertina interstage coupling caps....Morgan Jones Valve amps 4th ed, page 472.
I have a couple of 1955 des 88-50 o/p trannies made by Parmeko for the 50W Williamson design, and the phase margins are narrower than modern replicas by Sowter & others. (I daren't use the original transformers as it's pretty guaranteed the insulating varnish is peeling, hence the capacitance change).
The proper concertina "out cap" values should be around 10-22nF, and I regulary find the optimised circuit can take another +15dB of global nfb (on top of the 20dB) before instability starts with a step filter in the global nfb path.
In my 40 yrs I've never found it a "flighty" design, probably due to my design insistence of using stabilised power supplies for each section. Also, by putting a CCS in the Williamson driver tails does open up excellent peformance possibilities, bandwidth & top end performance.
I realise that I've optimised many of the disadvantages of these types of amps to make them into excellent runners.
richy
I still beg to disagree to others who find the o/p tranny criteria to be so critical. The LF pole becomes critical due to the absurd values of original concertina interstage coupling caps....Morgan Jones Valve amps 4th ed, page 472.
I have a couple of 1955 des 88-50 o/p trannies made by Parmeko for the 50W Williamson design, and the phase margins are narrower than modern replicas by Sowter & others. (I daren't use the original transformers as it's pretty guaranteed the insulating varnish is peeling, hence the capacitance change).
The proper concertina "out cap" values should be around 10-22nF, and I regulary find the optimised circuit can take another +15dB of global nfb (on top of the 20dB) before instability starts with a step filter in the global nfb path.
In my 40 yrs I've never found it a "flighty" design, probably due to my design insistence of using stabilised power supplies for each section. Also, by putting a CCS in the Williamson driver tails does open up excellent peformance possibilities, bandwidth & top end performance.
I realise that I've optimised many of the disadvantages of these types of amps to make them into excellent runners.
richy
So I drew up this schematic with some of the modifications I plan to do. Still a work in progress but it is getting there.
View attachment Williamson Amp 3 .bmp
View attachment Williamson Amp 3 .bmp
Last edited:
It's important to keep in mind that the lead and lag networks' values are for a completely different amplifier, and cannot be known yet. As others have already mentioned, pentode operation and a large RC timeconstant both exaggerate stability issues. Some Zobels around the output transformer "cain't hoit".
All good fortune,
Chris
All good fortune,
Chris
I have seen the word concertina around the forums a bit. Can anyone elaborate on what it means in the context of tube amps? Is the "out cap" the cap in the feedback loop?
"Concertina", AKA Cathodyne, AKA Split Load, is a phase splitter currently held in high regard. Its only practical weakness is limited voltage swing capability. Williamson topology deals with that weakness by following the splitter with a differential gain block.
The issue of phase shifts in a Williamson style amp comes to the fore because of the high pass poles associated with the coupling caps. between both splitter/diff. gain block and diff. gain block/"finals". It's all too easy to get low frequency phase shift oscillation, when Williamson style circuitry is employed.
Look up Barkhausen criteria. The O/P "iron" completes the 360o phase rotation.
The issue of phase shifts in a Williamson style amp comes to the fore because of the high pass poles associated with the coupling caps. between both splitter/diff. gain block and diff. gain block/"finals". It's all too easy to get low frequency phase shift oscillation, when Williamson style circuitry is employed.
Look up Barkhausen criteria. The O/P "iron" completes the 360o phase rotation.Unless the plate and cathode resistors are replaced by inductances (chokes or interstage transformers); but hardly ever practiced in real world designs.
However, the unity coupling output transformers of McIntosh amps are actually split load phase inverters as well.
Last edited:
Actually the RC time constants should be spread *further* apart, rather than being moved closer. It's also recommended to make the smallest RC be at the point of clipping, ei, at the output stage's g1. This minimizes recovery time from overload ("hang time"). This time constant also needs to be sufficiently far from the output transformer's RLprimary time constant under worst case conditions, which is usually near maximum signal.
The principle is very similar to single dominant pole in opamps.
All good fortune,
Chris
The principle is very similar to single dominant pole in opamps.
All good fortune,
Chris
Thank you everyone for your input. Some of this is over my head.
As for now I think I will stick to my schematic and then "tune" once it is built. I wanted a "sold" game plan to start with so I know what parts to buy and how I want to do the layout.
Any other ideas on improvements power supply or wherever, I am all ears.
. As for now I think I will stick to my schematic and then "tune" once it is built. I wanted a "sold" game plan to start with so I know what parts to buy and how I want to do the layout.
Any other ideas on improvements power supply or wherever, I am all ears.
You are absolutely correct! When converting to full bridge I did a positive voltage configuration . Thank you for spotting that.
As for the 117V no it is not necessary but it is an easy way to get the bias voltage. Plus I have some filament transformers laying around. That way I wont have to spend more on a PT with another tap or strain the 5V4 any more. When it is done, I plan tweak the power supply to run the 5V4 close to its capabilities to keep the B+ as stable as possible.
Im not great with power supplies so im not quite sure how much more I can raise the capacitances (most importantly before the choke), especially before my trannys come in.
. Thank you for spotting that. As for the 117V no it is not necessary but it is an easy way to get the bias voltage. Plus I have some filament transformers laying around. That way I wont have to spend more on a PT with another tap or strain the 5V4 any more. When it is done, I plan tweak the power supply to run the 5V4 close to its capabilities to keep the B+ as stable as possible.
Im not great with power supplies so im not quite sure how much more I can raise the capacitances (most importantly before the choke), especially before my trannys come in.
Last edited:
I think 40uf is max for most rectifier tubes. You can also add on the primary (wall outlet side) an inrush current limiter which will add a slow start feature. That will help extend the life of the rectifier tube. The inrush limiters look like MOV's and have different current ratings. match the current rating to the idle current of your amp.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.