Page 1 of 2

KT88 users - I have a question

PostPosted: Wed Jul 09, 2003 7:29 am
by Shannon Parks
I'm nearing the finish line on the Eicone rev a layout. Trying to keep alternate power tubes compatible with the board, I looked over the KT88 datasheet last night and see that pin 1 is tied to the can, not the cathode. Do people disconnect pin 1 in their 6550 amps when using the KT88's, or is having a small voltage on these cans no big deal? Actually, it be cool because you could check the biasing on the cans, methinks. Someone please set me straight here. Thanks!

Shannon

KT-88 Answer

PostPosted: Wed Jul 09, 2003 12:03 pm
by Thermion
I don't believe that pin 1 is connected to anything in either the KT-88 or the 6550. In the old Dynaco MkIIIs, pin 1 on the tube socket is used as an extra solder lug for wiring.

I like the idea of using Kt-88s in place of the EL34s in the Eiclone circuit. If someone needs the extra power, it would make a good option.

JT

re: KT88 & 6550

PostPosted: Wed Jul 09, 2003 4:52 pm
by Shannon Parks
Looks like some KT88/6550's connect to nothing, while others may be connected to the metal ring. FWIW, I'll put a jumper in there to pin 8 (for EL34, normal operation) so no cuts will have to be made.
BTW, JT - I think I'll layout the pseudo-triode resistors on the board like you mentioned the other day.

Shannon

Connected or Not?

PostPosted: Thu Jul 10, 2003 7:59 am
by Thermion
The discussion on this topic made me want to do a little more research. The old GEC spec for the KT-88 has pin 1 connected to the base ring. Everything I looked at for the 6550 show a NC for pin 1. BTW the KT-90 spec also shows no connection.

Next, I decided to check the KT-88s that I have around. I only have two different types, A Sovtek (Reflector) model from 1999, and a brand new JJ KT-88 just delivered last week. The Sovtek tube has pin 1 connected to the base ring, and the JJ tube is not connected. Go figure!

Makes me wonder why they would do this. My advice would be don't use pin 1 on the tube socket keeping the base ring unconnected.

FWIW, I have been using my MKIII's for about 12-13 years. At first I used GE and Russian 6550's. All of the 6550's required replacement after about 2000-3000 hours of operation. In fact, the last set of 6550's I used were Svetlana, and 3 of the 4 tested bad after about 2000 hours. I replaced those with the Sovtek KT-88s. They have been used for about 3000 hours over 4 years and they still test fine. I'm hoping the JJ's with be even better.

JT

KT88s do, some 6550s might

PostPosted: Fri Jul 11, 2003 5:19 am
by EWBrown
I looked thru my VTL "Pure Tube" book last night, the specs for the GEC KT88 indicate that pin 1 is connected to the metal base sleeve. Some 6550s have this, some don't, and I can also verify that EI KT90s do not.
My only SWAG for this is to provide some minor amount of shielding for the base connections if Pin 1 is grounded.

From what I read in the VTL book, KT88s can always replace 6550s, but 6550s can't always replace KT88s.

FWIW, KT = Kinkless Tetrode.

/ed B in NH

PostPosted: Tue Jul 15, 2003 7:01 pm
by Thermion
I've been burning in my new JJ Kt-88's for a few days now and I am very impressed. Much better than any 6550 I have used. Richer fuller sound, imaging and detail are impecable. And the low end is deep and tight, although it took a couple of days of burning in for it to develop. :razz:


The quad was very closely matched in transconductance, and the bias has been extremely stable after initial adjustment. These would be a good choice for anyone using a 6550/KT-88 amp.

JT

Aaaaacccckkkk!!!! (So Close!)

PostPosted: Tue Jul 15, 2003 7:49 pm
by Shannon Parks
Ok, so I read up Dynaco's 'Using KT88 tubes in the Mark II' (thank you Gary Kaufman) and according to them it was a plug and play swap job - IF the bias voltage could be cranked up (down?) to -55VDC. The Mark II needed a resistor swapped and some caps to get the lower voltage. I just ran out to the garage to see what I could adjust the Eiclones to, and I could only get -44VDC. Looks like I'll have to play with a few components.

BTW - I'm just about to Google this, but what is the word on 60Hz hum on a fixed bias circuit? I mean, it's half-wave rectified and you don't want too much capacitance in there because you want your bias up fast - 100uF seems to be the norm. Any wisdom here?
:-k

Shannon

Double it?

PostPosted: Wed Jul 16, 2003 5:27 am
by EWBrown
You could add another diode and cap to the bias circuit and make it a voltage doubler, then you-d have plenty of headroom for the -55VDC bias.


The grid current drawn is minimal. so a large filtering cap isn't really necessary, I haven't observed (or heard) hum on the ST-70.

Since the B+ (assuming tube rectification) and filaments come up slowly, I would presume that as long as the bias is up before everything else is, then a larger filteing cap could be used, perhaps 220 uF @ 100VDC.

In Yet ANother Project Amp, I am considering using a separate small 48 VAC xfmr and FWB rectification for the fixed bias. That way I can get away using Hammond 270 series iron and save a few bux. :crazyeyes:

Just an early morning SWAG :morning:

/ed B in NH :infinity:

Another good choice.

PostPosted: Wed Jul 16, 2003 7:37 am
by Thermion
I have seen where people have used an inexpensive 12 volt filament transformer for the bias. The trick is to connect the secondary of the transformer to the 6.3 volt tap of the power transformer and then rectify the primary for the bias supply. The current draw is minimal and there is usually spare current capacity on the 6.3 volt taps anyway. This will give you about -80 volts DC for the bias.

This is a good way to go if your power transformer doesn't have a bias tap.

JT

capacitive voltage divider C-

PostPosted: Wed Jul 16, 2003 8:38 am
by Shannon Parks
Hey guys,
The Eiclone uses a capacitive voltage divider off of one of the secondaries - highly recommended by Ned and it is very cool indeed.
See an example here:
http://www.triodeel.com/dynawil1.gif

I'll just change the divider values to get my higher voltage. I was just bummed I couldn't pop in some KT88s Thermion has lent me (ok, sure - what's 600mVDC on the cans gonna hurt?). It's a good time for me to look over the circuit - plan to buffet the pot wipers with a safety resistor. ;)

Shannon

Hum a few bars for me, then I'll just fake it.

PostPosted: Wed Jul 16, 2003 12:44 pm
by EWBrown
I was just looking over some schemo's in the old Acrosound pamphlet on Gary Kaufman's site, and on the circuit for the KT-88 / 6550 60 watt amp (Page 13) , they use a backwards 12V filament trannie to generate the bias.

The two filter caps are a whopping 20 uF each. As far as hum is concerned, any residual hum injected onto the PA grids will be cancelled out by the push-pull circuit. It would be in phase on the grids, and 180 degrees out of phase on the primary winding, and not seen or heard on the output.

Some of the Acrosound OPTs had a third separate winding for the screens, for those tubes (like 6146) that have substantially different voltages for the screens and plates (this could work for SV83s in the DIY35, if any suitable 3-winding trannies are available today).

I like that cap-coupled bias circuit, nice and simple.

If one likes the separate bias trannie approach, or needs an isolated low voltage for other requirements, AES has a nice cheap 48 and 15 VAC CT trannie, first one in their power transformer listing, for under $10, on sale for $7.95 until 9/15/03. Part number: P-TC-141-354

/ed B in NH

Simple is always better

PostPosted: Wed Jul 16, 2003 10:29 pm
by Thermion
separks wrote:Hey guys,
The Eiclone uses a capacitive voltage divider off of one of the secondaries - highly recommended by Ned and it is very cool indeed.
See an example here:
http://www.triodeel.com/dynawil1.gif

Yeah, I forgot you showed this to us the other day. A simple and elegant solution. I may even mod the bias supply on my ST70 to use this method. This should work very well in the Eiclones.

JT :lying:

XsubC

PostPosted: Thu Jul 17, 2003 9:52 am
by EWBrown
FWIW, The reactance of a 1 uF cap @ 60 Hz is 2652.5 Ohms,
1 / (2*Pi*F*C) F in Hz, C in Farads, for uF and 60 Hz, it works out to
1,000,000 / 377 X C (in uF)
Reactance is inversely proportional to capacitance. (and frequency)

So, now y'all can calculate what cap value is needed to do the job :idea:

/ed Brown in NH :crazyeyes: :splat: :infinity:

Yet another biased approach

PostPosted: Wed Jul 30, 2003 5:40 am
by EWBrown
If one is using 211 / VT4Cs, 845s, or 276As in a single ended amp, there is a fairly high negative grid bias voltage requirement, one alternate approach would be to add two more diodes, cathodes to the HV leads on the power tranny, cathodes tied together, going to the filter cap, then resistor divided as needed to obtained the required grid bias voltage.

These vintage power triodes require a hgh B+ voltage, 900 to 1250 VDC, the grid bias can be from -100 to -300 vdc, the 6SN7 input and driving stages run on a more "normal" 300-400 VDC, so take the bias as described above off the lower AC voltage power trannie.

One of these "beastatrons" is going to use a LOT of iron, two power trannies, a filter choke (or two) a separate 10 VAC filament trannie
and an 11 pound OPT for each channel, so this one is definitely NOT going to be built on a stock aluminum chassis. Nothing better to do during the long and cold winters around here, so why not...

/ed B in NH

Re: Another good choice.

PostPosted: Fri Aug 31, 2007 9:29 pm
by Ty_Bower
Thermion wrote:...This is a good way to go if your power transformer doesn't have a bias tap.


I was reading up on old threads, and came across this one discussing the clever bias supply in the Ikes. It made me think of this article I read about deriving the bias off the center tap of the full wave rectifier.

http://www.aikenamps.com/BackBiasing.html

Not really the same design as the Ike, but just another method for making a bias supply without a dedicated winding.