Craig NOS Valves||
October 18, 2013 - 10:58 am|
Really nice effort. I do have one question. Where did you source such a horribly mis-matched pair of JJ 7591's? I suggest you buy a good matching pair from some other source that does a good job weeding out the culls. I think your test results would change to some degree. The big problem with JJ 7591's is quality control. A well burned in and matched pair or quad would act very differently. The EH 7591 are horrible long term in an amplifier the test results and bias points after a few months of service change dramatically.
7591's are designed to be tested at low input signal levels for a reason......they are low signal level tubes. (Easy to drive). They were designed that way to allow the hi fi manufactures to keep the basic design of the EL84/7189 amplifier circuits and amplifier size relatively unchanged to implement more power.
The EH tube is absolutely not a true 7591 and it is absolutely why it fails the GM test in a vintage Hickok tube tester. The EH is not a horrible tube but the fact that it is not a true 7591 is pretty hard to dispute. That is why its in the larger EH/Reflector plant 6L6 bottle!
Fred Longworth - Classic Audio Repair, San Diego||
April 26, 2013 - 03:31 am|
|Stephen and Dave,|
Thanks for this. Very illuminating. My associate (Jordan) and I were discussing 7591A's this very Thursday afternoon, faced with the need to retube a Sansui 1000A.
December 24, 2012 - 04:27 pm|
|Hi Dave G.,|
Thanks for your great advice! I'm certainly going to put those resistors in place. The reason I mentioned the suppressor grid is because it is internally connected to the cathode and I thought its close proximity to the screen grid might promote the arcing. You are very correct about the operation of the screen grid near its max rating, and only being about 10 volts below the plate.
I live in Seattle, Washington, and our line voltage consistently runs about 120 VAC. I have a Variac transformer and I am going to use it to drop the line voltage to the amplifier down to 110 - 115 VAC range. I've already done some testing and have found dropping the line voltage to 110 VAC produces significantly lower B+ voltage. Combining the resistor mods and the Variac should provide a lot more reliability and hopefully extend the lifetime of the output tubes. Thanks again for your advise. Happy Holidays to you!
December 24, 2012 - 03:27 pm|
|Hi Dave --|
In reviewing the schematic of your Sherwood, it has all the classic elements in place that promote output tube arcing. In high fidelity applications, all such event invariably occur between the screen grid and the cathode -- not at the plate as is commonly thought. Per my article, the elements that promote these events include:
1. The use of high Gm tubes, of which the 7868 tube certainly qualifies.
2. Operation of the screen grid near its maximum rated voltage. These tubes have a Design Maximum screen voltage rating of 440 vdc, and your unit -- operating on the typically lower AC line voltages of its day -- operates the screen grids at 415 volts. It likely powers them to even higher levels on today's AC line voltages.
3. Operating the plate and screens at very similar voltage levels. Considering the drop produced through the primary winding of your output transformers, these are likely within 10 volts or so of each other.
4. Fixed bias operation.
These elements are all accumulative, such that when they are all in place, the installation of screen stability resistors becomes mandatory. They are all clearly present in the design of your Sherwood.
That they were not included in the original design often speaks to the higher ESR power supply caps that were produced when this equipment was manufactured, so that the events were less likely to occur. On the other hand, some models (from Eico for example) were known to eat output tubes from the get-go by way of arcing.
When the older gear is upgraded with today's power supply components, the lower ESR values can really act as a trigger when the four elements are present, unfortunately making what was almost a rare event all too common.
As Steve mentioned, my recommendation in this case would be the installation of screen stability resistors (100 ohm, 1/4 watt is fine), which have been shown now to all but eliminate such occurrences.
I would also strongly recommend including a small 10 ohm 1/4 watt resistor in the cathode circuit of each output tube as well. Without any means to adjust the DC bias or balance on your unit, matched output tubes are a must. The inclusion of these resistors will not only allow you to check how well your output tubes are balanced among themselves, but also determine the actual current level they are operating at, which is important for the health of your tubes and the unit in general.
None of 7591/7868 class tubes manufactured today will produce the absolute performance of the original American products. But installation of the screen stability resistors and ensuring that the tubes are operating well within their design ratings will likely let today's tubes operate very acceptably in your unit.
I hope this helps!
December 23, 2012 - 02:23 pm|
|Hi Dave W., (BTW, I only incorporate the initial, in case Dave Gillespie should join the thread.) Not having heard a response from him (yet), I will go ahead and mention the fact that he has seen a phenomenon of spontaneous arcing in pentodes/beam power tubes. In fact, it is covered in his article here: http://www.tronola.com/html/maximize_tube_life.html. |
He found that the solution was to add small stopper resistors in series with the screen grids. I later did some analysis and hypothesized that the phenomenon could be due to VHF oscillation resulting from the tube's interaction with stray capacitances and lead inductance. Unfortunately, attempts to produce the phenomenon in the lab were unsuccessful, so we couldn't investigate further. (On the bright side, it gave me ample fodder for teasing Dave about it endlessly :) I did find though, that such spontaneous arcing from RF parasitics had been documented as far back as the 1930's.
If you're curious about it, you could add 100ohm 1/2W resistors to the screens and see if the EH tubes still arc. I would be very interested in hearing the results, should you decide to experiment.
December 23, 2012 - 09:28 am|
|Hi Steve L. A correction to my last post: my amp is only 50 years old, not 60. I'm getting so old, I can't do simple arithmetic any more. Also, I shouldn't be surprised that solid-state rectifiers are used in the amplifier design - we started incorporating these devices into our vacuum-tube designs during the late 50's.|
I checked the getter flashing on each of the LH7591A's and they all look good - no whitening. So my conjecture about leakage seems to be wrong. Maybe the suppressor grid is too close to the screen grid, causing arcing? I dunno. I do know that I don't have any problem with my RCA 7591A's. BTW, the max plate voltage rating for a 7591A is 550 volts, not 500 volts as stated in my last post. Thanks again for your help.