Hi sorenj07,
I agree with Johan on your test equipment when he said .......
I am still fairly certain that your ground layout is causing your grief. You do not need designed in feedback to have positive feedback occurring in your system.
Keep high current ground currents isolated and direct to your last filter, your signal; grounds go there as well. Grounds or common points in a stage have your signal grounds for that stage connected closely together. Same for your B+ connections. Make sense?
-Chris
I agree with Johan on your test equipment when he said .......
I am still fairly certain that your ground layout is causing your grief. You do not need designed in feedback to have positive feedback occurring in your system.
Keep high current ground currents isolated and direct to your last filter, your signal; grounds go there as well. Grounds or common points in a stage have your signal grounds for that stage connected closely together. Same for your B+ connections. Make sense?
-Chris
got in a longish discussion about stuff over the phone with Eric Hayes from over at DiyTube. During the course of that discussion, I cut the wire connecting PSU ground to signal at the RCA jack and instead tapped off the main PSU cap and went straight up to the signal buss between signal ground and the first 10-ohm 6L6 cathode resistor. the problem disappeared. the tubes are still a bit microphonic but hey, that's life. i'll probably do a few more tweaks such as beefing up the 1W PSU resistors, and changing the forked last stages into a simple RCRC.
still, though, what kind of zener diode could I use to bypass the 10K bias resistor?
still, though, what kind of zener diode could I use to bypass the 10K bias resistor?
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
well, there they are. both work fine except for some substantial hum from the left one. bias is at 55mA each with a B+ of 500V. I don't exactly know what the bias voltage is, I just twiddled the pot until I got .55V across 10 ohms.
the amps seem extremely sensitive (they're a straight up williamson design). is there a way to skip a gainstage such as paralleling the first two 6SN7 sections to make one beefier phase splitter? I'm temped just to cut out the direct-coupled gainstage, because otherwise I have no volume control unless I build a unity-gain buffer driving a volume pot..
I am definitely not a valve amp expert, but came across this today while doing some research:
http://www.google.com.au/url?sa=t&c...Zjd8KDiMGTi4CjnCw&sig2=nUzI7d4CzESQ2QXtLcreXA
In his paper he describes the four stage Williamson designs as not at all suitable for DIY as they are on the edge of stability. Might be worth a read.
http://www.google.com.au/url?sa=t&c...Zjd8KDiMGTi4CjnCw&sig2=nUzI7d4CzESQ2QXtLcreXA
In his paper he describes the four stage Williamson designs as not at all suitable for DIY as they are on the edge of stability. Might be worth a read.
Normally this topology would be used with some global negative feedback - I'd suggest about 6dB which will also have the beneficial effect of lowering the output impedance by half.
Removing the input stage is not really viable with this topology as you will then have about 22dB less gain than currently and that will make it really difficult to drive.
OR you can however reduce the gain considerably in that first stage by using local feedback. (Cathode degeneration - which will require an input coupling cap and an additional resistor, other changes may also be required to maintain the right dc voltage at the plate.)
Removing the input stage is not really viable with this topology as you will then have about 22dB less gain than currently and that will make it really difficult to drive.
OR you can however reduce the gain considerably in that first stage by using local feedback. (Cathode degeneration - which will require an input coupling cap and an additional resistor, other changes may also be required to maintain the right dc voltage at the plate.)
Attachments
well, another "false alarm". turns out the sovtek tube i had in the left mono in the direct-coupled stage was hummy. subbed out each mono's Sov at this position with an OS RCA coin-base 6SN7GTB and all is good 
learned an important lesson - ALWAYS have spare tubes, preferably of a different brand. can't wait to try different 6L6's. anyone want to happen to try my Sovtek 6L6WXT+'s in exchange for some other brand? be warned, i'm kicking the pants out of them in this amp: 500VB+, 55mA per tube (27.5W dissipation). way above the GC's screen rating but they're doing OK and most decent tubes should live.
tweaks like NFB can wait for vacation - something I want to fine-tune, probably. thanks everyone!
learned an important lesson - ALWAYS have spare tubes, preferably of a different brand. can't wait to try different 6L6's. anyone want to happen to try my Sovtek 6L6WXT+'s in exchange for some other brand? be warned, i'm kicking the pants out of them in this amp: 500VB+, 55mA per tube (27.5W dissipation). way above the GC's screen rating but they're doing OK and most decent tubes should live.
tweaks like NFB can wait for vacation - something I want to fine-tune, probably. thanks everyone!
Hi sorenj07,
Your plate voltage is very high and so is your dissipation. I am not surprised at all that the Sovteks croaked in there. They are intended more for 400 VDC plate supplies as even 460 VDC is pushing them a bit too far.
What you really need for this service are real KT-66's, 7581A's will also work better than 6L6GC's (since you're above where most work). The final suggestion I have are 6L6EH's from New Sensor. They sound great too. You can buy matched sets and these have stood up to some of the sillier amps I have seen. I've been using these lately and am not disappointed at all.
What you really need is a new tube with a larger cathode and a carbon plate (or similar high temperature material), I think you will need gold plated grid wire also. Either than or simply use KT-88's or 6550's.
-Chris
No way!
Your plate voltage is very high and so is your dissipation. I am not surprised at all that the Sovteks croaked in there. They are intended more for 400 VDC plate supplies as even 460 VDC is pushing them a bit too far.
What you really need for this service are real KT-66's, 7581A's will also work better than 6L6GC's (since you're above where most work). The final suggestion I have are 6L6EH's from New Sensor. They sound great too. You can buy matched sets and these have stood up to some of the sillier amps I have seen. I've been using these lately and am not disappointed at all.
What you really need is a new tube with a larger cathode and a carbon plate (or similar high temperature material), I think you will need gold plated grid wire also. Either than or simply use KT-88's or 6550's.
-Chris
you think I'll burn the 6L6's out at this rate? I'm not going Tubelab-type crazy on them so I hope to have them last for at least a year. I'm seriously considering beefier tubes though, but starting at like 80 a quad for the bigger guys, I'm holding off a bit.
i should also probably get some bigger resistors for the RC's that feed those 6SN7's..
i should also probably get some bigger resistors for the RC's that feed those 6SN7's..
Hi sorenj07,
Your supply voltage is over the top for a 6L6GC. Fender Twin Reverbs are hard on tubes and they only run around 475 VDC. Those tubes I listed are the only ones I was able to get good life and good sound with these amps.
So yes, your circuit is expecting a bit much from these tubes. I think you are breaking new ground. The 7581A types may be your best bet, then the 6L6EH - if they last. If they do, and they might, kudos to New Sensor.
-Chris
Your supply voltage is over the top for a 6L6GC. Fender Twin Reverbs are hard on tubes and they only run around 475 VDC. Those tubes I listed are the only ones I was able to get good life and good sound with these amps.
So yes, your circuit is expecting a bit much from these tubes. I think you are breaking new ground. The 7581A types may be your best bet, then the 6L6EH - if they last. If they do, and they might, kudos to New Sensor.
-Chris
OK. A quick fix I have in mind is subbing in a 5R4-type rectifier instead of the 5U4GB, for more voltage drop.
Anyway, I have another problem:
I connected some NFB on one of the monoblocks and got a very painful oscillation once the amp heated all the way up. It sounded like a square wave at around 1-2KHz. I used a 22K 1W resistor from the 8-ohm speaker jack bypassed by a 120pf cap (had it lying around) and connected to the cathode side of the first 470 ohm resistor. What did I do wrong?
Anyway, I have another problem:
I connected some NFB on one of the monoblocks and got a very painful oscillation once the amp heated all the way up. It sounded like a square wave at around 1-2KHz. I used a 22K 1W resistor from the 8-ohm speaker jack bypassed by a 120pf cap (had it lying around) and connected to the cathode side of the first 470 ohm resistor. What did I do wrong?
Hi !
Which type of feedback did you use ?
You have a 50:50 chance to build a negative feedback. Otherwise you have a positive feedback. Seems you have the positive one. So please swap the feedback connection on your output transformer and it will work.
I'm very familiar with these "screams" when the feedback is positiv instead negative. It can set your teeth on edge
Regards, Simon
Which type of feedback did you use ?
You have a 50:50 chance to build a negative feedback. Otherwise you have a positive feedback. Seems you have the positive one. So please swap the feedback connection on your output transformer and it will work.
I'm very familiar with these "screams" when the feedback is positiv instead negative. It can set your teeth on edge
Regards, Simon
sorenj07 said:
Excessive?
Folks ....
We must have different sets of tube data. According to mine (RCA) this is within specs for the 6L6GC! Firstly, if the 55mA is cathode current, it includes screen dissipation, so the total allowed by RCA is 35W. Then, it is stated that if the operation is UL as here, the max. screen voltage is 500V.
One must also keep in mind that max. voltage specification is a somewhat vague term, dependent on conditions. Nothing is suddenly going to explode the moment the voltage is exceeded. If one looks at the spec. for sweeper use, the same internal construction is suddenly capable of handling 2-3 times higher voltage. Fine, for a limited time, but it does indicate that the tube will not spark internally. If one now keeps in mind that in audio peaks are also there for a relatively short time, then the max. voltage that can be tolerated will not exceed the tube limit even if the quiescent voltage is somewhat higher than spec. In fact, to my understanding it has relatively little to do with tube life - the dissipation has. Even that is based on a tube life of so long (not sure what for RCA); again it is not a "watershed" either side of which matters are quite different.
Chris, I am very careful to disagree with you and do so with eyes and reason wide open; are you saying that 6L6WXT is not a 6L6GC? I would not debate that; I had my reservations long ago, and specifically chopped open defunct 6L6GCs and 6L6WXTs. The ..XT's construction is not quite as rigid as a classic ..GC, but then even the latter differs from RCA to GE to Sylvania (all of which I have). I could not really find something to convince me that the ..XT is detectably worse than the ..GC.
At present I am operating 6L6GCs (GE NOS make) with 540Va-k and 450Vs-k, and at 50mA total current/tube. That (per pair each side of p.p. that is 4 x 6L6) gives me 120W output at the OPT secondary. With the GE tube the G1 cooling fin is quite larger than that for either ..XT or RCA, which is comforting. In future I would like to try 6L6EH which seems to get good report, since the ..GCs are getting very difficult to find.
So apology for the long narrative and hijacking the thread somewhat for 6L6 discussion; these days we are all sitting somewhat nervous as experience gained 40 years ago is just not valid any more (I heard JJs are not up to standard). Thus, having mentioned the above, I myself would still appreciate comments from others with experience. It is said that (own) experience is a good teacher ... but second hand is cheaper!
Regards.
Hi Johan,
I did try several Sovtek tubes and they can't hack those high voltages. I did find that the 7581A (same characteristics as the 6L6GC) would handle this with no trouble, their plates are rated for 35 W. Often we see 7581A's also marked with "KT66".
I was once unable to get 7581A's and so took a chance on a quad of 6L6EH's. These had plates about the same size as some old GE 7581A's and also some real KT-66's I had. Happily, these tubes have proven to be as reliable as the 7581A types. I haven't had a problem with any I have used. Since then I have tried other Electroharmonix branded tubes and have been very happy with them.
-Chris
I've seen that most new 6L6GC tubes either hum greatly (one of the pair is conducting too much current), or the plate(s) turn red (same thing). So I don't trust most new 6L6 designs past 400 VDC on the plate. Therefore, your manual is correct, but most newer manufacture tubes are not up to spec.
I did try several Sovtek tubes and they can't hack those high voltages. I did find that the 7581A (same characteristics as the 6L6GC) would handle this with no trouble, their plates are rated for 35 W. Often we see 7581A's also marked with "KT66".
I was once unable to get 7581A's and so took a chance on a quad of 6L6EH's. These had plates about the same size as some old GE 7581A's and also some real KT-66's I had. Happily, these tubes have proven to be as reliable as the 7581A types. I haven't had a problem with any I have used. Since then I have tried other Electroharmonix branded tubes and have been very happy with them.
-Chris
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.