Knight KA-95 red plating

[miles2go]

Good evening... I figured since cooler is arriving, I can put this amplifier into service, or at least that's the plan. The amp uses what looks to be an ultra-linear power amp with dual rectifiers. I will try to get a schematic up with the post. This was a kit amplifier and by all appearances, the builder did a good clean and tidy assembly. The electrolytics in the power supply were replaced just before I purchased the amp locally. The gentlemen was very honest and disclosed the bias issue as we corresponded, I ended up getting a considerable deal because of this.

Fast forward about six months to last night when I decided to take a closer look at the issue. I have a quad of Winged C EL-34's from another amp I used as test subjects and took some voltage readings after bring the amp up on the variac. It became evident once I hit around 100 VAC or so that the grids were hot. The schematic calls for 4V on the control grids and 40V on the cathodes. I took readings @ 115 VAC input and anywhere from 4.8V to 5.4V on the grids and 38V - 39V on the cathodes. There is a balancing circuit as part of the bias control which I believe may be the culprit since after balancing the pairs by way of the test socket yielded unbalanced readings directly at the grids to chassis ground. After measurements were taken, I dimmed the lights to look at the plates in darkness and noticed a glow like a line through the length of V11 and possibly a bit on V9 as well. My other concern is the screen grid measurements 5V higher than the plates, the schematic shows a 10V favor to the screens so this is obviously by design.

So my question is, what kind of precautions or modifications can be taken to avoid over dissipation? I am not opposed to cathode bias and by-passing the current bias control. I have looked at a number of Williamson type circuits and it seems that grid leak resistors are generally larger than the 150K found on this amp, also most schematics show a screen grid dropping resistor up to 1K in value. Would it make sense to add those to this design? Thoughts?

 

[gadget73]

Check the coupling caps into the grids of the output tubes. Any leakage there will give trouble.

This is actually a Mullard design, not a Williamson. The style of phase inverter and the pentode voltage amp are the giveaways there. The screen stoppers are a good idea, but I don't suspect thats the issue here.

This is also cathode bias. The balance circuit is just a simple voltage divider to allow current balancing, which is important on something like this. The voltage feeds back to the grids somewhat because the effective cathode resistor is actually too large. Its using tube heaters as the resistor, and it doesn't quite work out right. The cathode voltage back into the grid is the workaround. If the resistors in that circuit are off value it will mess with things and there are a bunch in there associated with the bias circuit.

oh, also worth a look, but are all the tube types correct? If someone stuck a preamp tube with a higher heater current in there it will mess with how things operate.

 

[miles2go]

Hi gadget, thanks for pointing that out... it's clear as day once you mentioned the pre-amp tubes acting as the bias string for the output tubes. This must have been common practice since I have at least two amps that used the same method. As for the coupling caps, I did change those out last night as it seemed like the first step in diagnosing excessive DC at the grids. The tubes are all original to the amp and have been verified good according to my tester. The only tubes I changed out were the output tubes because I didn't want to tax the originals until the bias got squared away.

 

[miles2go]

o.k. so it seems there is a discrepancy in the pre-amp section which could possibly be an issue. The circuit board which contains the pre-amp tubes is showing EF-86 as opposed to UF-86, which is what the schematic and part list shows as the voltage amplifier. I am not familiar with the UF-86 and I can't find much info regarding this tube. TDSL shows application data being the same, but no ratings available. It also states that the filament voltage is different to the EL-86. I wonder if this was an alternate circuit board that was changed at a later time, after the schematic was established.

Whatever the case, it is a more than a bit confusing and I wonder if this can even be compensated for very easily... Can't exactly refer to the schematic regarding the voltage amplifier.

 

[primosounds]

Yeah, i think it is supposed to be an EF86 which is the audio pentode that is commonly used in that circuit. The schematic you displayed is incorrect or is a bad copy. i checked another schematic as the one showed made me wonder and sho nuff, it shows an ef86.

 

[miles2go]

Thanks, so the schematic I posted is right out of assembly manual which was included with the amp. The builders check offs are seen throughout with a few notes as well, maybe there were variations as parts became more scarce and updates were made. The EF-86 is definitely far more common than the UF-86, which from what I found is the 12V version, regarding the filament supply. I noticed last night the resistors that run parallel to the filaments across V5 & V6 (R-125 & R-127) have been reduced to 91 ohms as opposed to the 220 ohms spec'd for UF-86.

I am not even sure this is the issue regarding the bit of red plating though, as the sum of filaments which make up the cathode voltage are right in the ballpark. I guess I need to check and verify the plate voltages back through the pre-amp sections, maybe the filaments as well as they cascade back to the input. Still don't understand why the control grids of the outputs are so out of whack.

 

[triode17]

The schematic calls for 4V on the control grids and 40V on the cathodes. I took readings @ 115 VAC input and anywhere from 4.8V to 5.4V on the grids and 38V - 39V on the cathodes.

Are you sure about this? 4v on the control grid?? Are you saying +4v or -4v? Either way, that's going to send the tube wild. Plus, 38v on the cathode shows a high current flow. The schematic you sent is unintelligible. Can you send a better photo?

 

[miles2go]

Are you sure about this? 4v on the control grid?? Are you saying +4v or -4v? Either way, that's going to send the tube wild. Plus, 38v on the cathode shows a high current flow. The schematic you sent is unintelligible. Can you send a better photo?

Sorry about the picture quality, here is another pic showing a close up of the schematic showing one side of the power amp. Hopefully this is better...

 

[gadget73]

It would be +4v. Thats actually proper. The schematic shows that some cathode voltage is sent back to the grid on purpose. Fisher did the same thing with some of their amps that ran tube heaters as a cathode resistor. Vk is some 40v, Vg is 4v, so you're effectively at -36v of bias.

http://oldtech.net/Amps.html

has the schematic in biggie size. That one calls for an EF86. UF86 would be the same tube with a 12.6v 100ma heater.

https://www.radiomuseum.org/tubes/tube_uf86.html

Possible both could be used, but it would change some values in the bias resistor chain. May be worth comparing the two schematics and the actual amp to see whats going on there.

 

[miles2go]

Dude you rock... I was hoping someone knew where to find the proper schematic. Trying to take pics of 50 some odd year yellowed paper with an Android doesn't seem to work that well The schematic you posted shows the voltage amps V5 & V6 with the proper resistor values for the EL86, which fortunately the original builder must have had access to since I have just now verified. This includes R-80 and R-89 as 3.3M instead of the 1M for the UF86, also the filaments on your link show V5 & V6 in series which has been correctly done as well. I will look it over more and see if there are other differences.

At this point I think I am going to focus on the output balance circuit, and perhaps replace the resistors with better quality, tighter tolerance metal film since it seems to affect the overall bias of the power tubes.

 

[miles2go]

The coupling caps C-49 thru C-52 being changed from. 25uF on the UF86 version to .005uF on the EL86 version seems extreme, I changed those out the night before last so they will probably remain at .25uF. I tested the amp with a Apex matched quad of JJ's a bit ago and aside from only running voltage to 112 VAC, it sounded quite nice. No signs of red plating, but still had grid voltage roughly at 4.75V and cathodes at 34 or 35V. Still a work in progress...

 

[gadget73]

.005 doesn't really seem sensible. That would have a -3db point around 200 hz. The amp would try to make full frequency but I'd expect a lot of LF distortion as it makes the attempt. A 0.05 would be a far more sensible value. 0.25 is probably OK as long as it doesn't have any low frequency stability problems.


The problem with this thing, the hotter the tubes bias, the more grid voltage, which makes them conduct more, etc. Its a positive feedback loop, which makes it prone to run away. Possible that new production tubes bias different enough that those 33K resistors need to be bumped in value to settle it down.

 

[miles2go]

.005 doesn't really seem sensible. That would have a -3db point around 200 hz. The amp would try to make full frequency but I'd expect a lot of LF distortion as it makes the attempt. A 0.05 would be a far more sensible value. 0.25 is probably OK as long as it doesn't have any low frequency stability problems.


The problem with this thing, the hotter the tubes bias, the more grid voltage, which makes them conduct more, etc. Its a positive feedback loop, which makes it prone to run away. Possible that new production tubes bias different enough that those 33K resistors need to be bumped in value to settle it down.

o.k. so maybe try 56K and see where that lands? This seems like a sensible approach to lowering the grids some, I would feel a heck of allot better maintaining the grids under +4V and also the 34V to 36V negative in relation to the cathode for a cooler bias. I think right now I am around 30V difference between grid and cathode, and if I am understanding the measurements correctly, around 65mA per tube since I am seeing .65V or so across the 10 ohm resistors on the test socket.. seems hot, though that may be for each pair. The target voltage is stated as .6V at 120VAC input.

 

[miles2go]

Oh yea, I just remembered, .22uF was the value I used for the couplers, .1 to .22 seems common so hopefully that will be o.k. The bass response seems tight and fairly focused as it is...

 

[gadget73]

yeah see what a 56k gets you.

Looks like its 10 ohms per tube, so 65ma each. If stock expects .6v, thats 60ma. If all the voltages come out exactly like stock, thats 26 watts plate+screen dissipation which is not unreasonable. Just the plate is probably around 23w Datasheet gives 25w for the plate using design center, and those are conservative ratings.

 

[miles2go]

yeah see what a 56k gets you.

Looks like its 10 ohms per tube, so 65ma each. If stock expects .6v, thats 60ma. If all the voltages come out exactly like stock, thats 26 watts plate+screen dissipation which is not unreasonable. Just the plate is probably around 23w Datasheet gives 25w for the plate using design center, and those are conservative ratings.

Hey gadget, I ended up replacing all the resistors in the output balance circuit. I selected the tightest ones I had, especially the 10 ohm ones since they measure the cathode current. I didn't stray to far from the stock resistors (R-104 & R-105) and (R-108 & R-109), I ended up using 39K instead. No doubt 47K or 56K would lower the grid voltages, but I didn't want to stray too far from the design specs if I could help it. It didn't change the power dissipation numbers, but did lower the grid voltages a bit. Right now I have from 4V to 4.25V which I am hoping will be fine. I have measured them at idle of course, and at various output levels and they don't seem to change.

I still plan on adding screen grid resistors, either 10 ohm or 100 ohm if I can fit the larger wattage ones in. I am not sure how i can even measure screen current right now. I suspect 10 ohm 1/2W would be o.k.

Maybe the voltages will settle in a bit more as the amp is used, plus the new power tubes will need more burning in I suppose. I get the feeling it has been stored for a long time and not used until the previous owner pulled it to sell and ended up changing the electrolytics. I am uploading a couple of pics, one of which shows the output balance panel on the rear of the amp.

Maybe one day I will recap the preamp/voltage amp board, that will take some planning though... So now I think I will enjoy the sound of the amp and the warmth it provides

Thanks again,
Miles

 

[gadget73]

So how do voltages compare with the schematic now? I usually read across the 6.3 vac heater winding and adjust the variac until I get 6.3v. Once thats right, if all is right with the world the rest of the voltages should line up reasonably well with the schematic. Most of these things have a 15% tolerance on voltages too, so if its in that window I generally call it good.

For the screen, 100 ohm will be fine, even 1/4w would be acceptable. Its not going to dissipate much at all. I usually use 1/2w because its what I have on hand. My Pilot SA-260 has a UL output stage, and the resistors are simply between the tube socket and the lead out of the transformer, with heat shrink over them to keep anything from shorting. Not fancy but I didn't have any better way of doing it.

 

[miles2go]

Ahhh... that is a good idea, I will try that. Right now I am just running it at 112VAC and the numbers are adding up to roughly 27W dissipation per tube. Plates: 428V, screens: 432V, cathodes: 35V and control grids from 3.6V to 4.2V. I am hoping 2 to 3 watts of that is through the screens.

 

[gadget73]

what are you getting across the 10 ohm resistors?

An actual EL34 will have more screen current than a 6CA7 so you're probably safe.

 

[miles2go]

o.k. so I posted averages earlier, here are the grid voltages per tube and the measurements across the 10 resistors:

V9 grid: 3.5V with .63V across 10 ohm
V10 grid: 4.1V with .64V across 10 ohm
V11 grid: 3.3V with .63V across 10 ohm
V12 grid: 3.8V with .60V across 10 ohm

If I adjust the control grids for balance, the current across the 10 ohm resistors for the pairs of power tubes becomes more imbalanced, which creates hum. As it is right now, it is very quite, but still seems a bit hot if I were to take it up to say 117VAC. I think if I could get more voltage at the cathodes, it would help. I am wondering if I adjusted R-124 & R-126 or R-106 & R-107, I could raise the cathodes V, but that would probably alter the grids V as well... not to mention, how would that affect the heaters down stream?