It is commonly recommended that if employing series connection of filaments in the amp design (2 * 6.3 V = 12.6 V transformer winding), that the tubes be matched and always replaced as a pair.
Does this also hold true for regulated DC filament supplies, whether those supplies be constant current or constant voltage?
Does this also hold true for regulated DC filament supplies, whether those supplies be constant current or constant voltage?
As far as the filaments are concerned the only matching necessary is the current draw of each 6.3V tube. That must be the same so that each tube receives equal division of the 12.6 volts. And you probably already know this.
Replacing one tube with the same type and not the other is of no consequence as far as the filament supply is concerned. Be it regulated or not. But keeping them matched for other reasons like gain balance might be necessary.
Victor
Replacing one tube with the same type and not the other is of no consequence as far as the filament supply is concerned. Be it regulated or not. But keeping them matched for other reasons like gain balance might be necessary.
Victor
serial
The problem is actually that filaments may have slightly different resistances, which results in different voltage drops across each tube.
Say, you have a 12V DC regulated filament suply and you wire two similar 6.3V tubes in series. Then you measure 5.9V across one tube and 6.1V across another tube. If they're "matched" you may measure something like 5.95 and 6.05.
But I'm afraid the term "matching" usually refers to the B+ currents, not the heaters.
Btw, some older tubes have other problems with series heater supply, for example check the 6C8G, but that's another story.
The problem is actually that filaments may have slightly different resistances, which results in different voltage drops across each tube.
Say, you have a 12V DC regulated filament suply and you wire two similar 6.3V tubes in series. Then you measure 5.9V across one tube and 6.1V across another tube. If they're "matched" you may measure something like 5.95 and 6.05.
But I'm afraid the term "matching" usually refers to the B+ currents, not the heaters.
Btw, some older tubes have other problems with series heater supply, for example check the 6C8G, but that's another story.
Now I think I am more confused. I actually had never meant to suggest that the tubes be matched for filament draw, just that they be matched for gain.
Are you suggesting that ensuring that the tube filaments having the same voltage across them is of no consequence, or that it is necessary for this to be the case to get matched gain from them?
Are you suggesting that ensuring that the tube filaments having the same voltage across them is of no consequence, or that it is necessary for this to be the case to get matched gain from them?
Morgan Jones addresses this in his "Valve Amplifiers" book. I don't have the exact statement in front of me, but basically he recommends heaters in parallel (whether in the same or different tube) be supplied with a voltage source, and heaters in series with a current source.
This will be your best defense against varying filament resistances, temperature coefficients, and the like.
While some designs may benefit from balancing triode sections for bias, gm, or mu, I don't hear of too much emphasis on filament balance. It would certainly be that much more difficult to balance both the triode sections and filament at the same time.
You mentioned "it is commonly recommended". I haven't heard this too much, but I can understand the concern about imbalance with a series connection and voltage source. Balance would be beneficial.
To answer your last question, it wouldn't change matters if you use AC or DC, regulated or unregulated, but it would change matters if your are using a voltage or current source (see first paragraph).
This will be your best defense against varying filament resistances, temperature coefficients, and the like.
While some designs may benefit from balancing triode sections for bias, gm, or mu, I don't hear of too much emphasis on filament balance. It would certainly be that much more difficult to balance both the triode sections and filament at the same time.
You mentioned "it is commonly recommended". I haven't heard this too much, but I can understand the concern about imbalance with a series connection and voltage source. Balance would be beneficial.
To answer your last question, it wouldn't change matters if you use AC or DC, regulated or unregulated, but it would change matters if your are using a voltage or current source (see first paragraph).
If 2 tubes of the same type, manufactured in the same time frame in the same plant, are wired with series heaters, there is nothing to worry about.
Series connection of disparate tubes is a problem unless the warmup time is controlled. The "A" suffix seen in 7 and 9 pin miniature types is an indication of controlled heater warmup time. The warmup time became a matter of concern in TV receivers that used length series heater strings.
Series connection of disparate tubes is a problem unless the warmup time is controlled. The "A" suffix seen in 7 and 9 pin miniature types is an indication of controlled heater warmup time. The warmup time became a matter of concern in TV receivers that used length series heater strings.
zigzagflux said:
Exactly!
So if you intend to wire something in series just build a current regulated PS, and don't worry about the voltage.
Yesterday I've missed the part of your post saying "constant current or constant voltage". Sorry for that! When I hear "regulated PS" I always think of regulated voltage.
The old apparatus that used series wired mains powered heaters sometimes had a current regulator tubes, and if you read the datasheets you may pay attention to the fact they didn't vorry about the voltage being 110V or 115V or whatever.
For example, read this: U30
Another important thing to remember if you are using series connected filaments, designed for series connection (*), but without a constant current source to provide a constant heater current for the chain, you need a 'buildout resistor'. Heater filaments tend to have a much lower resistanxce when cold, which results in a considerable current surge when the heater voltage is applied. In series heater chains, it is the relative differences of these resistances, as well as heating times of the actual tubes, that result in a seizable imbalance of voltages across series connected heaters, simultaneously with the aforementioned current surge. As a result , some tubes' heaters literally flash white hot. Although heaters generallly survive this for a long time (but shorter than under proper conditions), it causes filament material evaporationand contamination of the insides of the tube, hence deteriorated characteristics. It is often visible as blackened areas on the bulb glass between the pins and the bottom mica in tubes used in TVs, for instance. To lessen this effect, series connected strings were not directly connected to a voltage supp[ly equalling the approximate sum of all heater voltages, but rather to a slightly higher voltage, via a 'buildout resistor'. Because this resistor, unlike the resistance of the filament, is a constant, it acts as a current surge limiter, and considerably lessens the effect described above. Of course, the penalty is the ehat developed on this resistor. With a current source supplying the chain, no resistor is necessary, since the current source already limits the current. Use of it, however, does considerably prolong the ehating up time.
(*) Tubes not originally designed for series heater conenction whould in general NOT be connected in that manner as for them it is only guaranteed that the correct applied heater voltage will make the heater operate correctly. There can be considerable variations in the actual current even between different production dates of the same tube version, from the same manufacturer, not to mention across different versions, dates and manufacturers. If series connected, this will result in imporpper heater voltage distribution - one will be too high, one too low. If you must series connect, them you will have to chose and measure tubes to work properly so connected.
There are also a number of tubes that were designed with several heater connection syles in mind. 12A*7 are a good example - they can be connected as 6.3 or 12.6V parallel heated, and just happen to require 300mA and 150mA heater current respectively, which are standard series heater chain currents. Although this looks like great foresight on the part of the tube designers, you should not take this for granted. Many versions of the tube exists, and unless you have the correct datasheet to check if series heater chain application is designed for (keyword: controlled heater characteristics, or the data outright mentions it), do not assume anything. It largely depends on what market the particular version was made for. In Europe it is common to apply tubes like the ECC83 (~12AX7) in a 300mA series chain, also the EF80 and EF184, along with some ECH models that have 6.3V heaters that also pass 300mA under these conditions. But even themn there are great variations in actual heater construction, evident when you replace a tube in a series chain and get the above mentioned white hot filament 'flash'.
(*) Tubes not originally designed for series heater conenction whould in general NOT be connected in that manner as for them it is only guaranteed that the correct applied heater voltage will make the heater operate correctly. There can be considerable variations in the actual current even between different production dates of the same tube version, from the same manufacturer, not to mention across different versions, dates and manufacturers. If series connected, this will result in imporpper heater voltage distribution - one will be too high, one too low. If you must series connect, them you will have to chose and measure tubes to work properly so connected.
There are also a number of tubes that were designed with several heater connection syles in mind. 12A*7 are a good example - they can be connected as 6.3 or 12.6V parallel heated, and just happen to require 300mA and 150mA heater current respectively, which are standard series heater chain currents. Although this looks like great foresight on the part of the tube designers, you should not take this for granted. Many versions of the tube exists, and unless you have the correct datasheet to check if series heater chain application is designed for (keyword: controlled heater characteristics, or the data outright mentions it), do not assume anything. It largely depends on what market the particular version was made for. In Europe it is common to apply tubes like the ECC83 (~12AX7) in a 300mA series chain, also the EF80 and EF184, along with some ECH models that have 6.3V heaters that also pass 300mA under these conditions. But even themn there are great variations in actual heater construction, evident when you replace a tube in a series chain and get the above mentioned white hot filament 'flash'.
Eli Duttman said:
I had a Marantz 2 power amplifier with a pair of 6AU4 wired in series - why I will never know, but in any case I had a lot of rectifier tube failures in this amp due to uneven distribution of filament voltage across the pair. The solution was to find a match to one of the tubes installed in the amplifier such that the filament voltages were equal after some warm up time.. Note that several of the failed tubes met all of the above criteria.
Concerning the statements that parallel or series is better with current regulation or voltage regulation, I don't get it. Once everything is warmed up, all voltages and currents will be stable. The heaters don't know or care if current, voltage, or neither is actively regulated.
I looked through MJ's Valve Amplifiers 4th ed. for his comments, but couldn't find them. Perhaps someone could offer page numbers, or better yet an explanation.
-- Dave
I looked through MJ's Valve Amplifiers 4th ed. for his comments, but couldn't find them. Perhaps someone could offer page numbers, or better yet an explanation.
-- Dave
Dave Cigna said:
That is indeed true, but the question is what you design for in terms of tolerances, and what happens before everything is stable.
In tubes designed for series chains,m manufacturers guarantee that the heater will ehat correctly at a given current and the voltage is left as the variable with a large tolerance. For parallel heaters the oposite is true. So, if you are not taking the right tolerances into account, you may end up with either the need to select tubes, or with a possibility for under and/or overheated heaters in your tubes, or possible failures due to turn-on transients.
Series heaters- Unequal voltage, but equal current.
Parallel heaters- Unequal current, but equal voltage.
In either case power dissipation varies between heaters. So it doesn't really matter which way you go. If you go regulated, voltage regulation works best with parallel and current regulation works best with series. Series connection with current regulation also provides a soft turn on and can prolong life.
The only time I would recommend against series connection is if you have a plate matched pair of tubes. Assuming they were matched having an equal filament voltage at the time.
The best performance I've had was current-regulated, series fillaments. Warm up time may be a little slower but the tubes last longer.
Parallel heaters- Unequal current, but equal voltage.
In either case power dissipation varies between heaters. So it doesn't really matter which way you go. If you go regulated, voltage regulation works best with parallel and current regulation works best with series. Series connection with current regulation also provides a soft turn on and can prolong life.
The only time I would recommend against series connection is if you have a plate matched pair of tubes. Assuming they were matched having an equal filament voltage at the time.
The best performance I've had was current-regulated, series fillaments. Warm up time may be a little slower but the tubes last longer.
I understand that current regulation results in slow warmup which is presumably good.
And, I understand that some people find that current reg sounds better with directly heated tubes, but that's not what we're talking about here. We're talking about indirectly heated tubes. In particular, the subjective sound of different types of supplies is an entirely different discussion.
My wondering has to do with what's going on after everything is warmed up and stable. So long as nothing is out of whack, that the heaters themselves are close to their nominal specifications, that any regulation - voltage or current - is set where it's supposed to be, then I see no difference between current regulation and voltage regulation. I contend that it would be virtually impossible to tell which type was being used just by probing around with a DMM. (Perturbing the system in any way would be cheating.) I can think of no reason why the heaters themselves would care what type of regulation (or none at all) was being used.
So, I just can't find any basis in the claim that one method works better with parallel connection and the other with series.
-- Dave
And, I understand that some people find that current reg sounds better with directly heated tubes, but that's not what we're talking about here. We're talking about indirectly heated tubes. In particular, the subjective sound of different types of supplies is an entirely different discussion.
My wondering has to do with what's going on after everything is warmed up and stable. So long as nothing is out of whack, that the heaters themselves are close to their nominal specifications, that any regulation - voltage or current - is set where it's supposed to be, then I see no difference between current regulation and voltage regulation. I contend that it would be virtually impossible to tell which type was being used just by probing around with a DMM. (Perturbing the system in any way would be cheating.) I can think of no reason why the heaters themselves would care what type of regulation (or none at all) was being used.
So, I just can't find any basis in the claim that one method works better with parallel connection and the other with series.
-- Dave
Dave Cigna said:
I would say you are both right and wrong
If your assumptions (*) and (**) are true, then you are certainly right on accounts (***) and (****), but they are rather big assumptions. Although heater failures are really not that common, it is practically exclusively what happens BEFORE (*) that is to blame for them. And, let's face it, you have to think about that too. Especially with the more popular tubes that have amny NOS types as well as current - how many heater types have you seen? You know, the ones that heat up nice and evenly on one end of the scale, vs the ones that give a bright flash on the exposed part of the filament between cathode sleeve and pin, on the other? If you know in advance what you will be putting in a construction, certainly the assumptions (*) and (**) are easyer to satisfy. If you want to plan for some tolerance and flexibility, then it is best to design with what the heaters themselves were designed for - keeping in mind that in later tubes most of the heater design was unified, so both series and parallel heater style tubes got controlled heater characteristics. But, this is not true of all tubes out there so one should check.
Regarding voltage regulation for parallel connection, it does work best when everything is up and running - IF you really need regulation - but it is prudent to include some sort of current limit for the initial surge. Mostly with AC heating this is provided by dimensioning the transformer. And yes, with an oversize transformer, some parallel heater tubes do flash bright white on application of power! With series heating, again, if you must regulate, you would be safest with constant current, although constant voltage will work with a surge limit and if it's close enough, because the filaments are designed to counteract current variations. Keep in mind that with series chains they invariably DID start off with constant voltage, that of the mains supply usually - but there was always a build-out resistor that also served as a surge limiter. Check the definition of heat-up times for most tubes - there is something like 'when filament is connected in series with resistor of a value of the nominal heater resistance at operating temperature, to a voltage of twice the heater voltage' in there. This actually spells out surge limit by buildout resistor, and it also implies that initially the current is much bigger but not more than twice the nominal value.
Dave Cigna said:
I have the third edition, which I believe is the latest. Take a look at pp. 267-270, under the heading "Heaters and their supplies":
"...indicating that the heater behaves as a constant resistnace over this very limited range. Since P=I^2R and P=V^2/R, two conclusions emerge. First, parallel chains should be constant voltage regulated, and series chains should be constant current regulated.
Second, we should not mix topologies...because the 6SN7 double triode has its heaters internally wired in parallel it should only be used with a 6.3V Thevenin supply, whereas a 12SN7 (whose heaters are in series) would ideally be powered by a 300mA Norton supply."
There is much more in this section of interest, and I would say I have preserved MJ's context with these quotations. Essentially, any mismatch will be corrected a little with a type of negative feedback by sticking to the connections he specifies.
If you current regulate parallel connection and one of the filaments opens it will cram more current into the other filaments. Of course if the supply voltage feeding the CCS is just enough it will prevent this. But it's best not to design things that close.
Aside from that this is why it works best. However, I didn't say it was necessary as nothing is wrong with a little fluctuation, but hitting the ideal is best.
Voltage regulating parallel connection ensures an ideal voltage across all filaments. Only the current per tube varies. Current regulating wouldn't ensure either. Shure, all filament voltage would be the same, but there is no guarantee what that voltage would be.
Current regulating series connection ensures ideal current through all filaments. Only voltage across each filament varies. Voltage regulating wouldn't ensure either. Shure, all filament currents would be the same, but there is no guarantee what that current would be.
As stated earlier I've had the best results with series filaments with current regulation. Partially do to the soft start characteristic. Also because current regulation has a very good noise rejection and creates the lowest amount of ground noise. Because the amount of current flowing trough the ground conductors never varies due to the variable resistance of the tungsten or from what ripple there may be. While the noise improvements may be very small I'll take any improvements I can get when doing small signal stuff such as a phono stage.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.