|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
184.160.11.147
Hello all,It's been discussed a bit here and there, and i know Pete Millett has come with an elaborate and elegant solution. But for a simpler solution i also read that people are using integrated stereo IC amps (STK465 for example, see image and link) driven by some oscillator circuit. It seems those can go up to over 50Khz, which is plenty since tests made by Kurt S. around 2001 found out that upwards of 30khz was good, and that feasibility decreased drastically from that point anyways.
Even read reports of people driving their heaters from a regular mid-fi SS amp driven by a sine wave gen. I might try that, or buy one of those ICs to play with. IIRC correctly, Kurt even used a tube amp for heating the filaments of his test bed amp !!!
Anybody has experience along those lines? A 6P21S happens to have 8 ohms resistance... :-) I'm a bit concerned by the low cold filament resistance some tubes exhibit, but a pre-heat circuit is possible.
Joris
Edits: 10/27/14 10/27/14Follow Ups:
I have played with filament high frequency heating in 1998.
My design supplied a pure sine wave to the filament.
It worked nicely. The background was the blackest I have ever
heard with absolutely no noise. The tonality was different
from DC and 60Ha AC heating.
Just by chance I opened a box with 16 year old prototypes
two weeks ago. Last week I have improved the linear amplifier
to handle better the low resistance of the cold tube. I will
play more when I have some time.
I am sure we can all benefit from such hands-on experience.
Did you had a chance to compare square wave heating vs pure sine heating?
Regards,
Joris
The definition "blackest background" is exactly the right word for the high-frequency AC heating experience. Less noise or hum or anything than with any DC solution I have ever tried (including car battery... actually I might say "particularly than less than car battery" which might come as a surprise to some.
Electronic transformers for halogene bulbs, once modified (cap added across the rectifier bridge) output an "almost square" wave, or we might call it "not perfect square". After all, if it were a true square wave, the resulting voltage would be even higher.
Given that the pure sine wave principle requires a generator and an amplifier, I assume that the circuit is more complicated than the half-wave as found in electronic transformers. Probably less efficient as well.
But I am always curious to see how other people solve "problems" in some original way!
As for the sound, I guess that there should be no particular difference, or any difference - between the two HF AC solutions. The black background, which is (most probably) conducive to the "very clean sound", should be the most obvious common trait.
******
http://rh-amps.blogspot.com/
I feel more confident with this project now with all the support you gave and I know that help is nearby if i run into trouble.
This project is really important to me, since beside my current breadboarded amp i have this long-haul project involving the GM70, and i suspect hum will be a problem with its 20V filaments.
In that specific case, the raise in output volts (to about 16-17V) resulting from adding the capacitor will be a blessing, but the output will still need to be raised to 20V by addition of a few turns to the transformer secondary. Is that feasible?
As it happens, I am already getting late with my RH70 project: one version is supposed to be "standard RH", similar to the RH813. I have gathered some tubes and should start breadboarding the amp.
As a matter of fact, I plan to modify the unit pictured on my blog, because that is the fastest way to proceed. It has 4 turns for 10V: I plan to unwind those secondaries and wind new 8 turns secondaries for 20V. Since the current draw is lower at 3A each, I am going to use different (lower gauge, obviously stranded wire with a thinner isolation to occupy less space).
Alternatively I might use 3 parallel solid core wires AWG23, that arrangement is good up to 53kHz, and 1A per wire should not be a problem due to the low number of turns (short wire) allowing higher current density (more like chassis wiring than current transfer). Or strip the wire and use heat shrinking tube as thin isolation. Space is an issue here, not maximum power.
Ultimately, that is your answer: remove the original secondary and wind a new one with the appropriate number of turns. Feel free to ask: write an email if it's urgent ;)
******
http://rh-amps.blogspot.com/
Joris, those solutions are all much more costly than the electronic transformers being discussed. Maybe there's an advantage to driving the filament with a sinewave rather than squarewave, but I don't think anyone has yet produced any evidence to support that. Meanwhile, the little transformers can be easily modified to oscillate at 65kHz or so, and the versions capable of powering 211s and 845s cost less than $10 US. I was the first to use electronic transformers for this purpose as far as I know, and you can believe I've been keeping my eyes open for something better over the last ten years. I do wish something was available at better quality (i.e. not a Chinese import), but there's just nothing on the market being made in the West that's even remotely similar to the size and efficiency of these modules.
--------------------------
Buy Chinese. Bury freedom.
TK
These are from UK.
DanL
This is not an "electronic transformer for halogen bulbs", rather:
PS-65 SERIES - 65W SINGLE OUTPUT AC-DC OPEN FRAME POWER SUPPLY
The output is DC, and this is what is usually called SMPS. The output is DC, and the 65kHz refers to the oscillation frequency.
You could relatively easily add high-speed diodes (I guess UF400x would do) and the rest of the necessary components, transforming an "electronic transformer for halogen bulbs" into an SMPS device.
But that would be superfluous - HF AC can perform the task (heating direct heated tubes) more efficiently and hum/noise free. Definitely a better solution to the SMPS - the latter should be compared to other DC solutions.
******
http://rh-amps.blogspot.com/
If i'm not mistaken those do not produce a square wave output unless modified?
I do use 12V computer switching supply to heat my tubes presently, via addition of a current regulator.
Unlikely . Nothing gets made in the UK anymore apart from crap cars...
'MeanWell Enterprises Co., Ltd. is one of the leading switching power supply manufacturers in Taiwan'
From the website
Al
The "oriental" (Chinese) products are actually a rather simple elaboration to exploit a generic oscillator circuit - the half wave bridge. And, they are not doing it for electron tubes, but halogen light bulbs: otherwise it would be much more expensive (an up-market affair)?
As TK has noted his "fear" of the audibility of artefacts (after 10 years of use?! no way...) at 65kHz, as suggested by Lynn Olsen... and mentions David Berning and his allegedly 250kHz solution, this is a chance to point out the difference between "good idea" (stroke of genius?) and "over-engineering" (necessary if you are in the Hi-End business, for the marketing department to use for your advantage).
The circuit itself is rather simple. We could enhance it by including a regulation stage (with LM317 or TL783) that would cancel virtually all 100Hz ripple before feeding the oscillator stage. Adding a few elements like NTCs in the right spots, and a couple additional elements to provide fail-safe mechanisms, all that is left would be to increase further the oscillation frequency (different ratio in the first small transformer) and use better (more expensive) IGBTs... at that point we can design a new board and incorporate it in our amps, while the marketing department has a field day with our "enhanced, innovative, and unique" features -- while in practice the original Chinese unit was good enough after the necessary modifications that make it suitable for the purpose.
Even more so, the cost of the necessary added parts (mod - cap, NTC) can easily exceed the cost of the unit itself! If we were to source the various parts on our own (ferrite cores, wire, etc.) we would end up with a much higher price than if we buy the unit.
Being sort of isolated here in Serbia is a blessing in disguise. There is no Lynn Olsen to freighten me with the potential audibility of HF AC if frequencies are not at least 100kHz... so I can just do it - and once I am aware that I cannot hear any of it, nor can the people I ask to listen with me hear anything - at that point it's too late for the "Lynn Olsens" to make me change my mind, because I consider it a proven fact (for me and others who were satisfied with the result).
Thus: are multiplexing and heterodyne by-products possible? Well, most probably they are. Can those by-products be found on a scope? Well, probably they can. But can they be heard? I cannot hear them with 30-50kHz, TK has not heard them with 65kHz... why would we need 100kHz or 250kHz? Only for marketing purposes, the best of the best, the non-plus-ultra... no small wonder that DIY equipment more often than not sounds better than coveted High-End stuff.
******
http://rh-amps.blogspot.com/
Thanks Alex for your interesting reply.
I do have read your blog on the subject, and am interested in trying, but there were a few points that were unclear to me. I'd like to take this opportunity to ask you a couple of questions to make sure my understanding is correct.
When you say that the output of these transformers is between 30 and 60 kHz but “modulated” by the mains - do you mean that the output is a rectified sinewave at mains frequency that is "chopped" by some square-wave generator afterwards? So adding a cap at the bridge output basically gives a DC supply that is then chopped? Wouldn't an LC stage after the rectifier limit the raise in DC voltage?
And second point, once you have reduced the number of turns on the output toroid top compensate for the higher voltage, can you adjust it down to the required heater voltage as well or you still need to drop voltage with power resistors?
These questions might seem basic or downright dumb, but then i am just a self-taught hobbyist. HF in particular, is a bit out of my league - but it defenitely interests me in the goal of HF heating.
Thank you in advance.
Joris
"do you mean that the output is a rectified sinewave at mains frequency that is "chopped" by some square-wave generator afterwards? So adding a cap at the bridge output basically gives a DC supply that is then chopped?"
Yes, that's how these modules are configured. With no filter cap, the output from the module is essentially 120Hz chopped at the HF rate.
"Wouldn't an LC stage after the rectifier limit the raise in DC voltage?"
There are two issues with that. First, the choke would have to meet critical inductance at 120 Hz without incurring excessive loss. A value of 0.1 to 0.2H is probably about right for this. A more serious issue is the initial voltage overshoot when power is first applied to the LC. Depending on the DCR of the choke, that might be 50% or more of the static voltage. Personally, I wouldn't want to connect such a circuit to a filament. If regulation is your goal, a small SS device might be a better choice.
Remember that everything ahead of the HF output transformer is attached to the mains. These components must be well isolated from the chassis. Meeting that requirement will be more difficult as the circuit becomes physically larger.
"...you still need to drop voltage with power resistors?"
If the transformer is toroidal, and you don't mind rewinding both the primary and secondary, the exact ratio can probably be created for the filament. Otherwise, if you only unwind turns from the secondary, resistors will probably be required to "trim" the voltage.
--------------------------
Buy Chinese. Bury freedom.
Thanks TK,
I read most of the threads you posted on this subject. Just ordered three 20W units to play with. I hope their transformers will be of the toroidal variety.
Well TK has already given the answers. A few additional points might be added, but I guess at the moment we need not complicate further with the "what if" list ;)
Never heard of 20W units?! If there are any, I expect the transformers to be very tiny, which in turns means more difficult to work with.
There are two important ratings: maximum power, which is mostly related to the "primary side capability", or what is the maximum that you can get from it: I suggest to take at least 30% more powerful units than needed; and minimum power, i.e. the minimum load required for oscillation. For a 50W unit this is usually 20W minimum, but from experience I can say they will operate on 5W automotive bulbs without apparent flicker although the light may appear dim which means that ooeration is not stable, and 10W perfectly stable and shining with adequate light intensity... but this varies from make to make. You should always test this ahead with automotive bulbs (5, 10, 20, 40/45, 50/55 W units are available for car lighting. I use H7 bulbs as 813 filament simulation.
******
http://rh-amps.blogspot.com/
Thanks Alex for the car light trick! I take good note. I was thinking of using a 8 ohm power resistor as a load, as it has the same resistance as the target tube, and observing with my good ol' 50Mhz analog scope. Do you think it is ok? I do have a true RMS DMM, but i'm not sure if it's capable of reliable measurements up to those frequencies.
I think it will be quite educative for me, but when poking around with a scope probe I'll make sure to stay away from the input transformer primary :-) It'd be safe to use an isolation transformer as well, although the only one i have is only about 25VA. I'll try to rig some back-to-back trannies.
If the model I bought is too small i can always find another use for the units. I might even connect them to halogen lamps :-) Anyways, it's just 12$ for the three. On the picture for the 40W model we can see that the case has a screw for removal, so like you did i will put back the case on after modding, passing the capacitor's wires through the vent holes.
Thanks for your input!
Joris
Joris,
The resistor is a good option if it has the same resistance as the filament. The only problem might be power dissipation and the fact that it does not light-up (hopefully!).
The units look very similar to the ones I like to use: the case is "clip on" and the inside is actually upside down (i.e. the bottom is actually the cover). The screw has a different purpose, it is bolted to a piece of metal so the output transistors are cooled by the piece of metal and the box! In order to access the board you will need to undo the screw, of course - but you can open it without a screwdriver. Hopefully the transformer is a tiny toroid.
The scope is the best option, and the DMM needs to be capable of measuring at least 100kHz frequency. A Fluke 87 III and upwards will do. Once you start taking measurements and doing some math, you will see that the tolerances can be made pretty tight by calculation only, even without measurements.
Let us know!
******
http://rh-amps.blogspot.com/
Joris, don't forget that the capacitor you add is connected to the AC mains through the rectifier diodes. There's no isolation. The only safe way to use these units in an amplifier chassis is to mount everything inside another (larger) plastic box.
--------------------------
Buy Chinese. Bury freedom.
After the modification, the cap which is too large remains outside of the box of the original electronic transformer unit. I use connectors, and before installing it inside an amp isolate those with heat-shrinking tube, and of course fix everything in place so it does not move.
Another part that might protrude out of the box is the NTC that you are going to install in series with the AC input (where mains AC comes in). Eventually you can connect it instead of the small fuse / low power resistor inside the unit which is going to blow anyway once you connect the cap across the bridge (if such fuse/resistor is applicable to your unit).
Once the unit is fixed inside, you can treat it just like any other transformer - connect the mains connection to your mains input via the main switch or however you were planning to do it.
The cap to be connected across the rectifier bridge should be rated 400V for Europe and UK (220-240V mains) and 200V for US/Japan (120V mains approx.). Since the current draw of US/Japan units is double in the primary, the cap needs to be twice the capacity. Thus, if you would use a 200uF/400V cap for Europe, in the US you should use 2 200uF/200V caps in parallel.
The above cap values are what I feel to be a reasonable minimum, but the quality of the results (ripple cancellation for the initial rectification) obviously depends on the current draw. It's not the same when supplying 2x 813 with 100W output power, and supplying 2x 300B with 12W of output power: the first case obviously has almost 10x more current draw on the primary, thus more capacity is necessary.
The increase in output voltage after inserting the cap is a result of the ripple cancellation and subsequent DC RMS increase for the oscillating stage: this poses no problem to the oscillator or the output transformer primary, since peak voltage remains the same (for the same mains input). A byproduct of the cap is the "redundancy" of the diac in the oscillator circuit, since the voltage never falls below the level where the diac is activated again restarting the oscillation - while in normal halogen bulb mode the diac is active in every cycle.
While testing and fine-tuning this circuit, you must observe the usual rules when working with lethal voltages. Isolation is necessary, and careful what you do with the probes...
******
http://rh-amps.blogspot.com/
Alex, I agree with much of what you've said, but Lynn has many years' experience in this field, and I respect his opinions. Just to clarify the issue, Lynn's concern was not that the system would be problematic with 20-20kHz audio. Rather, he was considering the fact that digital sources can output spurious, low-level products above 20kHz. The concern is that these products might mix with the filament's AC frequency and the incoming baseband audio to create audible artifacts. The fact that this doesn't appear to have occurred in your system or mine doesn't mean we're not vulnerable, or that it might not happen to someone else who uses this technique.
--------------------------
Buy Chinese. Bury freedom.
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: