Edcor CXPP25-8-7.6K

[77seriesIII]

I think the above is good for the ST-35 Rev D board but want to ask here before I order it. How critical is the 8k input impedance? I know the manual states going down to 6k w/ some OPTs but want to check before slapping down some dosh.

I have to replace probably three...meaning 4 OPTs that just checked out or were DOA and just found out.

Erick

[Ty_Bower]

I used a pair of Hammond 1620 on a ST35. They are specified as 6.6K with 40% screen taps. I made no adjustments to the circuit other than setting the NFB loop resistor to 20.3K as indicated in the assembly manual. After later comparing to a ST35 with real Z565 outputs, I noticed the lower primary impedance of the 1620 resulted in an amp with more gain (it's louder). I also observed on the oscilloscope the Hammonds would saturate fairly early with low frequency material (maybe 4 watts at 20 Hz). Overall, the amp sounded very good.

I wouldn't be too concerned about a 7.6K primary impedance. It will work. Though it might not be perfectly optimal, it isn't too far off. You might want to ask if you can get those with a 25% screen tap instead of Edcor's usual 40%. The Z565 has a screen tap at roughly that point.

Edcor also makes the CXPP25-MS-8K and the CXPP21-MS-8.6K for just a few dollars more than the CXPP25-8-7.6K, though I've been led to believe that single secondary outputs have some intrinsic advantages over the multi-secondary types.

[dcgillespie]

From a transformer quality standpoint, the Edcor transformers are good, but the Dynaclone from Triode is definitely superior, most notably in low frequency performance.

From a circuit performance standpoint, the best performance with 6BQ5 tubes in UL with cathode bias will be obtained with a 10K ohm plate to plate load. While transformers of around 8K ohms will work ok, distortion will definitely be increased when using these transformers with cathode bias, as that form of bias limits the peak current needed to produce low distortion at this load impedance with 6BQ5 tubes.

Dave

[battradio]

Hi Erick ,

The anteks are some you should consider .

http://www.antekinc.com/pdf/AT-020L.pdf

Mark

[Blooze]

apurcell22 had some of the Edcors made with a 23% UL taps on them, so Edcor should have the specs already on file. The model number he gave me was: cxpp25-ms-8k-sp.

[77seriesIII]

Dave,

on the 10k input impedance for the EL84, how is that number arrived at? Is it on the tube spec sheets and I never noticed, ancient wisdom or ? Suffice to say I am @ 40 pages into the 3rd edition of Jones' Valve amplifiers and havent hit that section yet.

thx

./e

[dcgillespie]

Erick -- It is a somewhat complicated subject, but in general, it is based on the characteristics of each particular tube type, in conjunction with the way it is operated in a circuit. Therefore, there is no one impedance that is optimum for a given tube, but an optimum impedance for a given tube in a given circuit. Change the tube type or change the operating conditions, and it changes the optimum load impedance.

Each vacuum tube has published for it what is known as a "family of plate curves" -- among many curves published for it. It shows the dynamic relationship between the plate current a given tube can pull, and the plate voltage and grid voltage required to produce it. If it is a screen grid tube, then it is additional element on the curves controlling the plate current as well. These curves make up a primary part of a given tube's characteristics.

But circuit conditions also play a part in how a tube functions as well. For example, in triodes, raising or lowering the plate voltage has a significant impact on plate current. But in pentodes, plate voltage has little to do with plate current, since it is primarily a product of screen voltage. Use fixed bias and peak current draw capability is maximized. Use cathode bias, and it is restrained -- and we haven't even applied a signal to the tube yet! Ultra Linear operation changes things even more as now the screen becomes a dynamic element rather than a static element, as AC is now being introduced on it. The product of all these design variables come together in a given design to determine what the optimum load will be.

So the characteristics of a tube, and the conditions it operates under determines its optimum impedance to operate into. As a basic fact, when the external load impedance equals the load of the generator (the tubes in this case), maximum power transfer is achieved. But this usually does not produce the point of lowest distortion. So an engineer draws a set of what is known as "load lines" across the family of plate curves to determine what the power output and the distortion will be for a given tube under a given set of operating conditions. Then, based on his design goals for these qualities, he can chose the load line that best delivers the performance he is trying to achieve.

For the EL84 in a classic UL configuration with cathode bias, when it is operated at a typical B+ of 360-380 vdc, the best compromise between power output and distortion is achieved with a plate to plate load of 10K ohms, and a screen taping point of 50%. If fixed bias operation is used, then even better performance can be had with a plate to plate load of 8K ohms, with a taping point of 23-25%. If you are interested, all of this directly relates to a significant element forming the basis for the improved performance modification I introduced for the SCA35/ST35 amplifiers in the thread in this section titled as such. There is also a followup piece available to further help you, that directly compares the performance of two transformers optimized for these two types of operation.

Keep reading and studying Jone's book -- and there are so many other good books available as well. The more you understand about how a given tube interacts in a given circuit, the more you can improve your sound!

Dave

[77seriesIII]

Dave,

Thanks for the explanation! I'm going to read this a few times to digest it and make sure I am tracking correctly.

Battradio, I contacted Antek and they will no longer make that particular OPT, bummer. Their reasoning, too expensive.

./e