|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
75.168.141.10
In Reply to: RE: 36KD6 posted by Lew on June 09, 2020 at 16:20:54
Where is Lew and what have you done with him?
Seriously, after all the mods you've done, this one is pretty easy and might just solve the issue.
Follow Ups:
that I have stuck to every single element of the original schematic, which I was lucky to have obtained from Roger Modjeski a couple of years ago. There are no modifications, except new parts and better capacitors in some cases. I finally got out one of my oscilloscopes and did the dangling coil of wire trick to see if I can detect oscillation. By this criterion there is no oscillation, now. (I fixed one more iffy looking solder joint before that.) So, I am thinking maybe the very low voltage, high frequency AC I am picking up with my Fluke frequency counter is of such low magnitude that it won't be audible. Also, it may always have been there, since I have never used the Fluke in this way before.
What I might do is to move the 1K grid stop closer to the grids they serve. Beveridge built them sitting in space about half way from the input board to the grid pin of the output tube, about 1" to 1.5" from the grid of the driven output tube. I have seen many photos of other Beveridge direct-drive amplifiers; they were all built like that. Even the schematic depicts a distance between the 1K and the output tube grid. (But no distance is stipulated in the detailed schematic.) In this one of my two amplifiers, they are perhaps closer than normal to the input side, which leaves maybe 2" to the output tube. Back when Bill Thalmann did work on these, right after I bought them, he added some ferrite beads between the body of the 1K resistor and the input board, to suppress oscillation. Bill did the same on the 7.5M resistor lead. I haven't messed with that.
-
We'll see what that does. The bodies of the 1K resistors are now within an inch of the grids. Not "close enough" normally, but close for any Beveridge amplifier I've seen.
-
I think yes.
In the course of making repairs after the fuse blew for the second time, which event also took out some of the filter capacitors and some of the current sharing resistors in the power supply output, I substituted carbon comp resistors with metal film resistors. Because the latter are physically larger than the originals, I more conveniently mounted them on the backside of the circuitboard. As it turns out, that was a big no-no, and was causing the noise. Whether it can be called oscillation or just noise I don't know. An EE tells me that it doesn't meet some of the definition of an oscillation, because there are many many mixed frequencies going on at the same time. Anyway, when I remounted those metal film resistors on the other side of the circuitboard, the noise is definitely reduced by at least 90%. I also found some high-voltage wiring that was weakening at its solder joints, because it's very thin gauge and flexing was weakening it. I resoldered those joints, and that helped too.But those metal film resistors were definitely the major culprit.
If moving the resistors like that did it, it is without any doubt an oscillation. What happens is the oscillation frequency gets rectified in other parts of the amplifier and you get all kinds of intermodulations that basically sound like noise. That EE you spoke to does not have a lot of hands-on experience with things like this.
So, when I reported a 90% improvement, I hadn't yet moved all of the current sharing resistors to the safe side of the PCB. Three out of 8 were still on the sensitive side of the board. I finished the job yesterday and was hoping the result would be a small further lessening of the noise, which I am measuring by AC voltage on various parts. To my chagrin, that is not the case. In fact, after moving those last 3 resistors, the measurements are not quite as good as they were yesterday. I changed nothing else. It's not a complete regression, but not an improvement either. And for all I know, the amplifier would "work" if I connect it back to the speaker. It's getting to the point where I have to just try it.
I have one more trick to try first. Years ago when he was also trying to eliminate oscillation after I first purchased the system, Bill Thalmann put some ferrite beads on the leads of the 1K grid stop resistors, on the side nearest to the low voltage input stage. When I moved the bodies of the 1K resistors closer to the grids of the output tubes they serve, I also moved those ferrite beads closer to the HV output board. I wonder whether THAT is now contributing to the random noise. I will move them back toward the input board and see what happens.
Hm.
This really supports the idea that the layout and component choice has resulted in a circuit that is unstable and on the verge of oscillation on any given Sunday. Moving the current limiting devices should have had no effect whatsoever on the noise. But if you were having an RFI problem, this could have had a big effect- which later might not have appeared to do what it did earlier. RFI is like that- its inconsistent. An oscillation essentially produces RFI in the circuit- and as a result can act almost exactly the same, as the radiation gets everywhere in circuit.
I really would change out those stopping resistors, as 1K seems entirely inadequate. From your description (and what I've seen of them), the amp seems to have been designed around a 'by gosh and by golly' approach rather than the math actually worked out in all the areas. The 1K stops seem to me an example of that- more an acknowledgement that you need them but not actually a value that would do what they are meant to do. Its not a big deal- you've already moved them. If they don't calm the amp down a bit you can always change them back.
Noise went down when I moved the 1K resistors AND the ferrite beads back to where they were when this saga started. For what it's worth, now that the levels are lower, I can trace the noise to the top tube in the totem pole that feeds the screen. That tube has no direct connection to the 1K resistor, which of course supplies bias and signal to the grid of the bottom tube in that totem pole. But I can see that most remaining noise is on the 7.5M resistor that attaches from the input stage to the cathode of the top tube and then to the screen outputs. (I assume it's feedback, but I could be wrong.) Also, this is with nothing attached to the output. Could it be that loading the amp would reduce the electrical noise?
And yes, oscillation or at least noise, is a known problem with these amplifiers. If I did not love the sound of the system so much, I would not put up with it. But I do, and I am. Roger Modjeski had mods for the amplifiers, among which was using tubes in the input stage. But apart from that, I don't know what he did to them. Too bad, for us owners and for him, that he is no longer around.
I have a good schematic (bought from RAM) which I might be able to send to you, if you're interested.
I think I can give it a go on the speaker now.
I measured the speakers when all this started. From stators to screen the resistance is infinite. There is 1.5nF of capacitance from one stator across to the other, and as one would expect, there is about 3nF of capacitance between each stator and the screen. I bought some 6kV rated, 3nF capacitors off eBay with the idea to create a dummy load (inside an insulated box, of course). I'm a little afraid of trying that, however. Two caps in series where the stators connect at each end of the chain, and the screen attaches to the junction between the two capacitors. (On these speakers, the screen is low impedance and held at zero potential and is driven by signal voltage.) Each stator is at +/-3200V and also driven by signal voltage.
-
When installed, the amplifier is the the base upon which a 6-foot speaker cabinet rests. You can't see the tubes or anything else that is going on. You have to go on funny noises, either from the speaker or from the amp chassis.
-
The amp connects directly to the speaker via HV banana plugs. The stators receive +/-3200V. Can you imagine a "safe" cable for that? I've got 20kVDC wire, but not enough of it. I remember seeing a photo where some guy in Europe inverted his Beveridge speakers, so that the amplifier sits on top. Then he removed the "bottom" part of the clam shell chassis (which becomes the "top" part when the whole speaker/amplifier is inverted). Thus he could see and access the amplifier while playing music, if he stood on a step ladder, because the speaker alone is at least 6 ft tall. I honestly don't want to be standing on a step ladder next to +/-3200V that is riding on a structurally "tippy" cabinet. I guess for test purposes, you could lie the speaker on its backside and use 20kV wire terminated in HV banana plugs on the amplifier end and HV banana jacks on the speaker end, to link the banana jacks on the amplifier, which would be sitting on the floor next to the speaker, to the relevant male banana plugs on the speaker.
Up to 10kv test prod wire is readily available. You will be able to run 6-8' of cable between the amp and panel with banana plugs and jacks.This is what RM detailed to me when he was building and repair amps and testing with open panels.
I run about 6' of cable to my DD amps with high voltage silicone wire in as similar way
https://canada.newark.com/pomona/6733-0/test-lead-wire-silicone-18-awg/dp/63H2513
Edits: 06/26/20
Yes, I am sure that COULD be done, but I wouldn't want to do it except in the case where I am trouble-shooting. Otherwise, I find it nice and neat and safer to have the Beveridge amplifiers serve as the base for the speaker with no external wiring. This system is in our finished basement which I like to keep only dimly lit while listening to music. As it is, the IC from the preamp to the L hand Beveridge travel across the floor between me at my listening position and the speakers. More than once I have tripped on that particular IC on occasions where I was turning the system on or off. I'd hate to do that with speaker cables carrying +/-3200VDC.
Is there a particular reason why you have chosen to use external speaker cables? I guess it may be inherent to the fact that you converted Model 3s to DD amplification.
I'm only suggesting this as method to troubleshoot.
For my Model 3's the DD amps were never part of the structure so I use HV cable to tie the amps to the panels.
I appreciate your input greatly.
ebay link
-
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: