6550a question for engineer?

[KJ2005]

Hello ,

As I am about to build a PP amplifier for a pair of 6550a, and I would like to ask, if someone has experience to the drive circuit - input and driver.

The intension was to use the old circuit from Michaelson & Austin, but if some DIY people has experienced and tested a better configuration for the Tungsol 6550a tubes , I really would appreciate some help with the circuit stucture.

Has anyone out there some knowledge for a modern, or more up to date design, when we speak about the 6550a tubes?

I like symmetrical designs - Is D76/D79 from Audio Research a possible option? - and does it sound better than the mentioned circuit from M&A?

The plan is a fet regulated high volt psu for the input - and drive tubes.

Rgds

[HollowState]

I'm of the opinion that 6550s, like 6L6s, KT88s, EL34s etc, are very easy tubes to drive. They require no special drivers per-se. They can all be capacitor coupled right from the plates of the previous stage assuming that has the required gain and balance. Of course a cathode follower driver is good design, but not truly necessary. For push-pull, fixed bias is the way to go with not much more then 50K ohms in the grid to ground circuit. Although I have seen more used.

AR used a cross-coupled interstage tube to help balance the phases through it's positive feedback. Probably because they used a screw-ball unity gain stage to invert phase right at the input. It worked but was more complicated then need be. A well designed long-tailed pair with a CCS in the tail is a much better way to phase invert a signal.

There are many dozens (perhaps hundreds) of circuits for PP 6550s (or others) out there for the taking. Most will sound good if properly implemented. My advice is, first, use the very best output transformer that you get or afford. This is the heart of a tube amplifier and is the most important part for good performance. Second, chose a circuit that uses only one capacitor in the signal path. (each phase) Any more then one is unnecessary.

Regulate the input and driver if you want. But since they almost always run in class A, regulation won't buy you much and I doubt if you'll hear a big difference provided they are properly decoupled from one another. Meaning an R and C between feed points.

Victor

[KJ2005]

Thanks, - I was not sure whether the cross-coupled interstage tube , and the cathode follower design as in D79 was necessary in order to get the phase right, and secondly - was it really necessary to use that many tube stages to drive a pair of 6550's? A larger number of tubes was seen in McIntosh 275 as well ( one amp I have never heard) . I have listened to the D79 and the sound , after my opinion, was soft and still with good attack in the deep region.

I understand , and what I hoped for, that the long tailed pair and a high performance output transformer are the more important issues when building a good tubeamp.

Could you please let me know what "CCS" is? Please let me have your opinion at the attached TVA-1 design. THe circuit contain not two, but four caps ( two phace) , and you advised me of having only one pair. Third ( I promise to stop now) Could you recommend any output transformer of high quality - We got "Sowter", and "Lundahl" in Europe?

Best rgds

Kim

[richwalters]

There are so many issues; what B+ ; o/p power?; input sensitivity
are you after ?

A concertina phase splitter with Williamson interstage driver will give a lower output Z to the output tube grids hence better top end performance than the longtail phasesplitter , esp when quite low value grid leaks are used for the output stage. This is a tough challenge esp for distortion.The Svet 6550C version quotes around 100K for fixed bias, which is a darned sight better than 50K for the early A version but I’m not sure about the value with TungSol reissue tubes.
Read up Morgan Jones, 4th edit Valve amp p. 415 mentions the poor performance of the Mullard 5-20 which uses an ECC83 as LTP and one should avoid this signal tube for such tough tasks.
. If you are looking at a -3dB response at 50Khz with a good quality output transformer then expect a driver stage to consume current to deliver the goods and to avoid early response slewing and distortion.
There are other issues, a 3 stage design is more stable than a 4 but with correct optimisation a 4 can be made just as good.
The 3 stage configuration shown in the Radford STA25 Renaissance which uses an ECC88 cascade and 6U8A (ECF80/2) as a beefier longtail pair into 6550’s, could be the configuration you want. At the moment I cannot copy pics.
If you aim for good interstage driver performance then I find 6550’s give better sonic performance than KT88’s (O.M.V).
As Victor mentions, a top notch output tranny is instrumental to performance.
I regulary use an ECF80 as concertina p/s followed by Williamson config 12BY7’s as triodes (expect 30mA current per pair); driving 6550’s and I get very low thd.
A four stage amp using a concertina p/s with 20dB global nfb with give a signal/noise ratio around -60-70dB down whereas a 3 stage using a LTP using the same global nfb should acheive nearly - 85-90dB down.
The -60dB noise figure will just be noticeable from a tweeter in the average distance/ listening position esp when a pentode is used as the input gain stage. I often use a pentode duped as a triode, this is useful when a line sens approx 0dBv (0.77V) is only used, signifigantly reducing the circuit noise.
The beauty of the concertina arrangement is no circuit tweaking for balance, whereas a LTP requires a pot /present to balance out for min thd.
If you use the concertina p/s, make sure the B+ psu ripple rejection is high, as the anode has a far lower PSRR.

richj

[HollowState]

Originally posted by kimjul2005

Could you please let me know what "CCS" is? Please let me have your opinion at the attached TVA-1 design. THe circuit contain not two, but four caps ( two phace) , and you advised me of having only one pair. Third ( I promise to stop now) Could you recommend any output transformer of high quality - We got "Sowter", and "Lundahl" in Europe?

Hi Kim,

CCS means "constant current source". It is a means of controlling current flow through the tubes and keeping it at a constant level as well as providing signal isolation. It can be as simple as a single LED or a transistor (or two) biased by a diodes. It could even be another tube. By keeping the current through the long-tailed pair even, you get a better balance and more equal phase inversion. Remember, it is the current through the tube more then the voltage across it that establishes gain. This works well up to a point because very unbalanced tubes (or sections) will still suffer. But it's better then a purely resistive tail and easy to apply. If you search this forum for ccs you should find lots to read. I think even John Brosky's Tubecad journal has a blog about them.

Whether an amplifier uses two capacitors or more in the signal path is really a matter if design philosophy and designer's choice. I believe the best capacitor is no capacitor. By directly coupling the input to the driver (or inverter) a capacitor is eliminated. But this comes at a price. The next stage must be elevated more above ground to remain biased correctly. This requires a higher B+ level from the power supply and more cost.

More capacitors means more total phase shift through the amplifier which can translate to instiblity when global feedback is used. So the design becomes a little more fussy in that respect. But at the same time, another capacitor can make the design easier in certian ways.

Recommending an output transformer is tough. I always use NOS American because I have many. But I've heard good thing about Lundhal and Sowter. Also about Transendar, Electra-Print and Edcor here in the states.

Victor

[KJ2005]

For more than 10 years ago a copy( still not ready) of the TVA-1 was started to be build. I have two voltages to go for B+ 450V and B+ 560V. I hoped for approx. 70W with a sensitivity of approx. 1 to 1,5v.

The sensitivity is not critical - I have the Luxman C03 and several tube preamps to deliver more than 5 volt clean RMS.

I need the 70 Watt to drive AR3a .

I have just ordered Morgan Jones second edition because one of my friends advised me to.

Never looked at Radford STA25 Renaissance because I thought the amplifier had the same configuration as STA25 serie III. The last one includes EF86(input) and 6U8 ( penthode/triode) as driver. I did not like to use one penthode for one 6550 , and another triode to drive the second 6550. I may have to think twice after having read your notes. Then I have to see whether the 12BY7 is among the tubes I have in stock. I have a test pair of ECF80!

I would be very pleased, if you, along the road, could send me a copy of the mentioned circuit pigs. Don't hurry - still working on the speakers, and the amplifier comes next.

Thank you for spending your time on this.

Rgds Kim

[KJ2005]

Yes , Now I know, ( konstant støms træk in danish). Did'nt know the short form in english.

What you mean is, that a constant current flow to keep the longtailed pair even, is more important than a very steady or "stif" +B supply for the same doubletube. We like to keep the gain steady going by a constant current flow! I shall remember this.

If one uses only a pair of caps, the problems with direct coupling between input and driver tube will as result demand higher +B in order to bias the tube right. I understand .

I like to keep things simple, and as few caps in line as possible - I agree! When you mention total phase shift in the amplifier after a cap/resistor - you mean poles? On the other hand caps keep each tube DC balance.

Yes - I have heard good things about many manufactorers in the states as well, but the transport cost to Europe because of the weight is rather high. I have bought some 11" AR speakers by AB tech to a reasonable price - but the freight cost - jesus!

Thanks for your reply

Sincerely yours

And best regards

Kim

[ErikdeBest]

Hi Kim

I do not know if it is still possible, but I would recommend you to cancel the order for the second edition and go for the third edition of Morgan Jones. I do not know the second edition, but the third is said to be more complete.

Erik

[KJ2005]

Thanks - I shall try to change to the third edition - I think to remember that the internetshop had the third edition in stock as well.

Best rgds.

Kim

[richwalters]

Originally posted by kimjul2005
For more than 10 years ago a copy( still not ready) of the TVA-1 was started to be build. I have two voltages to go for B+ 450V and B+ 560V. I hoped for approx. 70W with a sensitivity of approx. 1 to 1,5v.

The sensitivity is not critical - I have the Luxman C03 and several tube preamps to deliver more than 5 volt clean RMS.

You mention two B+ voltages: in Ultra Linear mode the 560V is out of the question for 6550 (whatever brand) but can take this in true pentode. Choose mode !
The ECF80 triode is rather nice, cheap with a low mu and makes an excellent concertina p/s. The pentode isn't screened and requires respect, microphony, hum etc, so I always use this in triode mode with line inputs.
Since you will have a high B+ (460+ will be required to get 70W), the direct coupling from input to driver tube isn't an issue.
If you use global nfb, a high pass LF roll-off in circuit is essential to correct for phase shift below o/p tranny cutoff.
The 12BY7 is getting pricy. If you stay with LTP then there are loads of tubes around. These can be 6SN7, 6CG7 etc. The circuit of the Harmon Kardon is about the simplest LTP configuration.
Instead of the 12BY7, the sibling EF184 is cheaper and more available.
At present I cannot send pics as camera is xxx, but you can contact with me your fax num via members contact email.

richj