View Single Post
  #23   Report Post  
Posted to rec.audio.tubes
hugeshows hugeshows is offline
external usenet poster
 
Posts: 29
Default Restoring a Sherwood S-5000

Hello all,


So I am finally getting around to writing up the place where I decided to call it quits, electrically speaking. Generally, I find the coupling caps to be good in the S-5000, and so when there are no indications of problems, I leave them alone. I think to do otherwise changes the character of the amp, and that is not something that I either encourage or discourage, because it's a personal choice. However, I think that for my tastes, once the power supply is up to snuff and everything checks out electrically and the amp sounds good, that's when I usually stop.

First off, the power supply rebuild... As you know, I went with the reproduction twistlocks from CE, which I purchased from AES. These capacitors seem to be of high quality, though I did have issues with the 1" diameter can and had to get a replacement. Regrettably, even the second one was a bit out of round at the base as if the crimping machine were having issues with that size, or the die was out of round. Anyway, I decided to install it and it worked out fine. There was nothing remarkable about this cap except that it was a 20/20 cap and the stock was a 30/20. So I added a 10uf cap in parallel to the underside and wound up with over 35uf measured for that section.

The bias supply is what turned out to be the most challenging, since it is sort of an odd circuit, the way its positive side is floated from the chassis. The main issue is that I decided to rebuild the voltage dropping resistors using the stock parts and obviously values as well. The result of this is that the increased voltage from the silicon diodes in the bias circuit along with resistor values that were derived for a selenium rectifier caused an elevated bias voltage which made the output tubes run rather cold. Keep in mind that the bias adjustment pots on this amplifier merely adjust the balance of bias voltage on each channel, they do not control overall voltage and there is no way to match each tube's current draw across channels.

After rebuilding the circuit with the stock resistors and new capacitors (underneath) I started taking measurements. I was seeing that the output tubes were drawing less than 15ma plate current at idle, and that's cold for any 7189 amp even at 440v plate. Bias voltage was in the 24v range, and the DC filament supply for the phono section was at about 26.5. Clearly, the stock circuit could not supply the right voltages with a silicon rectifier. Rather than attempting to re-engineer the bias and DC filament circuits just now, I decided to experiment with putting different value resistors in series between the diodes and the first capacitor, and I arrived with an interesting solution.

https://www.dropbox.com/s/gzmtpckjoy...ias_Supply.jpg

Keep in mind that those two green ground connections were later deleted in regards to the mistakes I mentioned in my last post. Anyway, as you can see, I've taken a very old Ohmite 10w wirewound resistor of the screw-mount type and mounted it inside the chassis, bypassed by a 40 ohm 2w precision resistor that Stephie sent me many years back. The resulting parallel resistance is 8 ohms, and that turns out to put the voltages right where I want them.

Why chose two resistors instead of one? Well, I wanted to be able to adjust the bias voltage a bit. I found that 10 ohms ran it just about as hot as I'd ever want the tubes to run, around 24ma a piece with some tubes! 8 ohms got my used Amperex 7189s to run at 21ma a piece, which is about where these Sherwoods run stock. Yes, that's a LOT for this type of tube, which is why you have to stick with ones that can handle the higher plate voltage. Anyway, the 40 ohm resistor can be removed by me at any point if I want to run things a but hotter, or have tubes that need a little less bias to get their current up. If I find a 2w rheostat to replace it with that I can mount internally without drilling, then I might use that as the shunt. Basically, by using a big overrated resistor like that 10w, I guarantee that my bias voltage won't fail due to a crappy pot. If I put a rheostat where that 40 ohm resistor is now and it faults open, the bias circuit fails hot, but safe because the 10 ohm resistor is in parallel.

So, with this arrangement I wind up with 23.6 volts at the top of the DC filament supply, and around 19v bias finally making it to the tubes after the balance pots. This gives me around 19-21 ma idle current per tube at 440v plate.

So, I've also modified that cathode resistors, and done something new - I reused the AC accessory sockets for bias measurement points, and it works brilliantly! You can either measure each tube individually, or measure the balance of a pair by connecting the meter to both points. Since the S-5000 is setup with a balance type bias circuit, this works out great and can be used in place of or along with the hum balance test described in the manual.. FYI- Hum balance in the S-5000 is not the classic hum balance circuit you find in other amps, it's really a fixed bias balance adjustment using injected and then cancelled hum to determine the center point.

So the bias has been left stock, except there is now 8 ohms between the diodes and the rest of the circuit. The dual 12 ohm 1w cathode resistors have now been replaced with 4 individual 1w precision 10 ohm resistors. From the cathodes, I ran a 4 conductor solid core ribbon wire over to the AC sockets which were reclaimed in the AC supply rebuild (more on that later). Here's how I did it:

https://www.dropbox.com/s/o20s9dxgon...sistor_Mod.jpg

The ribbon cable simply goes down between the output tubes, where it ties to each cathode. From there, it runs under the lip of the chassis towards the sockets. Just before the sockets, there's a screw hole. At that spot, I carefully bent a dip in the cable (solid core) and wrapped it with a bit of friction tape to protect it from future screws. And then each channel's pair was soldered to a socket, I laid it out logically.

Now I can measure each tube's bias, correlate "hum balance" to real-world figures, and see just how hot I'm running things without even opening the chassis. If you measure from each terminal to chassis, you see a number which is converted into ma with no math at all, save moving a decimal twice.

https://www.dropbox.com/s/jic4damuuo..._terminals.jpg

Here is the AC power input and the bias measurement points again. This is after I restored the stock chassis-line capacitor arrangement, but now that I had re-purposed the AC receptacles that the cap was mounted to, I had to add a terminal strip to the bolt on the transformer. From there, I re-wired the power supply to put the hot leg of the AC through first the fuse, then the power switch, then the transformer primary. Jumpers were made to connect hot to the new terminal strip and the cap, where neutral and the other primary leg was connected. The cap was grounded to the chassis at that spot, which seems to have the same hum level as its old position so far, fairly low.

I must say that I find having bias measuring points far more useful than an accessory outlet that dirties the noise floor when used, and despite a nostalgia for keeping things stock, this modification works well for me.

I'm getting sleepy, so that's all for now.


Cheers,

-forkinthesocket