ok I Bought it ... what now :D

[Richard]

Looks within tolerance (certainly cathode voltages will vary with the current through the individual valve) but I'm a bit uneasy when the grid voltages aren't similar when 0v or from the same feed. What coupling caps are they George? Could they be leaking a little?

Rich

[Grog]

Hi Richard,
thanks again for your help, you should write a book on the KxT88

they are these ones, http://www.russandrews.com/product.a...JENVQCTHEMXETX

Metalised polypropylene so they shouldn't go leaky.

i'll go check the resistor values R8,R9,R30,R31 as these could affect grid voltage.

i did find that the HT varied between measuring the right and left channels, i guess it's mains variation, 508v when measuring the right channel (first) then left channel was 502V so i went back and checked the right which was 502V as well by then.
but this shouldn't really affect it much?

I hope the 16V caps will be ok with the heater bridge upgrade, i've just soldered them in
going for 4x1N4820-6800uf-1R-6800uf-1R-6800uf. i'm hoping the 2R total will be close enough to the suggested 2R2, if not i guess i'll add a R22 in there somewhere.

[Richard]

Yes the original was 16v 4700uF so they should be fine George. (I tried small 10v caps at the time and they popped the mains fuse twice at switch on so I just went to the next size available.)

The applied grid voltage goes nowhere, almost no current flows and it just sits there as potential, so those resistors should have no effect. Plastic coupling caps shouldn't be leaking, as you say.

0.8v is quite a lot of difference though and they're all down on the circuit -10v spec. I wondered if the valves may be drawing grid current over the grid leaks and creating a positive voltage. So I went to the data sheets and checked early and later Genelex KT88 sheets and a late GE 6550A sheet.

Lo and behold, max gridleak value for cathode bias is 220k and fixed bias 100k whereas these are 330k. (The 6550A sheets says 250k and 50k.) There is a plate dissipation proviso given in the later Genelex sheet but 330k still looks too high. Given that this amp is roughly 50/50 biased you'd expect to use 150k or 220k gridleaks.

You probably know that the reason to use the highest value pos is to allow a smaller value coupling cap. Halving the gridleak value would mean doubling the cc value to keep the same input filter frequency. Moving to 220k though will make very little difference moving it from 1Hz to 1.5Hz. Looking at that GEC circuit shows they used 0.5uF with 220k and full cathode bias.

This might also explain why your c. -9v reading is lower than the -10v on the circuit (pos bias being formed offsetting some of the applied neg bias). As the grid bias is now lower the valve draws more current and the cathode bias rises higher than the 10v shown. (We increased the cathode resistors to cause this also but perhaps we wouldn't have needed to if the grid bias was correct.)

Of course all this may not be a problem if all valves are near identical. If not though they would cause differences in bias which would cause difference in the currents through the optx and upset the balance. This might get worse as the valves wear and cause runaway which otherwise wouldn't happen with lower value gridleaks. So if it were me I'd lower the gridleaks a bit, 150k, and see if the grid voltages even up.

Rich

[Richard]

Just musing over the gridleaks again as 1) the designer Andy Grove is a very clever chap and 2) I'm surprised we missed this before if it was an error.

Anyway, the
1972 GE 6550A are cathode 250k, fixed 50k max and the
1959 Genelex KT88 are cathode 220k, fixed 100k max and going by them 330k is too high.

Looking at the 1974 Genelex figures though finds them changed depending on plate dissipation,



Kit88 has mixed bias but as it's able to control it's final bias setting automatically via the cathode resistors perhaps it can be considered cathode bias (I'm not sure of the reasoning behind the values). The amp ran its plates at a nominal 33W dissipation so this then would be OK going by the later KT88 spec.

Still, I'd still drop them to 220k next time it was convenient as I never used later Genelex valves and would want to use anything I fancied in the beast. I did use nos GE6550A, Svet 6550C and JJ KT88s which glowed like Hades but that was the mains voltage which threw all these figures out the window anyway and had the plates at over 40W.

All said though your currents look closely balanced (cathode v/cathodeR) and plate dis is around 35W so probably ok as it is and look elswhere for the hum.

Rich

[Black Stuart]

Richard,
a bit off-topic but your explanation about gridleak values made me think about my h/amp. There is a long running thread on Headfi re. the Bada PH12 h/amp and it was suggested that the driver 6SN7 would have a much easier life if the coupling cap was reduced in value.

This some of us did, using 0.47uF instead of 1uF, with no detriment to the sound or following components but here's the thing - I can keep my fingers on top of the driver tube because it never gets really hot and the particular 6SN7 I use produces lots of heat at the top of the glass but no way could I do that (well a masochist could) with the o/put valves.

I would imagine that the life of the driver tube is going to be far, far greater than the two o/put tubes. Many designs tend to thrash the tubes and of course seriously shorten tube life, fine when good tubes were a few quid each but not only are good NOS tubes bloody expensive but the supply is running out.

Perhaps you could start a thread and go into more depth on this whole issue, I for one would be extremely interested to get a clearer picture on this whole issue.

[Richard]

Hi Stuart,.

I'm not sure/familiar about the Bada coupling caps but you're right about valve life. WAD used to design for output valves to be run pretty much on their limit. This was stated at the time (Kit88 era) in the supplement articles as it was felt they sounded best that way. This caused problems though with mains voltages in some areas and different makes of valves etc - there was little margin for error - (compare that philosophy with the legendary reliability of Quad II and the way it ran the valves).

Later designs run things less hard but op valves do only last about 1/10 time of signal and drivers. 2000 hours is all Svet EL34s will manage if driven hard (according to Svet) but KT88s may be a bit better and nos KT66 were rated for huge hours up to 10,000 iirc. The exception with small valves might be 5687 which runs hard in many apps and the ECC82 in PreII which are run hard so perhaps change those two every couple of years.

Rich

[Black Stuart]

Richard,
re. the o/put tubes I intend to use 'directed airflow devices' ie. direct cooling onto the hottest part of the tubes but is there any component way I can acheive the same aim - maybe DAD is sufficient.

I was expecting maybe a fall-off in quality by using half the value of the specced c/caps but the reverse happened, probably because life is easier now for the driver tube.

I will have changed out virtually all the caps in the Bada (I do not like Solens in any way shape or form) but I'm not sure what to do about the tube reservoir caps, some have hugely increased the value from 56uF to 150uF but kept the voltage the same. I'm not sure this is the right way to go - any ideas here.

[Yellow]

Hi George,

Just to answer a couple of your original questions:

1. I know Hugo Cass through the old WAD days, and in fact bought some valves from him three or four years ago if I remember rightly.

2. The EH KT88 is a superb valve. I run a set in my KaT6550 amp from time to time and they are my preferred valve out of the ones I have which are all 6550 (Valve Art, Svetlana and Tung Sol.)

3. When I built my kit, I ran it upsidedown, propped up on books. Just keep the valves well away from any carpet and remember to have a dummy load or speakers connected - easy to forget!!!!

Regards,

Neil.

[Grog]

thanks for the advise,
i'm just putting the 220k grid leaks in now.
the heater mod is almost done.

i also noticed that the resistors to set the bias voltage R3, R5, are 0.25W not the specified 0.66W, although i've calculated teh power dissipated and it's well under the 0.25w
i assume this is un related?

[Grog]

powered it up and the same hum/buz was there (i did check to see there wasn't a stray bumble bee stuck in one of the valves , then my common sense woke up and kicked my brain for being so dumb!
seems the hum was down to me plugging the signal generator into a batery pre amp (so i had a proper analogue volume) then into the amp.
tried some shorted phono's and it was deadly silent......... until i plugged it into the klipschorns in, some quiet hum/buz just audible from the listening position.
with the test speakers (88db) i could hear the mains transformer hum over the speaker hum even with my ear against the speaker.