KT88 / 6550 Bias Questions

[bob p]

I've got a couple of questions about biasing the KT88:


First, the tube data sheets have what seems to be an odd footnote regarding the bias circuits for both cathode bias and fixed bias applications.

For cathode bias, the footnotes say that it is essential to use two separate cathode bias resistors.

For fixed bias, the footnotes say that it is essential to provide two separately adjustable bias sources.

I guess I must be missing something, as I've never seen this sort of footnote on a valve data sheet before. Can anyone explain it? Is this an artifact from the day when tubes were not sold in matched pairs, and the typical tubes were so poorly matched that bias had to be set individually?

I've been looking at some old HiFi schematics, and the cathode bias amps do all seem to use independent cathode bias resistors on each KT88. The fixed amps all use a separate bias control pot for each KT88, and sometimes they even have a 3rd pot to adjust the bias range of the other two pots.

In contrast, for other tubes cathode bias is typically accomplished with a common cathode resistor. Fixed bias is typically accomplished with a single bias range control, and if you're not planning on using matched tubes perhaps a balance pot is in the circuit. I'm wondering if there's something "different" about the KT88 that requires independent adjustment. I sure seem to be missing the point on this.


Second, I'm looking for a source for bias pots for fixed bias applications. If anyone can recommend what type of pots would be best in this role, I'd appreciate your help. If you could provide a Mouser part number, I'd be happier still!

TIA!

[Alex/Tubewonder]

Hmm. let me think...

I'd guess that KT88/6550 have substantial control grid leakage current when hot as they usually run with high plate currents. Grid leakage current will cause voltage drop on the grid leak resistor that will reduce the negative bias voltage and in effect cause the idle plate current to increase. Separate bias sources are required because even if the tubes are "matched" the grid leak currents are not so they give means to compensate for leakage besides setting bias separately for each tube. Same applies for cathode bias even though it's self regulating the effect of grid leak can be substantial. In fixed bias configuration adjust bias initially after ca 2 min warm up and readjust once tubes have warmed properly after about 15-20 min. That's at least what I think as I have not measured grid leakage on these tubes.

[Ray Ivers]

Bob,

As I understand it, the KT tube series was designed primarily for audio/hi-fi usage, so the circuits associated with the flagship KT88 tube tend to be hi-fi/audio related. These circuits often employ techniques known as "good design practices" which have been slow to catch on in the guitar-amp world.

Independent cathode-bias resistors and separate low-resistance adjustable fixed-bias circuits will work well with all tubes, not just KT88's - it's just that KT88's, having attracted the attention of many top 'golden era' amp designers and/or finding use in top-of-the-line equipment, tended to get this deluxe treatment more often than, say, EL84's - which the data sheets (being an essential element of the marketing/sales process) often reflect, with copious notes, high-power operating condition data, ultralinear curves, etc.

If you're into KT88's, as I am - this tube and the Type 1 Ei KT90 are my all-time-favorites for high-power stuff - an excellent resource is 'An Approach to Audio Frequency Amplifier Design' (ISBN 1-8825280-05-2, Audio Amateur Press reprint). This book has five different KT88-based output stages, from 30W to 400W - plus all sorts of other designs rarely (or never) seen in guitar amps. Perhaps you already own a copy - way cool IMO.

Mouser P/N's 652-3852A-162-104A 100K slotted-shaft cermet pots should work well for you; values are available from 1K to 100K (the 100K's are currently in stock). Sorry about the price ($9 each), but I would highly recommend using the best-quality bias pots you can find. I would also suggest using a 12AU7, 12AT7, or other lower-Rp driver.

Ray

[soundmasterg]

Most of the Sunn amps used the 6550/KT88 and only had one adjustment pot for the bias. They work fine, but the amp does benefit from seperate bias adjust pots, and it allows you to use unmatched tubes too.

[bob p]

Thanks everyone for their help!

Ray, I understand your point about "good design practices." there's a definite difference from what's good for HiFi and what's good enough for guitar amps, and we tend to take short cuts when things don't matter for MI applications. Because this build will be a HiFi amp and not a bass amp, i think that staying true to good design practices will be particularly important.

one thing that I had not really considered about "matched" tubes in my previous post is that matched tubes are only really "matched" when they are new. There are really no guarantees that they'll stay matched throughout their service life, so independent bias controls do seem more important from a HiFi perspective. i guess that's a pretty clear argument for independently adjustable fixed bias, or for a cathode bias circuit that employs separate cathode bias resistors and a "balance" pot.

If you're into KT88's, as I am - this tube and the Type 1 Ei KT90 are my all-time-favorites for high-power stuff - an excellent resource is 'An Approach to Audio Frequency Amplifier Design' (ISBN 1-8825280-05-2, Audio Amateur Press reprint). This book has five different KT88-based output stages, from 30W to 400W - plus all sorts of other designs rarely (or never) seen in guitar amps. Perhaps you already own a copy - way cool IMO.

if you are referring to the reprint of the 1957 publication by GEC, yes I have a copy!

One thing that I had noticed about the circuits in the GEC book was that they seemed to favor much higher voltage supplies, choke input filters, and small filter caps. Compare them to more "modern" designs that use lower voltage secondary supplies and high capacitance cap input filters. I wonder if this is just an artifact of the improvement in filter caps and their reduction in price over the years, with a resultant shift in design philosophy. It seems that with the advent of inexpensive high capacitance caps, 1400 VCT transformers like the one recommended for the 88-100 have become pretty hard to find.

I'm wondering if you'd have any comments about building the PS for the "88-100" with a 400-0-400 transformer and a cap input filter (with a big choke) vs. a 625-0-625 or a 750-0-750 with a choke input filter.

Mouser P/N's 652-3852A-162-104A 100K slotted-shaft cermet pots should work well for you; values are available from 1K to 100K (the 100K's are currently in stock). Sorry about the price ($9 each), but I would highly recommend using the best-quality bias pots you can find. I would also suggest using a 12AU7, 12AT7, or other lower-Rp driver.

Thanks forthe Mouser numbers! The $10 price per pot isn't at all scary. I had considered the cost of using wirewound rheostats, and THAT approach was rather pricey. In comparison, $10 for cermet pots seems inexpensive.

Back to the subject of tubes -- what are your preferences for new production tubes? I'm leaning toward the JJ KT88, but I don't have any first hand experience with new production KT88 and I'd appreciate any experience that you might have to offer.

Bob

[bob p]

I would also suggest using a 12AU7, 12AT7, or other lower-Rp driver.

Yes, I have been looking at designs that use 6SN7 or 12AU7 for PI, with or without 12BH7 + 12AT7 post-PI drivers. In addition to the GEC circuits in the "Approach" text, I've been looking at circuits like the Heath W-6A, Heath W6-M, and the Mac MC75 and MC275.

Do you have any experience with KT88-based HiFi amps? I'm wondering if you'd have any recommendations about which circuits might be preferable.

Thanks again!

Bob

[Ray Ivers]

Bob,

there's a definite difference from what's good for HiFi and what's good enough for guitar amps, and we tend to take short cuts when things don't matter for MI applications.

...not to mention falling far short of reaching a consensus regarding "what's good enough" and "what matters".

matched tubes are only really "matched" when they are new.

If you're referring to 'DC operating-point matching', I agree wholeheartedly. OTOH, transconductance-curve-tracer type matching apparently lasts through most of the flat portion of the 'bathtub' life-expectancy curve, or at least it seems that way to me. My take is that an enduring DC match relies more heavily on cathode emission staying constant than does Gm matching, which strikes me as more of an interelectrode-spacing type thing, although there's surely overlap between the two.

if you are referring to the reprint of the 1957 publication by GEC, yes I have a copy!

Cool - it's a great book! I got your message, and I have been to the Millett site - I never knew it as his, only as "that site that's filled up a massive chunk of my hard drive".

the circuits in the GEC book... seemed to favor much higher voltage supplies, choke input filters, and small filter caps.

Yes, it's definitely dated in this respect - but good PS regulation was definitely part of the GEC 'approach', at least with the higher-powered designs, so I guess they did what they had to do given the times. If semiconductors had been widely available when it was published, most of the higher-powered supplies would probably have incorporated at least some active regulation.

One thing I really love about the GEC book is that they favor two-tube push-pull over push-pull parallel, preferring to use two higher-output tubes and/or higher B+ - along with AB2 drive if necessary - to get more power, rather than just throwing more tubes at the task (the 10-KT88 PPP design is included as an appendix, not in the body of the text).

I'm wondering if you'd have any comments about building the PS for the "88-100" with a 400-0-400 transformer and a cap input filter (with a big choke) vs. a 625-0-625 or a 750-0-750 with a choke input filter.

Using a 400-0-400 (400mA) xfmr with this UL circuit, I'd want to go with a C-L-C (pi) filter, which should yield maybe 520 - 530VDC of pretty solid B+ at the 2nd cap and about 80W output from two KT88's; a 3.4K primary should be good for 100W or maybe even a bit more. I'd prefer a 460VAC/600mA/bridge/pi setup, though.

what are your preferences for new production tubes?

Well. I recently did some very basic power/waveform testing with the EH, SED, Valve Arts, and one other KT88 I can't recall - plus the KT90EH - in a Marshall Major UL circuit, under "semi-torture-test" conditions (two tubes, 8 ohm load on 8 ohm tap). IIRC the KT90 put out the most power with the best-looking waveform, followed by the rest in pretty much a 3-way tie. I've been using SED (Svetlanas when I bought them) KT88's in a '73 SVT for years with no problems and great sound. The rest of my KT88 amps all have either GECs or Nationals in them.

Regarding your post about KT88 hi-fi amps - all I have is one MC-75. As remarkable and high-performance as this circuit is, I don't think it's really one I could suggest as a new build unless one had the iron for it already in hand. I'd recommend either the W6A or the W6M (with PS upgrades) in a heartbeat, though.

Ray

[soundmasterg]

I've got SED 6550C's in my SVT and they sound and work fine for about 7 years now. I've got JJ KT88's in my Sunn things and I really like them a lot. Ray, on your recent test you didn't sample any of the JJ KT88's?

[Ray Ivers]

Greg,

OK, yeah - the fourth set were Groove Tubes with the 'winged' structures on the plate seams - those are JJ's, right? I've never gotten them from any other source, so I don't have any to compare with.

It just strikes me now that I should have run a 'control' test with both GEC's and early Chinese, to establish baselines at both ends of the performance scale; duh.

Every pair I tested worked fine, it just seemed the KT90's handled the heavy OT load a bit better than the others - which really only indicates that the KT90EH's are most probably real KT90's, not that the KT88's had any shortcomings, come to think of it.

Ray

[bob p]

If you're referring to 'DC operating-point matching', I agree wholeheartedly. OTOH, transconductance-curve-tracer type matching apparently lasts through most of the flat portion of the 'bathtub' life-expectancy curve, or at least it seems that way to me.

Ray, by any chance do you do your own Gm-curve-tracer matching? That's something that I had thought about doing, though I've never gotten around to doing anything about it.

Using a 400-0-400 (400mA) xfmr with this UL circuit, I'd want to go with a C-L-C (pi) filter, which should yield maybe 520 - 530VDC of pretty solid B+ at the 2nd cap and about 80W output from two KT88's; a 3.4K primary should be good for 100W or maybe even a bit more. I'd prefer a 460VAC/600mA/bridge/pi setup, though.

well, as you probably know the hard part about designing a PSU for an 88-100 type of amp is finding good iron to do the job. If you want to go with an L-C-L-C or an L-C-R-C type of filter, you have to track down a really high voltage transformer, and those are pretty hard to find. The only ones I know of are the Hammond 700 series. The voltage options aren't the best (too high or too low), the mA ratings often aren't very high (for a stereo application), and they're plate-only transformers -- no heaters, no rectifiers, no bias.

OTOH, if you want to do the more conventional C-L-C Pi filter, you still have problems finding a suitable high voltage winding that has enough current to do the job. Looking at Hammonds, for example, the only units with suitable current are the 400-0-400 trannies in the 200 and 300 series, which are a little on the low side voltage-wise to do the job. Spec'ing iron seems like the hardest part of this project.

The only other option I've come across is the One-Electron BFT-1B, and its secondaries are also too low. Its also missing a bias tap.

So I've got to ask -- if you were building an 88-100 HiFi amp, what would be your choice for PT iron? I'm having a hard time finding something that will give me the targeted voltages for 100 W for a pair of KT88. Of course, its harder still to find something capable of supplying a stereo application with two pair. Going one step further, I think its pointless to even think about the multiple pair amps like those listed in the appendix of the GEC book -- there just isn't any off the shelf iron suitable for the job.

Well. I recently did some very basic power/waveform testing with the EH, SED, Valve Arts, and one other KT88 I can't recall - plus the KT90EH - in a Marshall Major UL circuit, under "semi-torture-test" conditions (two tubes, 8 ohm load on 8 ohm tap). IIRC the KT90 put out the most power with the best-looking waveform, followed by the rest in pretty much a 3-way tie. I've been using SED (Svetlanas when I bought them) KT88's in a '73 SVT for years with no problems and great sound. The rest of my KT88 amps all have either GECs or Nationals in them.

I've got SED 6550C's in my SVT and they sound and work fine for about 7 years now. I've got JJ KT88's in my Sunn things and I really like them a lot. Ray, on your recent test you didn't sample any of the JJ KT88's?

Thanks for the reports. I have one pair of NOS Hewlett Packard 6550s that I'm keeping in reserve for a future project. Since I'll need a matched quad for the HiFi stereo project it looks like I'm going to be forced into using new production tubes. I'm hoping that the JJ KT88 will perform up to the traditional Tesla standards.

IIRC SED makes both 6550 and KT88 don't they? I wonder how different they really are.

Regarding your post about KT88 hi-fi amps - all I have is one MC-75. As remarkable and high-performance as this circuit is, I don't think it's really one I could suggest as a new build unless one had the iron for it already in hand. I'd recommend either the W6A or the W6M (with PS upgrades) in a heartbeat, though.

yes, the iron in the Mac definitely makes it an oddball. believe it or not there are people who have tackled the Mac/unity coupled combination of cathode feedback and screen grid feedback using a complicated combination of UL OTs and accessory transformers as plate chokes, but I think I'd prefer to stick with a traditional Hafler/Keroes UL design. It should be a lot simpler to implement.

One thing that struck me as odd about the MC75 and the Heath W6 designs is that they both used voltage doublers (which don't enjoy a good reputation for having good voltage regulation) in a demanding application with 6550s. Of course, the volage doubler is mated with a choke input filter, which helps to compensate for the problem, I guess. Personally, I'd prefer not to use a voltage doubler if I can avoid it, but as I mentioned earlier, the PT iron options seem very limiting.

[Ray Ivers]

Bob,

Ray, by any chance do you do your own Gm-curve-tracer matching?

God, no - I don't have a tracer, nor the time to use it if I did. I let Groove Tubes do mine.

So I've got to ask -- if you were building an 88-100 HiFi amp, what would be your choice for PT iron?

A custom-wind Heyboer, in a heartbeat. Here's what they made me for @ $110 each plus shipping (just as an example - the 88/100 circuit specs will no doubt be different):

Stand-up mounting, W 3.25", L 5.5", H 5.0"
Primary: 115/230VAC, 50/60Hz

Sec. Winding 1: 360 VAC CT, 900 mA 325VA
Sec. Winding 2: 45 VAC CT, 500mA 25VA
Sec. Winding 3: 6.1 VAC CT, 8A 50VA
Sec. Winding 4: 12.1 VAC CT, 4A 50VA, w/taps at 6.1 VAC CT, 4A

Total VA: 450
5% regulation, 12" wire leads, copper screen between primary & secondary

Alden H. was extremely helpful and patient with the whole thing too, despite several revisions due to critical size constraints; my experience with them was excellent (a good thing too, since I was having the exact same problem finding anything even ballpark-close off anyone's shelf).

FWIW - when you finalize your PS design, I'll be glad to suggest PT specs if it might be of help.

Ray

[soundmasterg]

You know Ray, I haven't looked close enough at the internals of the JJ to see if they are like you describe. You could try getting some from Bob at Eurotubes. I believe he does some matching like Groove Tubes and he's WAY into Sunn stuff too. He's less than a 20 minute drive from me and I've still never gone and met him though. One of these days I'll have to!

I'm always suspicious of new tube types and how close they are to the older ones. A good control test of the Genelex as compared to new production would be great! There are some new types that you're missing out on if you're only going through Groove Tubes for your supply. You could get the Tung-Sol 6550 and reissue Genelex KT88 from New Sensor. All the other Tung-Sol reissue types have been fabulous performers so far.

Bob, have you considered the Dynaco clone transformers that Triode Electronic is selling? I think they have one from two KT88's, but I don't think they have one for four. The Heyboer option Ray suggested is probably a great one too.

Greg

[bob p]

As it turns out, Triode is in the neighborhood. The biggest problem for me hasn't been sourcing a ganged OT -- its been sourcing a ganged PT. I'm putting a parts list together for a HiFi *STEREO* amp. To put it all in one chassis I need enough PT to power a quad of KT88, but I only need OTs to handle a pair in push-pull. AFAIK all of the KT88 iron that Triode has is geared toward monoblocks.

[bob p]

Thanks Ray. I hadn't even thought of having someone like Heyboer do a custom wind -- I always thought that was cost-prohibitive. Hmmm... maybe I could build that GEC 400W/ch stereo amp with 10 KT88. <evil snicker>

As I alluded to in my previous message, I'd really like to find a PT that can handle the supply duties for a 4xKT88 2x100W amp (a pair of KT88 in each channel). Off the cuff, I'd say something like 1400 VCT at about 650mA for a choke input filter, with adequate bias, filament and heater supplies. That would require about 8A for the 6.3V heaters or thereabouts. I dunno about the bias tap, but I could probably go with FREDs and omit the 5V windings. It would be even better if I could get the PT potted to match a pair of HiFi 4k OT for an AB1 P-P application. Is that too much to ask for?

[Ray Ivers]

Greg,

There are some new types that you're missing out on if you're only going through Groove Tubes for your supply.

If GT carries a certain tube type, I'll definitely try and get the GT's before going elsewhere - but that's not to say I only use GT's; I wish they made more of an effort to incorporate new tube offerings into their product line as they become available. In the last year or so, almost all my tube purchases have not been Grooves, simply because they don't carry any of the new stuff that's been appearing left and right (excepting KT77's).

You could get the Tung-Sol 6550 and reissue Genelex KT88 from New Sensor.

I think I've got at least one pair each of all the Tung-Sol and Mullard offerings from New Sensor - what I don't have is the time to 'taste-test' them. I've given up on the Gold Lions; apparently most of the first shipment was bought up by the guys selling them for $75 each on eBay, and I've been waiting almost 4 months for anything matched from NS ("uh... we've got some singles here... I think... " ). If and when they eventually straggle in, I'll try and get some to listen to - I see that at least they've finally added them to the web site ("OUT OF STOCK - LATE MARCH" - yeah, right ).

Ray