6550 Williamson - Low Bass Response...

[zman3]

Hello all,

Well, I've been puttering about with that "High Power Ultra-Linear Williamson" for a couple of weeks now, and I've learned quite a bit - mainly how much I don't know!

Everything seems nice and stable, both with or without feedback (note the 25k feedback pot). I'm using Chinese KT88's running 450 V on the plates at around 110 milliamps or so with -54 V on the grids. So far it sounds pretty good through my workshop speaker (a JBL bookshelf job), except that the higher frequencies seem rather bright, while the bass seems to be rather lacking. I can see the woofer moving, but the bass itself lacks "punch" ( the finished amps, by the way, will be driving a pair of ~96 dB Klipsch RF3's).

Below is the schematic of the current configuration as it sits on my workbench:



http://image72.webshots.com/172/5/4/...3XlUgSO_fs.jpg


I've read Talbot M. Wright's "Improving the Williamson Amplifier" article, but the major bass improvement involves a large cathode bypass capacitor for the output stage. I'm using fixed bias as you can see from the schematic above.

I know this sort of question has been asked before on this forum, but I wasn't quite sure how to apply that advice to my situation. Please pardon me if I'm out of line.

Oh - and Eli Duttman: You're absolutely right about good output iron. I didn't mean to sound like a know-it-all in my previous post. In fact I took your advice and bought decent iron - a pair of A431-S's from Triode Electronics. Should be there when I get home this evening. Woo Hoo!!!

Thanks!

[audiopro]

What is the purpose of the two 470k resistors that connect from the driver stage plate to the 0.1ufd/1k resistor junction of the output tubes. Not only are they going to inject positve voltage on the grid, but they are creating a peak in the frequency response. being in parallel with the 0.25 ufd cap.
Make the caps bigger too.

[zman3]

The basis for this project is an article titled High-Power Williamson Amplifier for Hi-Fi from the December 1955 edition of Radio-Electronics. The article states that those resistors assist in the reproduction of low frequencies by reversing the phase shift caused by the coupling caps. Apparently this RC network acts as direct coupling at low frequencies and as capacitive coupling at high frequencies.

Those resistors in the article's schematic differ from the ones shown on the schematic I posted above (47k and 470k vs. 100k and 1M). This is because I built from a different schematic (see page 13).

I'm not sure I understand all of this, but that's why I'm breadboarding.

[richwalters]

You mention the woofer moving.. Is that a very low sub audio freq movement irrespective of music ?
Ironically, to get more bass , reduce the value of 0.25uF(250nF) coupling caps to around 10-47nF. This may seem a backward step, but it isn't, as the LF circuit damping constants are (known problem in the Williamson) is too high, slow to decay and one has to reduce the loop gain at the low F end. This is one of the biggest misunderstandings.
I'd bet if you measured the above circuit thd at near full o/p say 20-30Hz with values in diagram, the figure will be quite low typ 0.2%. I find the optimum transient damping for these types of circuit requires far lower interstage cap value as values quoted. So loop gain is reduced, thd will rise around 1% at 20Hz or so, but you should find the bass fuller with much more attack.
Really the cathode bias o/p stage caps introduces a 6dB droop pole in the response, which helps.

Reference reading "valve amplifiers Morgan Jones 3rd ed" section classic power amps page 412.
This shows a modified Williamson circuit, is copyright protected so I cannot paste a diagram.

hope this helps.

richy

[Osprey]

I would definitely experiment with removing those 470k resistors. I would also be careful in using that pot for the feedback. Adjust your signal generator input V to give a 10V output and then adjust the pot to give you 2.5V out. That will give you a baseline of -12dB. That is much less than the original -20 dB if I remember correctly. That will help with stability and probably sound better too.

Another way to establish the feedback would be to adjust a 10V output to 1V using the feedback pot. Measure the resistance of the pot and then put a resistor of that resistance in series with the pot. You will then be sure to never increase the feedback more than 20dB. That will help to prevent any accidental oscillations (motorboating).

If you don't have a signal generator here's a cheap but effective one that goes on sale periodically:

http://www.mcmconnect.com/tenma/product/72-490/

[zman3]

Thanks for the feedback, guys.

Last night I swapped the Hammond 1650R for the new A-431-S. Boy did that make a difference! The bass was still rather lacking, but the rest of the music was much more clear and detailed. I then figured the culprit lay elsewhere so I implemented some of the modifications in the Talbot article mentioned previously. The schematic currently looks like this:


http://image66.webshots.com/566/8/77...3ArDGUB_fs.jpg

First I changed the values of the resistors (shown in red in the above schematic). This made the treble MUCH more clear and bright but did little to help the bass. Finally I placed a bypass cap across the 1.2k resistor in the input stage. That did it! I slowly added capacitance until I came up with the 870 uF combination shown. The amp sounds quite nice now and the bass is very well-defined with a lot of "punch". My only problem now is that the highs sound a bit harsh. It's not distorted, mind you; it's just that it seems to have a bit of "edge" that makes it uncomfortable at higher volume levels. So now I have more questions regarding smoothing out those highs:

• Should I go back to the original resistor values?
• Will adding a bypass cap across the feedback resistor help?
• Is there anything else I can do to smooth out the high frequency response?

Rich: To answer your question about the woofer moving, it was moving with the music. I can induce low frequency instability (something I don't like to do) by adjusting the feedback pot until I can see the woofer pumping in and out slo-o-o-wly, like less than 0.5 Hz. I'm just not sure if I'm increasing or decreasing the feedback as I do this.

Thanks again!

[richwalters]

[QUOTE]Originally posted by Osprey
[B]I would definitely experiment with removing those 470k resistors.
-----------------------------------------
>Nooo- don't do that... it will leave the 2nd 6sn7 stage grids without 0V bias reference-
>Remove the 870uF cathode bypass cap from the 1st stage. You are effectively shunting the global NFB into nothing.

The 3K9 and the 50uF cap in the anode B+ of concertina is a good thing as it improves the supply rejection performance of this node.
The extra 47K in the B+ first stage might upset the workings of the concertina. What DC voltage are you getting on the 22K concertina cathode ?? Reason: You may not be getting the optimum voltage swing.


richy

[HollowState]

By adding that 870 uFD cap across the first cathode resistor you have grounded out all your global feedback. By increasing the resistor's value you've altered (leaned out) the operating point of that stage. You've also increased the input sensitivity because now there is no cathode degeneration (feedback). Who are these X-spurts that write these dumb articles?

[Osprey]

Nooo- don't do that... it will leave the 2nd 6sn7 stage grids without 0V bias reference

The 0V bias reference is ground through the 47k and 100k resistors.

[richwalters]

Originally posted by Osprey


The 0V bias reference is ground through the 47k and 100k resistors.

sorrry... had all eyes on 2nd stage grid config... I was looking at the front end bits....also has no phase correction on first stage anode ?


richy