I used to use five turn 2W W/W pots for my bias adjustments (and other supporting parts), but I found that without servo based bias (or cathode), tubes would work as expected but bias would drift slowly and eventually the tube would runaway and red plate. George described similar results when he triode-strapped sweep tubes, too.
Now that I'm using auto grid bias in the design and providing a larger overhead for the bias voltage I haven't had any issues with runaway. I run the amp 24/7, too.
Now that I'm using auto grid bias in the design and providing a larger overhead for the bias voltage I haven't had any issues with runaway. I run the amp 24/7, too.
That could be a triode-strapped-sweep-tube problem, not typical of regular audio tubes whose emissions drop off slowly over time. One of the reasons George developed the UnSET - he’s convinced the runaway is inherent to switching tubes run at too high a g2. Some of them won’t even run 10 minutes in triode at full dissipation rating at higher plate voltage. Switching TRANSISTORS have that problem - it’s called second breakdown and responsible for a lot of blown Phase Linears. I’ve got sweeps running in regular pentode mode with 125V to 160V g2, but not enough hours on them to say that they’ll NEVER run away. Just no problems yet, and no more bias drift than a set of 6550’s.
It could be. The behaviour seems to be similar to using too high a grid leak resistor. I'm pretty sure if you took a KT88 and use fixed bias with a 1M grid leak, you'd have similar runaway problems.
The other thing I seem to remember was his experiments were single ended where my amps are all push pull. I doubt that matters unless he was running higher idle current than I do (even with 6P45S, I don't set it higher than 320V@120mA).
I also mostly design around the Soviet sweeps, and most of them have a higher rating for Vg2 than the US types I've played with.
That said, I've been running 12GT5 tubes at 320V B+, 65mA in this MA-1 24/7 for about a week now. So far so good.
I also built a client an MA-1 based on 12GE5 which had been working for him flawlessly since he got it (as it should). He doesn't run it 24/7 though.
Both the 12GT5 and 12GE5 are Sylvania if anyone was curious.
The other thing I seem to remember was his experiments were single ended where my amps are all push pull. I doubt that matters unless he was running higher idle current than I do (even with 6P45S, I don't set it higher than 320V@120mA).
I also mostly design around the Soviet sweeps, and most of them have a higher rating for Vg2 than the US types I've played with.
That said, I've been running 12GT5 tubes at 320V B+, 65mA in this MA-1 24/7 for about a week now. So far so good.
I also built a client an MA-1 based on 12GE5 which had been working for him flawlessly since he got it (as it should). He doesn't run it 24/7 though.
Both the 12GT5 and 12GE5 are Sylvania if anyone was curious.
I keep my grid leaks low in general - why I used direct coupled k-follower drivers on my big amp. I knew I was going to want to beat on it.
Horizontal sweeps are probably more prone to runaway than verticals, as the latter runs as a linear amplifier. Less total emission capability, too. No one needed 1 amp peak currents in the vertical sweep circuit. And runaway has a lot to do with quality, too. Back on the 7868s, old Sylvanias were still “working” but the cracked glass and almost exhausted getter were worrisome. Replacement Sylvania/PHILIPS were catastrophic. Despite perfectly matched cathode current initially (put in 10 ohm resistors to check). Every time Philips acquired something (Magnavox, Westinghouse lighting division, Advance ballasts) the quality seemed to go to hell in a handbasket. I figured it was part of the pattern. Everybody does that cost cutting crap today, but this was 35 years ago. ECG transistors have always been known to blow up, but a lot of it was just poor replacement suitability - they would often sub a switching type for a linear amp in their cross reference because it’s just based on superficial maximum ratings. That NEVER works, Philips or no Philips. Me and my old college buddy still Philips-bash today. Although my Sonicare still works as well as it did decades ago. I’ve acquired many Sylvania/Philips 21LG6’s and 26LW6’s, and a scattering of others and we’ll eventually see how those work out. At those prices it matters not. Nothing ventured, nothing gained. But if I keep G1 resistance low and Vg2 reasonable, they’ve got a chance.
Horizontal sweeps are probably more prone to runaway than verticals, as the latter runs as a linear amplifier. Less total emission capability, too. No one needed 1 amp peak currents in the vertical sweep circuit. And runaway has a lot to do with quality, too. Back on the 7868s, old Sylvanias were still “working” but the cracked glass and almost exhausted getter were worrisome. Replacement Sylvania/PHILIPS were catastrophic. Despite perfectly matched cathode current initially (put in 10 ohm resistors to check). Every time Philips acquired something (Magnavox, Westinghouse lighting division, Advance ballasts) the quality seemed to go to hell in a handbasket. I figured it was part of the pattern. Everybody does that cost cutting crap today, but this was 35 years ago. ECG transistors have always been known to blow up, but a lot of it was just poor replacement suitability - they would often sub a switching type for a linear amp in their cross reference because it’s just based on superficial maximum ratings. That NEVER works, Philips or no Philips. Me and my old college buddy still Philips-bash today. Although my Sonicare still works as well as it did decades ago. I’ve acquired many Sylvania/Philips 21LG6’s and 26LW6’s, and a scattering of others and we’ll eventually see how those work out. At those prices it matters not. Nothing ventured, nothing gained. But if I keep G1 resistance low and Vg2 reasonable, they’ve got a chance.
Ya for sweeps, I tend to aim for 100k. KT88 max is 51k I think.
I find 330k is ok for most Soviet tubes though.
I use the following poles.
First gain to splitter = 0.22u/510k
Spiltter to driver 1uF/510k
Driver to power tube 1uF/100k
This gives me a stable amplifier without "breathing" or "motorboating".
I find 330k is ok for most Soviet tubes though.
I use the following poles.
First gain to splitter = 0.22u/510k
Spiltter to driver 1uF/510k
Driver to power tube 1uF/100k
This gives me a stable amplifier without "breathing" or "motorboating".
Add a bias servo and you have *four* poles. Transformer makes five. You can direct couple the first two stages and eliminate one.
I was looking at two direct coupled LTPs driving a pair of 10JA5 (7591-like curves and lotsa gain) to see just how much global NFB I can muster, while keeping only two low frequency poles. One driver stage would be *enough* to get distortion below 1% with 6 -10 dB of feedback but where’s the challenge in that?
I was looking at two direct coupled LTPs driving a pair of 10JA5 (7591-like curves and lotsa gain) to see just how much global NFB I can muster, while keeping only two low frequency poles. One driver stage would be *enough* to get distortion below 1% with 6 -10 dB of feedback but where’s the challenge in that?
Philips was a Dutch company that loved living on IP, as a company model, until well after kill-off time. Remember the "pentode" wars, which the US makers destroyed en masse, via RCA and Sylvania?
A tiny sub sea country like NL should have seen that coming, but persisted with gross propaganda via Mullard's so called "superior quality" wars which persist to this day, and repeated ad nauseam on all audio forums (fori?).
The wheels started to come off on washing machines, so they started to hive stuff off at long last (anyone remember the "whirlpool" rebranding fiasco or the @&^#!*$ compact cassette??).
Legacy also persists with SPFIF (sony-philips) cr..p.
I will wave a happy goodbye to the day Philips no longer have any part in audio, as they were incapable of making microphones or mixers (the British, Germans and Danish sorted that out thank goodness!), while the Swiss fixed lots of other things inc recording and software.
Buy a DAC today, and we don't want anything Philips in there, I bought British made pentodes from GEC, and for the most part we use Swiss and French made hardware and software.
Go figure? Cartels are cartels no matter how you dress them up.
fact is the Westing 7591 and 8417 were a quantum leap better than anything from Europe until a (too late) belated attempt was made with the EL520 and some exceptional "TELE" stuff came out of Germany.
Last edited:
I agree with you, but to be fair, Philips can be separated into two for audio into either consumer grade (which I believe you are talking about) or over the top no expenses spared commercial-grade audio (which is rare over here in North America).
I would seriously love to own a Philip's EL6471 if nothing but to just to look at it... I probably wouldn't be able to power it up without adding some wiring and proper fuses. Anyway if I could, I am sure my wife would wonder where I got the speakers to handle the output and then say the setup is taking up too much space and why are the lights flickering to the music. So with that said, I would sell it and get something more living room acceptable and invest what is leftover.
In the end, I would sell Philips audio gear to someone who wants the troubles associated with it all.
I would seriously love to own a Philip's EL6471 if nothing but to just to look at it... I probably wouldn't be able to power it up without adding some wiring and proper fuses. Anyway if I could, I am sure my wife would wonder where I got the speakers to handle the output and then say the setup is taking up too much space and why are the lights flickering to the music. So with that said, I would sell it and get something more living room acceptable and invest what is leftover.
In the end, I would sell Philips audio gear to someone who wants the troubles associated with it all.
Ditto for Harmon International these days. Just because something says “JBL” on it no longer really means anything. You can still buy good JBL speakers, Crown amps, and dBx processing. You can’t buy those models from some place which just sells to anybody (at least not without some RIDICULOUS markup intended to be a deterrent).
Mullard today is just another New Sensor. I guess Electroharmonix is just as “good”, eh? At a few bucks less…..
Mullard today is just another New Sensor. I guess Electroharmonix is just as “good”, eh? At a few bucks less…..
Forum is second declension neuter, so the plural is Fora
4 years of Latin and that's all I remember.
Back to topic -- I wonder if we could "gel" and cross link 5 mils of acrylated silicone resin on a naked 7868 to improve the glass quality.
Could 7591's were used in place of 6L6's in the circuit below w/ no changes? 7591 spec sheet seems comparable to 6L6. Notice OT impedance is about 1/2 of non-quad PP.
The 'hot rod' ST70 mod article uses 1/2 the secondary load so the primary impedance is effectively 1/2- but wouldn't it be better to use OT's with 1/2 pri impedance to start with? Maybe cost was the reason. Would the high DCR values of original ST-70 OT's be a limiter on plate currents in quad design?
Jim
The 'hot rod' ST70 mod article uses 1/2 the secondary load so the primary impedance is effectively 1/2- but wouldn't it be better to use OT's with 1/2 pri impedance to start with? Maybe cost was the reason. Would the high DCR values of original ST-70 OT's be a limiter on plate currents in quad design?
Jim
Attachments
I don’t see what a treatment on the outside of the glass would do to the quality of the glass itself. The treatment wouldn’t be hermetic so no help there (a poor glass to metal seal would still be poor), and any impurities that make the glass ‘bad’ would be trapped inside still.
oemcar,
6L6 all variations:
Transconductance ~ 6,000 uMhos
u, g1 to g2 ~ 8
Maximum Plate dissipation 19 or 30 Watts, depending on the variation.
7591 all variations:
Transconductance ~ 10,000 uMhos
u, g1 to G2 ~ 17
Maximum Plate dissipation 19 Watts
The tubes are not nearly the same.
Yes, change some voltages, resistors, and output transformer primary circuit values, or put up with less than optimum performance.
Original 35 Watt transformers will saturate at low frequencies when used at 40, 60, or 75 Watts; Right?
Global negative feedback resistors and compensation capacitors will have to be adjusted when different output transformers are used.
Just my $2 opinion
6L6 all variations:
Transconductance ~ 6,000 uMhos
u, g1 to g2 ~ 8
Maximum Plate dissipation 19 or 30 Watts, depending on the variation.
7591 all variations:
Transconductance ~ 10,000 uMhos
u, g1 to G2 ~ 17
Maximum Plate dissipation 19 Watts
The tubes are not nearly the same.
Yes, change some voltages, resistors, and output transformer primary circuit values, or put up with less than optimum performance.
Original 35 Watt transformers will saturate at low frequencies when used at 40, 60, or 75 Watts; Right?
Global negative feedback resistors and compensation capacitors will have to be adjusted when different output transformers are used.
Just my $2 opinion
Last edited:
The 6L6, 5881, 6V6, KT66, KT77, 6550, and KT88 all use the same base connections.
The 7591 has a different base pinout connection diagram.
Do not make the same mistake I made once, I forgot, so I wired the 7591 connections as if it was a 6L6, 5881, 6V6, KT66, KT77, 6550, or KT88.
My bad. Live and learn.
I turned it off just in time to save the 7591 and surrounding circuit.
The 7591 has a different base pinout connection diagram.
Do not make the same mistake I made once, I forgot, so I wired the 7591 connections as if it was a 6L6, 5881, 6V6, KT66, KT77, 6550, or KT88.
My bad. Live and learn.
I turned it off just in time to save the 7591 and surrounding circuit.
Dropping the 7591 into the Williamson amp would let you eliminate the differential driver stage - split load inverter would be enough.
The "super" ST-70's output transformer would saturate at the same voltage as in a normal ST-70 - which would be twice the power. Yes, a transformer designed for PPP use would perform better, but doubling up the outputs just requires some available time and a chassis punch...
The "super" ST-70's output transformer would saturate at the same voltage as in a normal ST-70 - which would be twice the power. Yes, a transformer designed for PPP use would perform better, but doubling up the outputs just requires some available time and a chassis punch...
21HB5A gives you the same gm as the 7591 (10200)
(at the same plate current), with 2.7x the current capability (so you can use a lower primary Z OT, like 3.3K or 4K, and lower B+). As long as you are not using UL. Ie, as in the above schematics listed by Oemcar. These used to be $1, but are still relatively cheap. Plate size is the same as most 24 Watt TV Sweeps, and with NO plate cap.
(at the same plate current), with 2.7x the current capability (so you can use a lower primary Z OT, like 3.3K or 4K, and lower B+). As long as you are not using UL. Ie, as in the above schematics listed by Oemcar. These used to be $1, but are still relatively cheap. Plate size is the same as most 24 Watt TV Sweeps, and with NO plate cap.
Last edited:
That's a little lower gm than the 7591. 6573 gm at 60 mA versus 10200 gm for the 7591 at 60 mA.
35LR6 will get you 10500 gm at 60 mA, but it's a rather bigger 30 Watt Sweep tube, with a plate cap.
Two 21HB5A in parallel gives you an 8417 ( with more than twice the current capability and 1/4 the Rp, using half the Vg2, 1/2 the knee Volts for better efficiency, and better triode curves, for $2 a few years ago.)
Last edited: