Once again, thanks one and all! I will complete the tube amp on which I wish to listen to my old LPs on through an amplifier circuit that reproduces the sound as intended. Regarding audio 'listening' quality, I have built a 20 watt class A amp in which the distortion figures are in the .001 range. That said, records played through this amp sound 'clinical' and lack the more 'mellow' sound from a 'good' tube amplifier (I have compared them). I am aware that I am again wandering into dangerous ground but such is life!
Wishing one and all happy building🙂
Wishing one and all happy building🙂
Hmm...
Generously rated zeners are often not available in the exact voltage required--And are Expensive!
A Zener referenced MOSFET can be made trimmable to get exact bias, and with the appropriate Voltage rating, (suggest 1.5X B+ Supply) be reliable enough against your 'sub-microsecond shorts'--Of which I have never heard in all my 53 years!
--I suspect you mean some kind of flash-over or arc inside the tube--Probable in a high power Transmitter, but doubtful in an EL84 operated within its normal ratings.....
EL84--as far as I know, are not one of the usual suspects for interelectrode-shorts, not ever seen one fail that way--Yet!
Fuses make great safety devices too......
Generously rated zeners are often not available in the exact voltage required--And are Expensive!
A Zener referenced MOSFET can be made trimmable to get exact bias, and with the appropriate Voltage rating, (suggest 1.5X B+ Supply) be reliable enough against your 'sub-microsecond shorts'--Of which I have never heard in all my 53 years!
--I suspect you mean some kind of flash-over or arc inside the tube--Probable in a high power Transmitter, but doubtful in an EL84 operated within its normal ratings.....
EL84--as far as I know, are not one of the usual suspects for interelectrode-shorts, not ever seen one fail that way--Yet!
Fuses make great safety devices too......
Zeners commonly come in 5% tolerance. That is more than sufficiently accutate for tube biasing. Tubes aren't that critical. And if you are fussy, you can put a small trimpost in series with vety little effect on regulation. And if you are really really fussy you can put a shottky diode in series.
Let me be quite clear. Occaional sub-microsecond internal shorts in power tubes are quite normal. As I said they arise when stry electrons build up a charge on the inside of the glass. In power tubes that have seen a life of hard work, you can see the evidence of this in the form of burn marks on the inside of the glass adjacent to places where electrons can get out past the anode.
Unless these sub-micosecond shorts take out an external device, eg transistor, they cause no fault. The charge is discharged and the tube carries on. That's why you haven't noticed.
I'm not talking about the more dramatic internal arc that occur in tubes with loose material inside and/or have been operated well beyond ratings causing material to be knocked out of electrodes from ion bombardment. Such things do damage the tube.
As standard glass is a virtually perfect insulator, electrons striking glass buildup a negative charge that just keeps on increasing until breakdown and falsh-over occurs. The charge buildup is not directly dependent on teh applied anode or screen voltage, except in so far as the applied voltages cause electron acceleration.
Ever wondered why pre-war output tubes often had a grey coating (known as "dag" in the industry) inside the glass? It is a conductive coating to prevent charge build-up.
As I said before, after the War, a method of processing glass was developed that renders the inside very slightly conductive, so the dag was no longer required.
When a power tube is operated long enough without power off, and worked hard, charge buildup can still occur.
The problem is roughly proportional to tube size. Thus small power tubes like EL84 are not as susceptable as large tubes such as EL34 or KT88.
And as I said before, tubes made by US, British, and European tubes had good glass treatment. It appears Chines tubes do not. Russian tubes may not.
You may not have heard of it in 53 years. But it is certainly well known. TV set production went from all-tube designs, then in the early 1960's went to hybrid tube/transistor designs. Alls solid state designs came later. With hybrid designs, specific precautions were taken to prevent tube flashover from damaging semiconductors. Flashovers in the CRT gun, never noticed with all-tube TV designs, were particularly troublesome, but measures were also taken to protect transistors from power tubes such as frame output, audio output.
These measures were described in many technical journals and TV service manuals at the time. The protective measures ranged from subtle tings like carefull PCB layout (to prevent inductive coupling of the transient), choosing circuit configurations that separated transistors from tubes by transformers, spark-gaps, etc.
The occurance of sub-microsecond flashovers was also well known to tube based design engineers back in the tube days, as while the odd click can be tolerated or not even noticed in consumer gear, in recoding studio and other profesional gear it was not really toleratable. I recall reading about measures to address the problem in in Audio amplifiers in one of Neville Thiele's (a noted Australian electronics engineer with a world wide reputation) journal articles.
Fuses are useless from protecting semiconductors, and not terribly good at protecting tubes. Fuses when used properly protect against equipment faults causing a fire from grossly excessive power dissipation. Nothing more and nothing less.
It is bad engineering to insert a fuse to overcome a design fault that allows failure in one part to destroy another part, particularly a costly part. A good engineer will design out the cacading fault mode.
Last edited:
Mosquito asked what is the advantage of cross-couplinmg cathodes with two series capacitors with the centre point grounded via a high value resistor?
The resistor value is completely non-critical and it disspiates neglible power. Therefore it costs only cents and will just about never fail.
It's advantage is that it polarises the cross-coupling capacitors and thus permits them to be electrolytics, which are smaller and cheaper than non-electrolytics in the capacitance required.
However, I said in an earlier post, I prefer to use a non-electrolytic anyway - in which just a single capacitor between cathodes is required.
The resistor value is completely non-critical and it disspiates neglible power. Therefore it costs only cents and will just about never fail.
It's advantage is that it polarises the cross-coupling capacitors and thus permits them to be electrolytics, which are smaller and cheaper than non-electrolytics in the capacitance required.
However, I said in an earlier post, I prefer to use a non-electrolytic anyway - in which just a single capacitor between cathodes is required.
It must be a very long time since you looked at a modern FET line-up.
If one wished to design a FET-assisted zener for cathode biasing purposes, it can be achieved easily, using nFETs rated for higher than the dc supply voltage of any normal valve amplifier.
Dealing with sub-microsecond pulses is no problem, either, although I am assuming a competent designer.
The gate must be protected with the habitual zener, and proper precautions taken against oscillation, but these are very basic design tricks. With the Vds rating above the dc supply value - even 2x or 3x above, the task is quite manageable.
In this position, we don't need to care about Rds(on), so an nFET with a low-breakdown Vds rating is wholly pointless, even from a cost viewpoint. 900-1200V nFETs are a commonplace, and should be selected for any tube-amp design, if the designer or constructor needs a FET.
In all, this is just a scare-story, without any foundation in fact.
MOSFET in a Cathode circuit--wont fail if specced up correctly, especially when--As is usual practice when used in Cathode circuit--its decoupled by a cap--Cap will absorb/dissipate any small flashovers anyway.....
No bad design there. Been using MOSFETS and Tubes for years, both as plate-load and cathode riser--ONLY time I had a failure--was when I killed the MOSFET myself, on an A2 design I was potching with!
--The Fuse in the B+ was to protect the Tube/Output Transformer--In the Very unlikely event the MOSFET should short, or the tube be accidentally broken while on--Accidents Do happen.....
Adding a fuse I don't think constitutes Poor-Design in this instance.
Now you explained the phenomenon--Yes, Ive heard about this--But that charge is more akin to Static build-up, and has very little current behind it, easily quenched by the de-coupling-cap across the MOSFET, so to my mind is a non-issue....
Have you actually HAD a MOSFET in the cathode circuit fail that you can definitely say was due to this issue....?
No bad design there. Been using MOSFETS and Tubes for years, both as plate-load and cathode riser--ONLY time I had a failure--was when I killed the MOSFET myself, on an A2 design I was potching with!
--The Fuse in the B+ was to protect the Tube/Output Transformer--In the Very unlikely event the MOSFET should short, or the tube be accidentally broken while on--Accidents Do happen.....
Adding a fuse I don't think constitutes Poor-Design in this instance.
Now you explained the phenomenon--Yes, Ive heard about this--But that charge is more akin to Static build-up, and has very little current behind it, easily quenched by the de-coupling-cap across the MOSFET, so to my mind is a non-issue....
Have you actually HAD a MOSFET in the cathode circuit fail that you can definitely say was due to this issue....?
High voltage MOSFETS have been available from International Rectifier, Seimens etc for 35 years or more. But they wer not, and still aren't cheap. A simple diffused type bipolar transistor is comparitively cheap, vastly more tolerant of short spikes, and does just as good a job at regulating in this application.
In factory made equipment engineered for volume production and thouroughly acceptance tested, yes. As I said in an earlier post the problem was handled quite acceptably in TV sets - without using expensive high voltage MOSFETS.
But in home construction, with probably hobby-engineered layout etc, it is likley to be a problem.
You cannot possibly be serious. EL84's running on a HT of 300 V would allow you to choose a MSFET with 2x or 3x rating - at a price.
But a bigger tune running on 500 V HT or more rules MOSFETs out.
Note that an output tube under full drive sees its anode reaching twice the supply voltage due to transformer action. So chosing a MOSFET rated at only 2x the HT voltage has no margin at all.
Agreed. That's why we can use a simple diffused type bipolar. Cheap, more reliable, and does the job just as well. Even with a low collector voltage rating. There's no thin oxide layer to punch through, and no abrupt junctions.
Really?
Ever studdied application reports and engineering journal articles about it? Doesn't look like it.
In any case, as has been explained by myself and DF96, ccs cathode biasing does nothing usefull. It makes distortion worse. See earlier posts.
Last edited:
Yes, it is akin to staic buildup - that's really what it is.
But it is the voltage that punches through MOSFETs. Even in quite large power MOSFETs they are quite weak in this respect. In any case, if the tube flashes over cathode to screen, the current, although sub-microsecond, could be several amps, sourced from the HT filter cap.
No, because I am not dumb enough to put things in that don't do anything useful.
I do have lots of experience in the servicing of commercial hybrid tube-solid state equipment. Failure of semiconductors is very common in such equipment, even though the same parts when used in all-solid state equipment are very reliable.
Theere was an oscilloscope made by the now defunct Australian manufacturer BWD Electronics. In several models, the deflection amplifier was a cascode circuit - a pentode in the upper position driving a deflection plate, and a BC107 between cathode and ground. Signal input was to the BC107 base, with teh pentode operated "common grid". In solid state equipment, those old metal-can BC107's almost never ever failed. But in the large company I worked for, who had lots of thse CROs, dud BC107's were a routine replacement. Those BC107's were well within ratings, operating with only a small fraction of rated voltage current, power, and voltage. Evidence of tube flashover.
You may just be lucky with just one or two items in your home....
Last edited:
Yes, I delete the post because wrongly designed schematic. Anyway Here you're answering my question, and I'll try the modification in a Push Pull that I have, to see how this mod affect the performance.
Attachments
Last edited:
To get one thing straight right away - I am not arguing for or against cathode bias circuits of any kind.
For the rest, it seems your whole argument rests on cost.
This makes it especially easy to dismiss.
We only have to turn to the Mouser site and look at the first FET that popped out of a 1200V search:
IXTP06N120P IXYS | Mouser
The IXTP06N120P can withstand 1200V, handle 42W - but costs GBP 1.76 in tens. That's a little more than USD 2.50.
How cheap do you want?
So yes, I am serious - 500V dc supplies and all.
You are attributing quotes to me that were actually written by someone else - for the second time in the past few days.
Kindly take more care.
Go look up some bipolars. Since there is no oxide layer or abrupt junction, you don't need a big voltage rating. Any old 30V diffused device will do. $0.30 worth. Not a 10 cent planar though.
In any case, it is my view that if you must use constant voltage cathode biasing, you can just use a generously sized standard zener. Tubes don't need a precise voltage, and simple zeners are inherently rugged.
Last edited:
Sorry about that. Some sort of consequence of using the diyAudio quoting system.
OK--So you said--not a Plainar type--But Still a Semiconductor, just like a MOSFET.
MOSFET with an added 12-20V zener from S to G is very hard to kill--(Bit like this 'debate' it seems!)
Nah--its all to do with HOW you do summit with yer MOSFET as to how reliable its gonna be....
IF we were worried about every known (and unknown) phenomenon that can occur in electronic circuits--We would never end up building anything! By their nature--Tubes are less reliable than semis anyway--But we still build with 'em!
I'm out!😀
Last edited:
I buy 600V 3A STP3NK60Zs for near to $0.30 for industrial projects, with zener-protected gate, so for the EL84 application, you're actually perfectly safe with the FET at the same price - and not forgetting that any momentary slip or transient could blow a 30V bipolar transistor .
For DIY builds, using a 600-1200V FET is a safe route, and completely avoids playing roulette with marginal design.
But, the point I really want to make is that the suggestion that tube amps and the Power FET don't go together is now thoroughly out of date. A designer simply has to rate the Vds handling of the FET to the dc supply, plus some desired margin for safety.
The cost is a complete non-issue for DIY.
At higher currents, and high voltage, the Power FET actually has other advantages over BJTs- notably the freedom from second-breakdown.
Actually, choosing MOSFET with a high voltage rating doesn't help anywhere near as much as you apparently think.
Power MOSFETs have a very large internal capacitance between drain and gate. When external voltage spikes occur, as with tube flashover, the very rapid voltage rise gets coupled across into the gate, which is rated of course much mauch lower.
The positive voltage on the gate turns the MOSFET on of course, which should protect it. But the finite Rds(on) limits what it can do.
It's true that a home constructor does not have to justify everything to his boss or to the company bean counters - and doesn't have to show a profit.
But it's still silly to spend money unnecessarily. And where does it end? With a cost-doesn't-matter attitude, a home constructor could end up wasting money in all sorts of directions, in electronics projects and elsewhere.
You must be joking again.
2nd breakdown is simply NOT going to be an issue in tube biasing.
You and Alastair are ignoring that MOSFETs have an extremely thin oxide insulation layer. This layer is easily broken down by high voltage, and when it is the MOSFET is rendered useless.
MOSFETS incorporating internal gate protection zeners are proof against handling static (the early power MOSFETs required antistatic handling), 500V multi-ampere whacks from power tube flashower is beyond what they are meant to handle.
Bipolar transistors do not have this oxide layer. The chip can still be destroyed by transients, but it requires not just high voltage. it requires a certain ENERGY. Since the tube flashovers I've been talking about are sub-microsecond, a bipolar is inherently a better choice.
It is well known among engineers that bipolars made by the older diffusion process are more rugged. Silicon planar offers high performance - high gain and wide bandwidth, but its abrupt junctions and tiny dimensions make for a lower pulse energy required to destroy it. Oils ain't oils, and bipolars ain't bipolars.
Last edited:
For your possible interest:
6 x EL84 PPP with current source biasing.
http://www.keith-snook.info/amplifier-hifi-schematics/Beard Audio Service data P35 M70.pdf
I used "Ring of two" transistor current sources in my Baby Huey. A general purpose NPN small signal transistor with an MJE340 as the pass device.
The problem with current source biasing is overload recovery. The bypass caps are required to achieve Class AB. When these bypass caps charge up during momentary overload the time taken to bleed away the "extra" voltage is even longer than when using cathode bias resistors.
Cheers,
Ian
6 x EL84 PPP with current source biasing.
http://www.keith-snook.info/amplifier-hifi-schematics/Beard Audio Service data P35 M70.pdf
I used "Ring of two" transistor current sources in my Baby Huey. A general purpose NPN small signal transistor with an MJE340 as the pass device.
The problem with current source biasing is overload recovery. The bypass caps are required to achieve Class AB. When these bypass caps charge up during momentary overload the time taken to bleed away the "extra" voltage is even longer than when using cathode bias resistors.
Cheers,
Ian
I am thoroughly familiar with FET structure, thank you. I have been designing them into professional industrial equipment for more than 30 years.
Gate-source breakdown events are well understood. If you had read my words more carefully you would have seen the phrase "The gate must be protected with the habitual zener". Along with a precautionary Drain - Ground capacitor (stacked film), that is all that is required to manage your "sub-microsecond" strikes.
Protecting the gate in any application of MOSFETS is simply standard practice in competent design engineering.
We talking about paying USD 2.50 for a device than can withstand 1200V, versus your $0.50 for a BJT that can stand 30V.
In an equipment with a supply of 300-500V, with exposed terminals, built expressly for DIY tinkering, taking the 30V part is an obvious false economy. Accidental overvoltages are always likely.
Not talking about biasing - but general applications of FETs in Valve amps.
I am challenging your assertion that FETs are unsuitable for valve amps, which is patently untrue. The FETs must be dimensioned correctly, and circuits designed properly, but it is not overly difficult.
FETs are widely found in (for example) HV series regulators, anode loads, and all manner of applications.
The depletion-MOS device is almost universal in anode-load current source applications among the DIYers here on DIYaudio, and yet there are few, if any reports of the FETs failing.
I am not ignoring the structure, I am perfectly familiar with it.
With any semiconductor, reliable operation can be assured by simply keeping well within the data sheet ratings - accounting for known and measured transient events, and any that can be reasonably predicted.
With a FET, choosing one with a good margin for drain voltage vs supply voltage, and protecting the gate are the major precautions.
The rest of your paragraphs appear to be attempting to resurrect the scare-story that FETs can't be applied to valve amplifiers.
Well, a placebo doesn't actually change anything, so I'm not sure how this relates to an experiment where something was actually changed.
I personally find your comments and attitude to be rude and arrogant.
Back to topic, have you actually tried these output stage biasing methods?. If so, which CCS did you try and did you have means to balance the standing currents between the tubes, further.. were you using DHT output stage, or some other 'VW Beetle'.
One size does not fit all, and no need to throw away the foot if the shoe doesn't fit.
L.H
The late Allen Wright was truly on to something with his DPA amplifier. Common kathode CCS (without) bypass capacitor, and means to adjust for plate current between the two output tubes. I was following Lynn Olson and his Amity amplifier with PP drive to PP output, and preferred the SE to PP via a custom transformer in lieu of the Lundah, be it SE or PP drive. 5687/7119 with PP drive less good than 71A in SE drive, for me and me alone. Common kathode resistor without bypass capacitor was better, for me but I had means to adjust for plate current courtesy of the split secondary windings of the interstage. And of course, following on from that comparison between pure cathode bias (bypassed and not so) and fixed bias with grounded filament return. In the end, common CCS in the filament circuit, with means for DC adjustment via simply paralleled pots between the common CCS to signal common with wipers to grid returns (separate), worked very well and sounded about the same in magnitude of order over the aforementioned scheme.
But hey, nothing wrong with cathode bias in the strictest terms.. just to say, don't discount other things, and stop with the 'tubes weren't designed for fixed bias', indeed most if not all manuals, and I have some dating back to 1927, prescribe a C supply.
To be factual orders one to be factual.
L.H
But hey, nothing wrong with cathode bias in the strictest terms.. just to say, don't discount other things, and stop with the 'tubes weren't designed for fixed bias', indeed most if not all manuals, and I have some dating back to 1927, prescribe a C supply.
To be factual orders one to be factual.
L.H
- Status
- Not open for further replies.