KT88 SE Abdellah (modified by Alex Gendrano) + AK..

Hello everyone,

I am going to build the Abdellah SE KT88 as modified by Alex Gendrano.

I have searched the forums and have not been able to find a specific reference to the purpose of the 5H 250R choke and 10uF cap for the 6N1P plate as shown on the schematic:

http://diyaudioprojects.com/Tubes/KT88/

To the best of my knowledge this forms an LC filter and provides 29dB of ripple attenuation for the 120Hz component of the B+ and a 6.25 V B+ voltage drop.

Many other folks have constructed this circuit and not included these components.

Can someone please expand on the inclusion of these components?

Many thanks.

 

First off welcome to AK :thmbsp:

I started a thread on power supplies "A Different Approach"

Here

http://www.audiokarma.org/forums/showthread.php?t=257784

It is by no means done as I am have a lot to talk about yet!

My own amps use no chokes or resistors in the supply path any DCR has a dynamic effect this goes for power transformers as well as supply paths. When its high for output stages the amp is often lean sounding IMO to the point of boredom for me.

Though my approach is very controversial vs the classical approach it does yield me a level of performance that I was never ultimately prepared for.

SET12

 

Thanks for the welcome. I read your "A Different Approach". Interesting concept. Thanks.

I've been into electronics since about 1976. Unfortunately for me, a totally different field than Vtubes. I'm still trying to figure it all out.

 

The (5H 250R choke and 10uF cap for the 6N1P)

Form 2 functions, cleaning the signal for the driver tube, as single ended circuits have
very little if any rejection to power supply noise, so should utilize a clean supply to be
quiet on there output, also it forms isolation from the output stage.

Myself I would use the choke if I had one laying around, but a simple resistor in place
of the choke is fine, 250ohm resistor should keep the voltage the same, I'd say up to
1k is good there. (1/2 watt will do, 1 watt is more then enough)

 

Also myself I would run that circuit a bit different as well.

I'd drop the grid resistor down on the output tube to something in the 220k range,
I'd add a 1m grid resistor to the input tube for volume pot wiper safety, I would be
running in UL with feedback coming from the output plate to the driver plate like I
have it on the attached schem, I'd remove that bypass cap/resistor on your driver
tube an replace it with (2) 2v red LED's in series, also 1.2k screen resistor on KT88.

(actually 1.66-1.7v LED's instead of the 2v ones here to gain a bit of current now)

That is the basic circuit I've been running on SE Pentode amps running in UL mode.
(on the original amp I'd at least drop that KT88 grid resistor down to 220K)

'
'
(addition to the LED's, on the bottom side, replace 1 led with a 100ohm resistor)
Small say 100ohm unbypased cathode resistor on the driver may be a tweak to try.

 

Kegger's method of feedback is a great idea as it eliminates the speakers back EMF influence.

LED's are a great idea as well, I haven't used resistor and bypass caps in yrs after hearing LED's. Also you might want to pull a few ma more thru the LED's with a pull up resistor attached to the B+ as this will keep the forward bias voltage well into its knee keeping its forward voltage locked.

Personally I would use a slow turn on rectifier allowing for more capacitance. and smaller DCR in the supply path. A 250 ohm resistor makes the amp wimpy IMO.

Or if you wanted to use the same rectifier tube I would use a choke with the lowest DCR you can find in a 1H choke the thing to remember is its not the Henrie's its the DCR that effects dynamics and the 1H choke is equal to a 750 ohm resistor! Hammond has a 1H that is just 5.6 ohms! They have a 5H with 26 ohms as well.

SET12

 

I haven't checked "that" circuit, I believe at least 5ma should be going through the
LED's though, and if so that should be fine, yah if not you may want to supplement.
(use 1.66/1.7v LED's instead of 2v LED's to raise the current above say 4ma)

I agree I would use a 5V4G, 5AR4 or GZ37 rectifier instead of the 5U4 used original.

The 250ohm I was referring to is just the RC or LC stage used for driver decoupling.

 

Kegger-
You've got me interested in a few of these changes. I'm interested mostly in what the effect is of dropping the grid leak resistor to 220k on the KT88? I've seen this value on TubeLab's Simple SE design for a similar amp.

Glenn

 

Thanks very much for your input and explanation. I may have to try the original and these alternate values, and try with and without feedback for comparison.

 

That is generally a good way to approach it, an experiment for yourself along the way.



----------------------------

The grid leak resistor value is to put it in line with the actual tube specs, now here at the
"lowish" voltage and plate dissipation it may not make any difference on most tubes if any
but at higher levels of voltage an plate dissipation you'd run into tubes that just don't like
it and may runaway on you, the spec on a KT88 running cathode bias is 220K with a 20%
leeway, on this circuit with the previous stages output impedance being relatively low the
220K is a fine value for it to see so there is no harm in going down to 220K here anyway.
(in other-words a safer value that won't hurt the circuit, so might as well do it)

 

Kegger,

After delving into the KT88 further, I will use the 220K grid leak resistor.

If I am running in triode mode, do I need to use feedback, or should I switch it out along with the 1.2K?

 

No, no "need" for feedback in triode mode, yah if going triode I'd use around 100ohm for resistor on the screen grid.

 

I did do a quick check on your circuit, and I do have a suggested alteration.

(as SET12 alluded to)
The current on the driver is actually a little low, also if using a 6DJ8 instead
of the 6N1P the plate voltage is rather high, I'd suggest go 510ohm cathode
resistor instead of the 1k, and if using LED's use a pair of 1.66v series ones.

That should have both the 6N1P and the 6DJ8 work well in that circuit then.
(might even want to drop there plate load resistors down from 47K to 36K)

EDIT: Modified the previous circuit I posted to reflect the changes.

 

I think I'd like to try the LED bias to see what it sounds like.
Can you suggest a source for 1.66v LED's? I was looking at the red NTE 40031, and they list it at 1.86v typ. http://www.nteinc.com/specs/30000to30099/pdf/nte30037_43_45.pdf
Is this the type of component that you have to sort though to find the right forward voltage drop? (In circuit)

Also, I'm going to stick with the 6N1P (for now), so should I stick with the 2V LED's that you originally posted on your schematic?

I'm still learning about these different biasing approaches.

Thanks
Glenn

 

I believe I understand enough about this now to give it a go. It may be a while before I actually power-up since after pricing all components it appears that this is about a $600 project. So while I am collecting parts, I will be working on the chassis and wiring layout, and studying more about these concepts.

I want to thank you again, and everyone else, for the invaluable information.

Best regards.

 

"Typically" with the currents were dealing with the LED voltage drop will be
a bit less then that "normal" spec on the LED, so a 1.7v LED should do fine.

And yes sometimes you'll have to sift through the LED's to match them up.
(there cheap when bought in bulk, so I buy bulk lot's off Ebay often)

I'm sending you a PM with an Ebay link for 200 1.7v red LED's for $5.00 F,S.
(I looked up the PDF on these an look quite good for those low currents)

And yes I would use these even on the 6N1P, want to get that current up.
(you can clip-lead the LED's in till you got what your after in circuit)

I actually used those 2prong computer sockets for fans to "roll" LED's once.

Yu want low brightness/luminescence for LED's when there used for biasing.

 

FYI
I built the Abdella SEP amp last year and discovered he has a small error in the original schematic. (and continues onto the Gendrano version)
I used the suggested James 6123 OPT and discovered when set to Z=5K the UL screen tap is actually at 35% not 43% (It would be 43% when using the 3.5K taps)
HTH
WopOnTour

 

Hi Glenn,

You would be surprised what a couple of red leds on the cathode of the 6n1p will do.

I tried it, and it stayed on my SE KT88 design. You might also try an LM317 current sink on the cathode of the 6n1p.

HTH.

Alex F. Gendrano

 

I think 4ma per 6n1p triode or somewhere around 4ma, cathode of 6n1p relative to ground is around 4V. A couple of red leds can be used per cathode, slightly lower current, but better sounding.

I have tried 5AR4 on this circuit and somehow the sound became a bit clinical (for a lack of better term, a bit forward will also describe the sound), this might be due to the fact that the 5AR4 drops less voltage than the 5U4GB that I am using.

I have built monoblocks of this design, using EZ81 for rectifier, and I liked it better, I kept the B+ at 400VDC.

-Alex F Gendrano