Hello folks,
In an attempt to further my DIY "career"
, I've breadboarded the Acrosound "High Power Ultra-Linear Williamson" according to the following schematic:
http://www.webace.com.au/~electron/tubes/Circuits/achpwill.gif
I have made the following changes:
Please bear in mind that this is my first attempt to build an amp solely from a published schematic rather than a magazine article (I've built several that way).
With all that out of the way, I'm having a devil of a time trying to get this thing to behave! There are two main problems I'm having.
First, I suspect the voltage at that 175 uf cap (the one directly after the choke) is WAY too high. I'm reading 510V, and I'm guessing the value should be somewhere around 450V. This is even after running the whole works off a variac (my mains voltage runs about 122V). I realize most of this is because of poor transformer regulation - I'm presenting a 160-175 mA load to a transformer designed to take 465 mA. Any ideas on how to tame this beast? Would a couple of 50 ohm power resistors in series with the rectifier plates help?
Second, I can't seem to get the output stage to stabilize. Tubes are "matched", but even when I crank the plate voltage down to 450V, -60V is still required on the grids just to keep the thing from motorboating, howling and drawing excessive current. What to do?
Thanks folks!
In an attempt to further my DIY "career"
, I've breadboarded the Acrosound "High Power Ultra-Linear Williamson" according to the following schematic:http://www.webace.com.au/~electron/tubes/Circuits/achpwill.gif
I have made the following changes:
- The output transformer is a Hammond 1650R (5k 100W and configured for 8 ohm output) as opposed to the Acrosound TO-330 (3.8k). Not the same, I know, but it's what I happened to have in the "junk box".
- The Power Transformer is a Hammond 278CX (400-0-400 @ 465 mA). Grossly overrated, but (again) it's what I had on hand.
- Shuguang KT88 output tubes (these are newer versions and test ok on a Hickok 539A and in other circuits).
- The diode in the bias supply was replaced with a full-wave silicon rectifier.
- The 10 uf capacitor immediately after the 4H choke was increased to 175 uf.
Please bear in mind that this is my first attempt to build an amp solely from a published schematic rather than a magazine article (I've built several that way).
With all that out of the way, I'm having a devil of a time trying to get this thing to behave! There are two main problems I'm having.
First, I suspect the voltage at that 175 uf cap (the one directly after the choke) is WAY too high. I'm reading 510V, and I'm guessing the value should be somewhere around 450V. This is even after running the whole works off a variac (my mains voltage runs about 122V). I realize most of this is because of poor transformer regulation - I'm presenting a 160-175 mA load to a transformer designed to take 465 mA. Any ideas on how to tame this beast? Would a couple of 50 ohm power resistors in series with the rectifier plates help?
Second, I can't seem to get the output stage to stabilize. Tubes are "matched", but even when I crank the plate voltage down to 450V, -60V is still required on the grids just to keep the thing from motorboating, howling and drawing excessive current. What to do?
Thanks folks!
zman3 said:Hello folks,
In an attempt to further my DIY "career"
http://www.webace.com.au/~electron/tubes/Circuits/achpwill.gif
I have made the following changes:
- The output transformer is a Hammond 1650R (5k 100W and configured for 8 ohm output) as opposed to the Acrosound TO-330 (3.8k). Not the same, I know, but it's what I happened to have in the "junk box".
- The Power Transformer is a Hammond 278CX (400-0-400 @ 465 mA). Grossly overrated, but (again) it's what I had on hand.
- Shuguang KT88 output tubes (these are newer versions and test ok on a Hickok 539A and in other circuits).
- The diode in the bias supply was replaced with a full-wave silicon rectifier.
- The 10 uf capacitor immediately after the 4H choke was increased to 175 uf.
Please bear in mind that this is my first attempt to build an amp solely from a published schematic rather than a magazine article (I've built several that way).
With all that out of the way, I'm having a devil of a time trying to get this thing to behave! There are two main problems I'm having.
First, I suspect the voltage at that 175 uf cap (the one directly after the choke) is WAY too high. I'm reading 510V, and I'm guessing the value should be somewhere around 450V. This is even after running the whole works off a variac (my mains voltage runs about 122V). I realize most of this is because of poor transformer regulation - I'm presenting a 160-175 mA load to a transformer designed to take 465 mA. Any ideas on how to tame this beast? Would a couple of 50 ohm power resistors in series with the rectifier plates help?
Second, I can't seem to get the output stage to stabilize. Tubes are "matched", but even when I crank the plate voltage down to 450V, -60V is still required on the grids just to keep the thing from motorboating, howling and drawing excessive current. What to do?
Thanks folks!
Dude,
You are flat SOL!
Williamson style topology works ONLY when the O/P "iron" is creme de la creme. As you have found out, mundane stuff (like Hammond) gets you eaten alive by the phase shifts.Either you buy superior O/P trafos or you change topology. IMO, switching to Mullard style circuitry is your best option. CCS loaded 'SN7 sections as the voltage gain devices and 2X 12AT7s as the LTPs will give you a good gain structure and an excellent net HD spectrum. Don't forget to use a 10M45S as the LTP tail load, instead of a resistor. 12AT7 section voltage amplifiers and ECC99 LTPs are another workable small signal combo.
Running this thing open loop was going to be my next try.
As for using Hammond iron, I've always had decent success with it, considering its limitations. I built Rick Spencer's beginner's amp (that little 12V6/12SL7 job from a 2001 Audio Express article) and was extremely pleased with it. I've also breadboarded Still's 70W amp from the 1999 Glass Audio project book. This also worked, although it seemed to sound a bit more harsh.
Both of these ran in ultralinear mode, both used global feedback, and both used Hammond iron. Neither suffered from any instability problems (its worth noting the 70W job ran 550V on the plates, and the tubes were running at pretty close to their total dissipation limits).
My main focus is to build something around 50W that's SIMPLE, cool, reliable and relatively inexpensive - and maybe to learn a little in the process. I sort of figured I could worry about acquiring "boutique" iron later on.
Thanks!
As for using Hammond iron, I've always had decent success with it, considering its limitations. I built Rick Spencer's beginner's amp (that little 12V6/12SL7 job from a 2001 Audio Express article) and was extremely pleased with it. I've also breadboarded Still's 70W amp from the 1999 Glass Audio project book. This also worked, although it seemed to sound a bit more harsh.
Both of these ran in ultralinear mode, both used global feedback, and both used Hammond iron. Neither suffered from any instability problems (its worth noting the 70W job ran 550V on the plates, and the tubes were running at pretty close to their total dissipation limits).
My main focus is to build something around 50W that's SIMPLE, cool
, reliable and relatively inexpensive - and maybe to learn a little in the process. I sort of figured I could worry about acquiring "boutique" iron later on.Thanks!
Re: Re: Acrosound High Power Williamson - HELP!!!!
Hi Eli,
Hey, both problems mentioned by the OP, "motorboating" and "screaming", are not necessarily connected to using a different (inferior) xfrmr than the original suggested one, but also can be caused by many other factors. Think wiring regarding the "screaming", think changed time constants in the PSU (the OP mentioned this!) for motorboating.
For sure, even the original Williamson using the original Partridge OPT showed quite doubtful LF stability by itself - this has been pointed out by several people who should know what they are talking about (f.e, Morgan Jones, Claus Byrith, et al).
From own experience, you can get a Williamson unconditionally stable even with comparatively cheap Hammond iron, but I wholeheartily would agree that this is not a task for beginners, and "better" iron will help, indeed.
On the other hand, I doubt that more than say, 10% of any and all commercial designs and DIY amps featuring more than 2 stages plus a gNFB path would qualify for being "unconditionally stable"
Regards,
Tom Schlangen
Hi Eli,
You are flat SOL!
Hey, both problems mentioned by the OP, "motorboating" and "screaming", are not necessarily connected to using a different (inferior) xfrmr than the original suggested one, but also can be caused by many other factors. Think wiring regarding the "screaming", think changed time constants in the PSU (the OP mentioned this!) for motorboating.
For sure, even the original Williamson using the original Partridge OPT showed quite doubtful LF stability by itself - this has been pointed out by several people who should know what they are talking about (f.e, Morgan Jones, Claus Byrith, et al).
From own experience, you can get a Williamson unconditionally stable even with comparatively cheap Hammond iron, but I wholeheartily would agree that this is not a task for beginners, and "better" iron will help, indeed.
On the other hand, I doubt that more than say, 10% of any and all commercial designs and DIY amps featuring more than 2 stages plus a gNFB path would qualify for being "unconditionally stable"
Regards,
Tom Schlangen
Hi,
Do not exceed the maximum value of the capacitor-input filter for your particular rectifier tube. This value is stated in the data sheet.
The original Williamson used a 20H choke between the HV applied to the O/P transformer and the pre-amplifier stages. This normally avoids motor-boating (poor supply regulation).
See the rectifier tube data sheet if it is possible to increase the size of the 4H choke. The DTNW amp used 10H.
A choke-first filter is best. But you would have to increase the output voltage of your supply transformer. Unless you are happy with a bit less power.
Try removing the 47pF (47MMF) in the feedback loop. One of my DTNW clones oscillated because of such a cap.
Good luck.
Serge
Do not exceed the maximum value of the capacitor-input filter for your particular rectifier tube. This value is stated in the data sheet.
The original Williamson used a 20H choke between the HV applied to the O/P transformer and the pre-amplifier stages. This normally avoids motor-boating (poor supply regulation).
See the rectifier tube data sheet if it is possible to increase the size of the 4H choke. The DTNW amp used 10H.
A choke-first filter is best. But you would have to increase the output voltage of your supply transformer. Unless you are happy with a bit less power.
Try removing the 47pF (47MMF) in the feedback loop. One of my DTNW clones oscillated because of such a cap.
Good luck.
Serge
zman3 said:
Actually, that voltage doesn't seem to be too out of line.
First thing: break that gNFB loop and see what happens when running totally open loop. First, get it stable, then worry about adding gNFB. Secondly, when you connect the gNFB, connect it to the secondary tap that you are actually using. Simply connecting to the highest Z tap can cause lots of problems due to stray capacitance and the parasitic resonances they can cause.
Next, get the level of gNFB down. The original Williamson required these uber special OPTs since the original design used insane amounts of gNFB. No well designed hollow state rig requires that much gNFB these days. A big part of the problem was that back when the Williamson was designed, speaker design was very much a "black art", and lots of vintage speeks did wierd things at certain frequencies. The amps needed a lot of damping to tame those beasts. Modern speeks are designed better, since the advent of the Theil design methodology. Hollow state amps start sounding like solid state amps at ~12db(v) of gNFB. Once you get up to the 26db(v) that some Williamsons used, well, might as well go to Circuit City and pick up a SS amp: lots less frustration, and you won't notice the difference anyway.
Less gNFB means less stability troubles, and it'll sound better too.
If you're having problems getting the finals to stabilize, then you may have gassy finals. If this possibility is ruled out, then look for high frequency oscillations. If it's trying to make RF, that will cause this problem. You'll need to look for that with an o'scope, or perhaps a shortwave xcvr operated nearby. If it's making RF, you can pick it up on the SW. From what I'm seeing here, I suspect you have problems at both ends.
Whew doggies, Jethro!
It seems I've opened a can o' worms here. Much of what you guys are saying seems to make good sense.
I've had to suspend work on the amp this evening, as tomorrow is the first day of classes at my university. I'll probably get back to work on it tomorrow night and I'll definitely try out your suggestions - especially opening the NFB loop.
Thank you guys for all your help!
It seems I've opened a can o' worms here. Much of what you guys are saying seems to make good sense.
I've had to suspend work on the amp this evening, as tomorrow is the first day of classes at my university. I'll probably get back to work on it tomorrow night and I'll definitely try out your suggestions - especially opening the NFB loop.
Thank you guys for all your help!
Removing the NFB loop worked! I immediately noticed a difference upon power-up.
That nagging high plate voltage remains a problem, however. My main idea that it's running a bit hot comes from another version of the same schematic, this one from the Acrosound 1955 catalog:
Note the 450 volts entering the OPT. Those shunting resistors I spoke of earlier have increased to 1 megohm, and the balance pot is omitted entirely (they must have used one a well-matched pair of tubes!).
On my setup with ~500 volts Va, my plates begin to exhibit the red glow o' death with the bias set to -66 V, which is a far cry from the -48 V called for on the original schematic. Would a 400-ohm power resistor in series with the 4H choke be a good way to achieve that voltage drop?
That nagging high plate voltage remains a problem, however. My main idea that it's running a bit hot comes from another version of the same schematic, this one from the Acrosound 1955 catalog:
An externally hosted image should be here but it was not working when we last tested it.
Note the 450 volts entering the OPT. Those shunting resistors I spoke of earlier have increased to 1 megohm, and the balance pot is omitted entirely (they must have used one a well-matched pair of tubes!).
On my setup with ~500 volts Va, my plates begin to exhibit the red glow o' death with the bias set to -66 V, which is a far cry from the -48 V called for on the original schematic. Would a 400-ohm power resistor in series with the 4H choke be a good way to achieve that voltage drop?
Go to MOSFET Follies and scroll down to the "B+ Reducer", that'll be better than a series dropping resistor, which will cause a tremendous degradation of voltage regulation.
Thanks, Miles!
I didn't bookmark that page - I saved it! That is an excellent little tutorial on how to use such solid-state devices for hollow-state purposes.
I've seen schematics on other websites that exhibit strange hybrid topologies. This page gave me a really good insight into just what may be going on in some of those.
Donkey Shins!
I didn't bookmark that page - I saved it! That is an excellent little tutorial on how to use such solid-state devices for hollow-state purposes.
I've seen schematics on other websites that exhibit strange hybrid topologies. This page gave me a really good insight into just what may be going on in some of those.
Donkey Shins!
Yes, that's the way I'd prefer to do it, too.
This is a way of getting the benefits of DC coupling without having to cope with the crazy voltages that can result. The coupling cap itself is not actually eliminated but the effect is almost the same as if it had been. This can tame a Williamson, or any other amp design with 'too many' coupling caps for global NFB.
Ok, so now I've got a couple of options on B+ reduction.
IIRC these are rated for 1000V breakdown at 90W dissipation. Can I mount one of these to the chassis for heat dissipation, or will I have to use an isolated heatsink?
My apologies for the endless barrage of questions. With tubes I know just enough to be dangerous; when it comes to solid-state, I'm a veritable babe-in-the-woods.
Thanks again...
- Dropping the value of the input cap: My filter input cap consists of two 22 uf 450V caps in series, with 100k 1W equalizing/bleeder resistors. How far down can I go before my filter essentially becomes a choke-input filter? Not that I'm worried, mind you; My choke is 4H, which is within the ratings of the 5V4 I'm using.
- MOSFET/zener reduction: If I go this route, what specifics do I need to look for other than a high gate-source breakdown voltage? Mouser has these for 2.95 each:
IIRC these are rated for 1000V breakdown at 90W dissipation. Can I mount one of these to the chassis for heat dissipation, or will I have to use an isolated heatsink?
My apologies for the endless barrage of questions. With tubes I know just enough to be dangerous; when it comes to solid-state, I'm a veritable babe-in-the-woods.
Thanks again...
I wouldn't go much lower than that. If it hasn't been mentioned already you should download PSUD2 http://www.duncanamps.com/psud2/index.html and get to know it well.
The size of the heatsink depends on how much power is being dissipated by the FET. Look for something in a TO-220FP package. Its electrically isolated and can be chassis mounted with no worries. A quick search turned up this http://www.st.com/stonline/books/pdf/docs/9701.pdf Solid state isn't my strong point so maybe someone else could recommend something more tried and true.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.